Update messenger-backend submodule.

Changelog
Read contacts event if there are no phone numbers present, improve permission check for SDK<23.
2026-07-03 14:05:24 +02:00 · 2017-04-29 01:08:35 +02:00 · 2017-04-29 01:08:20 +02:00 · 2017-04-29 00:14:21 +02:00 · 2017-04-28 15:39:52 +02:00 · 2017-04-28 14:55:07 +02:00
1244 changed files with 22276 additions and 154553 deletions
@@ -6,3 +6,6 @@ obj/
 local.properties
 *.keystore
 libs/
+
+# ignore private scripts and directories, eg. local2github.prv.sh
+*.prv*
@@ -0,0 +1,137 @@
+# Delta Chat Changelog
+
+## v0.1.33
+2017-04-29
+
+* Better support for right-to-left (RTL) languages, taking advantage of
+  Android 4.2 (Jelly Bean MR1, API level 17).
+* Send PNG files without resizing and converting to JPEG
+* If JPEG files are send without compression, they still appear as image, not as attached files
+* Raise-to-speak defaults to false
+* Unify long click behaviour
+* Support Android's system function "Delete data"
+* Replies to messages pop up automatically even if send from other email addresses (typical scenario for alias addresses)
+* Fix group-replies from normal email-clients. 
+
+## v0.1.32
+2017-04-22
+
+* Update Spanish and Portuguese translations
+* Update internal sqlite library to version 3.18.0, released on 2017-03-28
+* Remove more of the custom language handling, use Android's routines instead
+* General code cleanup
+* Play GIF files
+* Option to disable autoplaying GIF files
+* When sending contacts, only use the names the receivers have set themselves
+* Show some hints when long-pressing icons in the action bar
+
+## v0.1.29
+2017-04-19
+
+* Add Russian translation
+* For outgoing (group-)messages, only use the names the receivers have set themselves
+
+## v0.1.28
+2017-04-14
+
+* Pimp notifications
+* Bug fixes
+
+## v0.1.27
+2017-04-12
+
+* Use a permanent foreground service for reliable notifications
+* Monitor the IMAP-IDLE thread and reconnect if IMAP-IDLE seems to hang
+* Various battery and background optimizations
+
+## v0.1.25
+2017-04-04
+
+* Use system or user selected video player.
+* Do not connect if not configured (avoids a warning on the first time startup)
+* Add  vertical scrollbar, eg. to settings activities.
+* Pimp GUI and logo.
+* Update Korean.
+
+## v0.1.24
+2017-03-31
+
+* Share images and documents from other apps to Delta Chat
+* Offer to mailto:-link-support to other apps
+* Ignore implausible sending time of incoming messages; use the receive time in these rare cases
+* Show errors only when Delta Chat is in foreground
+* Dynamically adapt video bitrate for longer videos to an attachment-size of max. 25 MB
+
+## v0.1.23
+2017-03-28
+
+* Retry connecting to IMAP if there is not network available on the first try
+* Notify about new messages if the app is not active for hours, optimize battery consumption
+
+## v0.1.22
+2017-03-22
+
+* Show HTML-only messages
+* Show connection errors
+* Add options for SSL/TLS and STARTTLS
+* Automatic account configuration, if possible
+* Recode large videos
+* Add Hungarian translation
+* Add Korean translation
+
+## v0.1.21
+2017-03-10
+
+* Record and send voice messages
+* Record and send videos
+* Send and play music
+* Send contacts and email addresses
+* Sending and opening attachments of any type
+* Share and open commands for all attachments
+* Accept VCards send to us by other apps
+* Clickable email addresses
+* Update Polish translation
+* Fix tablet startup bug
+* Close the app when using the lock-app-via-pincode function
+* Protect data by using a content provider for sharing
+* Try to clear the task switcher's screenshots when locking the app via pincode
+* Pimp GUI
+
+## v0.1.20
+2017-02-16
+
+* Avoid unwanted downloads of lots of old messages
+* Make the "Chats" folder visible if the server hides new folders by default
+* Fix a crash when the server returns empty folders
+* Update Polish and Portuguese translations
+* Use API level 25 (Nougat 7.1) as target
+
+## v0.1.18
+2017-02-11
+
+* Add Polish translation
+* Use a new default background for chats
+* Improve typography by using the system font instead of a custom resource font
+* Remove custom plural handling, use Android's routines instead
+* Remove unused source code and strings
+* More fixes of lint errors and warnings
+
+## v0.1.17
+2017-02-07
+
+* Drop two unnecessary permissions: ACCESS_COARSE_LOCATION and ACCESS_FINE_LOCATION
+* Really add French translation
+* Update Portuguese translation
+* Start fixing translation handling of the program
+* Remove special "foss" build, because the whole program is free now.
+
+## v0.1.16
+2017-02-06
+
+* Add French translation
+* Fix some lint errors and warnings
+
+## v0.1.15
+2017-01-31
+
+* Prepare for release on [F-Droid](https://f-droid.org/)
@@ -24,15 +24,15 @@ tasks.withType(JavaCompile) {
 }

 dependencies {
-    compile 'com.android.support:support-v4:23.4.0'
+    compile 'com.android.support:appcompat-v7:25.3.1'
    compile 'com.googlecode.mp4parser:isoparser:1.0.6'
    compile 'com.amulyakhare:com.amulyakhare.textdrawable:1.0.1'
    compile fileTree(dir: "$buildDir/native-libs", include: 'native-libs.jar')
 }

 android {
-    compileSdkVersion 23
-    buildToolsVersion '23.0.3'
+    compileSdkVersion 25
+    buildToolsVersion '25.0.2'

    useLibrary 'org.apache.http.legacy'
    defaultConfig.applicationId = "com.b44t.messenger"
@@ -74,34 +74,17 @@ android {
            minifyEnabled false
            proguardFiles getDefaultProguardFile('proguard-android.txt'), 'proguard-rules.pro'
        }
-
-        foss {
-            debuggable false
-            jniDebuggable false
-            signingConfig signingConfigs.release
-        }
    }

-    defaultConfig.versionCode = 14
+    defaultConfig.versionCode = 33

    sourceSets.main {
        jniLibs.srcDir 'libs'
        jni.srcDirs = [] //disable automatic ndk-build call
    }

-    sourceSets.debug {
-        manifest.srcFile 'config/debug/AndroidManifest.xml'
-    }
-
-    sourceSets.release {
-        manifest.srcFile 'config/release/AndroidManifest.xml'
-    }
-
-    sourceSets.foss {
-        manifest.srcFile 'config/foss/AndroidManifest.xml'
-    }
-
    productFlavors {
+        /*
        x86 {
            ndk {
                abiFilter "x86"
@@ -120,6 +103,7 @@ android {
            }
            versionCode = 1
        }
+        */
        fat {
            versionCode = 3
        }
@@ -131,9 +115,8 @@ android {
    }

    defaultConfig {
-        minSdkVersion 14 // 14: Android 4.0 Ice Cream Sandwich 2011 (Telegram default), 21: Android 5.0 Lollipop 2014 (recommended for InstantRun)
-        targetSdkVersion 23
-        versionName "0.1.15"
+        minSdkVersion 14     // 14: Android 4.0 Ice Cream Sandwich 2011 (Telegram default), 21: Android 5.0 Lollipop 2014 (recommended for InstantRun)
+        targetSdkVersion 25  // 25: Nougat. CAVE: Do NOT target "Andoid O" without checking the background tasks carefully, see https://developer.android.com/preview/behavior-changes.html#back-all . As long as we target "Nougat", everything works as expected even for "Andoid O" or later
+        versionName "0.1.33" // do NOT forget to increase defaultConfig.versionCode!
    }
 }
-
@@ -1,24 +0,0 @@
-<?xml version="1.0" encoding="utf-8"?>
-<manifest xmlns:android="http://schemas.android.com/apk/res/android"
-    xmlns:tools="http://schemas.android.com/tools"
-    package="com.b44t.messenger"
-    android:installLocation="auto">
-
-    <uses-feature android:name="android.hardware.location.gps" android:required="false" />
-    <uses-feature android:name="android.hardware.location.network" android:required="false" />
-    <uses-feature android:name="android.hardware.location" android:required="false" />
-    <uses-feature android:name="android.hardware.LOCATION" android:required="false" />
-
-    <uses-permission android:name="android.permission.ACCESS_COARSE_LOCATION" />
-    <uses-permission android:name="android.permission.ACCESS_FINE_LOCATION" />
-    <application
-        android:allowBackup="false"
-        android:icon="@drawable/ic_launcher"
-        android:label="@string/AppName"
-        android:theme="@style/Theme.MessengerProj.Start"
-        android:name=".ApplicationLoader"
-        android:hardwareAccelerated="@bool/useHardwareAcceleration"
-        android:largeHeap="true">
-    </application>
-
-</manifest>
@@ -1,17 +0,0 @@
-<?xml version="1.0" encoding="utf-8"?>
-<manifest xmlns:android="http://schemas.android.com/apk/res/android"
-    package="com.b44t.messenger"
-    android:installLocation="auto">
-
-    <application
-        android:allowBackup="false"
-        android:icon="@drawable/ic_launcher"
-        android:label="@string/AppName"
-        android:theme="@style/Theme.MessengerProj.Start"
-        android:name=".ApplicationLoader"
-        android:hardwareAccelerated="@bool/useHardwareAcceleration"
-        android:largeHeap="true">
-
-    </application>
-
-</manifest>
@@ -1,24 +0,0 @@
-<?xml version="1.0" encoding="utf-8"?>
-<manifest xmlns:android="http://schemas.android.com/apk/res/android"
-    xmlns:tools="http://schemas.android.com/tools"
-    package="com.b44t.messenger"
-    android:installLocation="auto">
-
-    <uses-feature android:name="android.hardware.location.gps" android:required="false" />
-    <uses-feature android:name="android.hardware.location.network" android:required="false" />
-    <uses-feature android:name="android.hardware.location" android:required="false" />
-    <uses-feature android:name="android.hardware.LOCATION" android:required="false" />
-
-    <uses-permission android:name="android.permission.ACCESS_COARSE_LOCATION" />
-    <uses-permission android:name="android.permission.ACCESS_FINE_LOCATION" />
-    <application
-        android:allowBackup="false"
-        android:icon="@drawable/ic_launcher"
-        android:label="@string/AppName"
-        android:theme="@style/Theme.MessengerProj.Start"
-        android:name=".ApplicationLoader"
-        android:hardwareAccelerated="@bool/useHardwareAcceleration"
-        android:largeHeap="true">
-    </application>
-
-</manifest>
@@ -961,88 +961,88 @@ include $(BUILD_STATIC_LIBRARY)
 ################################################################################


-include $(CLEAR_VARS)
-
-LOCAL_CFLAGS := -Wall -DANDROID -DHAVE_MALLOC_H -DHAVE_PTHREAD -DWEBP_USE_THREAD -finline-functions -ffast-math -ffunction-sections -fdata-sections -Os
-LOCAL_C_INCLUDES += $(LOCAL_PATH)/libwebp/src
-LOCAL_ARM_MODE := arm
-LOCAL_STATIC_LIBRARIES := cpufeatures
-LOCAL_MODULE := webp
-
-ifneq ($(findstring armeabi-v7a, $(TARGET_ARCH_ABI)),)
-  NEON := c.neon
-else
-  NEON := c
-endif
-
-LOCAL_SRC_FILES := \
-./libwebp/dec/alpha.c \
-./libwebp/dec/buffer.c \
-./libwebp/dec/frame.c \
-./libwebp/dec/idec.c \
-./libwebp/dec/io.c \
-./libwebp/dec/quant.c \
-./libwebp/dec/tree.c \
-./libwebp/dec/vp8.c \
-./libwebp/dec/vp8l.c \
-./libwebp/dec/webp.c \
-./libwebp/dsp/alpha_processing.c \
-./libwebp/dsp/alpha_processing_sse2.c \
-./libwebp/dsp/cpu.c \
-./libwebp/dsp/dec.c \
-./libwebp/dsp/dec_clip_tables.c \
-./libwebp/dsp/dec_mips32.c \
-./libwebp/dsp/dec_neon.$(NEON) \
-./libwebp/dsp/dec_sse2.c \
-./libwebp/dsp/enc.c \
-./libwebp/dsp/enc_avx2.c \
-./libwebp/dsp/enc_mips32.c \
-./libwebp/dsp/enc_neon.$(NEON) \
-./libwebp/dsp/enc_sse2.c \
-./libwebp/dsp/lossless.c \
-./libwebp/dsp/lossless_mips32.c \
-./libwebp/dsp/lossless_neon.$(NEON) \
-./libwebp/dsp/lossless_sse2.c \
-./libwebp/dsp/upsampling.c \
-./libwebp/dsp/upsampling_neon.$(NEON) \
-./libwebp/dsp/upsampling_sse2.c \
-./libwebp/dsp/yuv.c \
-./libwebp/dsp/yuv_mips32.c \
-./libwebp/dsp/yuv_sse2.c \
-./libwebp/enc/alpha.c \
-./libwebp/enc/analysis.c \
-./libwebp/enc/backward_references.c \
-./libwebp/enc/config.c \
-./libwebp/enc/cost.c \
-./libwebp/enc/filter.c \
-./libwebp/enc/frame.c \
-./libwebp/enc/histogram.c \
-./libwebp/enc/iterator.c \
-./libwebp/enc/picture.c \
-./libwebp/enc/picture_csp.c \
-./libwebp/enc/picture_psnr.c \
-./libwebp/enc/picture_rescale.c \
-./libwebp/enc/picture_tools.c \
-./libwebp/enc/quant.c \
-./libwebp/enc/syntax.c \
-./libwebp/enc/token.c \
-./libwebp/enc/tree.c \
-./libwebp/enc/vp8l.c \
-./libwebp/enc/webpenc.c \
-./libwebp/utils/bit_reader.c \
-./libwebp/utils/bit_writer.c \
-./libwebp/utils/color_cache.c \
-./libwebp/utils/filters.c \
-./libwebp/utils/huffman.c \
-./libwebp/utils/huffman_encode.c \
-./libwebp/utils/quant_levels.c \
-./libwebp/utils/quant_levels_dec.c \
-./libwebp/utils/random.c \
-./libwebp/utils/rescaler.c \
-./libwebp/utils/thread.c \
-./libwebp/utils/utils.c \
-
-include $(BUILD_STATIC_LIBRARY)
+#include $(CLEAR_VARS)
+#
+#LOCAL_CFLAGS := -Wall -DANDROID -DHAVE_MALLOC_H -DHAVE_PTHREAD -DWEBP_USE_THREAD -finline-functions -ffast-math -ffunction-sections -fdata-sections -Os
+#LOCAL_C_INCLUDES += $(LOCAL_PATH)/libwebp/src
+#LOCAL_ARM_MODE := arm
+#LOCAL_STATIC_LIBRARIES := cpufeatures
+#LOCAL_MODULE := webp
+#
+#ifneq ($(findstring armeabi-v7a, $(TARGET_ARCH_ABI)),)
+#  NEON := c.neon
+#else
+#  NEON := c
+#endif
+#
+#LOCAL_SRC_FILES := \
+#./libwebp/dec/alpha.c \
+#./libwebp/dec/buffer.c \
+#./libwebp/dec/frame.c \
+#./libwebp/dec/idec.c \
+#./libwebp/dec/io.c \
+#./libwebp/dec/quant.c \
+#./libwebp/dec/tree.c \
+#./libwebp/dec/vp8.c \
+#./libwebp/dec/vp8l.c \
+#./libwebp/dec/webp.c \
+#./libwebp/dsp/alpha_processing.c \
+#./libwebp/dsp/alpha_processing_sse2.c \
+#./libwebp/dsp/cpu.c \
+#./libwebp/dsp/dec.c \
+#./libwebp/dsp/dec_clip_tables.c \
+#./libwebp/dsp/dec_mips32.c \
+#./libwebp/dsp/dec_neon.$(NEON) \
+#./libwebp/dsp/dec_sse2.c \
+#./libwebp/dsp/enc.c \
+#./libwebp/dsp/enc_avx2.c \
+#./libwebp/dsp/enc_mips32.c \
+#./libwebp/dsp/enc_neon.$(NEON) \
+#./libwebp/dsp/enc_sse2.c \
+#./libwebp/dsp/lossless.c \
+#./libwebp/dsp/lossless_mips32.c \
+#./libwebp/dsp/lossless_neon.$(NEON) \
+#./libwebp/dsp/lossless_sse2.c \
+#./libwebp/dsp/upsampling.c \
+#./libwebp/dsp/upsampling_neon.$(NEON) \
+#./libwebp/dsp/upsampling_sse2.c \
+#./libwebp/dsp/yuv.c \
+#./libwebp/dsp/yuv_mips32.c \
+#./libwebp/dsp/yuv_sse2.c \
+#./libwebp/enc/alpha.c \
+#./libwebp/enc/analysis.c \
+#./libwebp/enc/backward_references.c \
+#./libwebp/enc/config.c \
+#./libwebp/enc/cost.c \
+#./libwebp/enc/filter.c \
+#./libwebp/enc/frame.c \
+#./libwebp/enc/histogram.c \
+#./libwebp/enc/iterator.c \
+#./libwebp/enc/picture.c \
+#./libwebp/enc/picture_csp.c \
+#./libwebp/enc/picture_psnr.c \
+#./libwebp/enc/picture_rescale.c \
+#./libwebp/enc/picture_tools.c \
+#./libwebp/enc/quant.c \
+#./libwebp/enc/syntax.c \
+#./libwebp/enc/token.c \
+#./libwebp/enc/tree.c \
+#./libwebp/enc/vp8l.c \
+#./libwebp/enc/webpenc.c \
+#./libwebp/utils/bit_reader.c \
+#./libwebp/utils/bit_writer.c \
+#./libwebp/utils/color_cache.c \
+#./libwebp/utils/filters.c \
+#./libwebp/utils/huffman.c \
+#./libwebp/utils/huffman_encode.c \
+#./libwebp/utils/quant_levels.c \
+#./libwebp/utils/quant_levels_dec.c \
+#./libwebp/utils/random.c \
+#./libwebp/utils/rescaler.c \
+#./libwebp/utils/thread.c \
+#./libwebp/utils/utils.c \
+#
+#include $(BUILD_STATIC_LIBRARY)


 ################################################################################
@@ -1313,7 +1313,7 @@ LOCAL_CFLAGS 	+= -Drestrict='' -D__EMX__ -DOPUS_BUILD -DFIXED_POINT -DUSE_ALLOCA
 LOCAL_CFLAGS 	+= -DANDROID_NDK -DDISABLE_IMPORTGL -fno-strict-aliasing -fprefetch-loop-arrays -DAVOID_TABLES -DANDROID_TILE_BASED_DECODE -DANDROID_ARMV6_IDCT -ffast-math -D__STDC_CONSTANT_MACROS
 LOCAL_CPPFLAGS 	:= -DBSD=1 -ffast-math -Os -funroll-loops -std=c++11
 LOCAL_LDLIBS 	:= -ljnigraphics -llog -lz -latomic
-LOCAL_STATIC_LIBRARIES :=  etpan sasl2 webp sqlite crypto avformat avcodec avutil libiconv
+LOCAL_STATIC_LIBRARIES :=  etpan sasl2 sqlite crypto avformat avcodec avutil libiconv
 # if you get "undefined reference" errors, the reason for this may be the _order_! Eg. libiconv as the first library does not work!
 # "breakpad" was placed after "crypto", NativeLoader.cpp after mrwrapper.c

@@ -1498,54 +1498,54 @@ $(LOCAL_PATH)/ffmpeg \
 $(LOCAL_PATH)/libetpan/include \
 $(LOCAL_PATH)/sqlite

-LOCAL_SRC_FILES     += \
-./libjpeg/jcapimin.c \
-./libjpeg/jcapistd.c \
-./libjpeg/armv6_idct.S \
-./libjpeg/jccoefct.c \
-./libjpeg/jccolor.c \
-./libjpeg/jcdctmgr.c \
-./libjpeg/jchuff.c \
-./libjpeg/jcinit.c \
-./libjpeg/jcmainct.c \
-./libjpeg/jcmarker.c \
-./libjpeg/jcmaster.c \
-./libjpeg/jcomapi.c \
-./libjpeg/jcparam.c \
-./libjpeg/jcphuff.c \
-./libjpeg/jcprepct.c \
-./libjpeg/jcsample.c \
-./libjpeg/jctrans.c \
-./libjpeg/jdapimin.c \
-./libjpeg/jdapistd.c \
-./libjpeg/jdatadst.c \
-./libjpeg/jdatasrc.c \
-./libjpeg/jdcoefct.c \
-./libjpeg/jdcolor.c \
-./libjpeg/jddctmgr.c \
-./libjpeg/jdhuff.c \
-./libjpeg/jdinput.c \
-./libjpeg/jdmainct.c \
-./libjpeg/jdmarker.c \
-./libjpeg/jdmaster.c \
-./libjpeg/jdmerge.c \
-./libjpeg/jdphuff.c \
-./libjpeg/jdpostct.c \
-./libjpeg/jdsample.c \
-./libjpeg/jdtrans.c \
-./libjpeg/jerror.c \
-./libjpeg/jfdctflt.c \
-./libjpeg/jfdctfst.c \
-./libjpeg/jfdctint.c \
-./libjpeg/jidctflt.c \
-./libjpeg/jidctfst.c \
-./libjpeg/jidctint.c \
-./libjpeg/jidctred.c \
-./libjpeg/jmemmgr.c \
-./libjpeg/jmemnobs.c \
-./libjpeg/jquant1.c \
-./libjpeg/jquant2.c \
-./libjpeg/jutils.c
+#LOCAL_SRC_FILES     += \
+#./libjpeg/jcapimin.c \
+#./libjpeg/jcapistd.c \
+#./libjpeg/armv6_idct.S \
+#./libjpeg/jccoefct.c \
+#./libjpeg/jccolor.c \
+#./libjpeg/jcdctmgr.c \
+#./libjpeg/jchuff.c \
+#./libjpeg/jcinit.c \
+#./libjpeg/jcmainct.c \
+#./libjpeg/jcmarker.c \
+#./libjpeg/jcmaster.c \
+#./libjpeg/jcomapi.c \
+#./libjpeg/jcparam.c \
+#./libjpeg/jcphuff.c \
+#./libjpeg/jcprepct.c \
+#./libjpeg/jcsample.c \
+#./libjpeg/jctrans.c \
+#./libjpeg/jdapimin.c \
+#./libjpeg/jdapistd.c \
+#./libjpeg/jdatadst.c \
+#./libjpeg/jdatasrc.c \
+#./libjpeg/jdcoefct.c \
+#./libjpeg/jdcolor.c \
+#./libjpeg/jddctmgr.c \
+#./libjpeg/jdhuff.c \
+#./libjpeg/jdinput.c \
+#./libjpeg/jdmainct.c \
+#./libjpeg/jdmarker.c \
+#./libjpeg/jdmaster.c \
+#./libjpeg/jdmerge.c \
+#./libjpeg/jdphuff.c \
+#./libjpeg/jdpostct.c \
+#./libjpeg/jdsample.c \
+#./libjpeg/jdtrans.c \
+#./libjpeg/jerror.c \
+#./libjpeg/jfdctflt.c \
+#./libjpeg/jfdctfst.c \
+#./libjpeg/jfdctint.c \
+#./libjpeg/jidctflt.c \
+#./libjpeg/jidctfst.c \
+#./libjpeg/jidctint.c \
+#./libjpeg/jidctred.c \
+#./libjpeg/jmemmgr.c \
+#./libjpeg/jmemnobs.c \
+#./libjpeg/jquant1.c \
+#./libjpeg/jquant2.c \
+#./libjpeg/jutils.c

 LOCAL_SRC_FILES     += \
 ./libyuv/source/compare_common.cc \
@@ -1598,9 +1598,8 @@ ifeq ($(TARGET_ARCH_ABI),armeabi-v7a)
 endif

 LOCAL_SRC_FILES     += \
-./jni.c \
+./mrjnimain.c \
 ./audio.c \
-./utils.c \
 ./image.c \
 ./video.c \
 ./gifvideo.cpp \
@@ -1611,14 +1610,16 @@ LOCAL_SRC_FILES     += \
 ./messenger-backend/src/mre2ee.c \
 ./messenger-backend/src/mrimap.c \
 ./messenger-backend/src/mrjob.c \
-./messenger-backend/src/mrlog.c \
 ./messenger-backend/src/mrloginparam.c \
 ./messenger-backend/src/mrmailbox.c \
+./messenger-backend/src/mrmailbox_configure.c \
+./messenger-backend/src/mrmailbox_log.c \
 ./messenger-backend/src/mrmimeparser.c \
 ./messenger-backend/src/mrmsg.c \
 ./messenger-backend/src/mrosnative.c \
 ./messenger-backend/src/mrparam.c \
 ./messenger-backend/src/mrpoortext.c \
+./messenger-backend/src/mrsaxparser.c \
 ./messenger-backend/src/mrsimplify.c \
 ./messenger-backend/src/mrsmtp.c \
 ./messenger-backend/src/mrsqlite3.c \
@@ -1,3 +1,26 @@
+/*******************************************************************************
+ *
+ *                          Messenger Android Frontend
+ *                        (C) 2013-2016 Nikolai Kudashov
+ *                           (C) 2017 Björn Petersen
+ *                    Contact: r10s@b44t.com, http://b44t.com
+ *
+ * This program is free software: you can redistribute it and/or modify it under
+ * the terms of the GNU General Public License as published by the Free Software
+ * Foundation, either version 3 of the License, or (at your option) any later
+ * version.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
+ * FOR A PARTICULAR PURPOSE.  See the GNU General Public License for more
+ * details.
+ *
+ * You should have received a copy of the GNU General Public License along with
+ * this program.  If not, see http://www.gnu.org/licenses/ .
+ *
+ ******************************************************************************/
+
+
 #include <jni.h>
 #include <ogg/ogg.h>
 #include <stdio.h>
@@ -6,7 +29,7 @@
 #include <time.h>
 #include <opusfile.h>
 #include <math.h>
-#include "utils.h"
+#include "mrjnimain.h"

 typedef struct {
    int version;
@@ -115,7 +138,7 @@ static int read_chars(ROPacket *p, unsigned char *str, int nb_chars)
    return 1;
 }

-int opus_header_to_packet(const OpusHeader *h, unsigned char *packet, int len) {
+static int opus_header_to_packet(const OpusHeader *h, unsigned char *packet, int len) {
    int i;
    Packet p;
    unsigned char ch;
@@ -219,34 +242,34 @@ static int writeOggPage(ogg_page *page, FILE *os) {
    return written;
 }

-const opus_int32 bitrate = 16000;
-const opus_int32 rate = 16000;
-const opus_int32 frame_size = 960;
-const int with_cvbr = 1;
-const int max_ogg_delay = 0;
-const int comment_padding = 512;
+static const opus_int32 bitrate = 16000;
+static const opus_int32 rate = 16000;
+static const opus_int32 frame_size = 960;
+static const int with_cvbr = 1;
+static const int max_ogg_delay = 0;
+static const int comment_padding = 512;

-opus_int32 coding_rate = 16000;
-ogg_int32_t _packetId;
-OpusEncoder *_encoder = 0;
-uint8_t *_packet = 0;
-ogg_stream_state os;
-FILE *_fileOs = 0;
-oe_enc_opt inopt;
-OpusHeader header;
-opus_int32 min_bytes;
-int max_frame_bytes;
-ogg_packet op;
-ogg_page og;
-opus_int64 bytes_written;
-opus_int64 pages_out;
-opus_int64 total_samples;
-ogg_int64_t enc_granulepos;
-ogg_int64_t last_granulepos;
-int size_segments;
-int last_segments;
+static opus_int32 coding_rate = 16000;
+static ogg_int32_t _packetId;
+static OpusEncoder *_encoder = 0;
+static uint8_t *_packet = 0;
+static ogg_stream_state os;
+static FILE *_fileOs = 0;
+static oe_enc_opt inopt;
+static OpusHeader header;
+static opus_int32 min_bytes;
+static int max_frame_bytes;
+static ogg_packet op;
+static ogg_page og;
+static opus_int64 bytes_written;
+static opus_int64 pages_out;
+static opus_int64 total_samples;
+static ogg_int64_t enc_granulepos;
+static ogg_int64_t last_granulepos;
+static int size_segments;
+static int last_segments;

-void cleanupRecorder() {
+static void cleanupRecorder() {
    
    ogg_stream_flush(&os, &og);
    
@@ -282,7 +305,7 @@ void cleanupRecorder() {
    memset(&og, 0, sizeof(ogg_page));
 }

-int initRecorder(const char *path) {
+static int initRecorder(const char *path) {
    cleanupRecorder();
    
    if (!path) {
@@ -421,7 +444,7 @@ int initRecorder(const char *path) {
    return 1;
 }

-int writeFrame(uint8_t *framePcmBytes, unsigned int frameByteCount) {
+static int writeFrame(uint8_t *framePcmBytes, unsigned int frameByteCount) {
    int cur_frame_size = frame_size;
    _packetId++;
    
@@ -527,15 +550,15 @@ JNIEXPORT void Java_com_b44t_messenger_MediaController_stopRecord(JNIEnv *env, j
 }

 //player
-OggOpusFile *_opusFile;
-int _isSeekable = 0;
-int64_t _totalPcmDuration = 0;
-int64_t _currentPcmOffset = 0;
-int _finished = 0;
+static OggOpusFile *_opusFile;
+static int _isSeekable = 0;
+static int64_t _totalPcmDuration = 0;
+static int64_t _currentPcmOffset = 0;
+static int _finished = 0;
 static const int playerBuffersCount = 3;
 static const int playerSampleRate = 48000;

-void cleanupPlayer() {
+static void cleanupPlayer() {
    if (_opusFile) {
        op_free(_opusFile);
        _opusFile = 0;
@@ -546,7 +569,7 @@ void cleanupPlayer() {
    _finished = 0;
 }

-int seekPlayer(float position) {
+static int seekPlayer(float position) {
    if (!_opusFile || !_isSeekable || position < 0) {
        return 0;
    }
@@ -559,7 +582,7 @@ int seekPlayer(float position) {
    return result == OPUS_OK;
 }

-int initPlayer(const char *path) {
+static int initPlayer(const char *path) {
    cleanupPlayer();
    
    int openError = OPUS_OK;
@@ -576,7 +599,7 @@ int initPlayer(const char *path) {
    return 1;
 }

-void fillBuffer(uint8_t *buffer, int capacity, int *args) {
+static void fillBuffer(uint8_t *buffer, int capacity, int *args) {
    if (_opusFile) {
        args[1] = max(0, op_pcm_tell(_opusFile));
        
@@ -735,7 +758,7 @@ JNIEXPORT jbyteArray Java_com_b44t_messenger_MediaController_getWaveform2(JNIEnv
    return result;
 }

-int16_t *sampleBuffer = NULL;
+static int16_t *sampleBuffer = NULL;


 JNIEXPORT jbyteArray Java_com_b44t_messenger_MediaController_getWaveform(JNIEnv *env, jclass class, jstring path) {
@@ -819,3 +842,4 @@ JNIEXPORT jbyteArray Java_com_b44t_messenger_MediaController_getWaveform(JNIEnv
    
    return result;
 }
+
@@ -1,5 +1,28 @@
+/*******************************************************************************
+ *
+ *                          Messenger Android Frontend
+ *                        (C) 2013-2016 Nikolai Kudashov
+ *                           (C) 2017 Björn Petersen
+ *                    Contact: r10s@b44t.com, http://b44t.com
+ *
+ * This program is free software: you can redistribute it and/or modify it under
+ * the terms of the GNU General Public License as published by the Free Software
+ * Foundation, either version 3 of the License, or (at your option) any later
+ * version.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
+ * FOR A PARTICULAR PURPOSE.  See the GNU General Public License for more
+ * details.
+ *
+ * You should have received a copy of the GNU General Public License along with
+ * this program.  If not, see http://www.gnu.org/licenses/ .
+ *
+ ******************************************************************************/
+
+
 #include <jni.h>
-#include <utils.h>
+#include "mrjnimain.h"
 #include <libyuv.h>
 #include <android/bitmap.h>
 #include <cstdint>
@@ -53,19 +76,14 @@ typedef struct VideoInfo {
    AVPacket orig_pkt;
 };

-jobject makeGlobarRef(JNIEnv *env, jobject object) {
-    if (object) {
-        return env->NewGlobalRef(object);
-    }
-    return 0;
-}

-int gifvideoOnJNILoad(JavaVM *vm, JNIEnv *env) {
+int gifvideoOnJNILoad(JavaVM *vm, JNIEnv *env) { /* called from JNI_OnLoad() */
    av_register_all();
    return 0;
 }

-int open_codec_context(int *stream_idx, AVFormatContext *fmt_ctx, enum AVMediaType type) {
+
+static int open_codec_context(int *stream_idx, AVFormatContext *fmt_ctx, enum AVMediaType type) {
    int ret;
    AVStream *st;
    AVCodecContext *dec_ctx = NULL;
@@ -97,7 +115,7 @@ int open_codec_context(int *stream_idx, AVFormatContext *fmt_ctx, enum AVMediaTy
    return 0;
 }
    
-int decode_packet(VideoInfo *info, int *got_frame) {
+static int decode_packet(VideoInfo *info, int *got_frame) {
    int ret = 0;
    int decoded = info->pkt.size;
    
@@ -263,4 +281,5 @@ jint Java_com_b44t_ui_Components_AnimatedFileDrawable_getVideoFrame(JNIEnv *env,
    }
    return 0;
 }
-}
+
+} /* extern "C" */
@@ -21,10 +21,13 @@
 ******************************************************************************/


-package com.b44t.messenger;
+#ifndef __GIFVIDEO_H__
+#define __GIFVIDEO_H__

-public class DownloadObject {
-    public TLObject object;
-    public int type;
-    public long id;
-}
+
+#include <jni.h>
+
+int gifvideoOnJNILoad(JavaVM *vm, JNIEnv *env);
+
+
+#endif /* __GIFVIDEO_H__ */
@@ -1,25 +1,62 @@
+/*******************************************************************************
+ *
+ *                          Messenger Android Frontend
+ *                        (C) 2013-2016 Nikolai Kudashov
+ *                           (C) 2017 Björn Petersen
+ *                    Contact: r10s@b44t.com, http://b44t.com
+ *
+ * This program is free software: you can redistribute it and/or modify it under
+ * the terms of the GNU General Public License as published by the Free Software
+ * Foundation, either version 3 of the License, or (at your option) any later
+ * version.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
+ * FOR A PARTICULAR PURPOSE.  See the GNU General Public License for more
+ * details.
+ *
+ * You should have received a copy of the GNU General Public License along with
+ * this program.  If not, see http://www.gnu.org/licenses/ .
+ *
+ ******************************************************************************/
+
+
 #include <jni.h>
 #include <stdio.h>
 #include <setjmp.h>
-#include <libjpeg/jpeglib.h>
+/*#include <libjpeg/jpeglib.h>*/
 #include <android/bitmap.h>
-#include <libwebp/webp/decode.h>
-#include <libwebp/webp/encode.h>
-#include "utils.h"
+/*#include <libwebp/webp/decode.h>
+#include <libwebp/webp/encode.h>*/
+#include "mrjnimain.h"
 #include "image.h"

-jclass jclass_NullPointerException;
-jclass jclass_RuntimeException;

-jclass jclass_Options;
-jfieldID jclass_Options_inJustDecodeBounds;
-jfieldID jclass_Options_outHeight;
-jfieldID jclass_Options_outWidth;
+static jclass jclass_NullPointerException;
+static jclass jclass_RuntimeException;

-const uint32_t PGPhotoEnhanceHistogramBins = 256;
-const uint32_t PGPhotoEnhanceSegments = 4;
+static jclass jclass_Options;
+static jfieldID jclass_Options_inJustDecodeBounds;
+static jfieldID jclass_Options_outHeight;
+static jfieldID jclass_Options_outWidth;

-jclass createGlobarRef(JNIEnv *env, jclass class) {
+static const uint32_t PGPhotoEnhanceHistogramBins = 256;
+static const uint32_t PGPhotoEnhanceSegments = 4;
+
+static void throwException(JNIEnv *env, char *format, ...) {
+    jclass exClass = (*env)->FindClass(env, "java/lang/UnsupportedOperationException");
+    if (!exClass) {
+        return;
+	}
+    char dest[256];
+    va_list argptr;
+    va_start(argptr, format);
+    vsprintf(dest, format, argptr);
+    va_end(argptr);
+    (*env)->ThrowNew(env, exClass, dest);
+}
+
+static jclass createGlobalRef(JNIEnv *env, jclass class) {
    if (class) {
        return (*env)->NewGlobalRef(env, class);
    }
@@ -27,16 +64,16 @@ jclass createGlobarRef(JNIEnv *env, jclass class) {
 }

 jint imageOnJNILoad(JavaVM *vm, void *reserved, JNIEnv *env) {
-    jclass_NullPointerException = createGlobarRef(env, (*env)->FindClass(env, "java/lang/NullPointerException"));
+    jclass_NullPointerException = createGlobalRef(env, (*env)->FindClass(env, "java/lang/NullPointerException"));
    if (jclass_NullPointerException == 0) {
        return -1;
    }
-    jclass_RuntimeException = createGlobarRef(env, (*env)->FindClass(env, "java/lang/RuntimeException"));
+    jclass_RuntimeException = createGlobalRef(env, (*env)->FindClass(env, "java/lang/RuntimeException"));
    if (jclass_RuntimeException == 0) {
        return -1;
    }
    
-    jclass_Options = createGlobarRef(env, (*env)->FindClass(env, "android/graphics/BitmapFactory$Options"));
+    jclass_Options = createGlobalRef(env, (*env)->FindClass(env, "android/graphics/BitmapFactory$Options"));
    if (jclass_Options == 0) {
        return -1;
    }
@@ -256,17 +293,17 @@ static void fastBlur(int imageWidth, int imageHeight, int imageStride, void *pix
    free(rgb);
 }

-typedef struct my_error_mgr {
+/*typedef struct my_error_mgr {
    struct jpeg_error_mgr pub;
    jmp_buf setjmp_buffer;
 } *my_error_ptr;


-METHODDEF(void) my_error_exit(j_common_ptr cinfo) {
+METHODDEF(void) my_jpeglib_error_exit(j_common_ptr cinfo) {
    my_error_ptr myerr = (my_error_ptr) cinfo->err;
    (*cinfo->err->output_message) (cinfo);
    longjmp(myerr->setjmp_buffer, 1);
-}
+}*/

 JNIEXPORT void Java_com_b44t_messenger_Utilities_blurBitmap(JNIEnv *env, jclass class, jobject bitmap, int radius, int unpin, int width, int height, int stride) {
    if (!bitmap) {
@@ -410,7 +447,7 @@ JNIEXPORT void Java_com_b44t_messenger_Utilities_unpinBitmap(JNIEnv *env, jclass
    AndroidBitmap_unlockPixels(env, bitmap);
 }

-JNIEXPORT void Java_com_b44t_messenger_Utilities_loadBitmap(JNIEnv *env, jclass class, jstring path, jobject bitmap, int scale, int width, int height, int stride) {
+/*JNIEXPORT void Java_com_b44t_messenger_Utilities_loadBitmap(JNIEnv *env, jclass class, jstring path, jobject bitmap, int scale, int width, int height, int stride) {
    
    AndroidBitmapInfo info;
    int i;
@@ -424,7 +461,7 @@ JNIEXPORT void Java_com_b44t_messenger_Utilities_loadBitmap(JNIEnv *env, jclass
            struct jpeg_decompress_struct cinfo;

            cinfo.err = jpeg_std_error(&jerr.pub);
-            jerr.pub.error_exit = my_error_exit;
+            jerr.pub.error_exit = my_jpeglib_error_exit;
            
            if (!setjmp(jerr.setjmp_buffer)) {
                jpeg_create_decompress(&cinfo);
@@ -493,9 +530,9 @@ JNIEXPORT void Java_com_b44t_messenger_Utilities_loadBitmap(JNIEnv *env, jclass
    } else {
        throwException(env, "AndroidBitmap_getInfo() failed ! error=%d", i);
    }
-}
+}*/

-JNIEXPORT jboolean Java_com_b44t_messenger_Utilities_loadWebpImage(JNIEnv *env, jclass class, jobject outputBitmap, jobject buffer, jint len, jobject options, jboolean unpin) {
+/*JNIEXPORT jboolean Java_com_b44t_messenger_Utilities_loadWebpImage(JNIEnv *env, jclass class, jobject outputBitmap, jobject buffer, jint len, jobject options, jboolean unpin) {
    if (!buffer) {
        (*env)->ThrowNew(env, jclass_NullPointerException, "Input buffer can not be null");
        return 0;
@@ -545,4 +582,4 @@ JNIEXPORT jboolean Java_com_b44t_messenger_Utilities_loadWebpImage(JNIEnv *env,
    }
    
    return 1;
-}
+}*/
@@ -1,8 +1,34 @@
-#ifndef image_h
-#define image_h
+/*******************************************************************************
+ *
+ *                          Messenger Android Frontend
+ *                        (C) 2013-2016 Nikolai Kudashov
+ *                           (C) 2017 Björn Petersen
+ *                    Contact: r10s@b44t.com, http://b44t.com
+ *
+ * This program is free software: you can redistribute it and/or modify it under
+ * the terms of the GNU General Public License as published by the Free Software
+ * Foundation, either version 3 of the License, or (at your option) any later
+ * version.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
+ * FOR A PARTICULAR PURPOSE.  See the GNU General Public License for more
+ * details.
+ *
+ * You should have received a copy of the GNU General Public License along with
+ * this program.  If not, see http://www.gnu.org/licenses/ .
+ *
+ ******************************************************************************/
+
+
+#ifndef __IMAGE_H__
+#define __IMAGE_H__
+

 #include <jni.h>

 jint imageOnJNILoad(JavaVM *vm, void *reserved, JNIEnv *env);

-#endif
+
+#endif /* __IMAGE_H__ */
+
@@ -1,67 +0,0 @@
-#include <stdio.h>
-#include <string.h>
-#include <jni.h>
-#include <sys/types.h>
-#include <inttypes.h>
-#include <stdlib.h>
-#include <openssl/aes.h>
-#include <unistd.h>
-#include "utils.h"
-#include "image.h"
-
-int gifvideoOnJNILoad(JavaVM *vm, JNIEnv *env);
-
-jint JNI_OnLoad(JavaVM *vm, void *reserved) { /* this function is called automatically by the JNI when the library gets loaded */
-	JNIEnv *env = 0;
-    srand(time(NULL));
-    
-	if ((*vm)->GetEnv(vm, (void **) &env, JNI_VERSION_1_6) != JNI_OK) {
-		return -1;
-	}
-    
-    if (imageOnJNILoad(vm, reserved, env) == -1) {
-        return -1;
-    }
-    
-    if (gifvideoOnJNILoad(vm, env) == -1) {
-        return -1;
-    }
-
-	return JNI_VERSION_1_6;
-}
-
-void JNI_OnUnload(JavaVM *vm, void *reserved) {
-
-}
-
-/*
-JNIEXPORT void Java_com_b44t_messenger_Utilities_aesIgeEncryption(JNIEnv *env, jclass class, jobject buffer, jbyteArray key, jbyteArray iv, jboolean encrypt, int offset, int length) {
-    jbyte *what = (*env)->GetDirectBufferAddress(env, buffer) + offset;
-    unsigned char *keyBuff = (unsigned char *)(*env)->GetByteArrayElements(env, key, NULL);
-    unsigned char *ivBuff = (unsigned char *)(*env)->GetByteArrayElements(env, iv, NULL);
-    
-    AES_KEY akey;
-    if (!encrypt) {
-        AES_set_decrypt_key(keyBuff, 32 * 8, &akey);
-        AES_ige_encrypt(what, what, length, &akey, ivBuff, AES_DECRYPT);
-    } else {
-        AES_set_encrypt_key(keyBuff, 32 * 8, &akey);
-        AES_ige_encrypt(what, what, length, &akey, ivBuff, AES_ENCRYPT);
-    }
-    (*env)->ReleaseByteArrayElements(env, key, keyBuff, JNI_ABORT);
-    (*env)->ReleaseByteArrayElements(env, iv, ivBuff, 0);
-}
-*/
-
-JNIEXPORT jstring Java_com_b44t_messenger_Utilities_readlink(JNIEnv *env, jclass class, jstring path) {
-    static char buf[1000];
-    char *fileName = (*env)->GetStringUTFChars(env, path, NULL);
-    int result = readlink(fileName, buf, 999);
-    jstring value = 0;
-    if (result != -1) {
-        buf[result] = '\0';
-        value = (*env)->NewStringUTF(env, buf);
-    }
-    (*env)->ReleaseStringUTFChars(env, path, fileName);
-    return value;
-}
@@ -122,13 +122,13 @@
 #define LIBETPAN_REENTRANT 1

 /* Define this to the version of libEtPan */
-#define LIBETPAN_VERSION "1.2-dev-20141203"
+//#define LIBETPAN_VERSION "1.2-dev-20141203"

 /* Define this to the major version of libEtPan */
-#define LIBETPAN_VERSION_MAJOR 1
+//#define LIBETPAN_VERSION_MAJOR 1

 /* Define this to the minor version of libEtPan */
-#define LIBETPAN_VERSION_MINOR 2
+//#define LIBETPAN_VERSION_MINOR 2

 /* Define to the sub-directory in which libtool stores uninstalled libraries.
   */
@@ -144,7 +144,7 @@
 #define PACKAGE_NAME "libetpan"

 /* Define to the full name and version of this package. */
-#define PACKAGE_STRING "libetpan 1.2"
+//#define PACKAGE_STRING "libetpan 1.2"

 /* Define to the one symbol short name of this package. */
 #define PACKAGE_TARNAME "libetpan"
@@ -153,7 +153,7 @@
 #define PACKAGE_URL ""

 /* Define to the version of this package. */
-#define PACKAGE_VERSION "1.2"
+//#define PACKAGE_VERSION "1.2"

 /* Define to 1 if you have the ANSI C header files. */
 #define STDC_HEADERS 1
@@ -171,7 +171,7 @@
 #define USE_SSL 1

 /* Version number of package */
-#define VERSION "1.2"
+//#define VERSION "1.2"

 /* Define to `__inline__' or `__inline' if that's what the C compiler
   calls it, or to nothing if 'inline' is not supported under any name.  */
@@ -1,94 +0,0 @@
-This software is based in part on the work of the Independent JPEG Group.
-
----------------------
-
-The authors make NO WARRANTY or representation, either express or implied,
-with respect to this software, its quality, accuracy, merchantability, or
-fitness for a particular purpose.  This software is provided "AS IS", and you,
-its user, assume the entire risk as to its quality and accuracy.
-
-This software is copyright (C) 1991-1998, Thomas G. Lane.
-All Rights Reserved except as specified below.
-
-Permission is hereby granted to use, copy, modify, and distribute this
-software (or portions thereof) for any purpose, without fee, subject to these
-conditions:
-(1) If any part of the source code for this software is distributed, then this
-README file must be included, with this copyright and no-warranty notice
-unaltered; and any additions, deletions, or changes to the original files
-must be clearly indicated in accompanying documentation.
-(2) If only executable code is distributed, then the accompanying
-documentation must state that "this software is based in part on the work of
-the Independent JPEG Group".
-(3) Permission for use of this software is granted only if the user accepts
-full responsibility for any undesirable consequences; the authors accept
-NO LIABILITY for damages of any kind.
-
-These conditions apply to any software derived from or based on the IJG code,
-not just to the unmodified library.  If you use our work, you ought to
-acknowledge us.
-
-Permission is NOT granted for the use of any IJG author's name or company name
-in advertising or publicity relating to this software or products derived from
-it.  This software may be referred to only as "the Independent JPEG Group's
-software".
-
-We specifically permit and encourage the use of this software as the basis of
-commercial products, provided that all warranty or liability claims are
-assumed by the product vendor.
-
-
----------------------
-
-
- ARM NEON optimizations for libjpeg-turbo
-
- Copyright (C) 2009-2011 Nokia Corporation and/or its subsidiary(-ies).
- All rights reserved.
- Contact: Alexander Bokovoy <alexander.bokovoy@nokia.com>
-
- This software is provided 'as-is', without any express or implied
- warranty.  In no event will the authors be held liable for any damages
- arising from the use of this software.
-
- Permission is granted to anyone to use this software for any purpose,
- including commercial applications, and to alter it and redistribute it
- freely, subject to the following restrictions:
-
- 1. The origin of this software must not be misrepresented; you must not
-    claim that you wrote the original software. If you use this software
-    in a product, an acknowledgment in the product documentation would be
-    appreciated but is not required.
- 2. Altered source versions must be plainly marked as such, and must not be
-    misrepresented as being the original software.
- 3. This notice may not be removed or altered from any source distribution.
-
-
- ----------------------
-
-
- Copyright (c) 2011,  NVIDIA CORPORATION. All rights reserved.
-
- Redistribution and use in source and binary forms, with or without
- modification, are permitted provided that the following conditions
- are met:
-
-  * Redistributions of source code must retain the above copyright
-    notice, this list of conditions and the following disclaimer.
-  * Redistributions in binary form must reproduce the above copyright
-    notice, this list of conditions and the following disclaimer in the
-    documentation and/or other materials provided with the distribution.
-  * Neither the name of the NVIDIA CORPORATION nor the names of its
-    contributors may be used to endorse or promote products derived
-    from this software without specific prior written permission.
- THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
- AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE
- LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
- INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
- CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
- THE POSSIBILITY OF SUCH DAMAGE.
@@ -1,366 +0,0 @@
-/*
- * Copyright (C) 2010 The Android Open Source Project
- *
- * Licensed under the Apache License, Version 2.0 (the "License");
- * you may not use this file except in compliance with the License.
- * You may obtain a copy of the License at
- *
- *      http://www.apache.org/licenses/LICENSE-2.0
- *
- * Unless required by applicable law or agreed to in writing, software
- * distributed under the License is distributed on an "AS IS" BASIS,
- * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
- * See the License for the specific language governing permissions and
- * limitations under the License.
- */
-
-/*
- * This is a fast-and-accurate implementation of inverse Discrete Cosine
- * Transform (IDCT) for ARMv6+. It also performs dequantization of the input
- * coefficients just like other methods.
- *
- * This implementation is based on the scaled 1-D DCT algorithm proposed by
- * Arai, Agui, and Nakajima. The following code is based on the figure 4-8
- * on page 52 of the JPEG textbook by Pennebaker and Mitchell. Coefficients
- * are (almost) directly mapped into registers.
- *
- * The accuracy is achieved by using SMULWy and SMLAWy instructions. Both
- * multiply 32 bits by 16 bits and store the top 32 bits of the result. It
- * makes 32-bit fixed-point arithmetic possible without overflow. That is
- * why jpeg_idct_ifast(), which is written in C, cannot be improved.
- *
- * More tricks are used to gain more speed. First of all, we use as many
- * registers as possible. ARM processor has 16 registers including sp (r13)
- * and pc (r15), so only 14 registers can be used without limitations. In
- * general, we let r0 to r7 hold the coefficients; r10 and r11 hold four
- * 16-bit constants; r12 and r14 hold two of the four arguments; and r8 hold
- * intermediate value. In the second pass, r9 is the loop counter. In the
- * first pass, r8 to r11 are used to hold quantization values, so the loop
- * counter is held by sp. Yes, the stack pointer. Since it must be aligned
- * to 4-byte boundary all the time, we align it to 32-byte boundary and use
- * bit 3 to bit 5. As the result, we actually use 14.1 registers. :-)
- *
- * Second, we rearrange quantization values to access them sequentially. The
- * table is first transposed, and the new columns are placed in the order of
- * 7, 5, 1, 3, 0, 2, 4, 6. Thus we can use LDMDB to load four values at a
- * time. Rearranging coefficients also helps, but that requires to change a
- * dozen of files, which seems not worth it. In addition, we choose to scale
- * up quantization values by 13 bits, so the coefficients are scaled up by
- * 16 bits after both passes. Then we can pack and saturate them two at a
- * time using PKHTB and USAT16 instructions.
- *
- * Third, we reorder the instructions to avoid bubbles in the pipeline. This
- * is done by hand accroding to the cycle timings and the interlock behavior
- * described in the technical reference manual of ARM1136JF-S. We also take
- * advantage of dual issue processors by interleaving instructions with
- * dependencies. It has been benchmarked on four devices and all the results
- * showed distinguishable improvements. Note that PLD instructions actually
- * slow things down, so they are removed at the last minute. In the future,
- * this might be futher improved using a system profiler.
- */
-
-#ifdef __arm__
-#include <machine/cpu-features.h>
-#endif
-
-#if __ARM_ARCH__ >= 6
-
-// void armv6_idct(short *coefs, int *quans, unsigned char *rows, int col)
-    .arm
-    .text
-    .align
-    .global armv6_idct
-    .func   armv6_idct
-
-armv6_idct:
-    // Push everything except sp (r13) and pc (r15).
-    stmdb   sp!, {r4, r5, r6, r7, r8, r9, r10, r11, r12, r14}
-
-    // r12 = quans, r14 = coefs.
-    sub     r4, sp, #236
-    bic     sp, r4, #31
-    add     r5, sp, #224
-    add     r12, r1, #256
-    stm     r5, {r2, r3, r4}
-    add     r14, r0, #16
-
-pass1_head:
-    // Load quantization values. (q[0, 2, 4, 6])
-    ldmdb   r12!, {r8, r9, r10, r11}
-
-    // Load coefficients. (c[4, 1, 2, 3, 0, 5, 6, 7])
-    ldrsh   r4, [r14, #-2] !
-    ldrsh   r1, [r14, #16]
-    ldrsh   r2, [r14, #32]
-    ldrsh   r3, [r14, #48]
-    ldrsh   r0, [r14, #64]
-    ldrsh   r5, [r14, #80]
-    ldrsh   r6, [r14, #96]
-    ldrsh   r7, [r14, #112]
-
-    // r4 = q[0] * c[0];
-    mul     r4, r8, r4
-
-    // Check if ACs are all zero.
-    cmp     r0, #0
-    orreqs  r8, r1, r2
-    orreqs  r8, r3, r5
-    orreqs  r8, r6, r7
-    beq     pass1_zero
-
-    // Step 1: Dequantizations.
-
-    // r2 = q[2] * c[2];
-    // r0 = q[4] * c[4] + r4;
-    // r6 = q[6] * c[6] + r2;
-    mul     r2, r9, r2
-    mla     r0, r10, r0, r4
-    mla     r6, r11, r6, r2
-
-    // Load quantization values. (q[7, 5, 1, 3])
-    ldmdb   r12!, {r8, r9, r10, r11}
-
-    // r4 = r4 * 2 - r0 = -(r0 - r4 * 2);
-    // r2 = r2 * 2 - r6 = -(r6 - r2 * 2);
-    rsb     r4, r0, r4, lsl #1
-    rsb     r2, r6, r2, lsl #1
-
-    // r7 = q[7] * c[7];
-    // r5 = q[5] * c[5];
-    // r1 = q[1] * c[1] + r7;
-    // r3 = q[3] * c[3] + r5;
-    mul     r7, r8, r7
-    mul     r5, r9, r5
-    mla     r1, r10, r1, r7
-    mla     r3, r11, r3, r5
-
-    // Load constants.
-    ldrd    r10, constants
-
-    // Step 2: Rotations and Butterflies.
-
-    // r7 = r1 - r7 * 2;
-    // r1 = r1 - r3;
-    // r5 = r5 * 2 - r3 = -(r3 - r5 * 2);
-    // r3 = r1 + r3 * 2;
-    // r8 = r5 + r7;
-    sub     r7, r1, r7, lsl #1
-    sub     r1, r1, r3
-    rsb     r5, r3, r5, lsl #1
-    add     r3, r1, r3, lsl #1
-    add     r8, r5, r7
-
-    // r2 = r2 * 1.41421 = r2 * 27146 / 65536 + r2;
-    // r8 = r8 * 1.84776 / 8 = r8 * 15137 / 65536;
-    // r1 = r1 * 1.41421 = r1 * 27146 / 65536 + r1;
-    smlawt  r2, r2, r10, r2
-    smulwb  r8, r8, r10
-    smlawt  r1, r1, r10, r1
-
-    // r0 = r0 + r6;
-    // r2 = r2 - r6;
-    // r6 = r0 - r6 * 2;
-    add     r0, r0, r6
-    sub     r2, r2, r6
-    sub     r6, r0, r6, lsl #1
-
-    // r5 = r5 * -2.61313 / 8 + r8 = r5 * -21407 / 65536 + r8;
-    // r8 = r7 * -1.08239 / 8 + r8 = r7 * -8867 / 65536 + r8;
-    smlawt  r5, r5, r11, r8
-    smlawb  r8, r7, r11, r8
-
-    // r4 = r4 + r2;
-    // r0 = r0 + r3;
-    // r2 = r4 - r2 * 2;
-    add     r4, r4, r2
-    add     r0, r0, r3
-    sub     r2, r4, r2, lsl #1
-
-    // r7 = r5 * 8 - r3 = -(r3 - r5 * 8);
-    // r3 = r0 - r3 * 2;
-    // r1 = r1 - r7;
-    // r4 = r4 + r7;
-    // r5 = r8 * 8 - r1 = -(r1 - r8 * 8);
-    // r7 = r4 - r7 * 2;
-    rsb     r7, r3, r5, lsl #3
-    sub     r3, r0, r3, lsl #1
-    sub     r1, r1, r7
-    add     r4, r4, r7
-    rsb     r5, r1, r8, lsl #3
-    sub     r7, r4, r7, lsl #1
-
-    // r2 = r2 + r1;
-    // r6 = r6 + r5;
-    // r1 = r2 - r1 * 2;
-    // r5 = r6 - r5 * 2;
-    add     r2, r2, r1
-    add     r6, r6, r5
-    sub     r1, r2, r1, lsl #1
-    sub     r5, r6, r5, lsl #1
-
-    // Step 3: Reorder and Save.
-
-    str     r0, [sp, #-4] !
-    str     r4, [sp, #32]
-    str     r2, [sp, #64]
-    str     r6, [sp, #96]
-    str     r5, [sp, #128]
-    str     r1, [sp, #160]
-    str     r7, [sp, #192]
-    str     r3, [sp, #224]
-    b       pass1_tail
-
-    // Precomputed 16-bit constants: 27146, 15137, -21407, -8867.
-    // Put them in the middle since LDRD only accepts offsets from -255 to 255.
-    .align  3
-constants:
-    .word   0x6a0a3b21
-    .word   0xac61dd5d
-
-pass1_zero:
-    str     r4, [sp, #-4] !
-    str     r4, [sp, #32]
-    str     r4, [sp, #64]
-    str     r4, [sp, #96]
-    str     r4, [sp, #128]
-    str     r4, [sp, #160]
-    str     r4, [sp, #192]
-    str     r4, [sp, #224]
-    sub     r12, r12, #16
-
-pass1_tail:
-    ands    r9, sp, #31
-    bne     pass1_head
-
-    // r12 = rows, r14 = col.
-    ldr     r12, [sp, #256]
-    ldr     r14, [sp, #260]
-
-    // Load constants.
-    ldrd    r10, constants
-
-pass2_head:
-    // Load coefficients. (c[0, 1, 2, 3, 4, 5, 6, 7])
-    ldmia   sp!, {r0, r1, r2, r3, r4, r5, r6, r7}
-
-    // r0 = r0 + 0x00808000;
-    add     r0, r0, #0x00800000
-    add     r0, r0, #0x00008000
-
-    // Step 1: Analog to the first pass.
-
-    // r0 = r0 + r4;
-    // r6 = r6 + r2;
-    add     r0, r0, r4
-    add     r6, r6, r2
-
-    // r4 = r0 - r4 * 2;
-    // r2 = r2 * 2 - r6 = -(r6 - r2 * 2);
-    sub     r4, r0, r4, lsl #1
-    rsb     r2, r6, r2, lsl #1
-
-    // r1 = r1 + r7;
-    // r3 = r3 + r5;
-    add     r1, r1, r7
-    add     r3, r3, r5
-
-    // Step 2: Rotations and Butterflies.
-
-    // r7 = r1 - r7 * 2;
-    // r1 = r1 - r3;
-    // r5 = r5 * 2 - r3 = -(r3 - r5 * 2);
-    // r3 = r1 + r3 * 2;
-    // r8 = r5 + r7;
-    sub     r7, r1, r7, lsl #1
-    sub     r1, r1, r3
-    rsb     r5, r3, r5, lsl #1
-    add     r3, r1, r3, lsl #1
-    add     r8, r5, r7
-
-    // r2 = r2 * 1.41421 = r2 * 27146 / 65536 + r2;
-    // r8 = r8 * 1.84776 / 8 = r8 * 15137 / 65536;
-    // r1 = r1 * 1.41421 = r1 * 27146 / 65536 + r1;
-    smlawt  r2, r2, r10, r2
-    smulwb  r8, r8, r10
-    smlawt  r1, r1, r10, r1
-
-    // r0 = r0 + r6;
-    // r2 = r2 - r6;
-    // r6 = r0 - r6 * 2;
-    add     r0, r0, r6
-    sub     r2, r2, r6
-    sub     r6, r0, r6, lsl #1
-
-    // r5 = r5 * -2.61313 / 8 + r8 = r5 * -21407 / 65536 + r8;
-    // r8 = r7 * -1.08239 / 8 + r8 = r7 * -8867 / 65536 + r8;
-    smlawt  r5, r5, r11, r8
-    smlawb  r8, r7, r11, r8
-
-    // r4 = r4 + r2;
-    // r0 = r0 + r3;
-    // r2 = r4 - r2 * 2;
-    add     r4, r4, r2
-    add     r0, r0, r3
-    sub     r2, r4, r2, lsl #1
-
-    // r7 = r5 * 8 - r3 = -(r3 - r5 * 8);
-    // r3 = r0 - r3 * 2;
-    // r1 = r1 - r7;
-    // r4 = r4 + r7;
-    // r5 = r8 * 8 - r1 = -(r1 - r8 * 8);
-    // r7 = r4 - r7 * 2;
-    rsb     r7, r3, r5, lsl #3
-    sub     r3, r0, r3, lsl #1
-    sub     r1, r1, r7
-    add     r4, r4, r7
-    rsb     r5, r1, r8, lsl #3
-    sub     r7, r4, r7, lsl #1
-
-    // r2 = r2 + r1;
-    // r6 = r6 + r5;
-    // r1 = r2 - r1 * 2;
-    // r5 = r6 - r5 * 2;
-    add     r2, r2, r1
-    add     r6, r6, r5
-    sub     r1, r2, r1, lsl #1
-    sub     r5, r6, r5, lsl #1
-
-    // Step 3: Reorder and Save.
-
-    // Load output pointer.
-    ldr     r8, [r12], #4
-
-    // For little endian: r6, r2, r4, r0, r3, r7, r1, r5.
-    pkhtb   r6, r6, r4, asr #16
-    pkhtb   r2, r2, r0, asr #16
-    pkhtb   r3, r3, r1, asr #16
-    pkhtb   r7, r7, r5, asr #16
-    usat16  r6, #8, r6
-    usat16  r2, #8, r2
-    usat16  r3, #8, r3
-    usat16  r7, #8, r7
-    orr     r0, r2, r6, lsl #8
-    orr     r1, r7, r3, lsl #8
-
-#ifdef __ARMEB__
-    // Reverse bytes for big endian.
-    rev     r0, r0
-    rev     r1, r1
-#endif
-
-    // Use STR instead of STRD to support unaligned access.
-    str     r0, [r8, r14] !
-    str     r1, [r8, #4]
-
-pass2_tail:
-    adds    r9, r9, #0x10000000
-    bpl     pass2_head
-
-    ldr     sp, [sp, #8]
-    add     sp, sp, #236
-
-    ldmia   sp!, {r4, r5, r6, r7, r8, r9, r10, r11, r12, r14}
-    bx      lr
-    .endfunc
-
-#endif
@@ -1,280 +0,0 @@
-/*
- * jcapimin.c
- *
- * Copyright (C) 1994-1998, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains application interface code for the compression half
- * of the JPEG library.  These are the "minimum" API routines that may be
- * needed in either the normal full-compression case or the transcoding-only
- * case.
- *
- * Most of the routines intended to be called directly by an application
- * are in this file or in jcapistd.c.  But also see jcparam.c for
- * parameter-setup helper routines, jcomapi.c for routines shared by
- * compression and decompression, and jctrans.c for the transcoding case.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-
-
-/*
- * Initialization of a JPEG compression object.
- * The error manager must already be set up (in case memory manager fails).
- */
-
-GLOBAL(void)
-jpeg_CreateCompress (j_compress_ptr cinfo, int version, size_t structsize)
-{
-  int i;
-
-  /* Guard against version mismatches between library and caller. */
-  cinfo->mem = NULL;		/* so jpeg_destroy knows mem mgr not called */
-  if (version != JPEG_LIB_VERSION)
-    ERREXIT2(cinfo, JERR_BAD_LIB_VERSION, JPEG_LIB_VERSION, version);
-  if (structsize != SIZEOF(struct jpeg_compress_struct))
-    ERREXIT2(cinfo, JERR_BAD_STRUCT_SIZE, 
-	     (int) SIZEOF(struct jpeg_compress_struct), (int) structsize);
-
-  /* For debugging purposes, we zero the whole master structure.
-   * But the application has already set the err pointer, and may have set
-   * client_data, so we have to save and restore those fields.
-   * Note: if application hasn't set client_data, tools like Purify may
-   * complain here.
-   */
-  {
-    struct jpeg_error_mgr * err = cinfo->err;
-    void * client_data = cinfo->client_data; /* ignore Purify complaint here */
-    MEMZERO(cinfo, SIZEOF(struct jpeg_compress_struct));
-    cinfo->err = err;
-    cinfo->client_data = client_data;
-  }
-  cinfo->is_decompressor = FALSE;
-
-  /* Initialize a memory manager instance for this object */
-  jinit_memory_mgr((j_common_ptr) cinfo);
-
-  /* Zero out pointers to permanent structures. */
-  cinfo->progress = NULL;
-  cinfo->dest = NULL;
-
-  cinfo->comp_info = NULL;
-
-  for (i = 0; i < NUM_QUANT_TBLS; i++)
-    cinfo->quant_tbl_ptrs[i] = NULL;
-
-  for (i = 0; i < NUM_HUFF_TBLS; i++) {
-    cinfo->dc_huff_tbl_ptrs[i] = NULL;
-    cinfo->ac_huff_tbl_ptrs[i] = NULL;
-  }
-
-  cinfo->script_space = NULL;
-
-  cinfo->input_gamma = 1.0;	/* in case application forgets */
-
-  /* OK, I'm ready */
-  cinfo->global_state = CSTATE_START;
-}
-
-
-/*
- * Destruction of a JPEG compression object
- */
-
-GLOBAL(void)
-jpeg_destroy_compress (j_compress_ptr cinfo)
-{
-  jpeg_destroy((j_common_ptr) cinfo); /* use common routine */
-}
-
-
-/*
- * Abort processing of a JPEG compression operation,
- * but don't destroy the object itself.
- */
-
-GLOBAL(void)
-jpeg_abort_compress (j_compress_ptr cinfo)
-{
-  jpeg_abort((j_common_ptr) cinfo); /* use common routine */
-}
-
-
-/*
- * Forcibly suppress or un-suppress all quantization and Huffman tables.
- * Marks all currently defined tables as already written (if suppress)
- * or not written (if !suppress).  This will control whether they get emitted
- * by a subsequent jpeg_start_compress call.
- *
- * This routine is exported for use by applications that want to produce
- * abbreviated JPEG datastreams.  It logically belongs in jcparam.c, but
- * since it is called by jpeg_start_compress, we put it here --- otherwise
- * jcparam.o would be linked whether the application used it or not.
- */
-
-GLOBAL(void)
-jpeg_suppress_tables (j_compress_ptr cinfo, boolean suppress)
-{
-  int i;
-  JQUANT_TBL * qtbl;
-  JHUFF_TBL * htbl;
-
-  for (i = 0; i < NUM_QUANT_TBLS; i++) {
-    if ((qtbl = cinfo->quant_tbl_ptrs[i]) != NULL)
-      qtbl->sent_table = suppress;
-  }
-
-  for (i = 0; i < NUM_HUFF_TBLS; i++) {
-    if ((htbl = cinfo->dc_huff_tbl_ptrs[i]) != NULL)
-      htbl->sent_table = suppress;
-    if ((htbl = cinfo->ac_huff_tbl_ptrs[i]) != NULL)
-      htbl->sent_table = suppress;
-  }
-}
-
-
-/*
- * Finish JPEG compression.
- *
- * If a multipass operating mode was selected, this may do a great deal of
- * work including most of the actual output.
- */
-
-GLOBAL(void)
-jpeg_finish_compress (j_compress_ptr cinfo)
-{
-  JDIMENSION iMCU_row;
-
-  if (cinfo->global_state == CSTATE_SCANNING ||
-      cinfo->global_state == CSTATE_RAW_OK) {
-    /* Terminate first pass */
-    if (cinfo->next_scanline < cinfo->image_height)
-      ERREXIT(cinfo, JERR_TOO_LITTLE_DATA);
-    (*cinfo->master->finish_pass) (cinfo);
-  } else if (cinfo->global_state != CSTATE_WRCOEFS)
-    ERREXIT1(cinfo, JERR_BAD_STATE, cinfo->global_state);
-  /* Perform any remaining passes */
-  while (! cinfo->master->is_last_pass) {
-    (*cinfo->master->prepare_for_pass) (cinfo);
-    for (iMCU_row = 0; iMCU_row < cinfo->total_iMCU_rows; iMCU_row++) {
-      if (cinfo->progress != NULL) {
-	cinfo->progress->pass_counter = (long) iMCU_row;
-	cinfo->progress->pass_limit = (long) cinfo->total_iMCU_rows;
-	(*cinfo->progress->progress_monitor) ((j_common_ptr) cinfo);
-      }
-      /* We bypass the main controller and invoke coef controller directly;
-       * all work is being done from the coefficient buffer.
-       */
-      if (! (*cinfo->coef->compress_data) (cinfo, (JSAMPIMAGE) NULL))
-	ERREXIT(cinfo, JERR_CANT_SUSPEND);
-    }
-    (*cinfo->master->finish_pass) (cinfo);
-  }
-  /* Write EOI, do final cleanup */
-  (*cinfo->marker->write_file_trailer) (cinfo);
-  (*cinfo->dest->term_destination) (cinfo);
-  /* We can use jpeg_abort to release memory and reset global_state */
-  jpeg_abort((j_common_ptr) cinfo);
-}
-
-
-/*
- * Write a special marker.
- * This is only recommended for writing COM or APPn markers.
- * Must be called after jpeg_start_compress() and before
- * first call to jpeg_write_scanlines() or jpeg_write_raw_data().
- */
-
-GLOBAL(void)
-jpeg_write_marker (j_compress_ptr cinfo, int marker,
-		   const JOCTET *dataptr, unsigned int datalen)
-{
-  JMETHOD(void, write_marker_byte, (j_compress_ptr info, int val));
-
-  if (cinfo->next_scanline != 0 ||
-      (cinfo->global_state != CSTATE_SCANNING &&
-       cinfo->global_state != CSTATE_RAW_OK &&
-       cinfo->global_state != CSTATE_WRCOEFS))
-    ERREXIT1(cinfo, JERR_BAD_STATE, cinfo->global_state);
-
-  (*cinfo->marker->write_marker_header) (cinfo, marker, datalen);
-  write_marker_byte = cinfo->marker->write_marker_byte;	/* copy for speed */
-  while (datalen--) {
-    (*write_marker_byte) (cinfo, *dataptr);
-    dataptr++;
-  }
-}
-
-/* Same, but piecemeal. */
-
-GLOBAL(void)
-jpeg_write_m_header (j_compress_ptr cinfo, int marker, unsigned int datalen)
-{
-  if (cinfo->next_scanline != 0 ||
-      (cinfo->global_state != CSTATE_SCANNING &&
-       cinfo->global_state != CSTATE_RAW_OK &&
-       cinfo->global_state != CSTATE_WRCOEFS))
-    ERREXIT1(cinfo, JERR_BAD_STATE, cinfo->global_state);
-
-  (*cinfo->marker->write_marker_header) (cinfo, marker, datalen);
-}
-
-GLOBAL(void)
-jpeg_write_m_byte (j_compress_ptr cinfo, int val)
-{
-  (*cinfo->marker->write_marker_byte) (cinfo, val);
-}
-
-
-/*
- * Alternate compression function: just write an abbreviated table file.
- * Before calling this, all parameters and a data destination must be set up.
- *
- * To produce a pair of files containing abbreviated tables and abbreviated
- * image data, one would proceed as follows:
- *
- *		initialize JPEG object
- *		set JPEG parameters
- *		set destination to table file
- *		jpeg_write_tables(cinfo);
- *		set destination to image file
- *		jpeg_start_compress(cinfo, FALSE);
- *		write data...
- *		jpeg_finish_compress(cinfo);
- *
- * jpeg_write_tables has the side effect of marking all tables written
- * (same as jpeg_suppress_tables(..., TRUE)).  Thus a subsequent start_compress
- * will not re-emit the tables unless it is passed write_all_tables=TRUE.
- */
-
-GLOBAL(void)
-jpeg_write_tables (j_compress_ptr cinfo)
-{
-  if (cinfo->global_state != CSTATE_START)
-    ERREXIT1(cinfo, JERR_BAD_STATE, cinfo->global_state);
-
-  /* (Re)initialize error mgr and destination modules */
-  (*cinfo->err->reset_error_mgr) ((j_common_ptr) cinfo);
-  (*cinfo->dest->init_destination) (cinfo);
-  /* Initialize the marker writer ... bit of a crock to do it here. */
-  jinit_marker_writer(cinfo);
-  /* Write them tables! */
-  (*cinfo->marker->write_tables_only) (cinfo);
-  /* And clean up. */
-  (*cinfo->dest->term_destination) (cinfo);
-  /*
-   * In library releases up through v6a, we called jpeg_abort() here to free
-   * any working memory allocated by the destination manager and marker
-   * writer.  Some applications had a problem with that: they allocated space
-   * of their own from the library memory manager, and didn't want it to go
-   * away during write_tables.  So now we do nothing.  This will cause a
-   * memory leak if an app calls write_tables repeatedly without doing a full
-   * compression cycle or otherwise resetting the JPEG object.  However, that
-   * seems less bad than unexpectedly freeing memory in the normal case.
-   * An app that prefers the old behavior can call jpeg_abort for itself after
-   * each call to jpeg_write_tables().
-   */
-}
@@ -1,161 +0,0 @@
-/*
- * jcapistd.c
- *
- * Copyright (C) 1994-1996, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains application interface code for the compression half
- * of the JPEG library.  These are the "standard" API routines that are
- * used in the normal full-compression case.  They are not used by a
- * transcoding-only application.  Note that if an application links in
- * jpeg_start_compress, it will end up linking in the entire compressor.
- * We thus must separate this file from jcapimin.c to avoid linking the
- * whole compression library into a transcoder.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-
-
-/*
- * Compression initialization.
- * Before calling this, all parameters and a data destination must be set up.
- *
- * We require a write_all_tables parameter as a failsafe check when writing
- * multiple datastreams from the same compression object.  Since prior runs
- * will have left all the tables marked sent_table=TRUE, a subsequent run
- * would emit an abbreviated stream (no tables) by default.  This may be what
- * is wanted, but for safety's sake it should not be the default behavior:
- * programmers should have to make a deliberate choice to emit abbreviated
- * images.  Therefore the documentation and examples should encourage people
- * to pass write_all_tables=TRUE; then it will take active thought to do the
- * wrong thing.
- */
-
-GLOBAL(void)
-jpeg_start_compress (j_compress_ptr cinfo, boolean write_all_tables)
-{
-  if (cinfo->global_state != CSTATE_START)
-    ERREXIT1(cinfo, JERR_BAD_STATE, cinfo->global_state);
-
-  if (write_all_tables)
-    jpeg_suppress_tables(cinfo, FALSE);	/* mark all tables to be written */
-
-  /* (Re)initialize error mgr and destination modules */
-  (*cinfo->err->reset_error_mgr) ((j_common_ptr) cinfo);
-  (*cinfo->dest->init_destination) (cinfo);
-  /* Perform master selection of active modules */
-  jinit_compress_master(cinfo);
-  /* Set up for the first pass */
-  (*cinfo->master->prepare_for_pass) (cinfo);
-  /* Ready for application to drive first pass through jpeg_write_scanlines
-   * or jpeg_write_raw_data.
-   */
-  cinfo->next_scanline = 0;
-  cinfo->global_state = (cinfo->raw_data_in ? CSTATE_RAW_OK : CSTATE_SCANNING);
-}
-
-
-/*
- * Write some scanlines of data to the JPEG compressor.
- *
- * The return value will be the number of lines actually written.
- * This should be less than the supplied num_lines only in case that
- * the data destination module has requested suspension of the compressor,
- * or if more than image_height scanlines are passed in.
- *
- * Note: we warn about excess calls to jpeg_write_scanlines() since
- * this likely signals an application programmer error.  However,
- * excess scanlines passed in the last valid call are *silently* ignored,
- * so that the application need not adjust num_lines for end-of-image
- * when using a multiple-scanline buffer.
- */
-
-GLOBAL(JDIMENSION)
-jpeg_write_scanlines (j_compress_ptr cinfo, JSAMPARRAY scanlines,
-		      JDIMENSION num_lines)
-{
-  JDIMENSION row_ctr, rows_left;
-
-  if (cinfo->global_state != CSTATE_SCANNING)
-    ERREXIT1(cinfo, JERR_BAD_STATE, cinfo->global_state);
-  if (cinfo->next_scanline >= cinfo->image_height)
-    WARNMS(cinfo, JWRN_TOO_MUCH_DATA);
-
-  /* Call progress monitor hook if present */
-  if (cinfo->progress != NULL) {
-    cinfo->progress->pass_counter = (long) cinfo->next_scanline;
-    cinfo->progress->pass_limit = (long) cinfo->image_height;
-    (*cinfo->progress->progress_monitor) ((j_common_ptr) cinfo);
-  }
-
-  /* Give master control module another chance if this is first call to
-   * jpeg_write_scanlines.  This lets output of the frame/scan headers be
-   * delayed so that application can write COM, etc, markers between
-   * jpeg_start_compress and jpeg_write_scanlines.
-   */
-  if (cinfo->master->call_pass_startup)
-    (*cinfo->master->pass_startup) (cinfo);
-
-  /* Ignore any extra scanlines at bottom of image. */
-  rows_left = cinfo->image_height - cinfo->next_scanline;
-  if (num_lines > rows_left)
-    num_lines = rows_left;
-
-  row_ctr = 0;
-  (*cinfo->main->process_data) (cinfo, scanlines, &row_ctr, num_lines);
-  cinfo->next_scanline += row_ctr;
-  return row_ctr;
-}
-
-
-/*
- * Alternate entry point to write raw data.
- * Processes exactly one iMCU row per call, unless suspended.
- */
-
-GLOBAL(JDIMENSION)
-jpeg_write_raw_data (j_compress_ptr cinfo, JSAMPIMAGE data,
-		     JDIMENSION num_lines)
-{
-  JDIMENSION lines_per_iMCU_row;
-
-  if (cinfo->global_state != CSTATE_RAW_OK)
-    ERREXIT1(cinfo, JERR_BAD_STATE, cinfo->global_state);
-  if (cinfo->next_scanline >= cinfo->image_height) {
-    WARNMS(cinfo, JWRN_TOO_MUCH_DATA);
-    return 0;
-  }
-
-  /* Call progress monitor hook if present */
-  if (cinfo->progress != NULL) {
-    cinfo->progress->pass_counter = (long) cinfo->next_scanline;
-    cinfo->progress->pass_limit = (long) cinfo->image_height;
-    (*cinfo->progress->progress_monitor) ((j_common_ptr) cinfo);
-  }
-
-  /* Give master control module another chance if this is first call to
-   * jpeg_write_raw_data.  This lets output of the frame/scan headers be
-   * delayed so that application can write COM, etc, markers between
-   * jpeg_start_compress and jpeg_write_raw_data.
-   */
-  if (cinfo->master->call_pass_startup)
-    (*cinfo->master->pass_startup) (cinfo);
-
-  /* Verify that at least one iMCU row has been passed. */
-  lines_per_iMCU_row = cinfo->max_v_samp_factor * DCTSIZE;
-  if (num_lines < lines_per_iMCU_row)
-    ERREXIT(cinfo, JERR_BUFFER_SIZE);
-
-  /* Directly compress the row. */
-  if (! (*cinfo->coef->compress_data) (cinfo, data)) {
-    /* If compressor did not consume the whole row, suspend processing. */
-    return 0;
-  }
-
-  /* OK, we processed one iMCU row. */
-  cinfo->next_scanline += lines_per_iMCU_row;
-  return lines_per_iMCU_row;
-}
@@ -1,449 +0,0 @@
-/*
- * jccoefct.c
- *
- * Copyright (C) 1994-1997, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains the coefficient buffer controller for compression.
- * This controller is the top level of the JPEG compressor proper.
- * The coefficient buffer lies between forward-DCT and entropy encoding steps.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-
-
-/* We use a full-image coefficient buffer when doing Huffman optimization,
- * and also for writing multiple-scan JPEG files.  In all cases, the DCT
- * step is run during the first pass, and subsequent passes need only read
- * the buffered coefficients.
- */
-#ifdef ENTROPY_OPT_SUPPORTED
-#define FULL_COEF_BUFFER_SUPPORTED
-#else
-#ifdef C_MULTISCAN_FILES_SUPPORTED
-#define FULL_COEF_BUFFER_SUPPORTED
-#endif
-#endif
-
-
-/* Private buffer controller object */
-
-typedef struct {
-  struct jpeg_c_coef_controller pub; /* public fields */
-
-  JDIMENSION iMCU_row_num;	/* iMCU row # within image */
-  JDIMENSION mcu_ctr;		/* counts MCUs processed in current row */
-  int MCU_vert_offset;		/* counts MCU rows within iMCU row */
-  int MCU_rows_per_iMCU_row;	/* number of such rows needed */
-
-  /* For single-pass compression, it's sufficient to buffer just one MCU
-   * (although this may prove a bit slow in practice).  We allocate a
-   * workspace of C_MAX_BLOCKS_IN_MCU coefficient blocks, and reuse it for each
-   * MCU constructed and sent.  (On 80x86, the workspace is FAR even though
-   * it's not really very big; this is to keep the module interfaces unchanged
-   * when a large coefficient buffer is necessary.)
-   * In multi-pass modes, this array points to the current MCU's blocks
-   * within the virtual arrays.
-   */
-  JBLOCKROW MCU_buffer[C_MAX_BLOCKS_IN_MCU];
-
-  /* In multi-pass modes, we need a virtual block array for each component. */
-  jvirt_barray_ptr whole_image[MAX_COMPONENTS];
-} my_coef_controller;
-
-typedef my_coef_controller * my_coef_ptr;
-
-
-/* Forward declarations */
-METHODDEF(boolean) compress_data
-    JPP((j_compress_ptr cinfo, JSAMPIMAGE input_buf));
-#ifdef FULL_COEF_BUFFER_SUPPORTED
-METHODDEF(boolean) compress_first_pass
-    JPP((j_compress_ptr cinfo, JSAMPIMAGE input_buf));
-METHODDEF(boolean) compress_output
-    JPP((j_compress_ptr cinfo, JSAMPIMAGE input_buf));
-#endif
-
-
-LOCAL(void)
-start_iMCU_row (j_compress_ptr cinfo)
-/* Reset within-iMCU-row counters for a new row */
-{
-  my_coef_ptr coef = (my_coef_ptr) cinfo->coef;
-
-  /* In an interleaved scan, an MCU row is the same as an iMCU row.
-   * In a noninterleaved scan, an iMCU row has v_samp_factor MCU rows.
-   * But at the bottom of the image, process only what's left.
-   */
-  if (cinfo->comps_in_scan > 1) {
-    coef->MCU_rows_per_iMCU_row = 1;
-  } else {
-    if (coef->iMCU_row_num < (cinfo->total_iMCU_rows-1))
-      coef->MCU_rows_per_iMCU_row = cinfo->cur_comp_info[0]->v_samp_factor;
-    else
-      coef->MCU_rows_per_iMCU_row = cinfo->cur_comp_info[0]->last_row_height;
-  }
-
-  coef->mcu_ctr = 0;
-  coef->MCU_vert_offset = 0;
-}
-
-
-/*
- * Initialize for a processing pass.
- */
-
-METHODDEF(void)
-start_pass_coef (j_compress_ptr cinfo, J_BUF_MODE pass_mode)
-{
-  my_coef_ptr coef = (my_coef_ptr) cinfo->coef;
-
-  coef->iMCU_row_num = 0;
-  start_iMCU_row(cinfo);
-
-  switch (pass_mode) {
-  case JBUF_PASS_THRU:
-    if (coef->whole_image[0] != NULL)
-      ERREXIT(cinfo, JERR_BAD_BUFFER_MODE);
-    coef->pub.compress_data = compress_data;
-    break;
-#ifdef FULL_COEF_BUFFER_SUPPORTED
-  case JBUF_SAVE_AND_PASS:
-    if (coef->whole_image[0] == NULL)
-      ERREXIT(cinfo, JERR_BAD_BUFFER_MODE);
-    coef->pub.compress_data = compress_first_pass;
-    break;
-  case JBUF_CRANK_DEST:
-    if (coef->whole_image[0] == NULL)
-      ERREXIT(cinfo, JERR_BAD_BUFFER_MODE);
-    coef->pub.compress_data = compress_output;
-    break;
-#endif
-  default:
-    ERREXIT(cinfo, JERR_BAD_BUFFER_MODE);
-    break;
-  }
-}
-
-
-/*
- * Process some data in the single-pass case.
- * We process the equivalent of one fully interleaved MCU row ("iMCU" row)
- * per call, ie, v_samp_factor block rows for each component in the image.
- * Returns TRUE if the iMCU row is completed, FALSE if suspended.
- *
- * NB: input_buf contains a plane for each component in image,
- * which we index according to the component's SOF position.
- */
-
-METHODDEF(boolean)
-compress_data (j_compress_ptr cinfo, JSAMPIMAGE input_buf)
-{
-  my_coef_ptr coef = (my_coef_ptr) cinfo->coef;
-  JDIMENSION MCU_col_num;	/* index of current MCU within row */
-  JDIMENSION last_MCU_col = cinfo->MCUs_per_row - 1;
-  JDIMENSION last_iMCU_row = cinfo->total_iMCU_rows - 1;
-  int blkn, bi, ci, yindex, yoffset, blockcnt;
-  JDIMENSION ypos, xpos;
-  jpeg_component_info *compptr;
-
-  /* Loop to write as much as one whole iMCU row */
-  for (yoffset = coef->MCU_vert_offset; yoffset < coef->MCU_rows_per_iMCU_row;
-       yoffset++) {
-    for (MCU_col_num = coef->mcu_ctr; MCU_col_num <= last_MCU_col;
-	 MCU_col_num++) {
-      /* Determine where data comes from in input_buf and do the DCT thing.
-       * Each call on forward_DCT processes a horizontal row of DCT blocks
-       * as wide as an MCU; we rely on having allocated the MCU_buffer[] blocks
-       * sequentially.  Dummy blocks at the right or bottom edge are filled in
-       * specially.  The data in them does not matter for image reconstruction,
-       * so we fill them with values that will encode to the smallest amount of
-       * data, viz: all zeroes in the AC entries, DC entries equal to previous
-       * block's DC value.  (Thanks to Thomas Kinsman for this idea.)
-       */
-      blkn = 0;
-      for (ci = 0; ci < cinfo->comps_in_scan; ci++) {
-	compptr = cinfo->cur_comp_info[ci];
-	blockcnt = (MCU_col_num < last_MCU_col) ? compptr->MCU_width
-						: compptr->last_col_width;
-	xpos = MCU_col_num * compptr->MCU_sample_width;
-	ypos = yoffset * DCTSIZE; /* ypos == (yoffset+yindex) * DCTSIZE */
-	for (yindex = 0; yindex < compptr->MCU_height; yindex++) {
-	  if (coef->iMCU_row_num < last_iMCU_row ||
-	      yoffset+yindex < compptr->last_row_height) {
-	    (*cinfo->fdct->forward_DCT) (cinfo, compptr,
-					 input_buf[compptr->component_index],
-					 coef->MCU_buffer[blkn],
-					 ypos, xpos, (JDIMENSION) blockcnt);
-	    if (blockcnt < compptr->MCU_width) {
-	      /* Create some dummy blocks at the right edge of the image. */
-	      jzero_far((void FAR *) coef->MCU_buffer[blkn + blockcnt],
-			(compptr->MCU_width - blockcnt) * SIZEOF(JBLOCK));
-	      for (bi = blockcnt; bi < compptr->MCU_width; bi++) {
-		coef->MCU_buffer[blkn+bi][0][0] = coef->MCU_buffer[blkn+bi-1][0][0];
-	      }
-	    }
-	  } else {
-	    /* Create a row of dummy blocks at the bottom of the image. */
-	    jzero_far((void FAR *) coef->MCU_buffer[blkn],
-		      compptr->MCU_width * SIZEOF(JBLOCK));
-	    for (bi = 0; bi < compptr->MCU_width; bi++) {
-	      coef->MCU_buffer[blkn+bi][0][0] = coef->MCU_buffer[blkn-1][0][0];
-	    }
-	  }
-	  blkn += compptr->MCU_width;
-	  ypos += DCTSIZE;
-	}
-      }
-      /* Try to write the MCU.  In event of a suspension failure, we will
-       * re-DCT the MCU on restart (a bit inefficient, could be fixed...)
-       */
-      if (! (*cinfo->entropy->encode_mcu) (cinfo, coef->MCU_buffer)) {
-	/* Suspension forced; update state counters and exit */
-	coef->MCU_vert_offset = yoffset;
-	coef->mcu_ctr = MCU_col_num;
-	return FALSE;
-      }
-    }
-    /* Completed an MCU row, but perhaps not an iMCU row */
-    coef->mcu_ctr = 0;
-  }
-  /* Completed the iMCU row, advance counters for next one */
-  coef->iMCU_row_num++;
-  start_iMCU_row(cinfo);
-  return TRUE;
-}
-
-
-#ifdef FULL_COEF_BUFFER_SUPPORTED
-
-/*
- * Process some data in the first pass of a multi-pass case.
- * We process the equivalent of one fully interleaved MCU row ("iMCU" row)
- * per call, ie, v_samp_factor block rows for each component in the image.
- * This amount of data is read from the source buffer, DCT'd and quantized,
- * and saved into the virtual arrays.  We also generate suitable dummy blocks
- * as needed at the right and lower edges.  (The dummy blocks are constructed
- * in the virtual arrays, which have been padded appropriately.)  This makes
- * it possible for subsequent passes not to worry about real vs. dummy blocks.
- *
- * We must also emit the data to the entropy encoder.  This is conveniently
- * done by calling compress_output() after we've loaded the current strip
- * of the virtual arrays.
- *
- * NB: input_buf contains a plane for each component in image.  All
- * components are DCT'd and loaded into the virtual arrays in this pass.
- * However, it may be that only a subset of the components are emitted to
- * the entropy encoder during this first pass; be careful about looking
- * at the scan-dependent variables (MCU dimensions, etc).
- */
-
-METHODDEF(boolean)
-compress_first_pass (j_compress_ptr cinfo, JSAMPIMAGE input_buf)
-{
-  my_coef_ptr coef = (my_coef_ptr) cinfo->coef;
-  JDIMENSION last_iMCU_row = cinfo->total_iMCU_rows - 1;
-  JDIMENSION blocks_across, MCUs_across, MCUindex;
-  int bi, ci, h_samp_factor, block_row, block_rows, ndummy;
-  JCOEF lastDC;
-  jpeg_component_info *compptr;
-  JBLOCKARRAY buffer;
-  JBLOCKROW thisblockrow, lastblockrow;
-
-  for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
-       ci++, compptr++) {
-    /* Align the virtual buffer for this component. */
-    buffer = (*cinfo->mem->access_virt_barray)
-      ((j_common_ptr) cinfo, coef->whole_image[ci],
-       coef->iMCU_row_num * compptr->v_samp_factor,
-       (JDIMENSION) compptr->v_samp_factor, TRUE);
-    /* Count non-dummy DCT block rows in this iMCU row. */
-    if (coef->iMCU_row_num < last_iMCU_row)
-      block_rows = compptr->v_samp_factor;
-    else {
-      /* NB: can't use last_row_height here, since may not be set! */
-      block_rows = (int) (compptr->height_in_blocks % compptr->v_samp_factor);
-      if (block_rows == 0) block_rows = compptr->v_samp_factor;
-    }
-    blocks_across = compptr->width_in_blocks;
-    h_samp_factor = compptr->h_samp_factor;
-    /* Count number of dummy blocks to be added at the right margin. */
-    ndummy = (int) (blocks_across % h_samp_factor);
-    if (ndummy > 0)
-      ndummy = h_samp_factor - ndummy;
-    /* Perform DCT for all non-dummy blocks in this iMCU row.  Each call
-     * on forward_DCT processes a complete horizontal row of DCT blocks.
-     */
-    for (block_row = 0; block_row < block_rows; block_row++) {
-      thisblockrow = buffer[block_row];
-      (*cinfo->fdct->forward_DCT) (cinfo, compptr,
-				   input_buf[ci], thisblockrow,
-				   (JDIMENSION) (block_row * DCTSIZE),
-				   (JDIMENSION) 0, blocks_across);
-      if (ndummy > 0) {
-	/* Create dummy blocks at the right edge of the image. */
-	thisblockrow += blocks_across; /* => first dummy block */
-	jzero_far((void FAR *) thisblockrow, ndummy * SIZEOF(JBLOCK));
-	lastDC = thisblockrow[-1][0];
-	for (bi = 0; bi < ndummy; bi++) {
-	  thisblockrow[bi][0] = lastDC;
-	}
-      }
-    }
-    /* If at end of image, create dummy block rows as needed.
-     * The tricky part here is that within each MCU, we want the DC values
-     * of the dummy blocks to match the last real block's DC value.
-     * This squeezes a few more bytes out of the resulting file...
-     */
-    if (coef->iMCU_row_num == last_iMCU_row) {
-      blocks_across += ndummy;	/* include lower right corner */
-      MCUs_across = blocks_across / h_samp_factor;
-      for (block_row = block_rows; block_row < compptr->v_samp_factor;
-	   block_row++) {
-	thisblockrow = buffer[block_row];
-	lastblockrow = buffer[block_row-1];
-	jzero_far((void FAR *) thisblockrow,
-		  (size_t) (blocks_across * SIZEOF(JBLOCK)));
-	for (MCUindex = 0; MCUindex < MCUs_across; MCUindex++) {
-	  lastDC = lastblockrow[h_samp_factor-1][0];
-	  for (bi = 0; bi < h_samp_factor; bi++) {
-	    thisblockrow[bi][0] = lastDC;
-	  }
-	  thisblockrow += h_samp_factor; /* advance to next MCU in row */
-	  lastblockrow += h_samp_factor;
-	}
-      }
-    }
-  }
-  /* NB: compress_output will increment iMCU_row_num if successful.
-   * A suspension return will result in redoing all the work above next time.
-   */
-
-  /* Emit data to the entropy encoder, sharing code with subsequent passes */
-  return compress_output(cinfo, input_buf);
-}
-
-
-/*
- * Process some data in subsequent passes of a multi-pass case.
- * We process the equivalent of one fully interleaved MCU row ("iMCU" row)
- * per call, ie, v_samp_factor block rows for each component in the scan.
- * The data is obtained from the virtual arrays and fed to the entropy coder.
- * Returns TRUE if the iMCU row is completed, FALSE if suspended.
- *
- * NB: input_buf is ignored; it is likely to be a NULL pointer.
- */
-
-METHODDEF(boolean)
-compress_output (j_compress_ptr cinfo, JSAMPIMAGE input_buf)
-{
-  my_coef_ptr coef = (my_coef_ptr) cinfo->coef;
-  JDIMENSION MCU_col_num;	/* index of current MCU within row */
-  int blkn, ci, xindex, yindex, yoffset;
-  JDIMENSION start_col;
-  JBLOCKARRAY buffer[MAX_COMPS_IN_SCAN];
-  JBLOCKROW buffer_ptr;
-  jpeg_component_info *compptr;
-
-  /* Align the virtual buffers for the components used in this scan.
-   * NB: during first pass, this is safe only because the buffers will
-   * already be aligned properly, so jmemmgr.c won't need to do any I/O.
-   */
-  for (ci = 0; ci < cinfo->comps_in_scan; ci++) {
-    compptr = cinfo->cur_comp_info[ci];
-    buffer[ci] = (*cinfo->mem->access_virt_barray)
-      ((j_common_ptr) cinfo, coef->whole_image[compptr->component_index],
-       coef->iMCU_row_num * compptr->v_samp_factor,
-       (JDIMENSION) compptr->v_samp_factor, FALSE);
-  }
-
-  /* Loop to process one whole iMCU row */
-  for (yoffset = coef->MCU_vert_offset; yoffset < coef->MCU_rows_per_iMCU_row;
-       yoffset++) {
-    for (MCU_col_num = coef->mcu_ctr; MCU_col_num < cinfo->MCUs_per_row;
-	 MCU_col_num++) {
-      /* Construct list of pointers to DCT blocks belonging to this MCU */
-      blkn = 0;			/* index of current DCT block within MCU */
-      for (ci = 0; ci < cinfo->comps_in_scan; ci++) {
-	compptr = cinfo->cur_comp_info[ci];
-	start_col = MCU_col_num * compptr->MCU_width;
-	for (yindex = 0; yindex < compptr->MCU_height; yindex++) {
-	  buffer_ptr = buffer[ci][yindex+yoffset] + start_col;
-	  for (xindex = 0; xindex < compptr->MCU_width; xindex++) {
-	    coef->MCU_buffer[blkn++] = buffer_ptr++;
-	  }
-	}
-      }
-      /* Try to write the MCU. */
-      if (! (*cinfo->entropy->encode_mcu) (cinfo, coef->MCU_buffer)) {
-	/* Suspension forced; update state counters and exit */
-	coef->MCU_vert_offset = yoffset;
-	coef->mcu_ctr = MCU_col_num;
-	return FALSE;
-      }
-    }
-    /* Completed an MCU row, but perhaps not an iMCU row */
-    coef->mcu_ctr = 0;
-  }
-  /* Completed the iMCU row, advance counters for next one */
-  coef->iMCU_row_num++;
-  start_iMCU_row(cinfo);
-  return TRUE;
-}
-
-#endif /* FULL_COEF_BUFFER_SUPPORTED */
-
-
-/*
- * Initialize coefficient buffer controller.
- */
-
-GLOBAL(void)
-jinit_c_coef_controller (j_compress_ptr cinfo, boolean need_full_buffer)
-{
-  my_coef_ptr coef;
-
-  coef = (my_coef_ptr)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				SIZEOF(my_coef_controller));
-  cinfo->coef = (struct jpeg_c_coef_controller *) coef;
-  coef->pub.start_pass = start_pass_coef;
-
-  /* Create the coefficient buffer. */
-  if (need_full_buffer) {
-#ifdef FULL_COEF_BUFFER_SUPPORTED
-    /* Allocate a full-image virtual array for each component, */
-    /* padded to a multiple of samp_factor DCT blocks in each direction. */
-    int ci;
-    jpeg_component_info *compptr;
-
-    for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
-	 ci++, compptr++) {
-      coef->whole_image[ci] = (*cinfo->mem->request_virt_barray)
-	((j_common_ptr) cinfo, JPOOL_IMAGE, FALSE,
-	 (JDIMENSION) jround_up((long) compptr->width_in_blocks,
-				(long) compptr->h_samp_factor),
-	 (JDIMENSION) jround_up((long) compptr->height_in_blocks,
-				(long) compptr->v_samp_factor),
-	 (JDIMENSION) compptr->v_samp_factor);
-    }
-#else
-    ERREXIT(cinfo, JERR_BAD_BUFFER_MODE);
-#endif
-  } else {
-    /* We only need a single-MCU buffer. */
-    JBLOCKROW buffer;
-    int i;
-
-    buffer = (JBLOCKROW)
-      (*cinfo->mem->alloc_large) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				  C_MAX_BLOCKS_IN_MCU * SIZEOF(JBLOCK));
-    for (i = 0; i < C_MAX_BLOCKS_IN_MCU; i++) {
-      coef->MCU_buffer[i] = buffer + i;
-    }
-    coef->whole_image[0] = NULL; /* flag for no virtual arrays */
-  }
-}
@@ -1,527 +0,0 @@
-/*
- * jccolor.c
- *
- * Copyright (C) 1991-1996, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains input colorspace conversion routines.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-
-// this enables unrolling null_convert's loop, and reading/write ints for speed
-#define ENABLE_ANDROID_NULL_CONVERT
-
-/* Private subobject */
-
-typedef struct {
-  struct jpeg_color_converter pub; /* public fields */
-
-  /* Private state for RGB->YCC conversion */
-  INT32 * rgb_ycc_tab;		/* => table for RGB to YCbCr conversion */
-} my_color_converter;
-
-typedef my_color_converter * my_cconvert_ptr;
-
-
-/**************** RGB -> YCbCr conversion: most common case **************/
-
-/*
- * YCbCr is defined per CCIR 601-1, except that Cb and Cr are
- * normalized to the range 0..MAXJSAMPLE rather than -0.5 .. 0.5.
- * The conversion equations to be implemented are therefore
- *	Y  =  0.29900 * R + 0.58700 * G + 0.11400 * B
- *	Cb = -0.16874 * R - 0.33126 * G + 0.50000 * B  + CENTERJSAMPLE
- *	Cr =  0.50000 * R - 0.41869 * G - 0.08131 * B  + CENTERJSAMPLE
- * (These numbers are derived from TIFF 6.0 section 21, dated 3-June-92.)
- * Note: older versions of the IJG code used a zero offset of MAXJSAMPLE/2,
- * rather than CENTERJSAMPLE, for Cb and Cr.  This gave equal positive and
- * negative swings for Cb/Cr, but meant that grayscale values (Cb=Cr=0)
- * were not represented exactly.  Now we sacrifice exact representation of
- * maximum red and maximum blue in order to get exact grayscales.
- *
- * To avoid floating-point arithmetic, we represent the fractional constants
- * as integers scaled up by 2^16 (about 4 digits precision); we have to divide
- * the products by 2^16, with appropriate rounding, to get the correct answer.
- *
- * For even more speed, we avoid doing any multiplications in the inner loop
- * by precalculating the constants times R,G,B for all possible values.
- * For 8-bit JSAMPLEs this is very reasonable (only 256 entries per table);
- * for 12-bit samples it is still acceptable.  It's not very reasonable for
- * 16-bit samples, but if you want lossless storage you shouldn't be changing
- * colorspace anyway.
- * The CENTERJSAMPLE offsets and the rounding fudge-factor of 0.5 are included
- * in the tables to save adding them separately in the inner loop.
- */
-
-#define SCALEBITS	16	/* speediest right-shift on some machines */
-#define CBCR_OFFSET	((INT32) CENTERJSAMPLE << SCALEBITS)
-#define ONE_HALF	((INT32) 1 << (SCALEBITS-1))
-#define FIX(x)		((INT32) ((x) * (1L<<SCALEBITS) + 0.5))
-
-/* We allocate one big table and divide it up into eight parts, instead of
- * doing eight alloc_small requests.  This lets us use a single table base
- * address, which can be held in a register in the inner loops on many
- * machines (more than can hold all eight addresses, anyway).
- */
-
-#define R_Y_OFF		0			/* offset to R => Y section */
-#define G_Y_OFF		(1*(MAXJSAMPLE+1))	/* offset to G => Y section */
-#define B_Y_OFF		(2*(MAXJSAMPLE+1))	/* etc. */
-#define R_CB_OFF	(3*(MAXJSAMPLE+1))
-#define G_CB_OFF	(4*(MAXJSAMPLE+1))
-#define B_CB_OFF	(5*(MAXJSAMPLE+1))
-#define R_CR_OFF	B_CB_OFF		/* B=>Cb, R=>Cr are the same */
-#define G_CR_OFF	(6*(MAXJSAMPLE+1))
-#define B_CR_OFF	(7*(MAXJSAMPLE+1))
-#define TABLE_SIZE	(8*(MAXJSAMPLE+1))
-
-
-/*
- * Initialize for RGB->YCC colorspace conversion.
- */
-
-METHODDEF(void)
-rgb_ycc_start (j_compress_ptr cinfo)
-{
-  my_cconvert_ptr cconvert = (my_cconvert_ptr) cinfo->cconvert;
-  INT32 * rgb_ycc_tab;
-  INT32 i;
-
-  /* Allocate and fill in the conversion tables. */
-  cconvert->rgb_ycc_tab = rgb_ycc_tab = (INT32 *)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				(TABLE_SIZE * SIZEOF(INT32)));
-
-  for (i = 0; i <= MAXJSAMPLE; i++) {
-    rgb_ycc_tab[i+R_Y_OFF] = FIX(0.29900) * i;
-    rgb_ycc_tab[i+G_Y_OFF] = FIX(0.58700) * i;
-    rgb_ycc_tab[i+B_Y_OFF] = FIX(0.11400) * i     + ONE_HALF;
-    rgb_ycc_tab[i+R_CB_OFF] = (-FIX(0.16874)) * i;
-    rgb_ycc_tab[i+G_CB_OFF] = (-FIX(0.33126)) * i;
-    /* We use a rounding fudge-factor of 0.5-epsilon for Cb and Cr.
-     * This ensures that the maximum output will round to MAXJSAMPLE
-     * not MAXJSAMPLE+1, and thus that we don't have to range-limit.
-     */
-    rgb_ycc_tab[i+B_CB_OFF] = FIX(0.50000) * i    + CBCR_OFFSET + ONE_HALF-1;
-/*  B=>Cb and R=>Cr tables are the same
-    rgb_ycc_tab[i+R_CR_OFF] = FIX(0.50000) * i    + CBCR_OFFSET + ONE_HALF-1;
-*/
-    rgb_ycc_tab[i+G_CR_OFF] = (-FIX(0.41869)) * i;
-    rgb_ycc_tab[i+B_CR_OFF] = (-FIX(0.08131)) * i;
-  }
-}
-
-
-/*
- * Convert some rows of samples to the JPEG colorspace.
- *
- * Note that we change from the application's interleaved-pixel format
- * to our internal noninterleaved, one-plane-per-component format.
- * The input buffer is therefore three times as wide as the output buffer.
- *
- * A starting row offset is provided only for the output buffer.  The caller
- * can easily adjust the passed input_buf value to accommodate any row
- * offset required on that side.
- */
-
-METHODDEF(void)
-rgb_ycc_convert (j_compress_ptr cinfo,
-		 JSAMPARRAY input_buf, JSAMPIMAGE output_buf,
-		 JDIMENSION output_row, int num_rows)
-{
-  my_cconvert_ptr cconvert = (my_cconvert_ptr) cinfo->cconvert;
-  register int r, g, b;
-  register INT32 * ctab = cconvert->rgb_ycc_tab;
-  register JSAMPROW inptr;
-  register JSAMPROW outptr0, outptr1, outptr2;
-  register JDIMENSION col;
-  JDIMENSION num_cols = cinfo->image_width;
-
-  while (--num_rows >= 0) {
-    inptr = *input_buf++;
-    outptr0 = output_buf[0][output_row];
-    outptr1 = output_buf[1][output_row];
-    outptr2 = output_buf[2][output_row];
-    output_row++;
-    for (col = 0; col < num_cols; col++) {
-      r = GETJSAMPLE(inptr[RGB_RED]);
-      g = GETJSAMPLE(inptr[RGB_GREEN]);
-      b = GETJSAMPLE(inptr[RGB_BLUE]);
-      inptr += RGB_PIXELSIZE;
-      /* If the inputs are 0..MAXJSAMPLE, the outputs of these equations
-       * must be too; we do not need an explicit range-limiting operation.
-       * Hence the value being shifted is never negative, and we don't
-       * need the general RIGHT_SHIFT macro.
-       */
-      /* Y */
-      outptr0[col] = (JSAMPLE)
-		((ctab[r+R_Y_OFF] + ctab[g+G_Y_OFF] + ctab[b+B_Y_OFF])
-		 >> SCALEBITS);
-      /* Cb */
-      outptr1[col] = (JSAMPLE)
-		((ctab[r+R_CB_OFF] + ctab[g+G_CB_OFF] + ctab[b+B_CB_OFF])
-		 >> SCALEBITS);
-      /* Cr */
-      outptr2[col] = (JSAMPLE)
-		((ctab[r+R_CR_OFF] + ctab[g+G_CR_OFF] + ctab[b+B_CR_OFF])
-		 >> SCALEBITS);
-    }
-  }
-}
-
-
-/**************** Cases other than RGB -> YCbCr **************/
-
-
-/*
- * Convert some rows of samples to the JPEG colorspace.
- * This version handles RGB->grayscale conversion, which is the same
- * as the RGB->Y portion of RGB->YCbCr.
- * We assume rgb_ycc_start has been called (we only use the Y tables).
- */
-
-METHODDEF(void)
-rgb_gray_convert (j_compress_ptr cinfo,
-		  JSAMPARRAY input_buf, JSAMPIMAGE output_buf,
-		  JDIMENSION output_row, int num_rows)
-{
-  my_cconvert_ptr cconvert = (my_cconvert_ptr) cinfo->cconvert;
-  register int r, g, b;
-  register INT32 * ctab = cconvert->rgb_ycc_tab;
-  register JSAMPROW inptr;
-  register JSAMPROW outptr;
-  register JDIMENSION col;
-  JDIMENSION num_cols = cinfo->image_width;
-
-  while (--num_rows >= 0) {
-    inptr = *input_buf++;
-    outptr = output_buf[0][output_row];
-    output_row++;
-    for (col = 0; col < num_cols; col++) {
-      r = GETJSAMPLE(inptr[RGB_RED]);
-      g = GETJSAMPLE(inptr[RGB_GREEN]);
-      b = GETJSAMPLE(inptr[RGB_BLUE]);
-      inptr += RGB_PIXELSIZE;
-      /* Y */
-      outptr[col] = (JSAMPLE)
-		((ctab[r+R_Y_OFF] + ctab[g+G_Y_OFF] + ctab[b+B_Y_OFF])
-		 >> SCALEBITS);
-    }
-  }
-}
-
-
-/*
- * Convert some rows of samples to the JPEG colorspace.
- * This version handles Adobe-style CMYK->YCCK conversion,
- * where we convert R=1-C, G=1-M, and B=1-Y to YCbCr using the same
- * conversion as above, while passing K (black) unchanged.
- * We assume rgb_ycc_start has been called.
- */
-
-METHODDEF(void)
-cmyk_ycck_convert (j_compress_ptr cinfo,
-		   JSAMPARRAY input_buf, JSAMPIMAGE output_buf,
-		   JDIMENSION output_row, int num_rows)
-{
-  my_cconvert_ptr cconvert = (my_cconvert_ptr) cinfo->cconvert;
-  register int r, g, b;
-  register INT32 * ctab = cconvert->rgb_ycc_tab;
-  register JSAMPROW inptr;
-  register JSAMPROW outptr0, outptr1, outptr2, outptr3;
-  register JDIMENSION col;
-  JDIMENSION num_cols = cinfo->image_width;
-
-  while (--num_rows >= 0) {
-    inptr = *input_buf++;
-    outptr0 = output_buf[0][output_row];
-    outptr1 = output_buf[1][output_row];
-    outptr2 = output_buf[2][output_row];
-    outptr3 = output_buf[3][output_row];
-    output_row++;
-    for (col = 0; col < num_cols; col++) {
-      r = MAXJSAMPLE - GETJSAMPLE(inptr[0]);
-      g = MAXJSAMPLE - GETJSAMPLE(inptr[1]);
-      b = MAXJSAMPLE - GETJSAMPLE(inptr[2]);
-      /* K passes through as-is */
-      outptr3[col] = inptr[3];	/* don't need GETJSAMPLE here */
-      inptr += 4;
-      /* If the inputs are 0..MAXJSAMPLE, the outputs of these equations
-       * must be too; we do not need an explicit range-limiting operation.
-       * Hence the value being shifted is never negative, and we don't
-       * need the general RIGHT_SHIFT macro.
-       */
-      /* Y */
-      outptr0[col] = (JSAMPLE)
-		((ctab[r+R_Y_OFF] + ctab[g+G_Y_OFF] + ctab[b+B_Y_OFF])
-		 >> SCALEBITS);
-      /* Cb */
-      outptr1[col] = (JSAMPLE)
-		((ctab[r+R_CB_OFF] + ctab[g+G_CB_OFF] + ctab[b+B_CB_OFF])
-		 >> SCALEBITS);
-      /* Cr */
-      outptr2[col] = (JSAMPLE)
-		((ctab[r+R_CR_OFF] + ctab[g+G_CR_OFF] + ctab[b+B_CR_OFF])
-		 >> SCALEBITS);
-    }
-  }
-}
-
-
-/*
- * Convert some rows of samples to the JPEG colorspace.
- * This version handles grayscale output with no conversion.
- * The source can be either plain grayscale or YCbCr (since Y == gray).
- */
-
-METHODDEF(void)
-grayscale_convert (j_compress_ptr cinfo,
-		   JSAMPARRAY input_buf, JSAMPIMAGE output_buf,
-		   JDIMENSION output_row, int num_rows)
-{
-  register JSAMPROW inptr;
-  register JSAMPROW outptr;
-  register JDIMENSION col;
-  JDIMENSION num_cols = cinfo->image_width;
-  int instride = cinfo->input_components;
-
-  while (--num_rows >= 0) {
-    inptr = *input_buf++;
-    outptr = output_buf[0][output_row];
-    output_row++;
-    for (col = 0; col < num_cols; col++) {
-      outptr[col] = inptr[0];	/* don't need GETJSAMPLE() here */
-      inptr += instride;
-    }
-  }
-}
-
-#ifdef ENABLE_ANDROID_NULL_CONVERT
-
-typedef unsigned long UINT32;
-
-#define B0(n)   ((n) & 0xFF)
-#define B1(n)   (((n) >> 8) & 0xFF)
-#define B2(n)   (((n) >> 16) & 0xFF)
-#define B3(n)   ((n) >> 24)
-
-#define PACK(a, b, c, d)    ((a) | ((b) << 8) | ((c) << 16) | ((d) << 24))
-
-static int ptr_is_quad(const void* p)
-{
-    return (((const char*)p - (const char*)0) & 3) == 0;
-}
-
-static void copyquads(const UINT32 in[], UINT32 out0[], UINT32 out1[], UINT32 out2[], int col4)
-{
-    do {
-        UINT32 src0 = *in++;
-        UINT32 src1 = *in++;
-        UINT32 src2 = *in++;
-        // LEndian
-        *out0++ = PACK(B0(src0), B3(src0), B2(src1), B1(src2));
-        *out1++ = PACK(B1(src0), B0(src1), B3(src1), B2(src2));
-        *out2++ = PACK(B2(src0), B1(src1), B0(src2), B3(src2));
-    } while (--col4 != 0);
-}
-
-#endif
-
-/*
- * Convert some rows of samples to the JPEG colorspace.
- * This version handles multi-component colorspaces without conversion.
- * We assume input_components == num_components.
- */
-
-METHODDEF(void)
-null_convert (j_compress_ptr cinfo,
-	      JSAMPARRAY input_buf, JSAMPIMAGE output_buf,
-	      JDIMENSION output_row, int num_rows)
-{
-  register JSAMPROW inptr;
-  register JSAMPROW outptr;
-  register JDIMENSION col;
-  register int ci;
-  int nc = cinfo->num_components;
-  JDIMENSION num_cols = cinfo->image_width;
-
-#ifdef ENABLE_ANDROID_NULL_CONVERT
-    if (1 == num_rows && 3 == nc && num_cols > 0) {
-        JSAMPROW inptr = *input_buf;
-        JSAMPROW outptr0 = output_buf[0][output_row];
-        JSAMPROW outptr1 = output_buf[1][output_row];
-        JSAMPROW outptr2 = output_buf[2][output_row];
-        
-        int col = num_cols;
-        int col4 = col >> 2;
-        if (col4 > 0 && ptr_is_quad(inptr) && ptr_is_quad(outptr0) &&
-                        ptr_is_quad(outptr1) && ptr_is_quad(outptr2)) {
-            
-            const UINT32* in = (const UINT32*)inptr;
-            UINT32* out0 = (UINT32*)outptr0;
-            UINT32* out1 = (UINT32*)outptr1;
-            UINT32* out2 = (UINT32*)outptr2;
-            copyquads(in, out0, out1, out2, col4);
-            col &= 3;
-            if (0 == col)
-                return;
-            col4 <<= 2;
-            inptr += col4 * 3;  /* we read this 3 times per in copyquads */
-            outptr0 += col4;
-            outptr1 += col4;
-            outptr2 += col4;
-            /* fall through to while-loop */
-        }
-        do {
-            *outptr0++ = *inptr++;
-            *outptr1++ = *inptr++;
-            *outptr2++ = *inptr++;
-        } while (--col != 0);
-        return;
-    }
-SLOW:
-#endif
-  while (--num_rows >= 0) {
-    /* It seems fastest to make a separate pass for each component. */
-    for (ci = 0; ci < nc; ci++) {
-      inptr = *input_buf;
-      outptr = output_buf[ci][output_row];
-      for (col = 0; col < num_cols; col++) {
-	outptr[col] = inptr[ci]; /* don't need GETJSAMPLE() here */
-	inptr += nc;
-      }
-    }
-    input_buf++;
-    output_row++;
-  }
-}
-
-
-/*
- * Empty method for start_pass.
- */
-
-METHODDEF(void)
-null_method (j_compress_ptr cinfo)
-{
-  /* no work needed */
-}
-
-
-/*
- * Module initialization routine for input colorspace conversion.
- */
-
-GLOBAL(void)
-jinit_color_converter (j_compress_ptr cinfo)
-{
-  my_cconvert_ptr cconvert;
-
-  cconvert = (my_cconvert_ptr)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				SIZEOF(my_color_converter));
-  cinfo->cconvert = (struct jpeg_color_converter *) cconvert;
-  /* set start_pass to null method until we find out differently */
-  cconvert->pub.start_pass = null_method;
-
-  /* Make sure input_components agrees with in_color_space */
-  switch (cinfo->in_color_space) {
-  case JCS_GRAYSCALE:
-    if (cinfo->input_components != 1)
-      ERREXIT(cinfo, JERR_BAD_IN_COLORSPACE);
-    break;
-
-  case JCS_RGB:
-#if RGB_PIXELSIZE != 3
-    if (cinfo->input_components != RGB_PIXELSIZE)
-      ERREXIT(cinfo, JERR_BAD_IN_COLORSPACE);
-    break;
-#endif /* else share code with YCbCr */
-
-  case JCS_YCbCr:
-    if (cinfo->input_components != 3)
-      ERREXIT(cinfo, JERR_BAD_IN_COLORSPACE);
-    break;
-
-  case JCS_CMYK:
-  case JCS_YCCK:
-    if (cinfo->input_components != 4)
-      ERREXIT(cinfo, JERR_BAD_IN_COLORSPACE);
-    break;
-
-  default:			/* JCS_UNKNOWN can be anything */
-    if (cinfo->input_components < 1)
-      ERREXIT(cinfo, JERR_BAD_IN_COLORSPACE);
-    break;
-  }
-
-  /* Check num_components, set conversion method based on requested space */
-  switch (cinfo->jpeg_color_space) {
-  case JCS_GRAYSCALE:
-    if (cinfo->num_components != 1)
-      ERREXIT(cinfo, JERR_BAD_J_COLORSPACE);
-    if (cinfo->in_color_space == JCS_GRAYSCALE)
-      cconvert->pub.color_convert = grayscale_convert;
-    else if (cinfo->in_color_space == JCS_RGB) {
-      cconvert->pub.start_pass = rgb_ycc_start;
-      cconvert->pub.color_convert = rgb_gray_convert;
-    } else if (cinfo->in_color_space == JCS_YCbCr)
-      cconvert->pub.color_convert = grayscale_convert;
-    else
-      ERREXIT(cinfo, JERR_CONVERSION_NOTIMPL);
-    break;
-
-  case JCS_RGB:
-    if (cinfo->num_components != 3)
-      ERREXIT(cinfo, JERR_BAD_J_COLORSPACE);
-    if (cinfo->in_color_space == JCS_RGB && RGB_PIXELSIZE == 3)
-      cconvert->pub.color_convert = null_convert;
-    else
-      ERREXIT(cinfo, JERR_CONVERSION_NOTIMPL);
-    break;
-
-  case JCS_YCbCr:
-    if (cinfo->num_components != 3)
-      ERREXIT(cinfo, JERR_BAD_J_COLORSPACE);
-    if (cinfo->in_color_space == JCS_RGB) {
-      cconvert->pub.start_pass = rgb_ycc_start;
-      cconvert->pub.color_convert = rgb_ycc_convert;
-    } else if (cinfo->in_color_space == JCS_YCbCr)
-      cconvert->pub.color_convert = null_convert;
-    else
-      ERREXIT(cinfo, JERR_CONVERSION_NOTIMPL);
-    break;
-
-  case JCS_CMYK:
-    if (cinfo->num_components != 4)
-      ERREXIT(cinfo, JERR_BAD_J_COLORSPACE);
-    if (cinfo->in_color_space == JCS_CMYK)
-      cconvert->pub.color_convert = null_convert;
-    else
-      ERREXIT(cinfo, JERR_CONVERSION_NOTIMPL);
-    break;
-
-  case JCS_YCCK:
-    if (cinfo->num_components != 4)
-      ERREXIT(cinfo, JERR_BAD_J_COLORSPACE);
-    if (cinfo->in_color_space == JCS_CMYK) {
-      cconvert->pub.start_pass = rgb_ycc_start;
-      cconvert->pub.color_convert = cmyk_ycck_convert;
-    } else if (cinfo->in_color_space == JCS_YCCK)
-      cconvert->pub.color_convert = null_convert;
-    else
-      ERREXIT(cinfo, JERR_CONVERSION_NOTIMPL);
-    break;
-
-  default:			/* allow null conversion of JCS_UNKNOWN */
-    if (cinfo->jpeg_color_space != cinfo->in_color_space ||
-	cinfo->num_components != cinfo->input_components)
-      ERREXIT(cinfo, JERR_CONVERSION_NOTIMPL);
-    cconvert->pub.color_convert = null_convert;
-    break;
-  }
-}
@@ -1,387 +0,0 @@
-/*
- * jcdctmgr.c
- *
- * Copyright (C) 1994-1996, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains the forward-DCT management logic.
- * This code selects a particular DCT implementation to be used,
- * and it performs related housekeeping chores including coefficient
- * quantization.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-#include "jdct.h"		/* Private declarations for DCT subsystem */
-
-
-/* Private subobject for this module */
-
-typedef struct {
-  struct jpeg_forward_dct pub;	/* public fields */
-
-  /* Pointer to the DCT routine actually in use */
-  forward_DCT_method_ptr do_dct;
-
-  /* The actual post-DCT divisors --- not identical to the quant table
-   * entries, because of scaling (especially for an unnormalized DCT).
-   * Each table is given in normal array order.
-   */
-  DCTELEM * divisors[NUM_QUANT_TBLS];
-
-#ifdef DCT_FLOAT_SUPPORTED
-  /* Same as above for the floating-point case. */
-  float_DCT_method_ptr do_float_dct;
-  FAST_FLOAT * float_divisors[NUM_QUANT_TBLS];
-#endif
-} my_fdct_controller;
-
-typedef my_fdct_controller * my_fdct_ptr;
-
-
-/*
- * Initialize for a processing pass.
- * Verify that all referenced Q-tables are present, and set up
- * the divisor table for each one.
- * In the current implementation, DCT of all components is done during
- * the first pass, even if only some components will be output in the
- * first scan.  Hence all components should be examined here.
- */
-
-METHODDEF(void)
-start_pass_fdctmgr (j_compress_ptr cinfo)
-{
-  my_fdct_ptr fdct = (my_fdct_ptr) cinfo->fdct;
-  int ci, qtblno, i;
-  jpeg_component_info *compptr;
-  JQUANT_TBL * qtbl;
-  DCTELEM * dtbl;
-
-  for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
-       ci++, compptr++) {
-    qtblno = compptr->quant_tbl_no;
-    /* Make sure specified quantization table is present */
-    if (qtblno < 0 || qtblno >= NUM_QUANT_TBLS ||
-	cinfo->quant_tbl_ptrs[qtblno] == NULL)
-      ERREXIT1(cinfo, JERR_NO_QUANT_TABLE, qtblno);
-    qtbl = cinfo->quant_tbl_ptrs[qtblno];
-    /* Compute divisors for this quant table */
-    /* We may do this more than once for same table, but it's not a big deal */
-    switch (cinfo->dct_method) {
-#ifdef DCT_ISLOW_SUPPORTED
-    case JDCT_ISLOW:
-      /* For LL&M IDCT method, divisors are equal to raw quantization
-       * coefficients multiplied by 8 (to counteract scaling).
-       */
-      if (fdct->divisors[qtblno] == NULL) {
-	fdct->divisors[qtblno] = (DCTELEM *)
-	  (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				      DCTSIZE2 * SIZEOF(DCTELEM));
-      }
-      dtbl = fdct->divisors[qtblno];
-      for (i = 0; i < DCTSIZE2; i++) {
-	dtbl[i] = ((DCTELEM) qtbl->quantval[i]) << 3;
-      }
-      break;
-#endif
-#ifdef DCT_IFAST_SUPPORTED
-    case JDCT_IFAST:
-      {
-	/* For AA&N IDCT method, divisors are equal to quantization
-	 * coefficients scaled by scalefactor[row]*scalefactor[col], where
-	 *   scalefactor[0] = 1
-	 *   scalefactor[k] = cos(k*PI/16) * sqrt(2)    for k=1..7
-	 * We apply a further scale factor of 8.
-	 */
-#define CONST_BITS 14
-	static const INT16 aanscales[DCTSIZE2] = {
-	  /* precomputed values scaled up by 14 bits */
-	  16384, 22725, 21407, 19266, 16384, 12873,  8867,  4520,
-	  22725, 31521, 29692, 26722, 22725, 17855, 12299,  6270,
-	  21407, 29692, 27969, 25172, 21407, 16819, 11585,  5906,
-	  19266, 26722, 25172, 22654, 19266, 15137, 10426,  5315,
-	  16384, 22725, 21407, 19266, 16384, 12873,  8867,  4520,
-	  12873, 17855, 16819, 15137, 12873, 10114,  6967,  3552,
-	   8867, 12299, 11585, 10426,  8867,  6967,  4799,  2446,
-	   4520,  6270,  5906,  5315,  4520,  3552,  2446,  1247
-	};
-	SHIFT_TEMPS
-
-	if (fdct->divisors[qtblno] == NULL) {
-	  fdct->divisors[qtblno] = (DCTELEM *)
-	    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-					DCTSIZE2 * SIZEOF(DCTELEM));
-	}
-	dtbl = fdct->divisors[qtblno];
-	for (i = 0; i < DCTSIZE2; i++) {
-	  dtbl[i] = (DCTELEM)
-	    DESCALE(MULTIPLY16V16((INT32) qtbl->quantval[i],
-				  (INT32) aanscales[i]),
-		    CONST_BITS-3);
-	}
-      }
-      break;
-#endif
-#ifdef DCT_FLOAT_SUPPORTED
-    case JDCT_FLOAT:
-      {
-	/* For float AA&N IDCT method, divisors are equal to quantization
-	 * coefficients scaled by scalefactor[row]*scalefactor[col], where
-	 *   scalefactor[0] = 1
-	 *   scalefactor[k] = cos(k*PI/16) * sqrt(2)    for k=1..7
-	 * We apply a further scale factor of 8.
-	 * What's actually stored is 1/divisor so that the inner loop can
-	 * use a multiplication rather than a division.
-	 */
-	FAST_FLOAT * fdtbl;
-	int row, col;
-	static const double aanscalefactor[DCTSIZE] = {
-	  1.0, 1.387039845, 1.306562965, 1.175875602,
-	  1.0, 0.785694958, 0.541196100, 0.275899379
-	};
-
-	if (fdct->float_divisors[qtblno] == NULL) {
-	  fdct->float_divisors[qtblno] = (FAST_FLOAT *)
-	    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-					DCTSIZE2 * SIZEOF(FAST_FLOAT));
-	}
-	fdtbl = fdct->float_divisors[qtblno];
-	i = 0;
-	for (row = 0; row < DCTSIZE; row++) {
-	  for (col = 0; col < DCTSIZE; col++) {
-	    fdtbl[i] = (FAST_FLOAT)
-	      (1.0 / (((double) qtbl->quantval[i] *
-		       aanscalefactor[row] * aanscalefactor[col] * 8.0)));
-	    i++;
-	  }
-	}
-      }
-      break;
-#endif
-    default:
-      ERREXIT(cinfo, JERR_NOT_COMPILED);
-      break;
-    }
-  }
-}
-
-
-/*
- * Perform forward DCT on one or more blocks of a component.
- *
- * The input samples are taken from the sample_data[] array starting at
- * position start_row/start_col, and moving to the right for any additional
- * blocks. The quantized coefficients are returned in coef_blocks[].
- */
-
-METHODDEF(void)
-forward_DCT (j_compress_ptr cinfo, jpeg_component_info * compptr,
-	     JSAMPARRAY sample_data, JBLOCKROW coef_blocks,
-	     JDIMENSION start_row, JDIMENSION start_col,
-	     JDIMENSION num_blocks)
-/* This version is used for integer DCT implementations. */
-{
-  /* This routine is heavily used, so it's worth coding it tightly. */
-  my_fdct_ptr fdct = (my_fdct_ptr) cinfo->fdct;
-  forward_DCT_method_ptr do_dct = fdct->do_dct;
-  DCTELEM * divisors = fdct->divisors[compptr->quant_tbl_no];
-  DCTELEM workspace[DCTSIZE2];	/* work area for FDCT subroutine */
-  JDIMENSION bi;
-
-  sample_data += start_row;	/* fold in the vertical offset once */
-
-  for (bi = 0; bi < num_blocks; bi++, start_col += DCTSIZE) {
-    /* Load data into workspace, applying unsigned->signed conversion */
-    { register DCTELEM *workspaceptr;
-      register JSAMPROW elemptr;
-      register int elemr;
-
-      workspaceptr = workspace;
-      for (elemr = 0; elemr < DCTSIZE; elemr++) {
-	elemptr = sample_data[elemr] + start_col;
-#if DCTSIZE == 8		/* unroll the inner loop */
-	*workspaceptr++ = GETJSAMPLE(*elemptr++) - CENTERJSAMPLE;
-	*workspaceptr++ = GETJSAMPLE(*elemptr++) - CENTERJSAMPLE;
-	*workspaceptr++ = GETJSAMPLE(*elemptr++) - CENTERJSAMPLE;
-	*workspaceptr++ = GETJSAMPLE(*elemptr++) - CENTERJSAMPLE;
-	*workspaceptr++ = GETJSAMPLE(*elemptr++) - CENTERJSAMPLE;
-	*workspaceptr++ = GETJSAMPLE(*elemptr++) - CENTERJSAMPLE;
-	*workspaceptr++ = GETJSAMPLE(*elemptr++) - CENTERJSAMPLE;
-	*workspaceptr++ = GETJSAMPLE(*elemptr++) - CENTERJSAMPLE;
-#else
-	{ register int elemc;
-	  for (elemc = DCTSIZE; elemc > 0; elemc--) {
-	    *workspaceptr++ = GETJSAMPLE(*elemptr++) - CENTERJSAMPLE;
-	  }
-	}
-#endif
-      }
-    }
-
-    /* Perform the DCT */
-    (*do_dct) (workspace);
-
-    /* Quantize/descale the coefficients, and store into coef_blocks[] */
-    { register DCTELEM temp, qval;
-      register int i;
-      register JCOEFPTR output_ptr = coef_blocks[bi];
-
-      for (i = 0; i < DCTSIZE2; i++) {
-	qval = divisors[i];
-	temp = workspace[i];
-	/* Divide the coefficient value by qval, ensuring proper rounding.
-	 * Since C does not specify the direction of rounding for negative
-	 * quotients, we have to force the dividend positive for portability.
-	 *
-	 * In most files, at least half of the output values will be zero
-	 * (at default quantization settings, more like three-quarters...)
-	 * so we should ensure that this case is fast.  On many machines,
-	 * a comparison is enough cheaper than a divide to make a special test
-	 * a win.  Since both inputs will be nonnegative, we need only test
-	 * for a < b to discover whether a/b is 0.
-	 * If your machine's division is fast enough, define FAST_DIVIDE.
-	 */
-#ifdef FAST_DIVIDE
-#define DIVIDE_BY(a,b)	a /= b
-#else
-#define DIVIDE_BY(a,b)	if (a >= b) a /= b; else a = 0
-#endif
-	if (temp < 0) {
-	  temp = -temp;
-	  temp += qval>>1;	/* for rounding */
-	  DIVIDE_BY(temp, qval);
-	  temp = -temp;
-	} else {
-	  temp += qval>>1;	/* for rounding */
-	  DIVIDE_BY(temp, qval);
-	}
-	output_ptr[i] = (JCOEF) temp;
-      }
-    }
-  }
-}
-
-
-#ifdef DCT_FLOAT_SUPPORTED
-
-METHODDEF(void)
-forward_DCT_float (j_compress_ptr cinfo, jpeg_component_info * compptr,
-		   JSAMPARRAY sample_data, JBLOCKROW coef_blocks,
-		   JDIMENSION start_row, JDIMENSION start_col,
-		   JDIMENSION num_blocks)
-/* This version is used for floating-point DCT implementations. */
-{
-  /* This routine is heavily used, so it's worth coding it tightly. */
-  my_fdct_ptr fdct = (my_fdct_ptr) cinfo->fdct;
-  float_DCT_method_ptr do_dct = fdct->do_float_dct;
-  FAST_FLOAT * divisors = fdct->float_divisors[compptr->quant_tbl_no];
-  FAST_FLOAT workspace[DCTSIZE2]; /* work area for FDCT subroutine */
-  JDIMENSION bi;
-
-  sample_data += start_row;	/* fold in the vertical offset once */
-
-  for (bi = 0; bi < num_blocks; bi++, start_col += DCTSIZE) {
-    /* Load data into workspace, applying unsigned->signed conversion */
-    { register FAST_FLOAT *workspaceptr;
-      register JSAMPROW elemptr;
-      register int elemr;
-
-      workspaceptr = workspace;
-      for (elemr = 0; elemr < DCTSIZE; elemr++) {
-	elemptr = sample_data[elemr] + start_col;
-#if DCTSIZE == 8		/* unroll the inner loop */
-	*workspaceptr++ = (FAST_FLOAT)(GETJSAMPLE(*elemptr++) - CENTERJSAMPLE);
-	*workspaceptr++ = (FAST_FLOAT)(GETJSAMPLE(*elemptr++) - CENTERJSAMPLE);
-	*workspaceptr++ = (FAST_FLOAT)(GETJSAMPLE(*elemptr++) - CENTERJSAMPLE);
-	*workspaceptr++ = (FAST_FLOAT)(GETJSAMPLE(*elemptr++) - CENTERJSAMPLE);
-	*workspaceptr++ = (FAST_FLOAT)(GETJSAMPLE(*elemptr++) - CENTERJSAMPLE);
-	*workspaceptr++ = (FAST_FLOAT)(GETJSAMPLE(*elemptr++) - CENTERJSAMPLE);
-	*workspaceptr++ = (FAST_FLOAT)(GETJSAMPLE(*elemptr++) - CENTERJSAMPLE);
-	*workspaceptr++ = (FAST_FLOAT)(GETJSAMPLE(*elemptr++) - CENTERJSAMPLE);
-#else
-	{ register int elemc;
-	  for (elemc = DCTSIZE; elemc > 0; elemc--) {
-	    *workspaceptr++ = (FAST_FLOAT)
-	      (GETJSAMPLE(*elemptr++) - CENTERJSAMPLE);
-	  }
-	}
-#endif
-      }
-    }
-
-    /* Perform the DCT */
-    (*do_dct) (workspace);
-
-    /* Quantize/descale the coefficients, and store into coef_blocks[] */
-    { register FAST_FLOAT temp;
-      register int i;
-      register JCOEFPTR output_ptr = coef_blocks[bi];
-
-      for (i = 0; i < DCTSIZE2; i++) {
-	/* Apply the quantization and scaling factor */
-	temp = workspace[i] * divisors[i];
-	/* Round to nearest integer.
-	 * Since C does not specify the direction of rounding for negative
-	 * quotients, we have to force the dividend positive for portability.
-	 * The maximum coefficient size is +-16K (for 12-bit data), so this
-	 * code should work for either 16-bit or 32-bit ints.
-	 */
-	output_ptr[i] = (JCOEF) ((int) (temp + (FAST_FLOAT) 16384.5) - 16384);
-      }
-    }
-  }
-}
-
-#endif /* DCT_FLOAT_SUPPORTED */
-
-
-/*
- * Initialize FDCT manager.
- */
-
-GLOBAL(void)
-jinit_forward_dct (j_compress_ptr cinfo)
-{
-  my_fdct_ptr fdct;
-  int i;
-
-  fdct = (my_fdct_ptr)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				SIZEOF(my_fdct_controller));
-  cinfo->fdct = (struct jpeg_forward_dct *) fdct;
-  fdct->pub.start_pass = start_pass_fdctmgr;
-
-  switch (cinfo->dct_method) {
-#ifdef DCT_ISLOW_SUPPORTED
-  case JDCT_ISLOW:
-    fdct->pub.forward_DCT = forward_DCT;
-    fdct->do_dct = jpeg_fdct_islow;
-    break;
-#endif
-#ifdef DCT_IFAST_SUPPORTED
-  case JDCT_IFAST:
-    fdct->pub.forward_DCT = forward_DCT;
-    fdct->do_dct = jpeg_fdct_ifast;
-    break;
-#endif
-#ifdef DCT_FLOAT_SUPPORTED
-  case JDCT_FLOAT:
-    fdct->pub.forward_DCT = forward_DCT_float;
-    fdct->do_float_dct = jpeg_fdct_float;
-    break;
-#endif
-  default:
-    ERREXIT(cinfo, JERR_NOT_COMPILED);
-    break;
-  }
-
-  /* Mark divisor tables unallocated */
-  for (i = 0; i < NUM_QUANT_TBLS; i++) {
-    fdct->divisors[i] = NULL;
-#ifdef DCT_FLOAT_SUPPORTED
-    fdct->float_divisors[i] = NULL;
-#endif
-  }
-}
@@ -1,909 +0,0 @@
-/*
- * jchuff.c
- *
- * Copyright (C) 1991-1997, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains Huffman entropy encoding routines.
- *
- * Much of the complexity here has to do with supporting output suspension.
- * If the data destination module demands suspension, we want to be able to
- * back up to the start of the current MCU.  To do this, we copy state
- * variables into local working storage, and update them back to the
- * permanent JPEG objects only upon successful completion of an MCU.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-#include "jchuff.h"		/* Declarations shared with jcphuff.c */
-
-
-/* Expanded entropy encoder object for Huffman encoding.
- *
- * The savable_state subrecord contains fields that change within an MCU,
- * but must not be updated permanently until we complete the MCU.
- */
-
-typedef struct {
-  INT32 put_buffer;		/* current bit-accumulation buffer */
-  int put_bits;			/* # of bits now in it */
-  int last_dc_val[MAX_COMPS_IN_SCAN]; /* last DC coef for each component */
-} savable_state;
-
-/* This macro is to work around compilers with missing or broken
- * structure assignment.  You'll need to fix this code if you have
- * such a compiler and you change MAX_COMPS_IN_SCAN.
- */
-
-#ifndef NO_STRUCT_ASSIGN
-#define ASSIGN_STATE(dest,src)  ((dest) = (src))
-#else
-#if MAX_COMPS_IN_SCAN == 4
-#define ASSIGN_STATE(dest,src)  \
-	((dest).put_buffer = (src).put_buffer, \
-	 (dest).put_bits = (src).put_bits, \
-	 (dest).last_dc_val[0] = (src).last_dc_val[0], \
-	 (dest).last_dc_val[1] = (src).last_dc_val[1], \
-	 (dest).last_dc_val[2] = (src).last_dc_val[2], \
-	 (dest).last_dc_val[3] = (src).last_dc_val[3])
-#endif
-#endif
-
-
-typedef struct {
-  struct jpeg_entropy_encoder pub; /* public fields */
-
-  savable_state saved;		/* Bit buffer & DC state at start of MCU */
-
-  /* These fields are NOT loaded into local working state. */
-  unsigned int restarts_to_go;	/* MCUs left in this restart interval */
-  int next_restart_num;		/* next restart number to write (0-7) */
-
-  /* Pointers to derived tables (these workspaces have image lifespan) */
-  c_derived_tbl * dc_derived_tbls[NUM_HUFF_TBLS];
-  c_derived_tbl * ac_derived_tbls[NUM_HUFF_TBLS];
-
-#ifdef ENTROPY_OPT_SUPPORTED	/* Statistics tables for optimization */
-  long * dc_count_ptrs[NUM_HUFF_TBLS];
-  long * ac_count_ptrs[NUM_HUFF_TBLS];
-#endif
-} huff_entropy_encoder;
-
-typedef huff_entropy_encoder * huff_entropy_ptr;
-
-/* Working state while writing an MCU.
- * This struct contains all the fields that are needed by subroutines.
- */
-
-typedef struct {
-  JOCTET * next_output_byte;	/* => next byte to write in buffer */
-  size_t free_in_buffer;	/* # of byte spaces remaining in buffer */
-  savable_state cur;		/* Current bit buffer & DC state */
-  j_compress_ptr cinfo;		/* dump_buffer needs access to this */
-} working_state;
-
-
-/* Forward declarations */
-METHODDEF(boolean) encode_mcu_huff JPP((j_compress_ptr cinfo,
-					JBLOCKROW *MCU_data));
-METHODDEF(void) finish_pass_huff JPP((j_compress_ptr cinfo));
-#ifdef ENTROPY_OPT_SUPPORTED
-METHODDEF(boolean) encode_mcu_gather JPP((j_compress_ptr cinfo,
-					  JBLOCKROW *MCU_data));
-METHODDEF(void) finish_pass_gather JPP((j_compress_ptr cinfo));
-#endif
-
-
-/*
- * Initialize for a Huffman-compressed scan.
- * If gather_statistics is TRUE, we do not output anything during the scan,
- * just count the Huffman symbols used and generate Huffman code tables.
- */
-
-METHODDEF(void)
-start_pass_huff (j_compress_ptr cinfo, boolean gather_statistics)
-{
-  huff_entropy_ptr entropy = (huff_entropy_ptr) cinfo->entropy;
-  int ci, dctbl, actbl;
-  jpeg_component_info * compptr;
-
-  if (gather_statistics) {
-#ifdef ENTROPY_OPT_SUPPORTED
-    entropy->pub.encode_mcu = encode_mcu_gather;
-    entropy->pub.finish_pass = finish_pass_gather;
-#else
-    ERREXIT(cinfo, JERR_NOT_COMPILED);
-#endif
-  } else {
-    entropy->pub.encode_mcu = encode_mcu_huff;
-    entropy->pub.finish_pass = finish_pass_huff;
-  }
-
-  for (ci = 0; ci < cinfo->comps_in_scan; ci++) {
-    compptr = cinfo->cur_comp_info[ci];
-    dctbl = compptr->dc_tbl_no;
-    actbl = compptr->ac_tbl_no;
-    if (gather_statistics) {
-#ifdef ENTROPY_OPT_SUPPORTED
-      /* Check for invalid table indexes */
-      /* (make_c_derived_tbl does this in the other path) */
-      if (dctbl < 0 || dctbl >= NUM_HUFF_TBLS)
-	ERREXIT1(cinfo, JERR_NO_HUFF_TABLE, dctbl);
-      if (actbl < 0 || actbl >= NUM_HUFF_TBLS)
-	ERREXIT1(cinfo, JERR_NO_HUFF_TABLE, actbl);
-      /* Allocate and zero the statistics tables */
-      /* Note that jpeg_gen_optimal_table expects 257 entries in each table! */
-      if (entropy->dc_count_ptrs[dctbl] == NULL)
-	entropy->dc_count_ptrs[dctbl] = (long *)
-	  (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				      257 * SIZEOF(long));
-      MEMZERO(entropy->dc_count_ptrs[dctbl], 257 * SIZEOF(long));
-      if (entropy->ac_count_ptrs[actbl] == NULL)
-	entropy->ac_count_ptrs[actbl] = (long *)
-	  (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				      257 * SIZEOF(long));
-      MEMZERO(entropy->ac_count_ptrs[actbl], 257 * SIZEOF(long));
-#endif
-    } else {
-      /* Compute derived values for Huffman tables */
-      /* We may do this more than once for a table, but it's not expensive */
-      jpeg_make_c_derived_tbl(cinfo, TRUE, dctbl,
-			      & entropy->dc_derived_tbls[dctbl]);
-      jpeg_make_c_derived_tbl(cinfo, FALSE, actbl,
-			      & entropy->ac_derived_tbls[actbl]);
-    }
-    /* Initialize DC predictions to 0 */
-    entropy->saved.last_dc_val[ci] = 0;
-  }
-
-  /* Initialize bit buffer to empty */
-  entropy->saved.put_buffer = 0;
-  entropy->saved.put_bits = 0;
-
-  /* Initialize restart stuff */
-  entropy->restarts_to_go = cinfo->restart_interval;
-  entropy->next_restart_num = 0;
-}
-
-
-/*
- * Compute the derived values for a Huffman table.
- * This routine also performs some validation checks on the table.
- *
- * Note this is also used by jcphuff.c.
- */
-
-GLOBAL(void)
-jpeg_make_c_derived_tbl (j_compress_ptr cinfo, boolean isDC, int tblno,
-			 c_derived_tbl ** pdtbl)
-{
-  JHUFF_TBL *htbl;
-  c_derived_tbl *dtbl;
-  int p, i, l, lastp, si, maxsymbol;
-  char huffsize[257];
-  unsigned int huffcode[257];
-  unsigned int code;
-
-  /* Note that huffsize[] and huffcode[] are filled in code-length order,
-   * paralleling the order of the symbols themselves in htbl->huffval[].
-   */
-
-  /* Find the input Huffman table */
-  if (tblno < 0 || tblno >= NUM_HUFF_TBLS)
-    ERREXIT1(cinfo, JERR_NO_HUFF_TABLE, tblno);
-  htbl =
-    isDC ? cinfo->dc_huff_tbl_ptrs[tblno] : cinfo->ac_huff_tbl_ptrs[tblno];
-  if (htbl == NULL)
-    ERREXIT1(cinfo, JERR_NO_HUFF_TABLE, tblno);
-
-  /* Allocate a workspace if we haven't already done so. */
-  if (*pdtbl == NULL)
-    *pdtbl = (c_derived_tbl *)
-      (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				  SIZEOF(c_derived_tbl));
-  dtbl = *pdtbl;
-  
-  /* Figure C.1: make table of Huffman code length for each symbol */
-
-  p = 0;
-  for (l = 1; l <= 16; l++) {
-    i = (int) htbl->bits[l];
-    if (i < 0 || p + i > 256)	/* protect against table overrun */
-      ERREXIT(cinfo, JERR_BAD_HUFF_TABLE);
-    while (i--)
-      huffsize[p++] = (char) l;
-  }
-  huffsize[p] = 0;
-  lastp = p;
-  
-  /* Figure C.2: generate the codes themselves */
-  /* We also validate that the counts represent a legal Huffman code tree. */
-
-  code = 0;
-  si = huffsize[0];
-  p = 0;
-  while (huffsize[p]) {
-    while (((int) huffsize[p]) == si) {
-      huffcode[p++] = code;
-      code++;
-    }
-    /* code is now 1 more than the last code used for codelength si; but
-     * it must still fit in si bits, since no code is allowed to be all ones.
-     */
-    if (((INT32) code) >= (((INT32) 1) << si))
-      ERREXIT(cinfo, JERR_BAD_HUFF_TABLE);
-    code <<= 1;
-    si++;
-  }
-  
-  /* Figure C.3: generate encoding tables */
-  /* These are code and size indexed by symbol value */
-
-  /* Set all codeless symbols to have code length 0;
-   * this lets us detect duplicate VAL entries here, and later
-   * allows emit_bits to detect any attempt to emit such symbols.
-   */
-  MEMZERO(dtbl->ehufsi, SIZEOF(dtbl->ehufsi));
-
-  /* This is also a convenient place to check for out-of-range
-   * and duplicated VAL entries.  We allow 0..255 for AC symbols
-   * but only 0..15 for DC.  (We could constrain them further
-   * based on data depth and mode, but this seems enough.)
-   */
-  maxsymbol = isDC ? 15 : 255;
-
-  for (p = 0; p < lastp; p++) {
-    i = htbl->huffval[p];
-    if (i < 0 || i > maxsymbol || dtbl->ehufsi[i])
-      ERREXIT(cinfo, JERR_BAD_HUFF_TABLE);
-    dtbl->ehufco[i] = huffcode[p];
-    dtbl->ehufsi[i] = huffsize[p];
-  }
-}
-
-
-/* Outputting bytes to the file */
-
-/* Emit a byte, taking 'action' if must suspend. */
-#define emit_byte(state,val,action)  \
-	{ *(state)->next_output_byte++ = (JOCTET) (val);  \
-	  if (--(state)->free_in_buffer == 0)  \
-	    if (! dump_buffer(state))  \
-	      { action; } }
-
-
-LOCAL(boolean)
-dump_buffer (working_state * state)
-/* Empty the output buffer; return TRUE if successful, FALSE if must suspend */
-{
-  struct jpeg_destination_mgr * dest = state->cinfo->dest;
-
-  if (! (*dest->empty_output_buffer) (state->cinfo))
-    return FALSE;
-  /* After a successful buffer dump, must reset buffer pointers */
-  state->next_output_byte = dest->next_output_byte;
-  state->free_in_buffer = dest->free_in_buffer;
-  return TRUE;
-}
-
-
-/* Outputting bits to the file */
-
-/* Only the right 24 bits of put_buffer are used; the valid bits are
- * left-justified in this part.  At most 16 bits can be passed to emit_bits
- * in one call, and we never retain more than 7 bits in put_buffer
- * between calls, so 24 bits are sufficient.
- */
-
-INLINE
-LOCAL(boolean)
-emit_bits (working_state * state, unsigned int code, int size)
-/* Emit some bits; return TRUE if successful, FALSE if must suspend */
-{
-  /* This routine is heavily used, so it's worth coding tightly. */
-  register INT32 put_buffer = (INT32) code;
-  register int put_bits = state->cur.put_bits;
-
-  /* if size is 0, caller used an invalid Huffman table entry */
-  if (size == 0)
-    ERREXIT(state->cinfo, JERR_HUFF_MISSING_CODE);
-
-  put_buffer &= (((INT32) 1)<<size) - 1; /* mask off any extra bits in code */
-  
-  put_bits += size;		/* new number of bits in buffer */
-  
-  put_buffer <<= 24 - put_bits; /* align incoming bits */
-
-  put_buffer |= state->cur.put_buffer; /* and merge with old buffer contents */
-  
-  while (put_bits >= 8) {
-    int c = (int) ((put_buffer >> 16) & 0xFF);
-    
-    emit_byte(state, c, return FALSE);
-    if (c == 0xFF) {		/* need to stuff a zero byte? */
-      emit_byte(state, 0, return FALSE);
-    }
-    put_buffer <<= 8;
-    put_bits -= 8;
-  }
-
-  state->cur.put_buffer = put_buffer; /* update state variables */
-  state->cur.put_bits = put_bits;
-
-  return TRUE;
-}
-
-
-LOCAL(boolean)
-flush_bits (working_state * state)
-{
-  if (! emit_bits(state, 0x7F, 7)) /* fill any partial byte with ones */
-    return FALSE;
-  state->cur.put_buffer = 0;	/* and reset bit-buffer to empty */
-  state->cur.put_bits = 0;
-  return TRUE;
-}
-
-
-/* Encode a single block's worth of coefficients */
-
-LOCAL(boolean)
-encode_one_block (working_state * state, JCOEFPTR block, int last_dc_val,
-		  c_derived_tbl *dctbl, c_derived_tbl *actbl)
-{
-  register int temp, temp2;
-  register int nbits;
-  register int k, r, i;
-  
-  /* Encode the DC coefficient difference per section F.1.2.1 */
-  
-  temp = temp2 = block[0] - last_dc_val;
-
-  if (temp < 0) {
-    temp = -temp;		/* temp is abs value of input */
-    /* For a negative input, want temp2 = bitwise complement of abs(input) */
-    /* This code assumes we are on a two's complement machine */
-    temp2--;
-  }
-  
-  /* Find the number of bits needed for the magnitude of the coefficient */
-  nbits = 0;
-  while (temp) {
-    nbits++;
-    temp >>= 1;
-  }
-  /* Check for out-of-range coefficient values.
-   * Since we're encoding a difference, the range limit is twice as much.
-   */
-  if (nbits > MAX_COEF_BITS+1)
-    ERREXIT(state->cinfo, JERR_BAD_DCT_COEF);
-  
-  /* Emit the Huffman-coded symbol for the number of bits */
-  if (! emit_bits(state, dctbl->ehufco[nbits], dctbl->ehufsi[nbits]))
-    return FALSE;
-
-  /* Emit that number of bits of the value, if positive, */
-  /* or the complement of its magnitude, if negative. */
-  if (nbits)			/* emit_bits rejects calls with size 0 */
-    if (! emit_bits(state, (unsigned int) temp2, nbits))
-      return FALSE;
-
-  /* Encode the AC coefficients per section F.1.2.2 */
-  
-  r = 0;			/* r = run length of zeros */
-  
-  for (k = 1; k < DCTSIZE2; k++) {
-    if ((temp = block[jpeg_natural_order[k]]) == 0) {
-      r++;
-    } else {
-      /* if run length > 15, must emit special run-length-16 codes (0xF0) */
-      while (r > 15) {
-	if (! emit_bits(state, actbl->ehufco[0xF0], actbl->ehufsi[0xF0]))
-	  return FALSE;
-	r -= 16;
-      }
-
-      temp2 = temp;
-      if (temp < 0) {
-	temp = -temp;		/* temp is abs value of input */
-	/* This code assumes we are on a two's complement machine */
-	temp2--;
-      }
-      
-      /* Find the number of bits needed for the magnitude of the coefficient */
-      nbits = 1;		/* there must be at least one 1 bit */
-      while ((temp >>= 1))
-	nbits++;
-      /* Check for out-of-range coefficient values */
-      if (nbits > MAX_COEF_BITS)
-	ERREXIT(state->cinfo, JERR_BAD_DCT_COEF);
-      
-      /* Emit Huffman symbol for run length / number of bits */
-      i = (r << 4) + nbits;
-      if (! emit_bits(state, actbl->ehufco[i], actbl->ehufsi[i]))
-	return FALSE;
-
-      /* Emit that number of bits of the value, if positive, */
-      /* or the complement of its magnitude, if negative. */
-      if (! emit_bits(state, (unsigned int) temp2, nbits))
-	return FALSE;
-      
-      r = 0;
-    }
-  }
-
-  /* If the last coef(s) were zero, emit an end-of-block code */
-  if (r > 0)
-    if (! emit_bits(state, actbl->ehufco[0], actbl->ehufsi[0]))
-      return FALSE;
-
-  return TRUE;
-}
-
-
-/*
- * Emit a restart marker & resynchronize predictions.
- */
-
-LOCAL(boolean)
-emit_restart (working_state * state, int restart_num)
-{
-  int ci;
-
-  if (! flush_bits(state))
-    return FALSE;
-
-  emit_byte(state, 0xFF, return FALSE);
-  emit_byte(state, JPEG_RST0 + restart_num, return FALSE);
-
-  /* Re-initialize DC predictions to 0 */
-  for (ci = 0; ci < state->cinfo->comps_in_scan; ci++)
-    state->cur.last_dc_val[ci] = 0;
-
-  /* The restart counter is not updated until we successfully write the MCU. */
-
-  return TRUE;
-}
-
-
-/*
- * Encode and output one MCU's worth of Huffman-compressed coefficients.
- */
-
-METHODDEF(boolean)
-encode_mcu_huff (j_compress_ptr cinfo, JBLOCKROW *MCU_data)
-{
-  huff_entropy_ptr entropy = (huff_entropy_ptr) cinfo->entropy;
-  working_state state;
-  int blkn, ci;
-  jpeg_component_info * compptr;
-
-  /* Load up working state */
-  state.next_output_byte = cinfo->dest->next_output_byte;
-  state.free_in_buffer = cinfo->dest->free_in_buffer;
-  ASSIGN_STATE(state.cur, entropy->saved);
-  state.cinfo = cinfo;
-
-  /* Emit restart marker if needed */
-  if (cinfo->restart_interval) {
-    if (entropy->restarts_to_go == 0)
-      if (! emit_restart(&state, entropy->next_restart_num))
-	return FALSE;
-  }
-
-  /* Encode the MCU data blocks */
-  for (blkn = 0; blkn < cinfo->blocks_in_MCU; blkn++) {
-    ci = cinfo->MCU_membership[blkn];
-    compptr = cinfo->cur_comp_info[ci];
-    if (! encode_one_block(&state,
-			   MCU_data[blkn][0], state.cur.last_dc_val[ci],
-			   entropy->dc_derived_tbls[compptr->dc_tbl_no],
-			   entropy->ac_derived_tbls[compptr->ac_tbl_no]))
-      return FALSE;
-    /* Update last_dc_val */
-    state.cur.last_dc_val[ci] = MCU_data[blkn][0][0];
-  }
-
-  /* Completed MCU, so update state */
-  cinfo->dest->next_output_byte = state.next_output_byte;
-  cinfo->dest->free_in_buffer = state.free_in_buffer;
-  ASSIGN_STATE(entropy->saved, state.cur);
-
-  /* Update restart-interval state too */
-  if (cinfo->restart_interval) {
-    if (entropy->restarts_to_go == 0) {
-      entropy->restarts_to_go = cinfo->restart_interval;
-      entropy->next_restart_num++;
-      entropy->next_restart_num &= 7;
-    }
-    entropy->restarts_to_go--;
-  }
-
-  return TRUE;
-}
-
-
-/*
- * Finish up at the end of a Huffman-compressed scan.
- */
-
-METHODDEF(void)
-finish_pass_huff (j_compress_ptr cinfo)
-{
-  huff_entropy_ptr entropy = (huff_entropy_ptr) cinfo->entropy;
-  working_state state;
-
-  /* Load up working state ... flush_bits needs it */
-  state.next_output_byte = cinfo->dest->next_output_byte;
-  state.free_in_buffer = cinfo->dest->free_in_buffer;
-  ASSIGN_STATE(state.cur, entropy->saved);
-  state.cinfo = cinfo;
-
-  /* Flush out the last data */
-  if (! flush_bits(&state))
-    ERREXIT(cinfo, JERR_CANT_SUSPEND);
-
-  /* Update state */
-  cinfo->dest->next_output_byte = state.next_output_byte;
-  cinfo->dest->free_in_buffer = state.free_in_buffer;
-  ASSIGN_STATE(entropy->saved, state.cur);
-}
-
-
-/*
- * Huffman coding optimization.
- *
- * We first scan the supplied data and count the number of uses of each symbol
- * that is to be Huffman-coded. (This process MUST agree with the code above.)
- * Then we build a Huffman coding tree for the observed counts.
- * Symbols which are not needed at all for the particular image are not
- * assigned any code, which saves space in the DHT marker as well as in
- * the compressed data.
- */
-
-#ifdef ENTROPY_OPT_SUPPORTED
-
-
-/* Process a single block's worth of coefficients */
-
-LOCAL(void)
-htest_one_block (j_compress_ptr cinfo, JCOEFPTR block, int last_dc_val,
-		 long dc_counts[], long ac_counts[])
-{
-  register int temp;
-  register int nbits;
-  register int k, r;
-  
-  /* Encode the DC coefficient difference per section F.1.2.1 */
-  
-  temp = block[0] - last_dc_val;
-  if (temp < 0)
-    temp = -temp;
-  
-  /* Find the number of bits needed for the magnitude of the coefficient */
-  nbits = 0;
-  while (temp) {
-    nbits++;
-    temp >>= 1;
-  }
-  /* Check for out-of-range coefficient values.
-   * Since we're encoding a difference, the range limit is twice as much.
-   */
-  if (nbits > MAX_COEF_BITS+1)
-    ERREXIT(cinfo, JERR_BAD_DCT_COEF);
-
-  /* Count the Huffman symbol for the number of bits */
-  dc_counts[nbits]++;
-  
-  /* Encode the AC coefficients per section F.1.2.2 */
-  
-  r = 0;			/* r = run length of zeros */
-  
-  for (k = 1; k < DCTSIZE2; k++) {
-    if ((temp = block[jpeg_natural_order[k]]) == 0) {
-      r++;
-    } else {
-      /* if run length > 15, must emit special run-length-16 codes (0xF0) */
-      while (r > 15) {
-	ac_counts[0xF0]++;
-	r -= 16;
-      }
-      
-      /* Find the number of bits needed for the magnitude of the coefficient */
-      if (temp < 0)
-	temp = -temp;
-      
-      /* Find the number of bits needed for the magnitude of the coefficient */
-      nbits = 1;		/* there must be at least one 1 bit */
-      while ((temp >>= 1))
-	nbits++;
-      /* Check for out-of-range coefficient values */
-      if (nbits > MAX_COEF_BITS)
-	ERREXIT(cinfo, JERR_BAD_DCT_COEF);
-      
-      /* Count Huffman symbol for run length / number of bits */
-      ac_counts[(r << 4) + nbits]++;
-      
-      r = 0;
-    }
-  }
-
-  /* If the last coef(s) were zero, emit an end-of-block code */
-  if (r > 0)
-    ac_counts[0]++;
-}
-
-
-/*
- * Trial-encode one MCU's worth of Huffman-compressed coefficients.
- * No data is actually output, so no suspension return is possible.
- */
-
-METHODDEF(boolean)
-encode_mcu_gather (j_compress_ptr cinfo, JBLOCKROW *MCU_data)
-{
-  huff_entropy_ptr entropy = (huff_entropy_ptr) cinfo->entropy;
-  int blkn, ci;
-  jpeg_component_info * compptr;
-
-  /* Take care of restart intervals if needed */
-  if (cinfo->restart_interval) {
-    if (entropy->restarts_to_go == 0) {
-      /* Re-initialize DC predictions to 0 */
-      for (ci = 0; ci < cinfo->comps_in_scan; ci++)
-	entropy->saved.last_dc_val[ci] = 0;
-      /* Update restart state */
-      entropy->restarts_to_go = cinfo->restart_interval;
-    }
-    entropy->restarts_to_go--;
-  }
-
-  for (blkn = 0; blkn < cinfo->blocks_in_MCU; blkn++) {
-    ci = cinfo->MCU_membership[blkn];
-    compptr = cinfo->cur_comp_info[ci];
-    htest_one_block(cinfo, MCU_data[blkn][0], entropy->saved.last_dc_val[ci],
-		    entropy->dc_count_ptrs[compptr->dc_tbl_no],
-		    entropy->ac_count_ptrs[compptr->ac_tbl_no]);
-    entropy->saved.last_dc_val[ci] = MCU_data[blkn][0][0];
-  }
-
-  return TRUE;
-}
-
-
-/*
- * Generate the best Huffman code table for the given counts, fill htbl.
- * Note this is also used by jcphuff.c.
- *
- * The JPEG standard requires that no symbol be assigned a codeword of all
- * one bits (so that padding bits added at the end of a compressed segment
- * can't look like a valid code).  Because of the canonical ordering of
- * codewords, this just means that there must be an unused slot in the
- * longest codeword length category.  Section K.2 of the JPEG spec suggests
- * reserving such a slot by pretending that symbol 256 is a valid symbol
- * with count 1.  In theory that's not optimal; giving it count zero but
- * including it in the symbol set anyway should give a better Huffman code.
- * But the theoretically better code actually seems to come out worse in
- * practice, because it produces more all-ones bytes (which incur stuffed
- * zero bytes in the final file).  In any case the difference is tiny.
- *
- * The JPEG standard requires Huffman codes to be no more than 16 bits long.
- * If some symbols have a very small but nonzero probability, the Huffman tree
- * must be adjusted to meet the code length restriction.  We currently use
- * the adjustment method suggested in JPEG section K.2.  This method is *not*
- * optimal; it may not choose the best possible limited-length code.  But
- * typically only very-low-frequency symbols will be given less-than-optimal
- * lengths, so the code is almost optimal.  Experimental comparisons against
- * an optimal limited-length-code algorithm indicate that the difference is
- * microscopic --- usually less than a hundredth of a percent of total size.
- * So the extra complexity of an optimal algorithm doesn't seem worthwhile.
- */
-
-GLOBAL(void)
-jpeg_gen_optimal_table (j_compress_ptr cinfo, JHUFF_TBL * htbl, long freq[])
-{
-#define MAX_CLEN 32		/* assumed maximum initial code length */
-  UINT8 bits[MAX_CLEN+1];	/* bits[k] = # of symbols with code length k */
-  int codesize[257];		/* codesize[k] = code length of symbol k */
-  int others[257];		/* next symbol in current branch of tree */
-  int c1, c2;
-  int p, i, j;
-  long v;
-
-  /* This algorithm is explained in section K.2 of the JPEG standard */
-
-  MEMZERO(bits, SIZEOF(bits));
-  MEMZERO(codesize, SIZEOF(codesize));
-  for (i = 0; i < 257; i++)
-    others[i] = -1;		/* init links to empty */
-  
-  freq[256] = 1;		/* make sure 256 has a nonzero count */
-  /* Including the pseudo-symbol 256 in the Huffman procedure guarantees
-   * that no real symbol is given code-value of all ones, because 256
-   * will be placed last in the largest codeword category.
-   */
-
-  /* Huffman's basic algorithm to assign optimal code lengths to symbols */
-
-  for (;;) {
-    /* Find the smallest nonzero frequency, set c1 = its symbol */
-    /* In case of ties, take the larger symbol number */
-    c1 = -1;
-    v = 1000000000L;
-    for (i = 0; i <= 256; i++) {
-      if (freq[i] && freq[i] <= v) {
-	v = freq[i];
-	c1 = i;
-      }
-    }
-
-    /* Find the next smallest nonzero frequency, set c2 = its symbol */
-    /* In case of ties, take the larger symbol number */
-    c2 = -1;
-    v = 1000000000L;
-    for (i = 0; i <= 256; i++) {
-      if (freq[i] && freq[i] <= v && i != c1) {
-	v = freq[i];
-	c2 = i;
-      }
-    }
-
-    /* Done if we've merged everything into one frequency */
-    if (c2 < 0)
-      break;
-    
-    /* Else merge the two counts/trees */
-    freq[c1] += freq[c2];
-    freq[c2] = 0;
-
-    /* Increment the codesize of everything in c1's tree branch */
-    codesize[c1]++;
-    while (others[c1] >= 0) {
-      c1 = others[c1];
-      codesize[c1]++;
-    }
-    
-    others[c1] = c2;		/* chain c2 onto c1's tree branch */
-    
-    /* Increment the codesize of everything in c2's tree branch */
-    codesize[c2]++;
-    while (others[c2] >= 0) {
-      c2 = others[c2];
-      codesize[c2]++;
-    }
-  }
-
-  /* Now count the number of symbols of each code length */
-  for (i = 0; i <= 256; i++) {
-    if (codesize[i]) {
-      /* The JPEG standard seems to think that this can't happen, */
-      /* but I'm paranoid... */
-      if (codesize[i] > MAX_CLEN)
-	ERREXIT(cinfo, JERR_HUFF_CLEN_OVERFLOW);
-
-      bits[codesize[i]]++;
-    }
-  }
-
-  /* JPEG doesn't allow symbols with code lengths over 16 bits, so if the pure
-   * Huffman procedure assigned any such lengths, we must adjust the coding.
-   * Here is what the JPEG spec says about how this next bit works:
-   * Since symbols are paired for the longest Huffman code, the symbols are
-   * removed from this length category two at a time.  The prefix for the pair
-   * (which is one bit shorter) is allocated to one of the pair; then,
-   * skipping the BITS entry for that prefix length, a code word from the next
-   * shortest nonzero BITS entry is converted into a prefix for two code words
-   * one bit longer.
-   */
-  
-  for (i = MAX_CLEN; i > 16; i--) {
-    while (bits[i] > 0) {
-      j = i - 2;		/* find length of new prefix to be used */
-      while (bits[j] == 0)
-	j--;
-      
-      bits[i] -= 2;		/* remove two symbols */
-      bits[i-1]++;		/* one goes in this length */
-      bits[j+1] += 2;		/* two new symbols in this length */
-      bits[j]--;		/* symbol of this length is now a prefix */
-    }
-  }
-
-  /* Remove the count for the pseudo-symbol 256 from the largest codelength */
-  while (bits[i] == 0)		/* find largest codelength still in use */
-    i--;
-  bits[i]--;
-  
-  /* Return final symbol counts (only for lengths 0..16) */
-  MEMCOPY(htbl->bits, bits, SIZEOF(htbl->bits));
-  
-  /* Return a list of the symbols sorted by code length */
-  /* It's not real clear to me why we don't need to consider the codelength
-   * changes made above, but the JPEG spec seems to think this works.
-   */
-  p = 0;
-  for (i = 1; i <= MAX_CLEN; i++) {
-    for (j = 0; j <= 255; j++) {
-      if (codesize[j] == i) {
-	htbl->huffval[p] = (UINT8) j;
-	p++;
-      }
-    }
-  }
-
-  /* Set sent_table FALSE so updated table will be written to JPEG file. */
-  htbl->sent_table = FALSE;
-}
-
-
-/*
- * Finish up a statistics-gathering pass and create the new Huffman tables.
- */
-
-METHODDEF(void)
-finish_pass_gather (j_compress_ptr cinfo)
-{
-  huff_entropy_ptr entropy = (huff_entropy_ptr) cinfo->entropy;
-  int ci, dctbl, actbl;
-  jpeg_component_info * compptr;
-  JHUFF_TBL **htblptr;
-  boolean did_dc[NUM_HUFF_TBLS];
-  boolean did_ac[NUM_HUFF_TBLS];
-
-  /* It's important not to apply jpeg_gen_optimal_table more than once
-   * per table, because it clobbers the input frequency counts!
-   */
-  MEMZERO(did_dc, SIZEOF(did_dc));
-  MEMZERO(did_ac, SIZEOF(did_ac));
-
-  for (ci = 0; ci < cinfo->comps_in_scan; ci++) {
-    compptr = cinfo->cur_comp_info[ci];
-    dctbl = compptr->dc_tbl_no;
-    actbl = compptr->ac_tbl_no;
-    if (! did_dc[dctbl]) {
-      htblptr = & cinfo->dc_huff_tbl_ptrs[dctbl];
-      if (*htblptr == NULL)
-	*htblptr = jpeg_alloc_huff_table((j_common_ptr) cinfo);
-      jpeg_gen_optimal_table(cinfo, *htblptr, entropy->dc_count_ptrs[dctbl]);
-      did_dc[dctbl] = TRUE;
-    }
-    if (! did_ac[actbl]) {
-      htblptr = & cinfo->ac_huff_tbl_ptrs[actbl];
-      if (*htblptr == NULL)
-	*htblptr = jpeg_alloc_huff_table((j_common_ptr) cinfo);
-      jpeg_gen_optimal_table(cinfo, *htblptr, entropy->ac_count_ptrs[actbl]);
-      did_ac[actbl] = TRUE;
-    }
-  }
-}
-
-
-#endif /* ENTROPY_OPT_SUPPORTED */
-
-
-/*
- * Module initialization routine for Huffman entropy encoding.
- */
-
-GLOBAL(void)
-jinit_huff_encoder (j_compress_ptr cinfo)
-{
-  huff_entropy_ptr entropy;
-  int i;
-
-  entropy = (huff_entropy_ptr)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				SIZEOF(huff_entropy_encoder));
-  cinfo->entropy = (struct jpeg_entropy_encoder *) entropy;
-  entropy->pub.start_pass = start_pass_huff;
-
-  /* Mark tables unallocated */
-  for (i = 0; i < NUM_HUFF_TBLS; i++) {
-    entropy->dc_derived_tbls[i] = entropy->ac_derived_tbls[i] = NULL;
-#ifdef ENTROPY_OPT_SUPPORTED
-    entropy->dc_count_ptrs[i] = entropy->ac_count_ptrs[i] = NULL;
-#endif
-  }
-}
@@ -1,47 +0,0 @@
-/*
- * jchuff.h
- *
- * Copyright (C) 1991-1997, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains declarations for Huffman entropy encoding routines
- * that are shared between the sequential encoder (jchuff.c) and the
- * progressive encoder (jcphuff.c).  No other modules need to see these.
- */
-
-/* The legal range of a DCT coefficient is
- *  -1024 .. +1023  for 8-bit data;
- * -16384 .. +16383 for 12-bit data.
- * Hence the magnitude should always fit in 10 or 14 bits respectively.
- */
-
-#if BITS_IN_JSAMPLE == 8
-#define MAX_COEF_BITS 10
-#else
-#define MAX_COEF_BITS 14
-#endif
-
-/* Derived data constructed for each Huffman table */
-
-typedef struct {
-  unsigned int ehufco[256];	/* code for each symbol */
-  char ehufsi[256];		/* length of code for each symbol */
-  /* If no code has been allocated for a symbol S, ehufsi[S] contains 0 */
-} c_derived_tbl;
-
-/* Short forms of external names for systems with brain-damaged linkers. */
-
-#ifdef NEED_SHORT_EXTERNAL_NAMES
-#define jpeg_make_c_derived_tbl	jMkCDerived
-#define jpeg_gen_optimal_table	jGenOptTbl
-#endif /* NEED_SHORT_EXTERNAL_NAMES */
-
-/* Expand a Huffman table definition into the derived format */
-EXTERN(void) jpeg_make_c_derived_tbl
-	JPP((j_compress_ptr cinfo, boolean isDC, int tblno,
-	     c_derived_tbl ** pdtbl));
-
-/* Generate an optimal table definition given the specified counts */
-EXTERN(void) jpeg_gen_optimal_table
-	JPP((j_compress_ptr cinfo, JHUFF_TBL * htbl, long freq[]));
@@ -1,72 +0,0 @@
-/*
- * jcinit.c
- *
- * Copyright (C) 1991-1997, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains initialization logic for the JPEG compressor.
- * This routine is in charge of selecting the modules to be executed and
- * making an initialization call to each one.
- *
- * Logically, this code belongs in jcmaster.c.  It's split out because
- * linking this routine implies linking the entire compression library.
- * For a transcoding-only application, we want to be able to use jcmaster.c
- * without linking in the whole library.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-
-
-/*
- * Master selection of compression modules.
- * This is done once at the start of processing an image.  We determine
- * which modules will be used and give them appropriate initialization calls.
- */
-
-GLOBAL(void)
-jinit_compress_master (j_compress_ptr cinfo)
-{
-  /* Initialize master control (includes parameter checking/processing) */
-  jinit_c_master_control(cinfo, FALSE /* full compression */);
-
-  /* Preprocessing */
-  if (! cinfo->raw_data_in) {
-    jinit_color_converter(cinfo);
-    jinit_downsampler(cinfo);
-    jinit_c_prep_controller(cinfo, FALSE /* never need full buffer here */);
-  }
-  /* Forward DCT */
-  jinit_forward_dct(cinfo);
-  /* Entropy encoding: either Huffman or arithmetic coding. */
-  if (cinfo->arith_code) {
-    ERREXIT(cinfo, JERR_ARITH_NOTIMPL);
-  } else {
-    if (cinfo->progressive_mode) {
-#ifdef C_PROGRESSIVE_SUPPORTED
-      jinit_phuff_encoder(cinfo);
-#else
-      ERREXIT(cinfo, JERR_NOT_COMPILED);
-#endif
-    } else
-      jinit_huff_encoder(cinfo);
-  }
-
-  /* Need a full-image coefficient buffer in any multi-pass mode. */
-  jinit_c_coef_controller(cinfo,
-		(boolean) (cinfo->num_scans > 1 || cinfo->optimize_coding));
-  jinit_c_main_controller(cinfo, FALSE /* never need full buffer here */);
-
-  jinit_marker_writer(cinfo);
-
-  /* We can now tell the memory manager to allocate virtual arrays. */
-  (*cinfo->mem->realize_virt_arrays) ((j_common_ptr) cinfo);
-
-  /* Write the datastream header (SOI) immediately.
-   * Frame and scan headers are postponed till later.
-   * This lets application insert special markers after the SOI.
-   */
-  (*cinfo->marker->write_file_header) (cinfo);
-}
@@ -1,293 +0,0 @@
-/*
- * jcmainct.c
- *
- * Copyright (C) 1994-1996, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains the main buffer controller for compression.
- * The main buffer lies between the pre-processor and the JPEG
- * compressor proper; it holds downsampled data in the JPEG colorspace.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-
-
-/* Note: currently, there is no operating mode in which a full-image buffer
- * is needed at this step.  If there were, that mode could not be used with
- * "raw data" input, since this module is bypassed in that case.  However,
- * we've left the code here for possible use in special applications.
- */
-#undef FULL_MAIN_BUFFER_SUPPORTED
-
-
-/* Private buffer controller object */
-
-typedef struct {
-  struct jpeg_c_main_controller pub; /* public fields */
-
-  JDIMENSION cur_iMCU_row;	/* number of current iMCU row */
-  JDIMENSION rowgroup_ctr;	/* counts row groups received in iMCU row */
-  boolean suspended;		/* remember if we suspended output */
-  J_BUF_MODE pass_mode;		/* current operating mode */
-
-  /* If using just a strip buffer, this points to the entire set of buffers
-   * (we allocate one for each component).  In the full-image case, this
-   * points to the currently accessible strips of the virtual arrays.
-   */
-  JSAMPARRAY buffer[MAX_COMPONENTS];
-
-#ifdef FULL_MAIN_BUFFER_SUPPORTED
-  /* If using full-image storage, this array holds pointers to virtual-array
-   * control blocks for each component.  Unused if not full-image storage.
-   */
-  jvirt_sarray_ptr whole_image[MAX_COMPONENTS];
-#endif
-} my_main_controller;
-
-typedef my_main_controller * my_main_ptr;
-
-
-/* Forward declarations */
-METHODDEF(void) process_data_simple_main
-	JPP((j_compress_ptr cinfo, JSAMPARRAY input_buf,
-	     JDIMENSION *in_row_ctr, JDIMENSION in_rows_avail));
-#ifdef FULL_MAIN_BUFFER_SUPPORTED
-METHODDEF(void) process_data_buffer_main
-	JPP((j_compress_ptr cinfo, JSAMPARRAY input_buf,
-	     JDIMENSION *in_row_ctr, JDIMENSION in_rows_avail));
-#endif
-
-
-/*
- * Initialize for a processing pass.
- */
-
-METHODDEF(void)
-start_pass_main (j_compress_ptr cinfo, J_BUF_MODE pass_mode)
-{
-  my_main_ptr main = (my_main_ptr) cinfo->main;
-
-  /* Do nothing in raw-data mode. */
-  if (cinfo->raw_data_in)
-    return;
-
-  main->cur_iMCU_row = 0;	/* initialize counters */
-  main->rowgroup_ctr = 0;
-  main->suspended = FALSE;
-  main->pass_mode = pass_mode;	/* save mode for use by process_data */
-
-  switch (pass_mode) {
-  case JBUF_PASS_THRU:
-#ifdef FULL_MAIN_BUFFER_SUPPORTED
-    if (main->whole_image[0] != NULL)
-      ERREXIT(cinfo, JERR_BAD_BUFFER_MODE);
-#endif
-    main->pub.process_data = process_data_simple_main;
-    break;
-#ifdef FULL_MAIN_BUFFER_SUPPORTED
-  case JBUF_SAVE_SOURCE:
-  case JBUF_CRANK_DEST:
-  case JBUF_SAVE_AND_PASS:
-    if (main->whole_image[0] == NULL)
-      ERREXIT(cinfo, JERR_BAD_BUFFER_MODE);
-    main->pub.process_data = process_data_buffer_main;
-    break;
-#endif
-  default:
-    ERREXIT(cinfo, JERR_BAD_BUFFER_MODE);
-    break;
-  }
-}
-
-
-/*
- * Process some data.
- * This routine handles the simple pass-through mode,
- * where we have only a strip buffer.
- */
-
-METHODDEF(void)
-process_data_simple_main (j_compress_ptr cinfo,
-			  JSAMPARRAY input_buf, JDIMENSION *in_row_ctr,
-			  JDIMENSION in_rows_avail)
-{
-  my_main_ptr main = (my_main_ptr) cinfo->main;
-
-  while (main->cur_iMCU_row < cinfo->total_iMCU_rows) {
-    /* Read input data if we haven't filled the main buffer yet */
-    if (main->rowgroup_ctr < DCTSIZE)
-      (*cinfo->prep->pre_process_data) (cinfo,
-					input_buf, in_row_ctr, in_rows_avail,
-					main->buffer, &main->rowgroup_ctr,
-					(JDIMENSION) DCTSIZE);
-
-    /* If we don't have a full iMCU row buffered, return to application for
-     * more data.  Note that preprocessor will always pad to fill the iMCU row
-     * at the bottom of the image.
-     */
-    if (main->rowgroup_ctr != DCTSIZE)
-      return;
-
-    /* Send the completed row to the compressor */
-    if (! (*cinfo->coef->compress_data) (cinfo, main->buffer)) {
-      /* If compressor did not consume the whole row, then we must need to
-       * suspend processing and return to the application.  In this situation
-       * we pretend we didn't yet consume the last input row; otherwise, if
-       * it happened to be the last row of the image, the application would
-       * think we were done.
-       */
-      if (! main->suspended) {
-	(*in_row_ctr)--;
-	main->suspended = TRUE;
-      }
-      return;
-    }
-    /* We did finish the row.  Undo our little suspension hack if a previous
-     * call suspended; then mark the main buffer empty.
-     */
-    if (main->suspended) {
-      (*in_row_ctr)++;
-      main->suspended = FALSE;
-    }
-    main->rowgroup_ctr = 0;
-    main->cur_iMCU_row++;
-  }
-}
-
-
-#ifdef FULL_MAIN_BUFFER_SUPPORTED
-
-/*
- * Process some data.
- * This routine handles all of the modes that use a full-size buffer.
- */
-
-METHODDEF(void)
-process_data_buffer_main (j_compress_ptr cinfo,
-			  JSAMPARRAY input_buf, JDIMENSION *in_row_ctr,
-			  JDIMENSION in_rows_avail)
-{
-  my_main_ptr main = (my_main_ptr) cinfo->main;
-  int ci;
-  jpeg_component_info *compptr;
-  boolean writing = (main->pass_mode != JBUF_CRANK_DEST);
-
-  while (main->cur_iMCU_row < cinfo->total_iMCU_rows) {
-    /* Realign the virtual buffers if at the start of an iMCU row. */
-    if (main->rowgroup_ctr == 0) {
-      for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
-	   ci++, compptr++) {
-	main->buffer[ci] = (*cinfo->mem->access_virt_sarray)
-	  ((j_common_ptr) cinfo, main->whole_image[ci],
-	   main->cur_iMCU_row * (compptr->v_samp_factor * DCTSIZE),
-	   (JDIMENSION) (compptr->v_samp_factor * DCTSIZE), writing);
-      }
-      /* In a read pass, pretend we just read some source data. */
-      if (! writing) {
-	*in_row_ctr += cinfo->max_v_samp_factor * DCTSIZE;
-	main->rowgroup_ctr = DCTSIZE;
-      }
-    }
-
-    /* If a write pass, read input data until the current iMCU row is full. */
-    /* Note: preprocessor will pad if necessary to fill the last iMCU row. */
-    if (writing) {
-      (*cinfo->prep->pre_process_data) (cinfo,
-					input_buf, in_row_ctr, in_rows_avail,
-					main->buffer, &main->rowgroup_ctr,
-					(JDIMENSION) DCTSIZE);
-      /* Return to application if we need more data to fill the iMCU row. */
-      if (main->rowgroup_ctr < DCTSIZE)
-	return;
-    }
-
-    /* Emit data, unless this is a sink-only pass. */
-    if (main->pass_mode != JBUF_SAVE_SOURCE) {
-      if (! (*cinfo->coef->compress_data) (cinfo, main->buffer)) {
-	/* If compressor did not consume the whole row, then we must need to
-	 * suspend processing and return to the application.  In this situation
-	 * we pretend we didn't yet consume the last input row; otherwise, if
-	 * it happened to be the last row of the image, the application would
-	 * think we were done.
-	 */
-	if (! main->suspended) {
-	  (*in_row_ctr)--;
-	  main->suspended = TRUE;
-	}
-	return;
-      }
-      /* We did finish the row.  Undo our little suspension hack if a previous
-       * call suspended; then mark the main buffer empty.
-       */
-      if (main->suspended) {
-	(*in_row_ctr)++;
-	main->suspended = FALSE;
-      }
-    }
-
-    /* If get here, we are done with this iMCU row.  Mark buffer empty. */
-    main->rowgroup_ctr = 0;
-    main->cur_iMCU_row++;
-  }
-}
-
-#endif /* FULL_MAIN_BUFFER_SUPPORTED */
-
-
-/*
- * Initialize main buffer controller.
- */
-
-GLOBAL(void)
-jinit_c_main_controller (j_compress_ptr cinfo, boolean need_full_buffer)
-{
-  my_main_ptr main;
-  int ci;
-  jpeg_component_info *compptr;
-
-  main = (my_main_ptr)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				SIZEOF(my_main_controller));
-  cinfo->main = (struct jpeg_c_main_controller *) main;
-  main->pub.start_pass = start_pass_main;
-
-  /* We don't need to create a buffer in raw-data mode. */
-  if (cinfo->raw_data_in)
-    return;
-
-  /* Create the buffer.  It holds downsampled data, so each component
-   * may be of a different size.
-   */
-  if (need_full_buffer) {
-#ifdef FULL_MAIN_BUFFER_SUPPORTED
-    /* Allocate a full-image virtual array for each component */
-    /* Note we pad the bottom to a multiple of the iMCU height */
-    for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
-	 ci++, compptr++) {
-      main->whole_image[ci] = (*cinfo->mem->request_virt_sarray)
-	((j_common_ptr) cinfo, JPOOL_IMAGE, FALSE,
-	 compptr->width_in_blocks * DCTSIZE,
-	 (JDIMENSION) jround_up((long) compptr->height_in_blocks,
-				(long) compptr->v_samp_factor) * DCTSIZE,
-	 (JDIMENSION) (compptr->v_samp_factor * DCTSIZE));
-    }
-#else
-    ERREXIT(cinfo, JERR_BAD_BUFFER_MODE);
-#endif
-  } else {
-#ifdef FULL_MAIN_BUFFER_SUPPORTED
-    main->whole_image[0] = NULL; /* flag for no virtual arrays */
-#endif
-    /* Allocate a strip buffer for each component */
-    for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
-	 ci++, compptr++) {
-      main->buffer[ci] = (*cinfo->mem->alloc_sarray)
-	((j_common_ptr) cinfo, JPOOL_IMAGE,
-	 compptr->width_in_blocks * DCTSIZE,
-	 (JDIMENSION) (compptr->v_samp_factor * DCTSIZE));
-    }
-  }
-}
@@ -1,664 +0,0 @@
-/*
- * jcmarker.c
- *
- * Copyright (C) 1991-1998, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains routines to write JPEG datastream markers.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-
-
-typedef enum {			/* JPEG marker codes */
-  M_SOF0  = 0xc0,
-  M_SOF1  = 0xc1,
-  M_SOF2  = 0xc2,
-  M_SOF3  = 0xc3,
-  
-  M_SOF5  = 0xc5,
-  M_SOF6  = 0xc6,
-  M_SOF7  = 0xc7,
-  
-  M_JPG   = 0xc8,
-  M_SOF9  = 0xc9,
-  M_SOF10 = 0xca,
-  M_SOF11 = 0xcb,
-  
-  M_SOF13 = 0xcd,
-  M_SOF14 = 0xce,
-  M_SOF15 = 0xcf,
-  
-  M_DHT   = 0xc4,
-  
-  M_DAC   = 0xcc,
-  
-  M_RST0  = 0xd0,
-  M_RST1  = 0xd1,
-  M_RST2  = 0xd2,
-  M_RST3  = 0xd3,
-  M_RST4  = 0xd4,
-  M_RST5  = 0xd5,
-  M_RST6  = 0xd6,
-  M_RST7  = 0xd7,
-  
-  M_SOI   = 0xd8,
-  M_EOI   = 0xd9,
-  M_SOS   = 0xda,
-  M_DQT   = 0xdb,
-  M_DNL   = 0xdc,
-  M_DRI   = 0xdd,
-  M_DHP   = 0xde,
-  M_EXP   = 0xdf,
-  
-  M_APP0  = 0xe0,
-  M_APP1  = 0xe1,
-  M_APP2  = 0xe2,
-  M_APP3  = 0xe3,
-  M_APP4  = 0xe4,
-  M_APP5  = 0xe5,
-  M_APP6  = 0xe6,
-  M_APP7  = 0xe7,
-  M_APP8  = 0xe8,
-  M_APP9  = 0xe9,
-  M_APP10 = 0xea,
-  M_APP11 = 0xeb,
-  M_APP12 = 0xec,
-  M_APP13 = 0xed,
-  M_APP14 = 0xee,
-  M_APP15 = 0xef,
-  
-  M_JPG0  = 0xf0,
-  M_JPG13 = 0xfd,
-  M_COM   = 0xfe,
-  
-  M_TEM   = 0x01,
-  
-  M_ERROR = 0x100
-} JPEG_MARKER;
-
-
-/* Private state */
-
-typedef struct {
-  struct jpeg_marker_writer pub; /* public fields */
-
-  unsigned int last_restart_interval; /* last DRI value emitted; 0 after SOI */
-} my_marker_writer;
-
-typedef my_marker_writer * my_marker_ptr;
-
-
-/*
- * Basic output routines.
- *
- * Note that we do not support suspension while writing a marker.
- * Therefore, an application using suspension must ensure that there is
- * enough buffer space for the initial markers (typ. 600-700 bytes) before
- * calling jpeg_start_compress, and enough space to write the trailing EOI
- * (a few bytes) before calling jpeg_finish_compress.  Multipass compression
- * modes are not supported at all with suspension, so those two are the only
- * points where markers will be written.
- */
-
-LOCAL(void)
-emit_byte (j_compress_ptr cinfo, int val)
-/* Emit a byte */
-{
-  struct jpeg_destination_mgr * dest = cinfo->dest;
-
-  *(dest->next_output_byte)++ = (JOCTET) val;
-  if (--dest->free_in_buffer == 0) {
-    if (! (*dest->empty_output_buffer) (cinfo))
-      ERREXIT(cinfo, JERR_CANT_SUSPEND);
-  }
-}
-
-
-LOCAL(void)
-emit_marker (j_compress_ptr cinfo, JPEG_MARKER mark)
-/* Emit a marker code */
-{
-  emit_byte(cinfo, 0xFF);
-  emit_byte(cinfo, (int) mark);
-}
-
-
-LOCAL(void)
-emit_2bytes (j_compress_ptr cinfo, int value)
-/* Emit a 2-byte integer; these are always MSB first in JPEG files */
-{
-  emit_byte(cinfo, (value >> 8) & 0xFF);
-  emit_byte(cinfo, value & 0xFF);
-}
-
-
-/*
- * Routines to write specific marker types.
- */
-
-LOCAL(int)
-emit_dqt (j_compress_ptr cinfo, int index)
-/* Emit a DQT marker */
-/* Returns the precision used (0 = 8bits, 1 = 16bits) for baseline checking */
-{
-  JQUANT_TBL * qtbl = cinfo->quant_tbl_ptrs[index];
-  int prec;
-  int i;
-
-  if (qtbl == NULL)
-    ERREXIT1(cinfo, JERR_NO_QUANT_TABLE, index);
-
-  prec = 0;
-  for (i = 0; i < DCTSIZE2; i++) {
-    if (qtbl->quantval[i] > 255)
-      prec = 1;
-  }
-
-  if (! qtbl->sent_table) {
-    emit_marker(cinfo, M_DQT);
-
-    emit_2bytes(cinfo, prec ? DCTSIZE2*2 + 1 + 2 : DCTSIZE2 + 1 + 2);
-
-    emit_byte(cinfo, index + (prec<<4));
-
-    for (i = 0; i < DCTSIZE2; i++) {
-      /* The table entries must be emitted in zigzag order. */
-      unsigned int qval = qtbl->quantval[jpeg_natural_order[i]];
-      if (prec)
-	emit_byte(cinfo, (int) (qval >> 8));
-      emit_byte(cinfo, (int) (qval & 0xFF));
-    }
-
-    qtbl->sent_table = TRUE;
-  }
-
-  return prec;
-}
-
-
-LOCAL(void)
-emit_dht (j_compress_ptr cinfo, int index, boolean is_ac)
-/* Emit a DHT marker */
-{
-  JHUFF_TBL * htbl;
-  int length, i;
-  
-  if (is_ac) {
-    htbl = cinfo->ac_huff_tbl_ptrs[index];
-    index += 0x10;		/* output index has AC bit set */
-  } else {
-    htbl = cinfo->dc_huff_tbl_ptrs[index];
-  }
-
-  if (htbl == NULL)
-    ERREXIT1(cinfo, JERR_NO_HUFF_TABLE, index);
-  
-  if (! htbl->sent_table) {
-    emit_marker(cinfo, M_DHT);
-    
-    length = 0;
-    for (i = 1; i <= 16; i++)
-      length += htbl->bits[i];
-    
-    emit_2bytes(cinfo, length + 2 + 1 + 16);
-    emit_byte(cinfo, index);
-    
-    for (i = 1; i <= 16; i++)
-      emit_byte(cinfo, htbl->bits[i]);
-    
-    for (i = 0; i < length; i++)
-      emit_byte(cinfo, htbl->huffval[i]);
-    
-    htbl->sent_table = TRUE;
-  }
-}
-
-
-LOCAL(void)
-emit_dac (j_compress_ptr cinfo)
-/* Emit a DAC marker */
-/* Since the useful info is so small, we want to emit all the tables in */
-/* one DAC marker.  Therefore this routine does its own scan of the table. */
-{
-#ifdef C_ARITH_CODING_SUPPORTED
-  char dc_in_use[NUM_ARITH_TBLS];
-  char ac_in_use[NUM_ARITH_TBLS];
-  int length, i;
-  jpeg_component_info *compptr;
-  
-  for (i = 0; i < NUM_ARITH_TBLS; i++)
-    dc_in_use[i] = ac_in_use[i] = 0;
-  
-  for (i = 0; i < cinfo->comps_in_scan; i++) {
-    compptr = cinfo->cur_comp_info[i];
-    dc_in_use[compptr->dc_tbl_no] = 1;
-    ac_in_use[compptr->ac_tbl_no] = 1;
-  }
-  
-  length = 0;
-  for (i = 0; i < NUM_ARITH_TBLS; i++)
-    length += dc_in_use[i] + ac_in_use[i];
-  
-  emit_marker(cinfo, M_DAC);
-  
-  emit_2bytes(cinfo, length*2 + 2);
-  
-  for (i = 0; i < NUM_ARITH_TBLS; i++) {
-    if (dc_in_use[i]) {
-      emit_byte(cinfo, i);
-      emit_byte(cinfo, cinfo->arith_dc_L[i] + (cinfo->arith_dc_U[i]<<4));
-    }
-    if (ac_in_use[i]) {
-      emit_byte(cinfo, i + 0x10);
-      emit_byte(cinfo, cinfo->arith_ac_K[i]);
-    }
-  }
-#endif /* C_ARITH_CODING_SUPPORTED */
-}
-
-
-LOCAL(void)
-emit_dri (j_compress_ptr cinfo)
-/* Emit a DRI marker */
-{
-  emit_marker(cinfo, M_DRI);
-  
-  emit_2bytes(cinfo, 4);	/* fixed length */
-
-  emit_2bytes(cinfo, (int) cinfo->restart_interval);
-}
-
-
-LOCAL(void)
-emit_sof (j_compress_ptr cinfo, JPEG_MARKER code)
-/* Emit a SOF marker */
-{
-  int ci;
-  jpeg_component_info *compptr;
-  
-  emit_marker(cinfo, code);
-  
-  emit_2bytes(cinfo, 3 * cinfo->num_components + 2 + 5 + 1); /* length */
-
-  /* Make sure image isn't bigger than SOF field can handle */
-  if ((long) cinfo->image_height > 65535L ||
-      (long) cinfo->image_width > 65535L)
-    ERREXIT1(cinfo, JERR_IMAGE_TOO_BIG, (unsigned int) 65535);
-
-  emit_byte(cinfo, cinfo->data_precision);
-  emit_2bytes(cinfo, (int) cinfo->image_height);
-  emit_2bytes(cinfo, (int) cinfo->image_width);
-
-  emit_byte(cinfo, cinfo->num_components);
-
-  for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
-       ci++, compptr++) {
-    emit_byte(cinfo, compptr->component_id);
-    emit_byte(cinfo, (compptr->h_samp_factor << 4) + compptr->v_samp_factor);
-    emit_byte(cinfo, compptr->quant_tbl_no);
-  }
-}
-
-
-LOCAL(void)
-emit_sos (j_compress_ptr cinfo)
-/* Emit a SOS marker */
-{
-  int i, td, ta;
-  jpeg_component_info *compptr;
-  
-  emit_marker(cinfo, M_SOS);
-  
-  emit_2bytes(cinfo, 2 * cinfo->comps_in_scan + 2 + 1 + 3); /* length */
-  
-  emit_byte(cinfo, cinfo->comps_in_scan);
-  
-  for (i = 0; i < cinfo->comps_in_scan; i++) {
-    compptr = cinfo->cur_comp_info[i];
-    emit_byte(cinfo, compptr->component_id);
-    td = compptr->dc_tbl_no;
-    ta = compptr->ac_tbl_no;
-    if (cinfo->progressive_mode) {
-      /* Progressive mode: only DC or only AC tables are used in one scan;
-       * furthermore, Huffman coding of DC refinement uses no table at all.
-       * We emit 0 for unused field(s); this is recommended by the P&M text
-       * but does not seem to be specified in the standard.
-       */
-      if (cinfo->Ss == 0) {
-	ta = 0;			/* DC scan */
-	if (cinfo->Ah != 0 && !cinfo->arith_code)
-	  td = 0;		/* no DC table either */
-      } else {
-	td = 0;			/* AC scan */
-      }
-    }
-    emit_byte(cinfo, (td << 4) + ta);
-  }
-
-  emit_byte(cinfo, cinfo->Ss);
-  emit_byte(cinfo, cinfo->Se);
-  emit_byte(cinfo, (cinfo->Ah << 4) + cinfo->Al);
-}
-
-
-LOCAL(void)
-emit_jfif_app0 (j_compress_ptr cinfo)
-/* Emit a JFIF-compliant APP0 marker */
-{
-  /*
-   * Length of APP0 block	(2 bytes)
-   * Block ID			(4 bytes - ASCII "JFIF")
-   * Zero byte			(1 byte to terminate the ID string)
-   * Version Major, Minor	(2 bytes - major first)
-   * Units			(1 byte - 0x00 = none, 0x01 = inch, 0x02 = cm)
-   * Xdpu			(2 bytes - dots per unit horizontal)
-   * Ydpu			(2 bytes - dots per unit vertical)
-   * Thumbnail X size		(1 byte)
-   * Thumbnail Y size		(1 byte)
-   */
-  
-  emit_marker(cinfo, M_APP0);
-  
-  emit_2bytes(cinfo, 2 + 4 + 1 + 2 + 1 + 2 + 2 + 1 + 1); /* length */
-
-  emit_byte(cinfo, 0x4A);	/* Identifier: ASCII "JFIF" */
-  emit_byte(cinfo, 0x46);
-  emit_byte(cinfo, 0x49);
-  emit_byte(cinfo, 0x46);
-  emit_byte(cinfo, 0);
-  emit_byte(cinfo, cinfo->JFIF_major_version); /* Version fields */
-  emit_byte(cinfo, cinfo->JFIF_minor_version);
-  emit_byte(cinfo, cinfo->density_unit); /* Pixel size information */
-  emit_2bytes(cinfo, (int) cinfo->X_density);
-  emit_2bytes(cinfo, (int) cinfo->Y_density);
-  emit_byte(cinfo, 0);		/* No thumbnail image */
-  emit_byte(cinfo, 0);
-}
-
-
-LOCAL(void)
-emit_adobe_app14 (j_compress_ptr cinfo)
-/* Emit an Adobe APP14 marker */
-{
-  /*
-   * Length of APP14 block	(2 bytes)
-   * Block ID			(5 bytes - ASCII "Adobe")
-   * Version Number		(2 bytes - currently 100)
-   * Flags0			(2 bytes - currently 0)
-   * Flags1			(2 bytes - currently 0)
-   * Color transform		(1 byte)
-   *
-   * Although Adobe TN 5116 mentions Version = 101, all the Adobe files
-   * now in circulation seem to use Version = 100, so that's what we write.
-   *
-   * We write the color transform byte as 1 if the JPEG color space is
-   * YCbCr, 2 if it's YCCK, 0 otherwise.  Adobe's definition has to do with
-   * whether the encoder performed a transformation, which is pretty useless.
-   */
-  
-  emit_marker(cinfo, M_APP14);
-  
-  emit_2bytes(cinfo, 2 + 5 + 2 + 2 + 2 + 1); /* length */
-
-  emit_byte(cinfo, 0x41);	/* Identifier: ASCII "Adobe" */
-  emit_byte(cinfo, 0x64);
-  emit_byte(cinfo, 0x6F);
-  emit_byte(cinfo, 0x62);
-  emit_byte(cinfo, 0x65);
-  emit_2bytes(cinfo, 100);	/* Version */
-  emit_2bytes(cinfo, 0);	/* Flags0 */
-  emit_2bytes(cinfo, 0);	/* Flags1 */
-  switch (cinfo->jpeg_color_space) {
-  case JCS_YCbCr:
-    emit_byte(cinfo, 1);	/* Color transform = 1 */
-    break;
-  case JCS_YCCK:
-    emit_byte(cinfo, 2);	/* Color transform = 2 */
-    break;
-  default:
-    emit_byte(cinfo, 0);	/* Color transform = 0 */
-    break;
-  }
-}
-
-
-/*
- * These routines allow writing an arbitrary marker with parameters.
- * The only intended use is to emit COM or APPn markers after calling
- * write_file_header and before calling write_frame_header.
- * Other uses are not guaranteed to produce desirable results.
- * Counting the parameter bytes properly is the caller's responsibility.
- */
-
-METHODDEF(void)
-write_marker_header (j_compress_ptr cinfo, int marker, unsigned int datalen)
-/* Emit an arbitrary marker header */
-{
-  if (datalen > (unsigned int) 65533)		/* safety check */
-    ERREXIT(cinfo, JERR_BAD_LENGTH);
-
-  emit_marker(cinfo, (JPEG_MARKER) marker);
-
-  emit_2bytes(cinfo, (int) (datalen + 2));	/* total length */
-}
-
-METHODDEF(void)
-write_marker_byte (j_compress_ptr cinfo, int val)
-/* Emit one byte of marker parameters following write_marker_header */
-{
-  emit_byte(cinfo, val);
-}
-
-
-/*
- * Write datastream header.
- * This consists of an SOI and optional APPn markers.
- * We recommend use of the JFIF marker, but not the Adobe marker,
- * when using YCbCr or grayscale data.  The JFIF marker should NOT
- * be used for any other JPEG colorspace.  The Adobe marker is helpful
- * to distinguish RGB, CMYK, and YCCK colorspaces.
- * Note that an application can write additional header markers after
- * jpeg_start_compress returns.
- */
-
-METHODDEF(void)
-write_file_header (j_compress_ptr cinfo)
-{
-  my_marker_ptr marker = (my_marker_ptr) cinfo->marker;
-
-  emit_marker(cinfo, M_SOI);	/* first the SOI */
-
-  /* SOI is defined to reset restart interval to 0 */
-  marker->last_restart_interval = 0;
-
-  if (cinfo->write_JFIF_header)	/* next an optional JFIF APP0 */
-    emit_jfif_app0(cinfo);
-  if (cinfo->write_Adobe_marker) /* next an optional Adobe APP14 */
-    emit_adobe_app14(cinfo);
-}
-
-
-/*
- * Write frame header.
- * This consists of DQT and SOFn markers.
- * Note that we do not emit the SOF until we have emitted the DQT(s).
- * This avoids compatibility problems with incorrect implementations that
- * try to error-check the quant table numbers as soon as they see the SOF.
- */
-
-METHODDEF(void)
-write_frame_header (j_compress_ptr cinfo)
-{
-  int ci, prec;
-  boolean is_baseline;
-  jpeg_component_info *compptr;
-  
-  /* Emit DQT for each quantization table.
-   * Note that emit_dqt() suppresses any duplicate tables.
-   */
-  prec = 0;
-  for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
-       ci++, compptr++) {
-    prec += emit_dqt(cinfo, compptr->quant_tbl_no);
-  }
-  /* now prec is nonzero iff there are any 16-bit quant tables. */
-
-  /* Check for a non-baseline specification.
-   * Note we assume that Huffman table numbers won't be changed later.
-   */
-  if (cinfo->arith_code || cinfo->progressive_mode ||
-      cinfo->data_precision != 8) {
-    is_baseline = FALSE;
-  } else {
-    is_baseline = TRUE;
-    for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
-	 ci++, compptr++) {
-      if (compptr->dc_tbl_no > 1 || compptr->ac_tbl_no > 1)
-	is_baseline = FALSE;
-    }
-    if (prec && is_baseline) {
-      is_baseline = FALSE;
-      /* If it's baseline except for quantizer size, warn the user */
-      TRACEMS(cinfo, 0, JTRC_16BIT_TABLES);
-    }
-  }
-
-  /* Emit the proper SOF marker */
-  if (cinfo->arith_code) {
-    emit_sof(cinfo, M_SOF9);	/* SOF code for arithmetic coding */
-  } else {
-    if (cinfo->progressive_mode)
-      emit_sof(cinfo, M_SOF2);	/* SOF code for progressive Huffman */
-    else if (is_baseline)
-      emit_sof(cinfo, M_SOF0);	/* SOF code for baseline implementation */
-    else
-      emit_sof(cinfo, M_SOF1);	/* SOF code for non-baseline Huffman file */
-  }
-}
-
-
-/*
- * Write scan header.
- * This consists of DHT or DAC markers, optional DRI, and SOS.
- * Compressed data will be written following the SOS.
- */
-
-METHODDEF(void)
-write_scan_header (j_compress_ptr cinfo)
-{
-  my_marker_ptr marker = (my_marker_ptr) cinfo->marker;
-  int i;
-  jpeg_component_info *compptr;
-
-  if (cinfo->arith_code) {
-    /* Emit arith conditioning info.  We may have some duplication
-     * if the file has multiple scans, but it's so small it's hardly
-     * worth worrying about.
-     */
-    emit_dac(cinfo);
-  } else {
-    /* Emit Huffman tables.
-     * Note that emit_dht() suppresses any duplicate tables.
-     */
-    for (i = 0; i < cinfo->comps_in_scan; i++) {
-      compptr = cinfo->cur_comp_info[i];
-      if (cinfo->progressive_mode) {
-	/* Progressive mode: only DC or only AC tables are used in one scan */
-	if (cinfo->Ss == 0) {
-	  if (cinfo->Ah == 0)	/* DC needs no table for refinement scan */
-	    emit_dht(cinfo, compptr->dc_tbl_no, FALSE);
-	} else {
-	  emit_dht(cinfo, compptr->ac_tbl_no, TRUE);
-	}
-      } else {
-	/* Sequential mode: need both DC and AC tables */
-	emit_dht(cinfo, compptr->dc_tbl_no, FALSE);
-	emit_dht(cinfo, compptr->ac_tbl_no, TRUE);
-      }
-    }
-  }
-
-  /* Emit DRI if required --- note that DRI value could change for each scan.
-   * We avoid wasting space with unnecessary DRIs, however.
-   */
-  if (cinfo->restart_interval != marker->last_restart_interval) {
-    emit_dri(cinfo);
-    marker->last_restart_interval = cinfo->restart_interval;
-  }
-
-  emit_sos(cinfo);
-}
-
-
-/*
- * Write datastream trailer.
- */
-
-METHODDEF(void)
-write_file_trailer (j_compress_ptr cinfo)
-{
-  emit_marker(cinfo, M_EOI);
-}
-
-
-/*
- * Write an abbreviated table-specification datastream.
- * This consists of SOI, DQT and DHT tables, and EOI.
- * Any table that is defined and not marked sent_table = TRUE will be
- * emitted.  Note that all tables will be marked sent_table = TRUE at exit.
- */
-
-METHODDEF(void)
-write_tables_only (j_compress_ptr cinfo)
-{
-  int i;
-
-  emit_marker(cinfo, M_SOI);
-
-  for (i = 0; i < NUM_QUANT_TBLS; i++) {
-    if (cinfo->quant_tbl_ptrs[i] != NULL)
-      (void) emit_dqt(cinfo, i);
-  }
-
-  if (! cinfo->arith_code) {
-    for (i = 0; i < NUM_HUFF_TBLS; i++) {
-      if (cinfo->dc_huff_tbl_ptrs[i] != NULL)
-	emit_dht(cinfo, i, FALSE);
-      if (cinfo->ac_huff_tbl_ptrs[i] != NULL)
-	emit_dht(cinfo, i, TRUE);
-    }
-  }
-
-  emit_marker(cinfo, M_EOI);
-}
-
-
-/*
- * Initialize the marker writer module.
- */
-
-GLOBAL(void)
-jinit_marker_writer (j_compress_ptr cinfo)
-{
-  my_marker_ptr marker;
-
-  /* Create the subobject */
-  marker = (my_marker_ptr)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				SIZEOF(my_marker_writer));
-  cinfo->marker = (struct jpeg_marker_writer *) marker;
-  /* Initialize method pointers */
-  marker->pub.write_file_header = write_file_header;
-  marker->pub.write_frame_header = write_frame_header;
-  marker->pub.write_scan_header = write_scan_header;
-  marker->pub.write_file_trailer = write_file_trailer;
-  marker->pub.write_tables_only = write_tables_only;
-  marker->pub.write_marker_header = write_marker_header;
-  marker->pub.write_marker_byte = write_marker_byte;
-  /* Initialize private state */
-  marker->last_restart_interval = 0;
-}
@@ -1,590 +0,0 @@
-/*
- * jcmaster.c
- *
- * Copyright (C) 1991-1997, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains master control logic for the JPEG compressor.
- * These routines are concerned with parameter validation, initial setup,
- * and inter-pass control (determining the number of passes and the work 
- * to be done in each pass).
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-
-
-/* Private state */
-
-typedef enum {
-	main_pass,		/* input data, also do first output step */
-	huff_opt_pass,		/* Huffman code optimization pass */
-	output_pass		/* data output pass */
-} c_pass_type;
-
-typedef struct {
-  struct jpeg_comp_master pub;	/* public fields */
-
-  c_pass_type pass_type;	/* the type of the current pass */
-
-  int pass_number;		/* # of passes completed */
-  int total_passes;		/* total # of passes needed */
-
-  int scan_number;		/* current index in scan_info[] */
-} my_comp_master;
-
-typedef my_comp_master * my_master_ptr;
-
-
-/*
- * Support routines that do various essential calculations.
- */
-
-LOCAL(void)
-initial_setup (j_compress_ptr cinfo)
-/* Do computations that are needed before master selection phase */
-{
-  int ci;
-  jpeg_component_info *compptr;
-  long samplesperrow;
-  JDIMENSION jd_samplesperrow;
-
-  /* Sanity check on image dimensions */
-  if (cinfo->image_height <= 0 || cinfo->image_width <= 0
-      || cinfo->num_components <= 0 || cinfo->input_components <= 0)
-    ERREXIT(cinfo, JERR_EMPTY_IMAGE);
-
-  /* Make sure image isn't bigger than I can handle */
-  if ((long) cinfo->image_height > (long) JPEG_MAX_DIMENSION ||
-      (long) cinfo->image_width > (long) JPEG_MAX_DIMENSION)
-    ERREXIT1(cinfo, JERR_IMAGE_TOO_BIG, (unsigned int) JPEG_MAX_DIMENSION);
-
-  /* Width of an input scanline must be representable as JDIMENSION. */
-  samplesperrow = (long) cinfo->image_width * (long) cinfo->input_components;
-  jd_samplesperrow = (JDIMENSION) samplesperrow;
-  if ((long) jd_samplesperrow != samplesperrow)
-    ERREXIT(cinfo, JERR_WIDTH_OVERFLOW);
-
-  /* For now, precision must match compiled-in value... */
-  if (cinfo->data_precision != BITS_IN_JSAMPLE)
-    ERREXIT1(cinfo, JERR_BAD_PRECISION, cinfo->data_precision);
-
-  /* Check that number of components won't exceed internal array sizes */
-  if (cinfo->num_components > MAX_COMPONENTS)
-    ERREXIT2(cinfo, JERR_COMPONENT_COUNT, cinfo->num_components,
-	     MAX_COMPONENTS);
-
-  /* Compute maximum sampling factors; check factor validity */
-  cinfo->max_h_samp_factor = 1;
-  cinfo->max_v_samp_factor = 1;
-  for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
-       ci++, compptr++) {
-    if (compptr->h_samp_factor<=0 || compptr->h_samp_factor>MAX_SAMP_FACTOR ||
-	compptr->v_samp_factor<=0 || compptr->v_samp_factor>MAX_SAMP_FACTOR)
-      ERREXIT(cinfo, JERR_BAD_SAMPLING);
-    cinfo->max_h_samp_factor = MAX(cinfo->max_h_samp_factor,
-				   compptr->h_samp_factor);
-    cinfo->max_v_samp_factor = MAX(cinfo->max_v_samp_factor,
-				   compptr->v_samp_factor);
-  }
-
-  /* Compute dimensions of components */
-  for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
-       ci++, compptr++) {
-    /* Fill in the correct component_index value; don't rely on application */
-    compptr->component_index = ci;
-    /* For compression, we never do DCT scaling. */
-    compptr->DCT_scaled_size = DCTSIZE;
-    /* Size in DCT blocks */
-    compptr->width_in_blocks = (JDIMENSION)
-      jdiv_round_up((long) cinfo->image_width * (long) compptr->h_samp_factor,
-		    (long) (cinfo->max_h_samp_factor * DCTSIZE));
-    compptr->height_in_blocks = (JDIMENSION)
-      jdiv_round_up((long) cinfo->image_height * (long) compptr->v_samp_factor,
-		    (long) (cinfo->max_v_samp_factor * DCTSIZE));
-    /* Size in samples */
-    compptr->downsampled_width = (JDIMENSION)
-      jdiv_round_up((long) cinfo->image_width * (long) compptr->h_samp_factor,
-		    (long) cinfo->max_h_samp_factor);
-    compptr->downsampled_height = (JDIMENSION)
-      jdiv_round_up((long) cinfo->image_height * (long) compptr->v_samp_factor,
-		    (long) cinfo->max_v_samp_factor);
-    /* Mark component needed (this flag isn't actually used for compression) */
-    compptr->component_needed = TRUE;
-  }
-
-  /* Compute number of fully interleaved MCU rows (number of times that
-   * main controller will call coefficient controller).
-   */
-  cinfo->total_iMCU_rows = (JDIMENSION)
-    jdiv_round_up((long) cinfo->image_height,
-		  (long) (cinfo->max_v_samp_factor*DCTSIZE));
-}
-
-
-#ifdef C_MULTISCAN_FILES_SUPPORTED
-
-LOCAL(void)
-validate_script (j_compress_ptr cinfo)
-/* Verify that the scan script in cinfo->scan_info[] is valid; also
- * determine whether it uses progressive JPEG, and set cinfo->progressive_mode.
- */
-{
-  const jpeg_scan_info * scanptr;
-  int scanno, ncomps, ci, coefi, thisi;
-  int Ss, Se, Ah, Al;
-  boolean component_sent[MAX_COMPONENTS];
-#ifdef C_PROGRESSIVE_SUPPORTED
-  int * last_bitpos_ptr;
-  int last_bitpos[MAX_COMPONENTS][DCTSIZE2];
-  /* -1 until that coefficient has been seen; then last Al for it */
-#endif
-
-  if (cinfo->num_scans <= 0)
-    ERREXIT1(cinfo, JERR_BAD_SCAN_SCRIPT, 0);
-
-  /* For sequential JPEG, all scans must have Ss=0, Se=DCTSIZE2-1;
-   * for progressive JPEG, no scan can have this.
-   */
-  scanptr = cinfo->scan_info;
-  if (scanptr->Ss != 0 || scanptr->Se != DCTSIZE2-1) {
-#ifdef C_PROGRESSIVE_SUPPORTED
-    cinfo->progressive_mode = TRUE;
-    last_bitpos_ptr = & last_bitpos[0][0];
-    for (ci = 0; ci < cinfo->num_components; ci++) 
-      for (coefi = 0; coefi < DCTSIZE2; coefi++)
-	*last_bitpos_ptr++ = -1;
-#else
-    ERREXIT(cinfo, JERR_NOT_COMPILED);
-#endif
-  } else {
-    cinfo->progressive_mode = FALSE;
-    for (ci = 0; ci < cinfo->num_components; ci++) 
-      component_sent[ci] = FALSE;
-  }
-
-  for (scanno = 1; scanno <= cinfo->num_scans; scanptr++, scanno++) {
-    /* Validate component indexes */
-    ncomps = scanptr->comps_in_scan;
-    if (ncomps <= 0 || ncomps > MAX_COMPS_IN_SCAN)
-      ERREXIT2(cinfo, JERR_COMPONENT_COUNT, ncomps, MAX_COMPS_IN_SCAN);
-    for (ci = 0; ci < ncomps; ci++) {
-      thisi = scanptr->component_index[ci];
-      if (thisi < 0 || thisi >= cinfo->num_components)
-	ERREXIT1(cinfo, JERR_BAD_SCAN_SCRIPT, scanno);
-      /* Components must appear in SOF order within each scan */
-      if (ci > 0 && thisi <= scanptr->component_index[ci-1])
-	ERREXIT1(cinfo, JERR_BAD_SCAN_SCRIPT, scanno);
-    }
-    /* Validate progression parameters */
-    Ss = scanptr->Ss;
-    Se = scanptr->Se;
-    Ah = scanptr->Ah;
-    Al = scanptr->Al;
-    if (cinfo->progressive_mode) {
-#ifdef C_PROGRESSIVE_SUPPORTED
-      /* The JPEG spec simply gives the ranges 0..13 for Ah and Al, but that
-       * seems wrong: the upper bound ought to depend on data precision.
-       * Perhaps they really meant 0..N+1 for N-bit precision.
-       * Here we allow 0..10 for 8-bit data; Al larger than 10 results in
-       * out-of-range reconstructed DC values during the first DC scan,
-       * which might cause problems for some decoders.
-       */
-#if BITS_IN_JSAMPLE == 8
-#define MAX_AH_AL 10
-#else
-#define MAX_AH_AL 13
-#endif
-      if (Ss < 0 || Ss >= DCTSIZE2 || Se < Ss || Se >= DCTSIZE2 ||
-	  Ah < 0 || Ah > MAX_AH_AL || Al < 0 || Al > MAX_AH_AL)
-	ERREXIT1(cinfo, JERR_BAD_PROG_SCRIPT, scanno);
-      if (Ss == 0) {
-	if (Se != 0)		/* DC and AC together not OK */
-	  ERREXIT1(cinfo, JERR_BAD_PROG_SCRIPT, scanno);
-      } else {
-	if (ncomps != 1)	/* AC scans must be for only one component */
-	  ERREXIT1(cinfo, JERR_BAD_PROG_SCRIPT, scanno);
-      }
-      for (ci = 0; ci < ncomps; ci++) {
-	last_bitpos_ptr = & last_bitpos[scanptr->component_index[ci]][0];
-	if (Ss != 0 && last_bitpos_ptr[0] < 0) /* AC without prior DC scan */
-	  ERREXIT1(cinfo, JERR_BAD_PROG_SCRIPT, scanno);
-	for (coefi = Ss; coefi <= Se; coefi++) {
-	  if (last_bitpos_ptr[coefi] < 0) {
-	    /* first scan of this coefficient */
-	    if (Ah != 0)
-	      ERREXIT1(cinfo, JERR_BAD_PROG_SCRIPT, scanno);
-	  } else {
-	    /* not first scan */
-	    if (Ah != last_bitpos_ptr[coefi] || Al != Ah-1)
-	      ERREXIT1(cinfo, JERR_BAD_PROG_SCRIPT, scanno);
-	  }
-	  last_bitpos_ptr[coefi] = Al;
-	}
-      }
-#endif
-    } else {
-      /* For sequential JPEG, all progression parameters must be these: */
-      if (Ss != 0 || Se != DCTSIZE2-1 || Ah != 0 || Al != 0)
-	ERREXIT1(cinfo, JERR_BAD_PROG_SCRIPT, scanno);
-      /* Make sure components are not sent twice */
-      for (ci = 0; ci < ncomps; ci++) {
-	thisi = scanptr->component_index[ci];
-	if (component_sent[thisi])
-	  ERREXIT1(cinfo, JERR_BAD_SCAN_SCRIPT, scanno);
-	component_sent[thisi] = TRUE;
-      }
-    }
-  }
-
-  /* Now verify that everything got sent. */
-  if (cinfo->progressive_mode) {
-#ifdef C_PROGRESSIVE_SUPPORTED
-    /* For progressive mode, we only check that at least some DC data
-     * got sent for each component; the spec does not require that all bits
-     * of all coefficients be transmitted.  Would it be wiser to enforce
-     * transmission of all coefficient bits??
-     */
-    for (ci = 0; ci < cinfo->num_components; ci++) {
-      if (last_bitpos[ci][0] < 0)
-	ERREXIT(cinfo, JERR_MISSING_DATA);
-    }
-#endif
-  } else {
-    for (ci = 0; ci < cinfo->num_components; ci++) {
-      if (! component_sent[ci])
-	ERREXIT(cinfo, JERR_MISSING_DATA);
-    }
-  }
-}
-
-#endif /* C_MULTISCAN_FILES_SUPPORTED */
-
-
-LOCAL(void)
-select_scan_parameters (j_compress_ptr cinfo)
-/* Set up the scan parameters for the current scan */
-{
-  int ci;
-
-#ifdef C_MULTISCAN_FILES_SUPPORTED
-  if (cinfo->scan_info != NULL) {
-    /* Prepare for current scan --- the script is already validated */
-    my_master_ptr master = (my_master_ptr) cinfo->master;
-    const jpeg_scan_info * scanptr = cinfo->scan_info + master->scan_number;
-
-    cinfo->comps_in_scan = scanptr->comps_in_scan;
-    for (ci = 0; ci < scanptr->comps_in_scan; ci++) {
-      cinfo->cur_comp_info[ci] =
-	&cinfo->comp_info[scanptr->component_index[ci]];
-    }
-    cinfo->Ss = scanptr->Ss;
-    cinfo->Se = scanptr->Se;
-    cinfo->Ah = scanptr->Ah;
-    cinfo->Al = scanptr->Al;
-  }
-  else
-#endif
-  {
-    /* Prepare for single sequential-JPEG scan containing all components */
-    if (cinfo->num_components > MAX_COMPS_IN_SCAN)
-      ERREXIT2(cinfo, JERR_COMPONENT_COUNT, cinfo->num_components,
-	       MAX_COMPS_IN_SCAN);
-    cinfo->comps_in_scan = cinfo->num_components;
-    for (ci = 0; ci < cinfo->num_components; ci++) {
-      cinfo->cur_comp_info[ci] = &cinfo->comp_info[ci];
-    }
-    cinfo->Ss = 0;
-    cinfo->Se = DCTSIZE2-1;
-    cinfo->Ah = 0;
-    cinfo->Al = 0;
-  }
-}
-
-
-LOCAL(void)
-per_scan_setup (j_compress_ptr cinfo)
-/* Do computations that are needed before processing a JPEG scan */
-/* cinfo->comps_in_scan and cinfo->cur_comp_info[] are already set */
-{
-  int ci, mcublks, tmp;
-  jpeg_component_info *compptr;
-  
-  if (cinfo->comps_in_scan == 1) {
-    
-    /* Noninterleaved (single-component) scan */
-    compptr = cinfo->cur_comp_info[0];
-    
-    /* Overall image size in MCUs */
-    cinfo->MCUs_per_row = compptr->width_in_blocks;
-    cinfo->MCU_rows_in_scan = compptr->height_in_blocks;
-    
-    /* For noninterleaved scan, always one block per MCU */
-    compptr->MCU_width = 1;
-    compptr->MCU_height = 1;
-    compptr->MCU_blocks = 1;
-    compptr->MCU_sample_width = DCTSIZE;
-    compptr->last_col_width = 1;
-    /* For noninterleaved scans, it is convenient to define last_row_height
-     * as the number of block rows present in the last iMCU row.
-     */
-    tmp = (int) (compptr->height_in_blocks % compptr->v_samp_factor);
-    if (tmp == 0) tmp = compptr->v_samp_factor;
-    compptr->last_row_height = tmp;
-    
-    /* Prepare array describing MCU composition */
-    cinfo->blocks_in_MCU = 1;
-    cinfo->MCU_membership[0] = 0;
-    
-  } else {
-    
-    /* Interleaved (multi-component) scan */
-    if (cinfo->comps_in_scan <= 0 || cinfo->comps_in_scan > MAX_COMPS_IN_SCAN)
-      ERREXIT2(cinfo, JERR_COMPONENT_COUNT, cinfo->comps_in_scan,
-	       MAX_COMPS_IN_SCAN);
-    
-    /* Overall image size in MCUs */
-    cinfo->MCUs_per_row = (JDIMENSION)
-      jdiv_round_up((long) cinfo->image_width,
-		    (long) (cinfo->max_h_samp_factor*DCTSIZE));
-    cinfo->MCU_rows_in_scan = (JDIMENSION)
-      jdiv_round_up((long) cinfo->image_height,
-		    (long) (cinfo->max_v_samp_factor*DCTSIZE));
-    
-    cinfo->blocks_in_MCU = 0;
-    
-    for (ci = 0; ci < cinfo->comps_in_scan; ci++) {
-      compptr = cinfo->cur_comp_info[ci];
-      /* Sampling factors give # of blocks of component in each MCU */
-      compptr->MCU_width = compptr->h_samp_factor;
-      compptr->MCU_height = compptr->v_samp_factor;
-      compptr->MCU_blocks = compptr->MCU_width * compptr->MCU_height;
-      compptr->MCU_sample_width = compptr->MCU_width * DCTSIZE;
-      /* Figure number of non-dummy blocks in last MCU column & row */
-      tmp = (int) (compptr->width_in_blocks % compptr->MCU_width);
-      if (tmp == 0) tmp = compptr->MCU_width;
-      compptr->last_col_width = tmp;
-      tmp = (int) (compptr->height_in_blocks % compptr->MCU_height);
-      if (tmp == 0) tmp = compptr->MCU_height;
-      compptr->last_row_height = tmp;
-      /* Prepare array describing MCU composition */
-      mcublks = compptr->MCU_blocks;
-      if (cinfo->blocks_in_MCU + mcublks > C_MAX_BLOCKS_IN_MCU)
-	ERREXIT(cinfo, JERR_BAD_MCU_SIZE);
-      while (mcublks-- > 0) {
-	cinfo->MCU_membership[cinfo->blocks_in_MCU++] = ci;
-      }
-    }
-    
-  }
-
-  /* Convert restart specified in rows to actual MCU count. */
-  /* Note that count must fit in 16 bits, so we provide limiting. */
-  if (cinfo->restart_in_rows > 0) {
-    long nominal = (long) cinfo->restart_in_rows * (long) cinfo->MCUs_per_row;
-    cinfo->restart_interval = (unsigned int) MIN(nominal, 65535L);
-  }
-}
-
-
-/*
- * Per-pass setup.
- * This is called at the beginning of each pass.  We determine which modules
- * will be active during this pass and give them appropriate start_pass calls.
- * We also set is_last_pass to indicate whether any more passes will be
- * required.
- */
-
-METHODDEF(void)
-prepare_for_pass (j_compress_ptr cinfo)
-{
-  my_master_ptr master = (my_master_ptr) cinfo->master;
-
-  switch (master->pass_type) {
-  case main_pass:
-    /* Initial pass: will collect input data, and do either Huffman
-     * optimization or data output for the first scan.
-     */
-    select_scan_parameters(cinfo);
-    per_scan_setup(cinfo);
-    if (! cinfo->raw_data_in) {
-      (*cinfo->cconvert->start_pass) (cinfo);
-      (*cinfo->downsample->start_pass) (cinfo);
-      (*cinfo->prep->start_pass) (cinfo, JBUF_PASS_THRU);
-    }
-    (*cinfo->fdct->start_pass) (cinfo);
-    (*cinfo->entropy->start_pass) (cinfo, cinfo->optimize_coding);
-    (*cinfo->coef->start_pass) (cinfo,
-				(master->total_passes > 1 ?
-				 JBUF_SAVE_AND_PASS : JBUF_PASS_THRU));
-    (*cinfo->main->start_pass) (cinfo, JBUF_PASS_THRU);
-    if (cinfo->optimize_coding) {
-      /* No immediate data output; postpone writing frame/scan headers */
-      master->pub.call_pass_startup = FALSE;
-    } else {
-      /* Will write frame/scan headers at first jpeg_write_scanlines call */
-      master->pub.call_pass_startup = TRUE;
-    }
-    break;
-#ifdef ENTROPY_OPT_SUPPORTED
-  case huff_opt_pass:
-    /* Do Huffman optimization for a scan after the first one. */
-    select_scan_parameters(cinfo);
-    per_scan_setup(cinfo);
-    if (cinfo->Ss != 0 || cinfo->Ah == 0 || cinfo->arith_code) {
-      (*cinfo->entropy->start_pass) (cinfo, TRUE);
-      (*cinfo->coef->start_pass) (cinfo, JBUF_CRANK_DEST);
-      master->pub.call_pass_startup = FALSE;
-      break;
-    }
-    /* Special case: Huffman DC refinement scans need no Huffman table
-     * and therefore we can skip the optimization pass for them.
-     */
-    master->pass_type = output_pass;
-    master->pass_number++;
-    /*FALLTHROUGH*/
-#endif
-  case output_pass:
-    /* Do a data-output pass. */
-    /* We need not repeat per-scan setup if prior optimization pass did it. */
-    if (! cinfo->optimize_coding) {
-      select_scan_parameters(cinfo);
-      per_scan_setup(cinfo);
-    }
-    (*cinfo->entropy->start_pass) (cinfo, FALSE);
-    (*cinfo->coef->start_pass) (cinfo, JBUF_CRANK_DEST);
-    /* We emit frame/scan headers now */
-    if (master->scan_number == 0)
-      (*cinfo->marker->write_frame_header) (cinfo);
-    (*cinfo->marker->write_scan_header) (cinfo);
-    master->pub.call_pass_startup = FALSE;
-    break;
-  default:
-    ERREXIT(cinfo, JERR_NOT_COMPILED);
-  }
-
-  master->pub.is_last_pass = (master->pass_number == master->total_passes-1);
-
-  /* Set up progress monitor's pass info if present */
-  if (cinfo->progress != NULL) {
-    cinfo->progress->completed_passes = master->pass_number;
-    cinfo->progress->total_passes = master->total_passes;
-  }
-}
-
-
-/*
- * Special start-of-pass hook.
- * This is called by jpeg_write_scanlines if call_pass_startup is TRUE.
- * In single-pass processing, we need this hook because we don't want to
- * write frame/scan headers during jpeg_start_compress; we want to let the
- * application write COM markers etc. between jpeg_start_compress and the
- * jpeg_write_scanlines loop.
- * In multi-pass processing, this routine is not used.
- */
-
-METHODDEF(void)
-pass_startup (j_compress_ptr cinfo)
-{
-  cinfo->master->call_pass_startup = FALSE; /* reset flag so call only once */
-
-  (*cinfo->marker->write_frame_header) (cinfo);
-  (*cinfo->marker->write_scan_header) (cinfo);
-}
-
-
-/*
- * Finish up at end of pass.
- */
-
-METHODDEF(void)
-finish_pass_master (j_compress_ptr cinfo)
-{
-  my_master_ptr master = (my_master_ptr) cinfo->master;
-
-  /* The entropy coder always needs an end-of-pass call,
-   * either to analyze statistics or to flush its output buffer.
-   */
-  (*cinfo->entropy->finish_pass) (cinfo);
-
-  /* Update state for next pass */
-  switch (master->pass_type) {
-  case main_pass:
-    /* next pass is either output of scan 0 (after optimization)
-     * or output of scan 1 (if no optimization).
-     */
-    master->pass_type = output_pass;
-    if (! cinfo->optimize_coding)
-      master->scan_number++;
-    break;
-  case huff_opt_pass:
-    /* next pass is always output of current scan */
-    master->pass_type = output_pass;
-    break;
-  case output_pass:
-    /* next pass is either optimization or output of next scan */
-    if (cinfo->optimize_coding)
-      master->pass_type = huff_opt_pass;
-    master->scan_number++;
-    break;
-  }
-
-  master->pass_number++;
-}
-
-
-/*
- * Initialize master compression control.
- */
-
-GLOBAL(void)
-jinit_c_master_control (j_compress_ptr cinfo, boolean transcode_only)
-{
-  my_master_ptr master;
-
-  master = (my_master_ptr)
-      (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				  SIZEOF(my_comp_master));
-  cinfo->master = (struct jpeg_comp_master *) master;
-  master->pub.prepare_for_pass = prepare_for_pass;
-  master->pub.pass_startup = pass_startup;
-  master->pub.finish_pass = finish_pass_master;
-  master->pub.is_last_pass = FALSE;
-
-  /* Validate parameters, determine derived values */
-  initial_setup(cinfo);
-
-  if (cinfo->scan_info != NULL) {
-#ifdef C_MULTISCAN_FILES_SUPPORTED
-    validate_script(cinfo);
-#else
-    ERREXIT(cinfo, JERR_NOT_COMPILED);
-#endif
-  } else {
-    cinfo->progressive_mode = FALSE;
-    cinfo->num_scans = 1;
-  }
-
-  if (cinfo->progressive_mode)	/*  TEMPORARY HACK ??? */
-    cinfo->optimize_coding = TRUE; /* assume default tables no good for progressive mode */
-
-  /* Initialize my private state */
-  if (transcode_only) {
-    /* no main pass in transcoding */
-    if (cinfo->optimize_coding)
-      master->pass_type = huff_opt_pass;
-    else
-      master->pass_type = output_pass;
-  } else {
-    /* for normal compression, first pass is always this type: */
-    master->pass_type = main_pass;
-  }
-  master->scan_number = 0;
-  master->pass_number = 0;
-  if (cinfo->optimize_coding)
-    master->total_passes = cinfo->num_scans * 2;
-  else
-    master->total_passes = cinfo->num_scans;
-}
@@ -1,106 +0,0 @@
-/*
- * jcomapi.c
- *
- * Copyright (C) 1994-1997, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains application interface routines that are used for both
- * compression and decompression.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-
-
-/*
- * Abort processing of a JPEG compression or decompression operation,
- * but don't destroy the object itself.
- *
- * For this, we merely clean up all the nonpermanent memory pools.
- * Note that temp files (virtual arrays) are not allowed to belong to
- * the permanent pool, so we will be able to close all temp files here.
- * Closing a data source or destination, if necessary, is the application's
- * responsibility.
- */
-
-GLOBAL(void)
-jpeg_abort (j_common_ptr cinfo)
-{
-  int pool;
-
-  /* Do nothing if called on a not-initialized or destroyed JPEG object. */
-  if (cinfo->mem == NULL)
-    return;
-
-  /* Releasing pools in reverse order might help avoid fragmentation
-   * with some (brain-damaged) malloc libraries.
-   */
-  for (pool = JPOOL_NUMPOOLS-1; pool > JPOOL_PERMANENT; pool--) {
-    (*cinfo->mem->free_pool) (cinfo, pool);
-  }
-
-  /* Reset overall state for possible reuse of object */
-  if (cinfo->is_decompressor) {
-    cinfo->global_state = DSTATE_START;
-    /* Try to keep application from accessing now-deleted marker list.
-     * A bit kludgy to do it here, but this is the most central place.
-     */
-    ((j_decompress_ptr) cinfo)->marker_list = NULL;
-  } else {
-    cinfo->global_state = CSTATE_START;
-  }
-}
-
-
-/*
- * Destruction of a JPEG object.
- *
- * Everything gets deallocated except the master jpeg_compress_struct itself
- * and the error manager struct.  Both of these are supplied by the application
- * and must be freed, if necessary, by the application.  (Often they are on
- * the stack and so don't need to be freed anyway.)
- * Closing a data source or destination, if necessary, is the application's
- * responsibility.
- */
-
-GLOBAL(void)
-jpeg_destroy (j_common_ptr cinfo)
-{
-  /* We need only tell the memory manager to release everything. */
-  /* NB: mem pointer is NULL if memory mgr failed to initialize. */
-  if (cinfo->mem != NULL)
-    (*cinfo->mem->self_destruct) (cinfo);
-  cinfo->mem = NULL;		/* be safe if jpeg_destroy is called twice */
-  cinfo->global_state = 0;	/* mark it destroyed */
-}
-
-
-/*
- * Convenience routines for allocating quantization and Huffman tables.
- * (Would jutils.c be a more reasonable place to put these?)
- */
-
-GLOBAL(JQUANT_TBL *)
-jpeg_alloc_quant_table (j_common_ptr cinfo)
-{
-  JQUANT_TBL *tbl;
-
-  tbl = (JQUANT_TBL *)
-    (*cinfo->mem->alloc_small) (cinfo, JPOOL_PERMANENT, SIZEOF(JQUANT_TBL));
-  tbl->sent_table = FALSE;	/* make sure this is false in any new table */
-  return tbl;
-}
-
-
-GLOBAL(JHUFF_TBL *)
-jpeg_alloc_huff_table (j_common_ptr cinfo)
-{
-  JHUFF_TBL *tbl;
-
-  tbl = (JHUFF_TBL *)
-    (*cinfo->mem->alloc_small) (cinfo, JPOOL_PERMANENT, SIZEOF(JHUFF_TBL));
-  tbl->sent_table = FALSE;	/* make sure this is false in any new table */
-  return tbl;
-}
@@ -1,156 +0,0 @@
-/* android jconfig.h */
-/*
- * jconfig.doc
- *
- * Copyright (C) 1991-1994, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file documents the configuration options that are required to
- * customize the JPEG software for a particular system.
- *
- * The actual configuration options for a particular installation are stored
- * in jconfig.h.  On many machines, jconfig.h can be generated automatically
- * or copied from one of the "canned" jconfig files that we supply.  But if
- * you need to generate a jconfig.h file by hand, this file tells you how.
- *
- * DO NOT EDIT THIS FILE --- IT WON'T ACCOMPLISH ANYTHING.
- * EDIT A COPY NAMED JCONFIG.H.
- */
-
-
-/*
- * These symbols indicate the properties of your machine or compiler.
- * #define the symbol if yes, #undef it if no.
- */
-
-/* Does your compiler support function prototypes?
- * (If not, you also need to use ansi2knr, see install.doc)
- */
-#define HAVE_PROTOTYPES
-
-/* Does your compiler support the declaration "unsigned char" ?
- * How about "unsigned short" ?
- */
-#define HAVE_UNSIGNED_CHAR
-#define HAVE_UNSIGNED_SHORT
-
-/* Define "void" as "char" if your compiler doesn't know about type void.
- * NOTE: be sure to define void such that "void *" represents the most general
- * pointer type, e.g., that returned by malloc().
- */
-/* #define void char */
-
-/* Define "const" as empty if your compiler doesn't know the "const" keyword.
- */
-/* #define const */
-
-/* Define this if an ordinary "char" type is unsigned.
- * If you're not sure, leaving it undefined will work at some cost in speed.
- * If you defined HAVE_UNSIGNED_CHAR then the speed difference is minimal.
- */
-#undef CHAR_IS_UNSIGNED
-
-/* Define this if your system has an ANSI-conforming <stddef.h> file.
- */
-#define HAVE_STDDEF_H
-
-/* Define this if your system has an ANSI-conforming <stdlib.h> file.
- */
-#define HAVE_STDLIB_H
-
-/* Define this if your system does not have an ANSI/SysV <string.h>,
- * but does have a BSD-style <strings.h>.
- */
-#undef NEED_BSD_STRINGS
-
-/* Define this if your system does not provide typedef size_t in any of the
- * ANSI-standard places (stddef.h, stdlib.h, or stdio.h), but places it in
- * <sys/types.h> instead.
- */
-#undef NEED_SYS_TYPES_H
-
-/* For 80x86 machines, you need to define NEED_FAR_POINTERS,
- * unless you are using a large-data memory model or 80386 flat-memory mode.
- * On less brain-damaged CPUs this symbol must not be defined.
- * (Defining this symbol causes large data structures to be referenced through
- * "far" pointers and to be allocated with a special version of malloc.)
- */
-#undef NEED_FAR_POINTERS
-
-/* Define this if your linker needs global names to be unique in less
- * than the first 15 characters.
- */
-#undef NEED_SHORT_EXTERNAL_NAMES
-
-/* Although a real ANSI C compiler can deal perfectly well with pointers to
- * unspecified structures (see "incomplete types" in the spec), a few pre-ANSI
- * and pseudo-ANSI compilers get confused.  To keep one of these bozos happy,
- * define INCOMPLETE_TYPES_BROKEN.  This is not recommended unless you
- * actually get "missing structure definition" warnings or errors while
- * compiling the JPEG code.
- */
-#undef INCOMPLETE_TYPES_BROKEN
-
-
-/*
- * The following options affect code selection within the JPEG library,
- * but they don't need to be visible to applications using the library.
- * To minimize application namespace pollution, the symbols won't be
- * defined unless JPEG_INTERNALS has been defined.
- */
-
-#ifdef JPEG_INTERNALS
-
-/* Define this if your compiler implements ">>" on signed values as a logical
- * (unsigned) shift; leave it undefined if ">>" is a signed (arithmetic) shift,
- * which is the normal and rational definition.
- */
-#undef RIGHT_SHIFT_IS_UNSIGNED
-
-
-#endif /* JPEG_INTERNALS */
-
-
-/*
- * The remaining options do not affect the JPEG library proper,
- * but only the sample applications cjpeg/djpeg (see cjpeg.c, djpeg.c).
- * Other applications can ignore these.
- */
-
-#ifdef JPEG_CJPEG_DJPEG
-
-/* These defines indicate which image (non-JPEG) file formats are allowed. */
-
-#define BMP_SUPPORTED		/* BMP image file format */
-#define GIF_SUPPORTED		/* GIF image file format */
-#define PPM_SUPPORTED		/* PBMPLUS PPM/PGM image file format */
-#undef RLE_SUPPORTED		/* Utah RLE image file format */
-#define TARGA_SUPPORTED		/* Targa image file format */
-
-/* Define this if you want to name both input and output files on the command
- * line, rather than using stdout and optionally stdin.  You MUST do this if
- * your system can't cope with binary I/O to stdin/stdout.  See comments at
- * head of cjpeg.c or djpeg.c.
- */
-#undef TWO_FILE_COMMANDLINE
-
-/* Define this if your system needs explicit cleanup of temporary files.
- * This is crucial under MS-DOS, where the temporary "files" may be areas
- * of extended memory; on most other systems it's not as important.
- */
-#undef NEED_SIGNAL_CATCHER
-
-/* By default, we open image files with fopen(...,"rb") or fopen(...,"wb").
- * This is necessary on systems that distinguish text files from binary files,
- * and is harmless on most systems that don't.  If you have one of the rare
- * systems that complains about the "b" spec, define this symbol.
- */
-#undef DONT_USE_B_MODE
-
-/* Define this if you want percent-done progress reports from cjpeg/djpeg.
- */
-#undef PROGRESS_REPORT
-
-
-#endif /* JPEG_CJPEG_DJPEG */
@@ -1,610 +0,0 @@
-/*
- * jcparam.c
- *
- * Copyright (C) 1991-1998, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains optional default-setting code for the JPEG compressor.
- * Applications do not have to use this file, but those that don't use it
- * must know a lot more about the innards of the JPEG code.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-
-
-/*
- * Quantization table setup routines
- */
-
-GLOBAL(void)
-jpeg_add_quant_table (j_compress_ptr cinfo, int which_tbl,
-		      const unsigned int *basic_table,
-		      int scale_factor, boolean force_baseline)
-/* Define a quantization table equal to the basic_table times
- * a scale factor (given as a percentage).
- * If force_baseline is TRUE, the computed quantization table entries
- * are limited to 1..255 for JPEG baseline compatibility.
- */
-{
-  JQUANT_TBL ** qtblptr;
-  int i;
-  long temp;
-
-  /* Safety check to ensure start_compress not called yet. */
-  if (cinfo->global_state != CSTATE_START)
-    ERREXIT1(cinfo, JERR_BAD_STATE, cinfo->global_state);
-
-  if (which_tbl < 0 || which_tbl >= NUM_QUANT_TBLS)
-    ERREXIT1(cinfo, JERR_DQT_INDEX, which_tbl);
-
-  qtblptr = & cinfo->quant_tbl_ptrs[which_tbl];
-
-  if (*qtblptr == NULL)
-    *qtblptr = jpeg_alloc_quant_table((j_common_ptr) cinfo);
-
-  for (i = 0; i < DCTSIZE2; i++) {
-    temp = ((long) basic_table[i] * scale_factor + 50L) / 100L;
-    /* limit the values to the valid range */
-    if (temp <= 0L) temp = 1L;
-    if (temp > 32767L) temp = 32767L; /* max quantizer needed for 12 bits */
-    if (force_baseline && temp > 255L)
-      temp = 255L;		/* limit to baseline range if requested */
-    (*qtblptr)->quantval[i] = (UINT16) temp;
-  }
-
-  /* Initialize sent_table FALSE so table will be written to JPEG file. */
-  (*qtblptr)->sent_table = FALSE;
-}
-
-
-GLOBAL(void)
-jpeg_set_linear_quality (j_compress_ptr cinfo, int scale_factor,
-			 boolean force_baseline)
-/* Set or change the 'quality' (quantization) setting, using default tables
- * and a straight percentage-scaling quality scale.  In most cases it's better
- * to use jpeg_set_quality (below); this entry point is provided for
- * applications that insist on a linear percentage scaling.
- */
-{
-  /* These are the sample quantization tables given in JPEG spec section K.1.
-   * The spec says that the values given produce "good" quality, and
-   * when divided by 2, "very good" quality.
-   */
-  static const unsigned int std_luminance_quant_tbl[DCTSIZE2] = {
-    16,  11,  10,  16,  24,  40,  51,  61,
-    12,  12,  14,  19,  26,  58,  60,  55,
-    14,  13,  16,  24,  40,  57,  69,  56,
-    14,  17,  22,  29,  51,  87,  80,  62,
-    18,  22,  37,  56,  68, 109, 103,  77,
-    24,  35,  55,  64,  81, 104, 113,  92,
-    49,  64,  78,  87, 103, 121, 120, 101,
-    72,  92,  95,  98, 112, 100, 103,  99
-  };
-  static const unsigned int std_chrominance_quant_tbl[DCTSIZE2] = {
-    17,  18,  24,  47,  99,  99,  99,  99,
-    18,  21,  26,  66,  99,  99,  99,  99,
-    24,  26,  56,  99,  99,  99,  99,  99,
-    47,  66,  99,  99,  99,  99,  99,  99,
-    99,  99,  99,  99,  99,  99,  99,  99,
-    99,  99,  99,  99,  99,  99,  99,  99,
-    99,  99,  99,  99,  99,  99,  99,  99,
-    99,  99,  99,  99,  99,  99,  99,  99
-  };
-
-  /* Set up two quantization tables using the specified scaling */
-  jpeg_add_quant_table(cinfo, 0, std_luminance_quant_tbl,
-		       scale_factor, force_baseline);
-  jpeg_add_quant_table(cinfo, 1, std_chrominance_quant_tbl,
-		       scale_factor, force_baseline);
-}
-
-
-GLOBAL(int)
-jpeg_quality_scaling (int quality)
-/* Convert a user-specified quality rating to a percentage scaling factor
- * for an underlying quantization table, using our recommended scaling curve.
- * The input 'quality' factor should be 0 (terrible) to 100 (very good).
- */
-{
-  /* Safety limit on quality factor.  Convert 0 to 1 to avoid zero divide. */
-  if (quality <= 0) quality = 1;
-  if (quality > 100) quality = 100;
-
-  /* The basic table is used as-is (scaling 100) for a quality of 50.
-   * Qualities 50..100 are converted to scaling percentage 200 - 2*Q;
-   * note that at Q=100 the scaling is 0, which will cause jpeg_add_quant_table
-   * to make all the table entries 1 (hence, minimum quantization loss).
-   * Qualities 1..50 are converted to scaling percentage 5000/Q.
-   */
-  if (quality < 50)
-    quality = 5000 / quality;
-  else
-    quality = 200 - quality*2;
-
-  return quality;
-}
-
-
-GLOBAL(void)
-jpeg_set_quality (j_compress_ptr cinfo, int quality, boolean force_baseline)
-/* Set or change the 'quality' (quantization) setting, using default tables.
- * This is the standard quality-adjusting entry point for typical user
- * interfaces; only those who want detailed control over quantization tables
- * would use the preceding three routines directly.
- */
-{
-  /* Convert user 0-100 rating to percentage scaling */
-  quality = jpeg_quality_scaling(quality);
-
-  /* Set up standard quality tables */
-  jpeg_set_linear_quality(cinfo, quality, force_baseline);
-}
-
-
-/*
- * Huffman table setup routines
- */
-
-LOCAL(void)
-add_huff_table (j_compress_ptr cinfo,
-		JHUFF_TBL **htblptr, const UINT8 *bits, const UINT8 *val)
-/* Define a Huffman table */
-{
-  int nsymbols, len;
-
-  if (*htblptr == NULL)
-    *htblptr = jpeg_alloc_huff_table((j_common_ptr) cinfo);
-
-  /* Copy the number-of-symbols-of-each-code-length counts */
-  MEMCOPY((*htblptr)->bits, bits, SIZEOF((*htblptr)->bits));
-
-  /* Validate the counts.  We do this here mainly so we can copy the right
-   * number of symbols from the val[] array, without risking marching off
-   * the end of memory.  jchuff.c will do a more thorough test later.
-   */
-  nsymbols = 0;
-  for (len = 1; len <= 16; len++)
-    nsymbols += bits[len];
-  if (nsymbols < 1 || nsymbols > 256)
-    ERREXIT(cinfo, JERR_BAD_HUFF_TABLE);
-
-  MEMCOPY((*htblptr)->huffval, val, nsymbols * SIZEOF(UINT8));
-
-  /* Initialize sent_table FALSE so table will be written to JPEG file. */
-  (*htblptr)->sent_table = FALSE;
-}
-
-
-LOCAL(void)
-std_huff_tables (j_compress_ptr cinfo)
-/* Set up the standard Huffman tables (cf. JPEG standard section K.3) */
-/* IMPORTANT: these are only valid for 8-bit data precision! */
-{
-  static const UINT8 bits_dc_luminance[17] =
-    { /* 0-base */ 0, 0, 1, 5, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0 };
-  static const UINT8 val_dc_luminance[] =
-    { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 };
-  
-  static const UINT8 bits_dc_chrominance[17] =
-    { /* 0-base */ 0, 0, 3, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0 };
-  static const UINT8 val_dc_chrominance[] =
-    { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 };
-  
-  static const UINT8 bits_ac_luminance[17] =
-    { /* 0-base */ 0, 0, 2, 1, 3, 3, 2, 4, 3, 5, 5, 4, 4, 0, 0, 1, 0x7d };
-  static const UINT8 val_ac_luminance[] =
-    { 0x01, 0x02, 0x03, 0x00, 0x04, 0x11, 0x05, 0x12,
-      0x21, 0x31, 0x41, 0x06, 0x13, 0x51, 0x61, 0x07,
-      0x22, 0x71, 0x14, 0x32, 0x81, 0x91, 0xa1, 0x08,
-      0x23, 0x42, 0xb1, 0xc1, 0x15, 0x52, 0xd1, 0xf0,
-      0x24, 0x33, 0x62, 0x72, 0x82, 0x09, 0x0a, 0x16,
-      0x17, 0x18, 0x19, 0x1a, 0x25, 0x26, 0x27, 0x28,
-      0x29, 0x2a, 0x34, 0x35, 0x36, 0x37, 0x38, 0x39,
-      0x3a, 0x43, 0x44, 0x45, 0x46, 0x47, 0x48, 0x49,
-      0x4a, 0x53, 0x54, 0x55, 0x56, 0x57, 0x58, 0x59,
-      0x5a, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68, 0x69,
-      0x6a, 0x73, 0x74, 0x75, 0x76, 0x77, 0x78, 0x79,
-      0x7a, 0x83, 0x84, 0x85, 0x86, 0x87, 0x88, 0x89,
-      0x8a, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97, 0x98,
-      0x99, 0x9a, 0xa2, 0xa3, 0xa4, 0xa5, 0xa6, 0xa7,
-      0xa8, 0xa9, 0xaa, 0xb2, 0xb3, 0xb4, 0xb5, 0xb6,
-      0xb7, 0xb8, 0xb9, 0xba, 0xc2, 0xc3, 0xc4, 0xc5,
-      0xc6, 0xc7, 0xc8, 0xc9, 0xca, 0xd2, 0xd3, 0xd4,
-      0xd5, 0xd6, 0xd7, 0xd8, 0xd9, 0xda, 0xe1, 0xe2,
-      0xe3, 0xe4, 0xe5, 0xe6, 0xe7, 0xe8, 0xe9, 0xea,
-      0xf1, 0xf2, 0xf3, 0xf4, 0xf5, 0xf6, 0xf7, 0xf8,
-      0xf9, 0xfa };
-  
-  static const UINT8 bits_ac_chrominance[17] =
-    { /* 0-base */ 0, 0, 2, 1, 2, 4, 4, 3, 4, 7, 5, 4, 4, 0, 1, 2, 0x77 };
-  static const UINT8 val_ac_chrominance[] =
-    { 0x00, 0x01, 0x02, 0x03, 0x11, 0x04, 0x05, 0x21,
-      0x31, 0x06, 0x12, 0x41, 0x51, 0x07, 0x61, 0x71,
-      0x13, 0x22, 0x32, 0x81, 0x08, 0x14, 0x42, 0x91,
-      0xa1, 0xb1, 0xc1, 0x09, 0x23, 0x33, 0x52, 0xf0,
-      0x15, 0x62, 0x72, 0xd1, 0x0a, 0x16, 0x24, 0x34,
-      0xe1, 0x25, 0xf1, 0x17, 0x18, 0x19, 0x1a, 0x26,
-      0x27, 0x28, 0x29, 0x2a, 0x35, 0x36, 0x37, 0x38,
-      0x39, 0x3a, 0x43, 0x44, 0x45, 0x46, 0x47, 0x48,
-      0x49, 0x4a, 0x53, 0x54, 0x55, 0x56, 0x57, 0x58,
-      0x59, 0x5a, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68,
-      0x69, 0x6a, 0x73, 0x74, 0x75, 0x76, 0x77, 0x78,
-      0x79, 0x7a, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87,
-      0x88, 0x89, 0x8a, 0x92, 0x93, 0x94, 0x95, 0x96,
-      0x97, 0x98, 0x99, 0x9a, 0xa2, 0xa3, 0xa4, 0xa5,
-      0xa6, 0xa7, 0xa8, 0xa9, 0xaa, 0xb2, 0xb3, 0xb4,
-      0xb5, 0xb6, 0xb7, 0xb8, 0xb9, 0xba, 0xc2, 0xc3,
-      0xc4, 0xc5, 0xc6, 0xc7, 0xc8, 0xc9, 0xca, 0xd2,
-      0xd3, 0xd4, 0xd5, 0xd6, 0xd7, 0xd8, 0xd9, 0xda,
-      0xe2, 0xe3, 0xe4, 0xe5, 0xe6, 0xe7, 0xe8, 0xe9,
-      0xea, 0xf2, 0xf3, 0xf4, 0xf5, 0xf6, 0xf7, 0xf8,
-      0xf9, 0xfa };
-  
-  add_huff_table(cinfo, &cinfo->dc_huff_tbl_ptrs[0],
-		 bits_dc_luminance, val_dc_luminance);
-  add_huff_table(cinfo, &cinfo->ac_huff_tbl_ptrs[0],
-		 bits_ac_luminance, val_ac_luminance);
-  add_huff_table(cinfo, &cinfo->dc_huff_tbl_ptrs[1],
-		 bits_dc_chrominance, val_dc_chrominance);
-  add_huff_table(cinfo, &cinfo->ac_huff_tbl_ptrs[1],
-		 bits_ac_chrominance, val_ac_chrominance);
-}
-
-
-/*
- * Default parameter setup for compression.
- *
- * Applications that don't choose to use this routine must do their
- * own setup of all these parameters.  Alternately, you can call this
- * to establish defaults and then alter parameters selectively.  This
- * is the recommended approach since, if we add any new parameters,
- * your code will still work (they'll be set to reasonable defaults).
- */
-
-GLOBAL(void)
-jpeg_set_defaults (j_compress_ptr cinfo)
-{
-  int i;
-
-  /* Safety check to ensure start_compress not called yet. */
-  if (cinfo->global_state != CSTATE_START)
-    ERREXIT1(cinfo, JERR_BAD_STATE, cinfo->global_state);
-
-  /* Allocate comp_info array large enough for maximum component count.
-   * Array is made permanent in case application wants to compress
-   * multiple images at same param settings.
-   */
-  if (cinfo->comp_info == NULL)
-    cinfo->comp_info = (jpeg_component_info *)
-      (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_PERMANENT,
-				  MAX_COMPONENTS * SIZEOF(jpeg_component_info));
-
-  /* Initialize everything not dependent on the color space */
-
-  cinfo->data_precision = BITS_IN_JSAMPLE;
-  /* Set up two quantization tables using default quality of 75 */
-  jpeg_set_quality(cinfo, 75, TRUE);
-  /* Set up two Huffman tables */
-  std_huff_tables(cinfo);
-
-  /* Initialize default arithmetic coding conditioning */
-  for (i = 0; i < NUM_ARITH_TBLS; i++) {
-    cinfo->arith_dc_L[i] = 0;
-    cinfo->arith_dc_U[i] = 1;
-    cinfo->arith_ac_K[i] = 5;
-  }
-
-  /* Default is no multiple-scan output */
-  cinfo->scan_info = NULL;
-  cinfo->num_scans = 0;
-
-  /* Expect normal source image, not raw downsampled data */
-  cinfo->raw_data_in = FALSE;
-
-  /* Use Huffman coding, not arithmetic coding, by default */
-  cinfo->arith_code = FALSE;
-
-  /* By default, don't do extra passes to optimize entropy coding */
-  cinfo->optimize_coding = FALSE;
-  /* The standard Huffman tables are only valid for 8-bit data precision.
-   * If the precision is higher, force optimization on so that usable
-   * tables will be computed.  This test can be removed if default tables
-   * are supplied that are valid for the desired precision.
-   */
-  if (cinfo->data_precision > 8)
-    cinfo->optimize_coding = TRUE;
-
-  /* By default, use the simpler non-cosited sampling alignment */
-  cinfo->CCIR601_sampling = FALSE;
-
-  /* No input smoothing */
-  cinfo->smoothing_factor = 0;
-
-  /* DCT algorithm preference */
-  cinfo->dct_method = JDCT_DEFAULT;
-
-  /* No restart markers */
-  cinfo->restart_interval = 0;
-  cinfo->restart_in_rows = 0;
-
-  /* Fill in default JFIF marker parameters.  Note that whether the marker
-   * will actually be written is determined by jpeg_set_colorspace.
-   *
-   * By default, the library emits JFIF version code 1.01.
-   * An application that wants to emit JFIF 1.02 extension markers should set
-   * JFIF_minor_version to 2.  We could probably get away with just defaulting
-   * to 1.02, but there may still be some decoders in use that will complain
-   * about that; saying 1.01 should minimize compatibility problems.
-   */
-  cinfo->JFIF_major_version = 1; /* Default JFIF version = 1.01 */
-  cinfo->JFIF_minor_version = 1;
-  cinfo->density_unit = 0;	/* Pixel size is unknown by default */
-  cinfo->X_density = 1;		/* Pixel aspect ratio is square by default */
-  cinfo->Y_density = 1;
-
-  /* Choose JPEG colorspace based on input space, set defaults accordingly */
-
-  jpeg_default_colorspace(cinfo);
-}
-
-
-/*
- * Select an appropriate JPEG colorspace for in_color_space.
- */
-
-GLOBAL(void)
-jpeg_default_colorspace (j_compress_ptr cinfo)
-{
-  switch (cinfo->in_color_space) {
-  case JCS_GRAYSCALE:
-    jpeg_set_colorspace(cinfo, JCS_GRAYSCALE);
-    break;
-  case JCS_RGB:
-    jpeg_set_colorspace(cinfo, JCS_YCbCr);
-    break;
-  case JCS_YCbCr:
-    jpeg_set_colorspace(cinfo, JCS_YCbCr);
-    break;
-  case JCS_CMYK:
-    jpeg_set_colorspace(cinfo, JCS_CMYK); /* By default, no translation */
-    break;
-  case JCS_YCCK:
-    jpeg_set_colorspace(cinfo, JCS_YCCK);
-    break;
-  case JCS_UNKNOWN:
-    jpeg_set_colorspace(cinfo, JCS_UNKNOWN);
-    break;
-  default:
-    ERREXIT(cinfo, JERR_BAD_IN_COLORSPACE);
-  }
-}
-
-
-/*
- * Set the JPEG colorspace, and choose colorspace-dependent default values.
- */
-
-GLOBAL(void)
-jpeg_set_colorspace (j_compress_ptr cinfo, J_COLOR_SPACE colorspace)
-{
-  jpeg_component_info * compptr;
-  int ci;
-
-#define SET_COMP(index,id,hsamp,vsamp,quant,dctbl,actbl)  \
-  (compptr = &cinfo->comp_info[index], \
-   compptr->component_id = (id), \
-   compptr->h_samp_factor = (hsamp), \
-   compptr->v_samp_factor = (vsamp), \
-   compptr->quant_tbl_no = (quant), \
-   compptr->dc_tbl_no = (dctbl), \
-   compptr->ac_tbl_no = (actbl) )
-
-  /* Safety check to ensure start_compress not called yet. */
-  if (cinfo->global_state != CSTATE_START)
-    ERREXIT1(cinfo, JERR_BAD_STATE, cinfo->global_state);
-
-  /* For all colorspaces, we use Q and Huff tables 0 for luminance components,
-   * tables 1 for chrominance components.
-   */
-
-  cinfo->jpeg_color_space = colorspace;
-
-  cinfo->write_JFIF_header = FALSE; /* No marker for non-JFIF colorspaces */
-  cinfo->write_Adobe_marker = FALSE; /* write no Adobe marker by default */
-
-  switch (colorspace) {
-  case JCS_GRAYSCALE:
-    cinfo->write_JFIF_header = TRUE; /* Write a JFIF marker */
-    cinfo->num_components = 1;
-    /* JFIF specifies component ID 1 */
-    SET_COMP(0, 1, 1,1, 0, 0,0);
-    break;
-  case JCS_RGB:
-    cinfo->write_Adobe_marker = TRUE; /* write Adobe marker to flag RGB */
-    cinfo->num_components = 3;
-    SET_COMP(0, 0x52 /* 'R' */, 1,1, 0, 0,0);
-    SET_COMP(1, 0x47 /* 'G' */, 1,1, 0, 0,0);
-    SET_COMP(2, 0x42 /* 'B' */, 1,1, 0, 0,0);
-    break;
-  case JCS_YCbCr:
-    cinfo->write_JFIF_header = TRUE; /* Write a JFIF marker */
-    cinfo->num_components = 3;
-    /* JFIF specifies component IDs 1,2,3 */
-    /* We default to 2x2 subsamples of chrominance */
-    SET_COMP(0, 1, 2,2, 0, 0,0);
-    SET_COMP(1, 2, 1,1, 1, 1,1);
-    SET_COMP(2, 3, 1,1, 1, 1,1);
-    break;
-  case JCS_CMYK:
-    cinfo->write_Adobe_marker = TRUE; /* write Adobe marker to flag CMYK */
-    cinfo->num_components = 4;
-    SET_COMP(0, 0x43 /* 'C' */, 1,1, 0, 0,0);
-    SET_COMP(1, 0x4D /* 'M' */, 1,1, 0, 0,0);
-    SET_COMP(2, 0x59 /* 'Y' */, 1,1, 0, 0,0);
-    SET_COMP(3, 0x4B /* 'K' */, 1,1, 0, 0,0);
-    break;
-  case JCS_YCCK:
-    cinfo->write_Adobe_marker = TRUE; /* write Adobe marker to flag YCCK */
-    cinfo->num_components = 4;
-    SET_COMP(0, 1, 2,2, 0, 0,0);
-    SET_COMP(1, 2, 1,1, 1, 1,1);
-    SET_COMP(2, 3, 1,1, 1, 1,1);
-    SET_COMP(3, 4, 2,2, 0, 0,0);
-    break;
-  case JCS_UNKNOWN:
-    cinfo->num_components = cinfo->input_components;
-    if (cinfo->num_components < 1 || cinfo->num_components > MAX_COMPONENTS)
-      ERREXIT2(cinfo, JERR_COMPONENT_COUNT, cinfo->num_components,
-	       MAX_COMPONENTS);
-    for (ci = 0; ci < cinfo->num_components; ci++) {
-      SET_COMP(ci, ci, 1,1, 0, 0,0);
-    }
-    break;
-  default:
-    ERREXIT(cinfo, JERR_BAD_J_COLORSPACE);
-  }
-}
-
-
-#ifdef C_PROGRESSIVE_SUPPORTED
-
-LOCAL(jpeg_scan_info *)
-fill_a_scan (jpeg_scan_info * scanptr, int ci,
-	     int Ss, int Se, int Ah, int Al)
-/* Support routine: generate one scan for specified component */
-{
-  scanptr->comps_in_scan = 1;
-  scanptr->component_index[0] = ci;
-  scanptr->Ss = Ss;
-  scanptr->Se = Se;
-  scanptr->Ah = Ah;
-  scanptr->Al = Al;
-  scanptr++;
-  return scanptr;
-}
-
-LOCAL(jpeg_scan_info *)
-fill_scans (jpeg_scan_info * scanptr, int ncomps,
-	    int Ss, int Se, int Ah, int Al)
-/* Support routine: generate one scan for each component */
-{
-  int ci;
-
-  for (ci = 0; ci < ncomps; ci++) {
-    scanptr->comps_in_scan = 1;
-    scanptr->component_index[0] = ci;
-    scanptr->Ss = Ss;
-    scanptr->Se = Se;
-    scanptr->Ah = Ah;
-    scanptr->Al = Al;
-    scanptr++;
-  }
-  return scanptr;
-}
-
-LOCAL(jpeg_scan_info *)
-fill_dc_scans (jpeg_scan_info * scanptr, int ncomps, int Ah, int Al)
-/* Support routine: generate interleaved DC scan if possible, else N scans */
-{
-  int ci;
-
-  if (ncomps <= MAX_COMPS_IN_SCAN) {
-    /* Single interleaved DC scan */
-    scanptr->comps_in_scan = ncomps;
-    for (ci = 0; ci < ncomps; ci++)
-      scanptr->component_index[ci] = ci;
-    scanptr->Ss = scanptr->Se = 0;
-    scanptr->Ah = Ah;
-    scanptr->Al = Al;
-    scanptr++;
-  } else {
-    /* Noninterleaved DC scan for each component */
-    scanptr = fill_scans(scanptr, ncomps, 0, 0, Ah, Al);
-  }
-  return scanptr;
-}
-
-
-/*
- * Create a recommended progressive-JPEG script.
- * cinfo->num_components and cinfo->jpeg_color_space must be correct.
- */
-
-GLOBAL(void)
-jpeg_simple_progression (j_compress_ptr cinfo)
-{
-  int ncomps = cinfo->num_components;
-  int nscans;
-  jpeg_scan_info * scanptr;
-
-  /* Safety check to ensure start_compress not called yet. */
-  if (cinfo->global_state != CSTATE_START)
-    ERREXIT1(cinfo, JERR_BAD_STATE, cinfo->global_state);
-
-  /* Figure space needed for script.  Calculation must match code below! */
-  if (ncomps == 3 && cinfo->jpeg_color_space == JCS_YCbCr) {
-    /* Custom script for YCbCr color images. */
-    nscans = 10;
-  } else {
-    /* All-purpose script for other color spaces. */
-    if (ncomps > MAX_COMPS_IN_SCAN)
-      nscans = 6 * ncomps;	/* 2 DC + 4 AC scans per component */
-    else
-      nscans = 2 + 4 * ncomps;	/* 2 DC scans; 4 AC scans per component */
-  }
-
-  /* Allocate space for script.
-   * We need to put it in the permanent pool in case the application performs
-   * multiple compressions without changing the settings.  To avoid a memory
-   * leak if jpeg_simple_progression is called repeatedly for the same JPEG
-   * object, we try to re-use previously allocated space, and we allocate
-   * enough space to handle YCbCr even if initially asked for grayscale.
-   */
-  if (cinfo->script_space == NULL || cinfo->script_space_size < nscans) {
-    cinfo->script_space_size = MAX(nscans, 10);
-    cinfo->script_space = (jpeg_scan_info *)
-      (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_PERMANENT,
-			cinfo->script_space_size * SIZEOF(jpeg_scan_info));
-  }
-  scanptr = cinfo->script_space;
-  cinfo->scan_info = scanptr;
-  cinfo->num_scans = nscans;
-
-  if (ncomps == 3 && cinfo->jpeg_color_space == JCS_YCbCr) {
-    /* Custom script for YCbCr color images. */
-    /* Initial DC scan */
-    scanptr = fill_dc_scans(scanptr, ncomps, 0, 1);
-    /* Initial AC scan: get some luma data out in a hurry */
-    scanptr = fill_a_scan(scanptr, 0, 1, 5, 0, 2);
-    /* Chroma data is too small to be worth expending many scans on */
-    scanptr = fill_a_scan(scanptr, 2, 1, 63, 0, 1);
-    scanptr = fill_a_scan(scanptr, 1, 1, 63, 0, 1);
-    /* Complete spectral selection for luma AC */
-    scanptr = fill_a_scan(scanptr, 0, 6, 63, 0, 2);
-    /* Refine next bit of luma AC */
-    scanptr = fill_a_scan(scanptr, 0, 1, 63, 2, 1);
-    /* Finish DC successive approximation */
-    scanptr = fill_dc_scans(scanptr, ncomps, 1, 0);
-    /* Finish AC successive approximation */
-    scanptr = fill_a_scan(scanptr, 2, 1, 63, 1, 0);
-    scanptr = fill_a_scan(scanptr, 1, 1, 63, 1, 0);
-    /* Luma bottom bit comes last since it's usually largest scan */
-    scanptr = fill_a_scan(scanptr, 0, 1, 63, 1, 0);
-  } else {
-    /* All-purpose script for other color spaces. */
-    /* Successive approximation first pass */
-    scanptr = fill_dc_scans(scanptr, ncomps, 0, 1);
-    scanptr = fill_scans(scanptr, ncomps, 1, 5, 0, 2);
-    scanptr = fill_scans(scanptr, ncomps, 6, 63, 0, 2);
-    /* Successive approximation second pass */
-    scanptr = fill_scans(scanptr, ncomps, 1, 63, 2, 1);
-    /* Successive approximation final pass */
-    scanptr = fill_dc_scans(scanptr, ncomps, 1, 0);
-    scanptr = fill_scans(scanptr, ncomps, 1, 63, 1, 0);
-  }
-}
-
-#endif /* C_PROGRESSIVE_SUPPORTED */
@@ -1,833 +0,0 @@
-/*
- * jcphuff.c
- *
- * Copyright (C) 1995-1997, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains Huffman entropy encoding routines for progressive JPEG.
- *
- * We do not support output suspension in this module, since the library
- * currently does not allow multiple-scan files to be written with output
- * suspension.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-#include "jchuff.h"		/* Declarations shared with jchuff.c */
-
-#ifdef C_PROGRESSIVE_SUPPORTED
-
-/* Expanded entropy encoder object for progressive Huffman encoding. */
-
-typedef struct {
-  struct jpeg_entropy_encoder pub; /* public fields */
-
-  /* Mode flag: TRUE for optimization, FALSE for actual data output */
-  boolean gather_statistics;
-
-  /* Bit-level coding status.
-   * next_output_byte/free_in_buffer are local copies of cinfo->dest fields.
-   */
-  JOCTET * next_output_byte;	/* => next byte to write in buffer */
-  size_t free_in_buffer;	/* # of byte spaces remaining in buffer */
-  INT32 put_buffer;		/* current bit-accumulation buffer */
-  int put_bits;			/* # of bits now in it */
-  j_compress_ptr cinfo;		/* link to cinfo (needed for dump_buffer) */
-
-  /* Coding status for DC components */
-  int last_dc_val[MAX_COMPS_IN_SCAN]; /* last DC coef for each component */
-
-  /* Coding status for AC components */
-  int ac_tbl_no;		/* the table number of the single component */
-  unsigned int EOBRUN;		/* run length of EOBs */
-  unsigned int BE;		/* # of buffered correction bits before MCU */
-  char * bit_buffer;		/* buffer for correction bits (1 per char) */
-  /* packing correction bits tightly would save some space but cost time... */
-
-  unsigned int restarts_to_go;	/* MCUs left in this restart interval */
-  int next_restart_num;		/* next restart number to write (0-7) */
-
-  /* Pointers to derived tables (these workspaces have image lifespan).
-   * Since any one scan codes only DC or only AC, we only need one set
-   * of tables, not one for DC and one for AC.
-   */
-  c_derived_tbl * derived_tbls[NUM_HUFF_TBLS];
-
-  /* Statistics tables for optimization; again, one set is enough */
-  long * count_ptrs[NUM_HUFF_TBLS];
-} phuff_entropy_encoder;
-
-typedef phuff_entropy_encoder * phuff_entropy_ptr;
-
-/* MAX_CORR_BITS is the number of bits the AC refinement correction-bit
- * buffer can hold.  Larger sizes may slightly improve compression, but
- * 1000 is already well into the realm of overkill.
- * The minimum safe size is 64 bits.
- */
-
-#define MAX_CORR_BITS  1000	/* Max # of correction bits I can buffer */
-
-/* IRIGHT_SHIFT is like RIGHT_SHIFT, but works on int rather than INT32.
- * We assume that int right shift is unsigned if INT32 right shift is,
- * which should be safe.
- */
-
-#ifdef RIGHT_SHIFT_IS_UNSIGNED
-#define ISHIFT_TEMPS	int ishift_temp;
-#define IRIGHT_SHIFT(x,shft)  \
-	((ishift_temp = (x)) < 0 ? \
-	 (ishift_temp >> (shft)) | ((~0) << (16-(shft))) : \
-	 (ishift_temp >> (shft)))
-#else
-#define ISHIFT_TEMPS
-#define IRIGHT_SHIFT(x,shft)	((x) >> (shft))
-#endif
-
-/* Forward declarations */
-METHODDEF(boolean) encode_mcu_DC_first JPP((j_compress_ptr cinfo,
-					    JBLOCKROW *MCU_data));
-METHODDEF(boolean) encode_mcu_AC_first JPP((j_compress_ptr cinfo,
-					    JBLOCKROW *MCU_data));
-METHODDEF(boolean) encode_mcu_DC_refine JPP((j_compress_ptr cinfo,
-					     JBLOCKROW *MCU_data));
-METHODDEF(boolean) encode_mcu_AC_refine JPP((j_compress_ptr cinfo,
-					     JBLOCKROW *MCU_data));
-METHODDEF(void) finish_pass_phuff JPP((j_compress_ptr cinfo));
-METHODDEF(void) finish_pass_gather_phuff JPP((j_compress_ptr cinfo));
-
-
-/*
- * Initialize for a Huffman-compressed scan using progressive JPEG.
- */
-
-METHODDEF(void)
-start_pass_phuff (j_compress_ptr cinfo, boolean gather_statistics)
-{  
-  phuff_entropy_ptr entropy = (phuff_entropy_ptr) cinfo->entropy;
-  boolean is_DC_band;
-  int ci, tbl;
-  jpeg_component_info * compptr;
-
-  entropy->cinfo = cinfo;
-  entropy->gather_statistics = gather_statistics;
-
-  is_DC_band = (cinfo->Ss == 0);
-
-  /* We assume jcmaster.c already validated the scan parameters. */
-
-  /* Select execution routines */
-  if (cinfo->Ah == 0) {
-    if (is_DC_band)
-      entropy->pub.encode_mcu = encode_mcu_DC_first;
-    else
-      entropy->pub.encode_mcu = encode_mcu_AC_first;
-  } else {
-    if (is_DC_band)
-      entropy->pub.encode_mcu = encode_mcu_DC_refine;
-    else {
-      entropy->pub.encode_mcu = encode_mcu_AC_refine;
-      /* AC refinement needs a correction bit buffer */
-      if (entropy->bit_buffer == NULL)
-	entropy->bit_buffer = (char *)
-	  (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				      MAX_CORR_BITS * SIZEOF(char));
-    }
-  }
-  if (gather_statistics)
-    entropy->pub.finish_pass = finish_pass_gather_phuff;
-  else
-    entropy->pub.finish_pass = finish_pass_phuff;
-
-  /* Only DC coefficients may be interleaved, so cinfo->comps_in_scan = 1
-   * for AC coefficients.
-   */
-  for (ci = 0; ci < cinfo->comps_in_scan; ci++) {
-    compptr = cinfo->cur_comp_info[ci];
-    /* Initialize DC predictions to 0 */
-    entropy->last_dc_val[ci] = 0;
-    /* Get table index */
-    if (is_DC_band) {
-      if (cinfo->Ah != 0)	/* DC refinement needs no table */
-	continue;
-      tbl = compptr->dc_tbl_no;
-    } else {
-      entropy->ac_tbl_no = tbl = compptr->ac_tbl_no;
-    }
-    if (gather_statistics) {
-      /* Check for invalid table index */
-      /* (make_c_derived_tbl does this in the other path) */
-      if (tbl < 0 || tbl >= NUM_HUFF_TBLS)
-        ERREXIT1(cinfo, JERR_NO_HUFF_TABLE, tbl);
-      /* Allocate and zero the statistics tables */
-      /* Note that jpeg_gen_optimal_table expects 257 entries in each table! */
-      if (entropy->count_ptrs[tbl] == NULL)
-	entropy->count_ptrs[tbl] = (long *)
-	  (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				      257 * SIZEOF(long));
-      MEMZERO(entropy->count_ptrs[tbl], 257 * SIZEOF(long));
-    } else {
-      /* Compute derived values for Huffman table */
-      /* We may do this more than once for a table, but it's not expensive */
-      jpeg_make_c_derived_tbl(cinfo, is_DC_band, tbl,
-			      & entropy->derived_tbls[tbl]);
-    }
-  }
-
-  /* Initialize AC stuff */
-  entropy->EOBRUN = 0;
-  entropy->BE = 0;
-
-  /* Initialize bit buffer to empty */
-  entropy->put_buffer = 0;
-  entropy->put_bits = 0;
-
-  /* Initialize restart stuff */
-  entropy->restarts_to_go = cinfo->restart_interval;
-  entropy->next_restart_num = 0;
-}
-
-
-/* Outputting bytes to the file.
- * NB: these must be called only when actually outputting,
- * that is, entropy->gather_statistics == FALSE.
- */
-
-/* Emit a byte */
-#define emit_byte(entropy,val)  \
-	{ *(entropy)->next_output_byte++ = (JOCTET) (val);  \
-	  if (--(entropy)->free_in_buffer == 0)  \
-	    dump_buffer(entropy); }
-
-
-LOCAL(void)
-dump_buffer (phuff_entropy_ptr entropy)
-/* Empty the output buffer; we do not support suspension in this module. */
-{
-  struct jpeg_destination_mgr * dest = entropy->cinfo->dest;
-
-  if (! (*dest->empty_output_buffer) (entropy->cinfo))
-    ERREXIT(entropy->cinfo, JERR_CANT_SUSPEND);
-  /* After a successful buffer dump, must reset buffer pointers */
-  entropy->next_output_byte = dest->next_output_byte;
-  entropy->free_in_buffer = dest->free_in_buffer;
-}
-
-
-/* Outputting bits to the file */
-
-/* Only the right 24 bits of put_buffer are used; the valid bits are
- * left-justified in this part.  At most 16 bits can be passed to emit_bits
- * in one call, and we never retain more than 7 bits in put_buffer
- * between calls, so 24 bits are sufficient.
- */
-
-INLINE
-LOCAL(void)
-emit_bits (phuff_entropy_ptr entropy, unsigned int code, int size)
-/* Emit some bits, unless we are in gather mode */
-{
-  /* This routine is heavily used, so it's worth coding tightly. */
-  register INT32 put_buffer = (INT32) code;
-  register int put_bits = entropy->put_bits;
-
-  /* if size is 0, caller used an invalid Huffman table entry */
-  if (size == 0)
-    ERREXIT(entropy->cinfo, JERR_HUFF_MISSING_CODE);
-
-  if (entropy->gather_statistics)
-    return;			/* do nothing if we're only getting stats */
-
-  put_buffer &= (((INT32) 1)<<size) - 1; /* mask off any extra bits in code */
-  
-  put_bits += size;		/* new number of bits in buffer */
-  
-  put_buffer <<= 24 - put_bits; /* align incoming bits */
-
-  put_buffer |= entropy->put_buffer; /* and merge with old buffer contents */
-
-  while (put_bits >= 8) {
-    int c = (int) ((put_buffer >> 16) & 0xFF);
-    
-    emit_byte(entropy, c);
-    if (c == 0xFF) {		/* need to stuff a zero byte? */
-      emit_byte(entropy, 0);
-    }
-    put_buffer <<= 8;
-    put_bits -= 8;
-  }
-
-  entropy->put_buffer = put_buffer; /* update variables */
-  entropy->put_bits = put_bits;
-}
-
-
-LOCAL(void)
-flush_bits (phuff_entropy_ptr entropy)
-{
-  emit_bits(entropy, 0x7F, 7); /* fill any partial byte with ones */
-  entropy->put_buffer = 0;     /* and reset bit-buffer to empty */
-  entropy->put_bits = 0;
-}
-
-
-/*
- * Emit (or just count) a Huffman symbol.
- */
-
-INLINE
-LOCAL(void)
-emit_symbol (phuff_entropy_ptr entropy, int tbl_no, int symbol)
-{
-  if (entropy->gather_statistics)
-    entropy->count_ptrs[tbl_no][symbol]++;
-  else {
-    c_derived_tbl * tbl = entropy->derived_tbls[tbl_no];
-    emit_bits(entropy, tbl->ehufco[symbol], tbl->ehufsi[symbol]);
-  }
-}
-
-
-/*
- * Emit bits from a correction bit buffer.
- */
-
-LOCAL(void)
-emit_buffered_bits (phuff_entropy_ptr entropy, char * bufstart,
-		    unsigned int nbits)
-{
-  if (entropy->gather_statistics)
-    return;			/* no real work */
-
-  while (nbits > 0) {
-    emit_bits(entropy, (unsigned int) (*bufstart), 1);
-    bufstart++;
-    nbits--;
-  }
-}
-
-
-/*
- * Emit any pending EOBRUN symbol.
- */
-
-LOCAL(void)
-emit_eobrun (phuff_entropy_ptr entropy)
-{
-  register int temp, nbits;
-
-  if (entropy->EOBRUN > 0) {	/* if there is any pending EOBRUN */
-    temp = entropy->EOBRUN;
-    nbits = 0;
-    while ((temp >>= 1))
-      nbits++;
-    /* safety check: shouldn't happen given limited correction-bit buffer */
-    if (nbits > 14)
-      ERREXIT(entropy->cinfo, JERR_HUFF_MISSING_CODE);
-
-    emit_symbol(entropy, entropy->ac_tbl_no, nbits << 4);
-    if (nbits)
-      emit_bits(entropy, entropy->EOBRUN, nbits);
-
-    entropy->EOBRUN = 0;
-
-    /* Emit any buffered correction bits */
-    emit_buffered_bits(entropy, entropy->bit_buffer, entropy->BE);
-    entropy->BE = 0;
-  }
-}
-
-
-/*
- * Emit a restart marker & resynchronize predictions.
- */
-
-LOCAL(void)
-emit_restart (phuff_entropy_ptr entropy, int restart_num)
-{
-  int ci;
-
-  emit_eobrun(entropy);
-
-  if (! entropy->gather_statistics) {
-    flush_bits(entropy);
-    emit_byte(entropy, 0xFF);
-    emit_byte(entropy, JPEG_RST0 + restart_num);
-  }
-
-  if (entropy->cinfo->Ss == 0) {
-    /* Re-initialize DC predictions to 0 */
-    for (ci = 0; ci < entropy->cinfo->comps_in_scan; ci++)
-      entropy->last_dc_val[ci] = 0;
-  } else {
-    /* Re-initialize all AC-related fields to 0 */
-    entropy->EOBRUN = 0;
-    entropy->BE = 0;
-  }
-}
-
-
-/*
- * MCU encoding for DC initial scan (either spectral selection,
- * or first pass of successive approximation).
- */
-
-METHODDEF(boolean)
-encode_mcu_DC_first (j_compress_ptr cinfo, JBLOCKROW *MCU_data)
-{
-  phuff_entropy_ptr entropy = (phuff_entropy_ptr) cinfo->entropy;
-  register int temp, temp2;
-  register int nbits;
-  int blkn, ci;
-  int Al = cinfo->Al;
-  JBLOCKROW block;
-  jpeg_component_info * compptr;
-  ISHIFT_TEMPS
-
-  entropy->next_output_byte = cinfo->dest->next_output_byte;
-  entropy->free_in_buffer = cinfo->dest->free_in_buffer;
-
-  /* Emit restart marker if needed */
-  if (cinfo->restart_interval)
-    if (entropy->restarts_to_go == 0)
-      emit_restart(entropy, entropy->next_restart_num);
-
-  /* Encode the MCU data blocks */
-  for (blkn = 0; blkn < cinfo->blocks_in_MCU; blkn++) {
-    block = MCU_data[blkn];
-    ci = cinfo->MCU_membership[blkn];
-    compptr = cinfo->cur_comp_info[ci];
-
-    /* Compute the DC value after the required point transform by Al.
-     * This is simply an arithmetic right shift.
-     */
-    temp2 = IRIGHT_SHIFT((int) ((*block)[0]), Al);
-
-    /* DC differences are figured on the point-transformed values. */
-    temp = temp2 - entropy->last_dc_val[ci];
-    entropy->last_dc_val[ci] = temp2;
-
-    /* Encode the DC coefficient difference per section G.1.2.1 */
-    temp2 = temp;
-    if (temp < 0) {
-      temp = -temp;		/* temp is abs value of input */
-      /* For a negative input, want temp2 = bitwise complement of abs(input) */
-      /* This code assumes we are on a two's complement machine */
-      temp2--;
-    }
-    
-    /* Find the number of bits needed for the magnitude of the coefficient */
-    nbits = 0;
-    while (temp) {
-      nbits++;
-      temp >>= 1;
-    }
-    /* Check for out-of-range coefficient values.
-     * Since we're encoding a difference, the range limit is twice as much.
-     */
-    if (nbits > MAX_COEF_BITS+1)
-      ERREXIT(cinfo, JERR_BAD_DCT_COEF);
-    
-    /* Count/emit the Huffman-coded symbol for the number of bits */
-    emit_symbol(entropy, compptr->dc_tbl_no, nbits);
-    
-    /* Emit that number of bits of the value, if positive, */
-    /* or the complement of its magnitude, if negative. */
-    if (nbits)			/* emit_bits rejects calls with size 0 */
-      emit_bits(entropy, (unsigned int) temp2, nbits);
-  }
-
-  cinfo->dest->next_output_byte = entropy->next_output_byte;
-  cinfo->dest->free_in_buffer = entropy->free_in_buffer;
-
-  /* Update restart-interval state too */
-  if (cinfo->restart_interval) {
-    if (entropy->restarts_to_go == 0) {
-      entropy->restarts_to_go = cinfo->restart_interval;
-      entropy->next_restart_num++;
-      entropy->next_restart_num &= 7;
-    }
-    entropy->restarts_to_go--;
-  }
-
-  return TRUE;
-}
-
-
-/*
- * MCU encoding for AC initial scan (either spectral selection,
- * or first pass of successive approximation).
- */
-
-METHODDEF(boolean)
-encode_mcu_AC_first (j_compress_ptr cinfo, JBLOCKROW *MCU_data)
-{
-  phuff_entropy_ptr entropy = (phuff_entropy_ptr) cinfo->entropy;
-  register int temp, temp2;
-  register int nbits;
-  register int r, k;
-  int Se = cinfo->Se;
-  int Al = cinfo->Al;
-  JBLOCKROW block;
-
-  entropy->next_output_byte = cinfo->dest->next_output_byte;
-  entropy->free_in_buffer = cinfo->dest->free_in_buffer;
-
-  /* Emit restart marker if needed */
-  if (cinfo->restart_interval)
-    if (entropy->restarts_to_go == 0)
-      emit_restart(entropy, entropy->next_restart_num);
-
-  /* Encode the MCU data block */
-  block = MCU_data[0];
-
-  /* Encode the AC coefficients per section G.1.2.2, fig. G.3 */
-  
-  r = 0;			/* r = run length of zeros */
-   
-  for (k = cinfo->Ss; k <= Se; k++) {
-    if ((temp = (*block)[jpeg_natural_order[k]]) == 0) {
-      r++;
-      continue;
-    }
-    /* We must apply the point transform by Al.  For AC coefficients this
-     * is an integer division with rounding towards 0.  To do this portably
-     * in C, we shift after obtaining the absolute value; so the code is
-     * interwoven with finding the abs value (temp) and output bits (temp2).
-     */
-    if (temp < 0) {
-      temp = -temp;		/* temp is abs value of input */
-      temp >>= Al;		/* apply the point transform */
-      /* For a negative coef, want temp2 = bitwise complement of abs(coef) */
-      temp2 = ~temp;
-    } else {
-      temp >>= Al;		/* apply the point transform */
-      temp2 = temp;
-    }
-    /* Watch out for case that nonzero coef is zero after point transform */
-    if (temp == 0) {
-      r++;
-      continue;
-    }
-
-    /* Emit any pending EOBRUN */
-    if (entropy->EOBRUN > 0)
-      emit_eobrun(entropy);
-    /* if run length > 15, must emit special run-length-16 codes (0xF0) */
-    while (r > 15) {
-      emit_symbol(entropy, entropy->ac_tbl_no, 0xF0);
-      r -= 16;
-    }
-
-    /* Find the number of bits needed for the magnitude of the coefficient */
-    nbits = 1;			/* there must be at least one 1 bit */
-    while ((temp >>= 1))
-      nbits++;
-    /* Check for out-of-range coefficient values */
-    if (nbits > MAX_COEF_BITS)
-      ERREXIT(cinfo, JERR_BAD_DCT_COEF);
-
-    /* Count/emit Huffman symbol for run length / number of bits */
-    emit_symbol(entropy, entropy->ac_tbl_no, (r << 4) + nbits);
-
-    /* Emit that number of bits of the value, if positive, */
-    /* or the complement of its magnitude, if negative. */
-    emit_bits(entropy, (unsigned int) temp2, nbits);
-
-    r = 0;			/* reset zero run length */
-  }
-
-  if (r > 0) {			/* If there are trailing zeroes, */
-    entropy->EOBRUN++;		/* count an EOB */
-    if (entropy->EOBRUN == 0x7FFF)
-      emit_eobrun(entropy);	/* force it out to avoid overflow */
-  }
-
-  cinfo->dest->next_output_byte = entropy->next_output_byte;
-  cinfo->dest->free_in_buffer = entropy->free_in_buffer;
-
-  /* Update restart-interval state too */
-  if (cinfo->restart_interval) {
-    if (entropy->restarts_to_go == 0) {
-      entropy->restarts_to_go = cinfo->restart_interval;
-      entropy->next_restart_num++;
-      entropy->next_restart_num &= 7;
-    }
-    entropy->restarts_to_go--;
-  }
-
-  return TRUE;
-}
-
-
-/*
- * MCU encoding for DC successive approximation refinement scan.
- * Note: we assume such scans can be multi-component, although the spec
- * is not very clear on the point.
- */
-
-METHODDEF(boolean)
-encode_mcu_DC_refine (j_compress_ptr cinfo, JBLOCKROW *MCU_data)
-{
-  phuff_entropy_ptr entropy = (phuff_entropy_ptr) cinfo->entropy;
-  register int temp;
-  int blkn;
-  int Al = cinfo->Al;
-  JBLOCKROW block;
-
-  entropy->next_output_byte = cinfo->dest->next_output_byte;
-  entropy->free_in_buffer = cinfo->dest->free_in_buffer;
-
-  /* Emit restart marker if needed */
-  if (cinfo->restart_interval)
-    if (entropy->restarts_to_go == 0)
-      emit_restart(entropy, entropy->next_restart_num);
-
-  /* Encode the MCU data blocks */
-  for (blkn = 0; blkn < cinfo->blocks_in_MCU; blkn++) {
-    block = MCU_data[blkn];
-
-    /* We simply emit the Al'th bit of the DC coefficient value. */
-    temp = (*block)[0];
-    emit_bits(entropy, (unsigned int) (temp >> Al), 1);
-  }
-
-  cinfo->dest->next_output_byte = entropy->next_output_byte;
-  cinfo->dest->free_in_buffer = entropy->free_in_buffer;
-
-  /* Update restart-interval state too */
-  if (cinfo->restart_interval) {
-    if (entropy->restarts_to_go == 0) {
-      entropy->restarts_to_go = cinfo->restart_interval;
-      entropy->next_restart_num++;
-      entropy->next_restart_num &= 7;
-    }
-    entropy->restarts_to_go--;
-  }
-
-  return TRUE;
-}
-
-
-/*
- * MCU encoding for AC successive approximation refinement scan.
- */
-
-METHODDEF(boolean)
-encode_mcu_AC_refine (j_compress_ptr cinfo, JBLOCKROW *MCU_data)
-{
-  phuff_entropy_ptr entropy = (phuff_entropy_ptr) cinfo->entropy;
-  register int temp;
-  register int r, k;
-  int EOB;
-  char *BR_buffer;
-  unsigned int BR;
-  int Se = cinfo->Se;
-  int Al = cinfo->Al;
-  JBLOCKROW block;
-  int absvalues[DCTSIZE2];
-
-  entropy->next_output_byte = cinfo->dest->next_output_byte;
-  entropy->free_in_buffer = cinfo->dest->free_in_buffer;
-
-  /* Emit restart marker if needed */
-  if (cinfo->restart_interval)
-    if (entropy->restarts_to_go == 0)
-      emit_restart(entropy, entropy->next_restart_num);
-
-  /* Encode the MCU data block */
-  block = MCU_data[0];
-
-  /* It is convenient to make a pre-pass to determine the transformed
-   * coefficients' absolute values and the EOB position.
-   */
-  EOB = 0;
-  for (k = cinfo->Ss; k <= Se; k++) {
-    temp = (*block)[jpeg_natural_order[k]];
-    /* We must apply the point transform by Al.  For AC coefficients this
-     * is an integer division with rounding towards 0.  To do this portably
-     * in C, we shift after obtaining the absolute value.
-     */
-    if (temp < 0)
-      temp = -temp;		/* temp is abs value of input */
-    temp >>= Al;		/* apply the point transform */
-    absvalues[k] = temp;	/* save abs value for main pass */
-    if (temp == 1)
-      EOB = k;			/* EOB = index of last newly-nonzero coef */
-  }
-
-  /* Encode the AC coefficients per section G.1.2.3, fig. G.7 */
-  
-  r = 0;			/* r = run length of zeros */
-  BR = 0;			/* BR = count of buffered bits added now */
-  BR_buffer = entropy->bit_buffer + entropy->BE; /* Append bits to buffer */
-
-  for (k = cinfo->Ss; k <= Se; k++) {
-    if ((temp = absvalues[k]) == 0) {
-      r++;
-      continue;
-    }
-
-    /* Emit any required ZRLs, but not if they can be folded into EOB */
-    while (r > 15 && k <= EOB) {
-      /* emit any pending EOBRUN and the BE correction bits */
-      emit_eobrun(entropy);
-      /* Emit ZRL */
-      emit_symbol(entropy, entropy->ac_tbl_no, 0xF0);
-      r -= 16;
-      /* Emit buffered correction bits that must be associated with ZRL */
-      emit_buffered_bits(entropy, BR_buffer, BR);
-      BR_buffer = entropy->bit_buffer; /* BE bits are gone now */
-      BR = 0;
-    }
-
-    /* If the coef was previously nonzero, it only needs a correction bit.
-     * NOTE: a straight translation of the spec's figure G.7 would suggest
-     * that we also need to test r > 15.  But if r > 15, we can only get here
-     * if k > EOB, which implies that this coefficient is not 1.
-     */
-    if (temp > 1) {
-      /* The correction bit is the next bit of the absolute value. */
-      BR_buffer[BR++] = (char) (temp & 1);
-      continue;
-    }
-
-    /* Emit any pending EOBRUN and the BE correction bits */
-    emit_eobrun(entropy);
-
-    /* Count/emit Huffman symbol for run length / number of bits */
-    emit_symbol(entropy, entropy->ac_tbl_no, (r << 4) + 1);
-
-    /* Emit output bit for newly-nonzero coef */
-    temp = ((*block)[jpeg_natural_order[k]] < 0) ? 0 : 1;
-    emit_bits(entropy, (unsigned int) temp, 1);
-
-    /* Emit buffered correction bits that must be associated with this code */
-    emit_buffered_bits(entropy, BR_buffer, BR);
-    BR_buffer = entropy->bit_buffer; /* BE bits are gone now */
-    BR = 0;
-    r = 0;			/* reset zero run length */
-  }
-
-  if (r > 0 || BR > 0) {	/* If there are trailing zeroes, */
-    entropy->EOBRUN++;		/* count an EOB */
-    entropy->BE += BR;		/* concat my correction bits to older ones */
-    /* We force out the EOB if we risk either:
-     * 1. overflow of the EOB counter;
-     * 2. overflow of the correction bit buffer during the next MCU.
-     */
-    if (entropy->EOBRUN == 0x7FFF || entropy->BE > (MAX_CORR_BITS-DCTSIZE2+1))
-      emit_eobrun(entropy);
-  }
-
-  cinfo->dest->next_output_byte = entropy->next_output_byte;
-  cinfo->dest->free_in_buffer = entropy->free_in_buffer;
-
-  /* Update restart-interval state too */
-  if (cinfo->restart_interval) {
-    if (entropy->restarts_to_go == 0) {
-      entropy->restarts_to_go = cinfo->restart_interval;
-      entropy->next_restart_num++;
-      entropy->next_restart_num &= 7;
-    }
-    entropy->restarts_to_go--;
-  }
-
-  return TRUE;
-}
-
-
-/*
- * Finish up at the end of a Huffman-compressed progressive scan.
- */
-
-METHODDEF(void)
-finish_pass_phuff (j_compress_ptr cinfo)
-{   
-  phuff_entropy_ptr entropy = (phuff_entropy_ptr) cinfo->entropy;
-
-  entropy->next_output_byte = cinfo->dest->next_output_byte;
-  entropy->free_in_buffer = cinfo->dest->free_in_buffer;
-
-  /* Flush out any buffered data */
-  emit_eobrun(entropy);
-  flush_bits(entropy);
-
-  cinfo->dest->next_output_byte = entropy->next_output_byte;
-  cinfo->dest->free_in_buffer = entropy->free_in_buffer;
-}
-
-
-/*
- * Finish up a statistics-gathering pass and create the new Huffman tables.
- */
-
-METHODDEF(void)
-finish_pass_gather_phuff (j_compress_ptr cinfo)
-{
-  phuff_entropy_ptr entropy = (phuff_entropy_ptr) cinfo->entropy;
-  boolean is_DC_band;
-  int ci, tbl;
-  jpeg_component_info * compptr;
-  JHUFF_TBL **htblptr;
-  boolean did[NUM_HUFF_TBLS];
-
-  /* Flush out buffered data (all we care about is counting the EOB symbol) */
-  emit_eobrun(entropy);
-
-  is_DC_band = (cinfo->Ss == 0);
-
-  /* It's important not to apply jpeg_gen_optimal_table more than once
-   * per table, because it clobbers the input frequency counts!
-   */
-  MEMZERO(did, SIZEOF(did));
-
-  for (ci = 0; ci < cinfo->comps_in_scan; ci++) {
-    compptr = cinfo->cur_comp_info[ci];
-    if (is_DC_band) {
-      if (cinfo->Ah != 0)	/* DC refinement needs no table */
-	continue;
-      tbl = compptr->dc_tbl_no;
-    } else {
-      tbl = compptr->ac_tbl_no;
-    }
-    if (! did[tbl]) {
-      if (is_DC_band)
-        htblptr = & cinfo->dc_huff_tbl_ptrs[tbl];
-      else
-        htblptr = & cinfo->ac_huff_tbl_ptrs[tbl];
-      if (*htblptr == NULL)
-        *htblptr = jpeg_alloc_huff_table((j_common_ptr) cinfo);
-      jpeg_gen_optimal_table(cinfo, *htblptr, entropy->count_ptrs[tbl]);
-      did[tbl] = TRUE;
-    }
-  }
-}
-
-
-/*
- * Module initialization routine for progressive Huffman entropy encoding.
- */
-
-GLOBAL(void)
-jinit_phuff_encoder (j_compress_ptr cinfo)
-{
-  phuff_entropy_ptr entropy;
-  int i;
-
-  entropy = (phuff_entropy_ptr)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				SIZEOF(phuff_entropy_encoder));
-  cinfo->entropy = (struct jpeg_entropy_encoder *) entropy;
-  entropy->pub.start_pass = start_pass_phuff;
-
-  /* Mark tables unallocated */
-  for (i = 0; i < NUM_HUFF_TBLS; i++) {
-    entropy->derived_tbls[i] = NULL;
-    entropy->count_ptrs[i] = NULL;
-  }
-  entropy->bit_buffer = NULL;	/* needed only in AC refinement scan */
-}
-
-#endif /* C_PROGRESSIVE_SUPPORTED */
@@ -1,354 +0,0 @@
-/*
- * jcprepct.c
- *
- * Copyright (C) 1994-1996, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains the compression preprocessing controller.
- * This controller manages the color conversion, downsampling,
- * and edge expansion steps.
- *
- * Most of the complexity here is associated with buffering input rows
- * as required by the downsampler.  See the comments at the head of
- * jcsample.c for the downsampler's needs.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-
-
-/* At present, jcsample.c can request context rows only for smoothing.
- * In the future, we might also need context rows for CCIR601 sampling
- * or other more-complex downsampling procedures.  The code to support
- * context rows should be compiled only if needed.
- */
-#ifdef INPUT_SMOOTHING_SUPPORTED
-#define CONTEXT_ROWS_SUPPORTED
-#endif
-
-
-/*
- * For the simple (no-context-row) case, we just need to buffer one
- * row group's worth of pixels for the downsampling step.  At the bottom of
- * the image, we pad to a full row group by replicating the last pixel row.
- * The downsampler's last output row is then replicated if needed to pad
- * out to a full iMCU row.
- *
- * When providing context rows, we must buffer three row groups' worth of
- * pixels.  Three row groups are physically allocated, but the row pointer
- * arrays are made five row groups high, with the extra pointers above and
- * below "wrapping around" to point to the last and first real row groups.
- * This allows the downsampler to access the proper context rows.
- * At the top and bottom of the image, we create dummy context rows by
- * copying the first or last real pixel row.  This copying could be avoided
- * by pointer hacking as is done in jdmainct.c, but it doesn't seem worth the
- * trouble on the compression side.
- */
-
-
-/* Private buffer controller object */
-
-typedef struct {
-  struct jpeg_c_prep_controller pub; /* public fields */
-
-  /* Downsampling input buffer.  This buffer holds color-converted data
-   * until we have enough to do a downsample step.
-   */
-  JSAMPARRAY color_buf[MAX_COMPONENTS];
-
-  JDIMENSION rows_to_go;	/* counts rows remaining in source image */
-  int next_buf_row;		/* index of next row to store in color_buf */
-
-#ifdef CONTEXT_ROWS_SUPPORTED	/* only needed for context case */
-  int this_row_group;		/* starting row index of group to process */
-  int next_buf_stop;		/* downsample when we reach this index */
-#endif
-} my_prep_controller;
-
-typedef my_prep_controller * my_prep_ptr;
-
-
-/*
- * Initialize for a processing pass.
- */
-
-METHODDEF(void)
-start_pass_prep (j_compress_ptr cinfo, J_BUF_MODE pass_mode)
-{
-  my_prep_ptr prep = (my_prep_ptr) cinfo->prep;
-
-  if (pass_mode != JBUF_PASS_THRU)
-    ERREXIT(cinfo, JERR_BAD_BUFFER_MODE);
-
-  /* Initialize total-height counter for detecting bottom of image */
-  prep->rows_to_go = cinfo->image_height;
-  /* Mark the conversion buffer empty */
-  prep->next_buf_row = 0;
-#ifdef CONTEXT_ROWS_SUPPORTED
-  /* Preset additional state variables for context mode.
-   * These aren't used in non-context mode, so we needn't test which mode.
-   */
-  prep->this_row_group = 0;
-  /* Set next_buf_stop to stop after two row groups have been read in. */
-  prep->next_buf_stop = 2 * cinfo->max_v_samp_factor;
-#endif
-}
-
-
-/*
- * Expand an image vertically from height input_rows to height output_rows,
- * by duplicating the bottom row.
- */
-
-LOCAL(void)
-expand_bottom_edge (JSAMPARRAY image_data, JDIMENSION num_cols,
-		    int input_rows, int output_rows)
-{
-  register int row;
-
-  for (row = input_rows; row < output_rows; row++) {
-    jcopy_sample_rows(image_data, input_rows-1, image_data, row,
-		      1, num_cols);
-  }
-}
-
-
-/*
- * Process some data in the simple no-context case.
- *
- * Preprocessor output data is counted in "row groups".  A row group
- * is defined to be v_samp_factor sample rows of each component.
- * Downsampling will produce this much data from each max_v_samp_factor
- * input rows.
- */
-
-METHODDEF(void)
-pre_process_data (j_compress_ptr cinfo,
-		  JSAMPARRAY input_buf, JDIMENSION *in_row_ctr,
-		  JDIMENSION in_rows_avail,
-		  JSAMPIMAGE output_buf, JDIMENSION *out_row_group_ctr,
-		  JDIMENSION out_row_groups_avail)
-{
-  my_prep_ptr prep = (my_prep_ptr) cinfo->prep;
-  int numrows, ci;
-  JDIMENSION inrows;
-  jpeg_component_info * compptr;
-
-  while (*in_row_ctr < in_rows_avail &&
-	 *out_row_group_ctr < out_row_groups_avail) {
-    /* Do color conversion to fill the conversion buffer. */
-    inrows = in_rows_avail - *in_row_ctr;
-    numrows = cinfo->max_v_samp_factor - prep->next_buf_row;
-    numrows = (int) MIN((JDIMENSION) numrows, inrows);
-    (*cinfo->cconvert->color_convert) (cinfo, input_buf + *in_row_ctr,
-				       prep->color_buf,
-				       (JDIMENSION) prep->next_buf_row,
-				       numrows);
-    *in_row_ctr += numrows;
-    prep->next_buf_row += numrows;
-    prep->rows_to_go -= numrows;
-    /* If at bottom of image, pad to fill the conversion buffer. */
-    if (prep->rows_to_go == 0 &&
-	prep->next_buf_row < cinfo->max_v_samp_factor) {
-      for (ci = 0; ci < cinfo->num_components; ci++) {
-	expand_bottom_edge(prep->color_buf[ci], cinfo->image_width,
-			   prep->next_buf_row, cinfo->max_v_samp_factor);
-      }
-      prep->next_buf_row = cinfo->max_v_samp_factor;
-    }
-    /* If we've filled the conversion buffer, empty it. */
-    if (prep->next_buf_row == cinfo->max_v_samp_factor) {
-      (*cinfo->downsample->downsample) (cinfo,
-					prep->color_buf, (JDIMENSION) 0,
-					output_buf, *out_row_group_ctr);
-      prep->next_buf_row = 0;
-      (*out_row_group_ctr)++;
-    }
-    /* If at bottom of image, pad the output to a full iMCU height.
-     * Note we assume the caller is providing a one-iMCU-height output buffer!
-     */
-    if (prep->rows_to_go == 0 &&
-	*out_row_group_ctr < out_row_groups_avail) {
-      for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
-	   ci++, compptr++) {
-	expand_bottom_edge(output_buf[ci],
-			   compptr->width_in_blocks * DCTSIZE,
-			   (int) (*out_row_group_ctr * compptr->v_samp_factor),
-			   (int) (out_row_groups_avail * compptr->v_samp_factor));
-      }
-      *out_row_group_ctr = out_row_groups_avail;
-      break;			/* can exit outer loop without test */
-    }
-  }
-}
-
-
-#ifdef CONTEXT_ROWS_SUPPORTED
-
-/*
- * Process some data in the context case.
- */
-
-METHODDEF(void)
-pre_process_context (j_compress_ptr cinfo,
-		     JSAMPARRAY input_buf, JDIMENSION *in_row_ctr,
-		     JDIMENSION in_rows_avail,
-		     JSAMPIMAGE output_buf, JDIMENSION *out_row_group_ctr,
-		     JDIMENSION out_row_groups_avail)
-{
-  my_prep_ptr prep = (my_prep_ptr) cinfo->prep;
-  int numrows, ci;
-  int buf_height = cinfo->max_v_samp_factor * 3;
-  JDIMENSION inrows;
-
-  while (*out_row_group_ctr < out_row_groups_avail) {
-    if (*in_row_ctr < in_rows_avail) {
-      /* Do color conversion to fill the conversion buffer. */
-      inrows = in_rows_avail - *in_row_ctr;
-      numrows = prep->next_buf_stop - prep->next_buf_row;
-      numrows = (int) MIN((JDIMENSION) numrows, inrows);
-      (*cinfo->cconvert->color_convert) (cinfo, input_buf + *in_row_ctr,
-					 prep->color_buf,
-					 (JDIMENSION) prep->next_buf_row,
-					 numrows);
-      /* Pad at top of image, if first time through */
-      if (prep->rows_to_go == cinfo->image_height) {
-	for (ci = 0; ci < cinfo->num_components; ci++) {
-	  int row;
-	  for (row = 1; row <= cinfo->max_v_samp_factor; row++) {
-	    jcopy_sample_rows(prep->color_buf[ci], 0,
-			      prep->color_buf[ci], -row,
-			      1, cinfo->image_width);
-	  }
-	}
-      }
-      *in_row_ctr += numrows;
-      prep->next_buf_row += numrows;
-      prep->rows_to_go -= numrows;
-    } else {
-      /* Return for more data, unless we are at the bottom of the image. */
-      if (prep->rows_to_go != 0)
-	break;
-      /* When at bottom of image, pad to fill the conversion buffer. */
-      if (prep->next_buf_row < prep->next_buf_stop) {
-	for (ci = 0; ci < cinfo->num_components; ci++) {
-	  expand_bottom_edge(prep->color_buf[ci], cinfo->image_width,
-			     prep->next_buf_row, prep->next_buf_stop);
-	}
-	prep->next_buf_row = prep->next_buf_stop;
-      }
-    }
-    /* If we've gotten enough data, downsample a row group. */
-    if (prep->next_buf_row == prep->next_buf_stop) {
-      (*cinfo->downsample->downsample) (cinfo,
-					prep->color_buf,
-					(JDIMENSION) prep->this_row_group,
-					output_buf, *out_row_group_ctr);
-      (*out_row_group_ctr)++;
-      /* Advance pointers with wraparound as necessary. */
-      prep->this_row_group += cinfo->max_v_samp_factor;
-      if (prep->this_row_group >= buf_height)
-	prep->this_row_group = 0;
-      if (prep->next_buf_row >= buf_height)
-	prep->next_buf_row = 0;
-      prep->next_buf_stop = prep->next_buf_row + cinfo->max_v_samp_factor;
-    }
-  }
-}
-
-
-/*
- * Create the wrapped-around downsampling input buffer needed for context mode.
- */
-
-LOCAL(void)
-create_context_buffer (j_compress_ptr cinfo)
-{
-  my_prep_ptr prep = (my_prep_ptr) cinfo->prep;
-  int rgroup_height = cinfo->max_v_samp_factor;
-  int ci, i;
-  jpeg_component_info * compptr;
-  JSAMPARRAY true_buffer, fake_buffer;
-
-  /* Grab enough space for fake row pointers for all the components;
-   * we need five row groups' worth of pointers for each component.
-   */
-  fake_buffer = (JSAMPARRAY)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				(cinfo->num_components * 5 * rgroup_height) *
-				SIZEOF(JSAMPROW));
-
-  for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
-       ci++, compptr++) {
-    /* Allocate the actual buffer space (3 row groups) for this component.
-     * We make the buffer wide enough to allow the downsampler to edge-expand
-     * horizontally within the buffer, if it so chooses.
-     */
-    true_buffer = (*cinfo->mem->alloc_sarray)
-      ((j_common_ptr) cinfo, JPOOL_IMAGE,
-       (JDIMENSION) (((long) compptr->width_in_blocks * DCTSIZE *
-		      cinfo->max_h_samp_factor) / compptr->h_samp_factor),
-       (JDIMENSION) (3 * rgroup_height));
-    /* Copy true buffer row pointers into the middle of the fake row array */
-    MEMCOPY(fake_buffer + rgroup_height, true_buffer,
-	    3 * rgroup_height * SIZEOF(JSAMPROW));
-    /* Fill in the above and below wraparound pointers */
-    for (i = 0; i < rgroup_height; i++) {
-      fake_buffer[i] = true_buffer[2 * rgroup_height + i];
-      fake_buffer[4 * rgroup_height + i] = true_buffer[i];
-    }
-    prep->color_buf[ci] = fake_buffer + rgroup_height;
-    fake_buffer += 5 * rgroup_height; /* point to space for next component */
-  }
-}
-
-#endif /* CONTEXT_ROWS_SUPPORTED */
-
-
-/*
- * Initialize preprocessing controller.
- */
-
-GLOBAL(void)
-jinit_c_prep_controller (j_compress_ptr cinfo, boolean need_full_buffer)
-{
-  my_prep_ptr prep;
-  int ci;
-  jpeg_component_info * compptr;
-
-  if (need_full_buffer)		/* safety check */
-    ERREXIT(cinfo, JERR_BAD_BUFFER_MODE);
-
-  prep = (my_prep_ptr)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				SIZEOF(my_prep_controller));
-  cinfo->prep = (struct jpeg_c_prep_controller *) prep;
-  prep->pub.start_pass = start_pass_prep;
-
-  /* Allocate the color conversion buffer.
-   * We make the buffer wide enough to allow the downsampler to edge-expand
-   * horizontally within the buffer, if it so chooses.
-   */
-  if (cinfo->downsample->need_context_rows) {
-    /* Set up to provide context rows */
-#ifdef CONTEXT_ROWS_SUPPORTED
-    prep->pub.pre_process_data = pre_process_context;
-    create_context_buffer(cinfo);
-#else
-    ERREXIT(cinfo, JERR_NOT_COMPILED);
-#endif
-  } else {
-    /* No context, just make it tall enough for one row group */
-    prep->pub.pre_process_data = pre_process_data;
-    for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
-	 ci++, compptr++) {
-      prep->color_buf[ci] = (*cinfo->mem->alloc_sarray)
-	((j_common_ptr) cinfo, JPOOL_IMAGE,
-	 (JDIMENSION) (((long) compptr->width_in_blocks * DCTSIZE *
-			cinfo->max_h_samp_factor) / compptr->h_samp_factor),
-	 (JDIMENSION) cinfo->max_v_samp_factor);
-    }
-  }
-}
@@ -1,519 +0,0 @@
-/*
- * jcsample.c
- *
- * Copyright (C) 1991-1996, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains downsampling routines.
- *
- * Downsampling input data is counted in "row groups".  A row group
- * is defined to be max_v_samp_factor pixel rows of each component,
- * from which the downsampler produces v_samp_factor sample rows.
- * A single row group is processed in each call to the downsampler module.
- *
- * The downsampler is responsible for edge-expansion of its output data
- * to fill an integral number of DCT blocks horizontally.  The source buffer
- * may be modified if it is helpful for this purpose (the source buffer is
- * allocated wide enough to correspond to the desired output width).
- * The caller (the prep controller) is responsible for vertical padding.
- *
- * The downsampler may request "context rows" by setting need_context_rows
- * during startup.  In this case, the input arrays will contain at least
- * one row group's worth of pixels above and below the passed-in data;
- * the caller will create dummy rows at image top and bottom by replicating
- * the first or last real pixel row.
- *
- * An excellent reference for image resampling is
- *   Digital Image Warping, George Wolberg, 1990.
- *   Pub. by IEEE Computer Society Press, Los Alamitos, CA. ISBN 0-8186-8944-7.
- *
- * The downsampling algorithm used here is a simple average of the source
- * pixels covered by the output pixel.  The hi-falutin sampling literature
- * refers to this as a "box filter".  In general the characteristics of a box
- * filter are not very good, but for the specific cases we normally use (1:1
- * and 2:1 ratios) the box is equivalent to a "triangle filter" which is not
- * nearly so bad.  If you intend to use other sampling ratios, you'd be well
- * advised to improve this code.
- *
- * A simple input-smoothing capability is provided.  This is mainly intended
- * for cleaning up color-dithered GIF input files (if you find it inadequate,
- * we suggest using an external filtering program such as pnmconvol).  When
- * enabled, each input pixel P is replaced by a weighted sum of itself and its
- * eight neighbors.  P's weight is 1-8*SF and each neighbor's weight is SF,
- * where SF = (smoothing_factor / 1024).
- * Currently, smoothing is only supported for 2h2v sampling factors.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-
-
-/* Pointer to routine to downsample a single component */
-typedef JMETHOD(void, downsample1_ptr,
-		(j_compress_ptr cinfo, jpeg_component_info * compptr,
-		 JSAMPARRAY input_data, JSAMPARRAY output_data));
-
-/* Private subobject */
-
-typedef struct {
-  struct jpeg_downsampler pub;	/* public fields */
-
-  /* Downsampling method pointers, one per component */
-  downsample1_ptr methods[MAX_COMPONENTS];
-} my_downsampler;
-
-typedef my_downsampler * my_downsample_ptr;
-
-
-/*
- * Initialize for a downsampling pass.
- */
-
-METHODDEF(void)
-start_pass_downsample (j_compress_ptr cinfo)
-{
-  /* no work for now */
-}
-
-
-/*
- * Expand a component horizontally from width input_cols to width output_cols,
- * by duplicating the rightmost samples.
- */
-
-LOCAL(void)
-expand_right_edge (JSAMPARRAY image_data, int num_rows,
-		   JDIMENSION input_cols, JDIMENSION output_cols)
-{
-  register JSAMPROW ptr;
-  register JSAMPLE pixval;
-  register int count;
-  int row;
-  int numcols = (int) (output_cols - input_cols);
-
-  if (numcols > 0) {
-    for (row = 0; row < num_rows; row++) {
-      ptr = image_data[row] + input_cols;
-      pixval = ptr[-1];		/* don't need GETJSAMPLE() here */
-      for (count = numcols; count > 0; count--)
-	*ptr++ = pixval;
-    }
-  }
-}
-
-
-/*
- * Do downsampling for a whole row group (all components).
- *
- * In this version we simply downsample each component independently.
- */
-
-METHODDEF(void)
-sep_downsample (j_compress_ptr cinfo,
-		JSAMPIMAGE input_buf, JDIMENSION in_row_index,
-		JSAMPIMAGE output_buf, JDIMENSION out_row_group_index)
-{
-  my_downsample_ptr downsample = (my_downsample_ptr) cinfo->downsample;
-  int ci;
-  jpeg_component_info * compptr;
-  JSAMPARRAY in_ptr, out_ptr;
-
-  for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
-       ci++, compptr++) {
-    in_ptr = input_buf[ci] + in_row_index;
-    out_ptr = output_buf[ci] + (out_row_group_index * compptr->v_samp_factor);
-    (*downsample->methods[ci]) (cinfo, compptr, in_ptr, out_ptr);
-  }
-}
-
-
-/*
- * Downsample pixel values of a single component.
- * One row group is processed per call.
- * This version handles arbitrary integral sampling ratios, without smoothing.
- * Note that this version is not actually used for customary sampling ratios.
- */
-
-METHODDEF(void)
-int_downsample (j_compress_ptr cinfo, jpeg_component_info * compptr,
-		JSAMPARRAY input_data, JSAMPARRAY output_data)
-{
-  int inrow, outrow, h_expand, v_expand, numpix, numpix2, h, v;
-  JDIMENSION outcol, outcol_h;	/* outcol_h == outcol*h_expand */
-  JDIMENSION output_cols = compptr->width_in_blocks * DCTSIZE;
-  JSAMPROW inptr, outptr;
-  INT32 outvalue;
-
-  h_expand = cinfo->max_h_samp_factor / compptr->h_samp_factor;
-  v_expand = cinfo->max_v_samp_factor / compptr->v_samp_factor;
-  numpix = h_expand * v_expand;
-  numpix2 = numpix/2;
-
-  /* Expand input data enough to let all the output samples be generated
-   * by the standard loop.  Special-casing padded output would be more
-   * efficient.
-   */
-  expand_right_edge(input_data, cinfo->max_v_samp_factor,
-		    cinfo->image_width, output_cols * h_expand);
-
-  inrow = 0;
-  for (outrow = 0; outrow < compptr->v_samp_factor; outrow++) {
-    outptr = output_data[outrow];
-    for (outcol = 0, outcol_h = 0; outcol < output_cols;
-	 outcol++, outcol_h += h_expand) {
-      outvalue = 0;
-      for (v = 0; v < v_expand; v++) {
-	inptr = input_data[inrow+v] + outcol_h;
-	for (h = 0; h < h_expand; h++) {
-	  outvalue += (INT32) GETJSAMPLE(*inptr++);
-	}
-      }
-      *outptr++ = (JSAMPLE) ((outvalue + numpix2) / numpix);
-    }
-    inrow += v_expand;
-  }
-}
-
-
-/*
- * Downsample pixel values of a single component.
- * This version handles the special case of a full-size component,
- * without smoothing.
- */
-
-METHODDEF(void)
-fullsize_downsample (j_compress_ptr cinfo, jpeg_component_info * compptr,
-		     JSAMPARRAY input_data, JSAMPARRAY output_data)
-{
-  /* Copy the data */
-  jcopy_sample_rows(input_data, 0, output_data, 0,
-		    cinfo->max_v_samp_factor, cinfo->image_width);
-  /* Edge-expand */
-  expand_right_edge(output_data, cinfo->max_v_samp_factor,
-		    cinfo->image_width, compptr->width_in_blocks * DCTSIZE);
-}
-
-
-/*
- * Downsample pixel values of a single component.
- * This version handles the common case of 2:1 horizontal and 1:1 vertical,
- * without smoothing.
- *
- * A note about the "bias" calculations: when rounding fractional values to
- * integer, we do not want to always round 0.5 up to the next integer.
- * If we did that, we'd introduce a noticeable bias towards larger values.
- * Instead, this code is arranged so that 0.5 will be rounded up or down at
- * alternate pixel locations (a simple ordered dither pattern).
- */
-
-METHODDEF(void)
-h2v1_downsample (j_compress_ptr cinfo, jpeg_component_info * compptr,
-		 JSAMPARRAY input_data, JSAMPARRAY output_data)
-{
-  int outrow;
-  JDIMENSION outcol;
-  JDIMENSION output_cols = compptr->width_in_blocks * DCTSIZE;
-  register JSAMPROW inptr, outptr;
-  register int bias;
-
-  /* Expand input data enough to let all the output samples be generated
-   * by the standard loop.  Special-casing padded output would be more
-   * efficient.
-   */
-  expand_right_edge(input_data, cinfo->max_v_samp_factor,
-		    cinfo->image_width, output_cols * 2);
-
-  for (outrow = 0; outrow < compptr->v_samp_factor; outrow++) {
-    outptr = output_data[outrow];
-    inptr = input_data[outrow];
-    bias = 0;			/* bias = 0,1,0,1,... for successive samples */
-    for (outcol = 0; outcol < output_cols; outcol++) {
-      *outptr++ = (JSAMPLE) ((GETJSAMPLE(*inptr) + GETJSAMPLE(inptr[1])
-			      + bias) >> 1);
-      bias ^= 1;		/* 0=>1, 1=>0 */
-      inptr += 2;
-    }
-  }
-}
-
-
-/*
- * Downsample pixel values of a single component.
- * This version handles the standard case of 2:1 horizontal and 2:1 vertical,
- * without smoothing.
- */
-
-METHODDEF(void)
-h2v2_downsample (j_compress_ptr cinfo, jpeg_component_info * compptr,
-		 JSAMPARRAY input_data, JSAMPARRAY output_data)
-{
-  int inrow, outrow;
-  JDIMENSION outcol;
-  JDIMENSION output_cols = compptr->width_in_blocks * DCTSIZE;
-  register JSAMPROW inptr0, inptr1, outptr;
-  register int bias;
-
-  /* Expand input data enough to let all the output samples be generated
-   * by the standard loop.  Special-casing padded output would be more
-   * efficient.
-   */
-  expand_right_edge(input_data, cinfo->max_v_samp_factor,
-		    cinfo->image_width, output_cols * 2);
-
-  inrow = 0;
-  for (outrow = 0; outrow < compptr->v_samp_factor; outrow++) {
-    outptr = output_data[outrow];
-    inptr0 = input_data[inrow];
-    inptr1 = input_data[inrow+1];
-    bias = 1;			/* bias = 1,2,1,2,... for successive samples */
-    for (outcol = 0; outcol < output_cols; outcol++) {
-      *outptr++ = (JSAMPLE) ((GETJSAMPLE(*inptr0) + GETJSAMPLE(inptr0[1]) +
-			      GETJSAMPLE(*inptr1) + GETJSAMPLE(inptr1[1])
-			      + bias) >> 2);
-      bias ^= 3;		/* 1=>2, 2=>1 */
-      inptr0 += 2; inptr1 += 2;
-    }
-    inrow += 2;
-  }
-}
-
-
-#ifdef INPUT_SMOOTHING_SUPPORTED
-
-/*
- * Downsample pixel values of a single component.
- * This version handles the standard case of 2:1 horizontal and 2:1 vertical,
- * with smoothing.  One row of context is required.
- */
-
-METHODDEF(void)
-h2v2_smooth_downsample (j_compress_ptr cinfo, jpeg_component_info * compptr,
-			JSAMPARRAY input_data, JSAMPARRAY output_data)
-{
-  int inrow, outrow;
-  JDIMENSION colctr;
-  JDIMENSION output_cols = compptr->width_in_blocks * DCTSIZE;
-  register JSAMPROW inptr0, inptr1, above_ptr, below_ptr, outptr;
-  INT32 membersum, neighsum, memberscale, neighscale;
-
-  /* Expand input data enough to let all the output samples be generated
-   * by the standard loop.  Special-casing padded output would be more
-   * efficient.
-   */
-  expand_right_edge(input_data - 1, cinfo->max_v_samp_factor + 2,
-		    cinfo->image_width, output_cols * 2);
-
-  /* We don't bother to form the individual "smoothed" input pixel values;
-   * we can directly compute the output which is the average of the four
-   * smoothed values.  Each of the four member pixels contributes a fraction
-   * (1-8*SF) to its own smoothed image and a fraction SF to each of the three
-   * other smoothed pixels, therefore a total fraction (1-5*SF)/4 to the final
-   * output.  The four corner-adjacent neighbor pixels contribute a fraction
-   * SF to just one smoothed pixel, or SF/4 to the final output; while the
-   * eight edge-adjacent neighbors contribute SF to each of two smoothed
-   * pixels, or SF/2 overall.  In order to use integer arithmetic, these
-   * factors are scaled by 2^16 = 65536.
-   * Also recall that SF = smoothing_factor / 1024.
-   */
-
-  memberscale = 16384 - cinfo->smoothing_factor * 80; /* scaled (1-5*SF)/4 */
-  neighscale = cinfo->smoothing_factor * 16; /* scaled SF/4 */
-
-  inrow = 0;
-  for (outrow = 0; outrow < compptr->v_samp_factor; outrow++) {
-    outptr = output_data[outrow];
-    inptr0 = input_data[inrow];
-    inptr1 = input_data[inrow+1];
-    above_ptr = input_data[inrow-1];
-    below_ptr = input_data[inrow+2];
-
-    /* Special case for first column: pretend column -1 is same as column 0 */
-    membersum = GETJSAMPLE(*inptr0) + GETJSAMPLE(inptr0[1]) +
-		GETJSAMPLE(*inptr1) + GETJSAMPLE(inptr1[1]);
-    neighsum = GETJSAMPLE(*above_ptr) + GETJSAMPLE(above_ptr[1]) +
-	       GETJSAMPLE(*below_ptr) + GETJSAMPLE(below_ptr[1]) +
-	       GETJSAMPLE(*inptr0) + GETJSAMPLE(inptr0[2]) +
-	       GETJSAMPLE(*inptr1) + GETJSAMPLE(inptr1[2]);
-    neighsum += neighsum;
-    neighsum += GETJSAMPLE(*above_ptr) + GETJSAMPLE(above_ptr[2]) +
-		GETJSAMPLE(*below_ptr) + GETJSAMPLE(below_ptr[2]);
-    membersum = membersum * memberscale + neighsum * neighscale;
-    *outptr++ = (JSAMPLE) ((membersum + 32768) >> 16);
-    inptr0 += 2; inptr1 += 2; above_ptr += 2; below_ptr += 2;
-
-    for (colctr = output_cols - 2; colctr > 0; colctr--) {
-      /* sum of pixels directly mapped to this output element */
-      membersum = GETJSAMPLE(*inptr0) + GETJSAMPLE(inptr0[1]) +
-		  GETJSAMPLE(*inptr1) + GETJSAMPLE(inptr1[1]);
-      /* sum of edge-neighbor pixels */
-      neighsum = GETJSAMPLE(*above_ptr) + GETJSAMPLE(above_ptr[1]) +
-		 GETJSAMPLE(*below_ptr) + GETJSAMPLE(below_ptr[1]) +
-		 GETJSAMPLE(inptr0[-1]) + GETJSAMPLE(inptr0[2]) +
-		 GETJSAMPLE(inptr1[-1]) + GETJSAMPLE(inptr1[2]);
-      /* The edge-neighbors count twice as much as corner-neighbors */
-      neighsum += neighsum;
-      /* Add in the corner-neighbors */
-      neighsum += GETJSAMPLE(above_ptr[-1]) + GETJSAMPLE(above_ptr[2]) +
-		  GETJSAMPLE(below_ptr[-1]) + GETJSAMPLE(below_ptr[2]);
-      /* form final output scaled up by 2^16 */
-      membersum = membersum * memberscale + neighsum * neighscale;
-      /* round, descale and output it */
-      *outptr++ = (JSAMPLE) ((membersum + 32768) >> 16);
-      inptr0 += 2; inptr1 += 2; above_ptr += 2; below_ptr += 2;
-    }
-
-    /* Special case for last column */
-    membersum = GETJSAMPLE(*inptr0) + GETJSAMPLE(inptr0[1]) +
-		GETJSAMPLE(*inptr1) + GETJSAMPLE(inptr1[1]);
-    neighsum = GETJSAMPLE(*above_ptr) + GETJSAMPLE(above_ptr[1]) +
-	       GETJSAMPLE(*below_ptr) + GETJSAMPLE(below_ptr[1]) +
-	       GETJSAMPLE(inptr0[-1]) + GETJSAMPLE(inptr0[1]) +
-	       GETJSAMPLE(inptr1[-1]) + GETJSAMPLE(inptr1[1]);
-    neighsum += neighsum;
-    neighsum += GETJSAMPLE(above_ptr[-1]) + GETJSAMPLE(above_ptr[1]) +
-		GETJSAMPLE(below_ptr[-1]) + GETJSAMPLE(below_ptr[1]);
-    membersum = membersum * memberscale + neighsum * neighscale;
-    *outptr = (JSAMPLE) ((membersum + 32768) >> 16);
-
-    inrow += 2;
-  }
-}
-
-
-/*
- * Downsample pixel values of a single component.
- * This version handles the special case of a full-size component,
- * with smoothing.  One row of context is required.
- */
-
-METHODDEF(void)
-fullsize_smooth_downsample (j_compress_ptr cinfo, jpeg_component_info *compptr,
-			    JSAMPARRAY input_data, JSAMPARRAY output_data)
-{
-  int outrow;
-  JDIMENSION colctr;
-  JDIMENSION output_cols = compptr->width_in_blocks * DCTSIZE;
-  register JSAMPROW inptr, above_ptr, below_ptr, outptr;
-  INT32 membersum, neighsum, memberscale, neighscale;
-  int colsum, lastcolsum, nextcolsum;
-
-  /* Expand input data enough to let all the output samples be generated
-   * by the standard loop.  Special-casing padded output would be more
-   * efficient.
-   */
-  expand_right_edge(input_data - 1, cinfo->max_v_samp_factor + 2,
-		    cinfo->image_width, output_cols);
-
-  /* Each of the eight neighbor pixels contributes a fraction SF to the
-   * smoothed pixel, while the main pixel contributes (1-8*SF).  In order
-   * to use integer arithmetic, these factors are multiplied by 2^16 = 65536.
-   * Also recall that SF = smoothing_factor / 1024.
-   */
-
-  memberscale = 65536L - cinfo->smoothing_factor * 512L; /* scaled 1-8*SF */
-  neighscale = cinfo->smoothing_factor * 64; /* scaled SF */
-
-  for (outrow = 0; outrow < compptr->v_samp_factor; outrow++) {
-    outptr = output_data[outrow];
-    inptr = input_data[outrow];
-    above_ptr = input_data[outrow-1];
-    below_ptr = input_data[outrow+1];
-
-    /* Special case for first column */
-    colsum = GETJSAMPLE(*above_ptr++) + GETJSAMPLE(*below_ptr++) +
-	     GETJSAMPLE(*inptr);
-    membersum = GETJSAMPLE(*inptr++);
-    nextcolsum = GETJSAMPLE(*above_ptr) + GETJSAMPLE(*below_ptr) +
-		 GETJSAMPLE(*inptr);
-    neighsum = colsum + (colsum - membersum) + nextcolsum;
-    membersum = membersum * memberscale + neighsum * neighscale;
-    *outptr++ = (JSAMPLE) ((membersum + 32768) >> 16);
-    lastcolsum = colsum; colsum = nextcolsum;
-
-    for (colctr = output_cols - 2; colctr > 0; colctr--) {
-      membersum = GETJSAMPLE(*inptr++);
-      above_ptr++; below_ptr++;
-      nextcolsum = GETJSAMPLE(*above_ptr) + GETJSAMPLE(*below_ptr) +
-		   GETJSAMPLE(*inptr);
-      neighsum = lastcolsum + (colsum - membersum) + nextcolsum;
-      membersum = membersum * memberscale + neighsum * neighscale;
-      *outptr++ = (JSAMPLE) ((membersum + 32768) >> 16);
-      lastcolsum = colsum; colsum = nextcolsum;
-    }
-
-    /* Special case for last column */
-    membersum = GETJSAMPLE(*inptr);
-    neighsum = lastcolsum + (colsum - membersum) + colsum;
-    membersum = membersum * memberscale + neighsum * neighscale;
-    *outptr = (JSAMPLE) ((membersum + 32768) >> 16);
-
-  }
-}
-
-#endif /* INPUT_SMOOTHING_SUPPORTED */
-
-
-/*
- * Module initialization routine for downsampling.
- * Note that we must select a routine for each component.
- */
-
-GLOBAL(void)
-jinit_downsampler (j_compress_ptr cinfo)
-{
-  my_downsample_ptr downsample;
-  int ci;
-  jpeg_component_info * compptr;
-  boolean smoothok = TRUE;
-
-  downsample = (my_downsample_ptr)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				SIZEOF(my_downsampler));
-  cinfo->downsample = (struct jpeg_downsampler *) downsample;
-  downsample->pub.start_pass = start_pass_downsample;
-  downsample->pub.downsample = sep_downsample;
-  downsample->pub.need_context_rows = FALSE;
-
-  if (cinfo->CCIR601_sampling)
-    ERREXIT(cinfo, JERR_CCIR601_NOTIMPL);
-
-  /* Verify we can handle the sampling factors, and set up method pointers */
-  for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
-       ci++, compptr++) {
-    if (compptr->h_samp_factor == cinfo->max_h_samp_factor &&
-	compptr->v_samp_factor == cinfo->max_v_samp_factor) {
-#ifdef INPUT_SMOOTHING_SUPPORTED
-      if (cinfo->smoothing_factor) {
-	downsample->methods[ci] = fullsize_smooth_downsample;
-	downsample->pub.need_context_rows = TRUE;
-      } else
-#endif
-	downsample->methods[ci] = fullsize_downsample;
-    } else if (compptr->h_samp_factor * 2 == cinfo->max_h_samp_factor &&
-	       compptr->v_samp_factor == cinfo->max_v_samp_factor) {
-      smoothok = FALSE;
-      downsample->methods[ci] = h2v1_downsample;
-    } else if (compptr->h_samp_factor * 2 == cinfo->max_h_samp_factor &&
-	       compptr->v_samp_factor * 2 == cinfo->max_v_samp_factor) {
-#ifdef INPUT_SMOOTHING_SUPPORTED
-      if (cinfo->smoothing_factor) {
-	downsample->methods[ci] = h2v2_smooth_downsample;
-	downsample->pub.need_context_rows = TRUE;
-      } else
-#endif
-	downsample->methods[ci] = h2v2_downsample;
-    } else if ((cinfo->max_h_samp_factor % compptr->h_samp_factor) == 0 &&
-	       (cinfo->max_v_samp_factor % compptr->v_samp_factor) == 0) {
-      smoothok = FALSE;
-      downsample->methods[ci] = int_downsample;
-    } else
-      ERREXIT(cinfo, JERR_FRACT_SAMPLE_NOTIMPL);
-  }
-
-#ifdef INPUT_SMOOTHING_SUPPORTED
-  if (cinfo->smoothing_factor && !smoothok)
-    TRACEMS(cinfo, 0, JTRC_SMOOTH_NOTIMPL);
-#endif
-}
@@ -1,388 +0,0 @@
-/*
- * jctrans.c
- *
- * Copyright (C) 1995-1998, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains library routines for transcoding compression,
- * that is, writing raw DCT coefficient arrays to an output JPEG file.
- * The routines in jcapimin.c will also be needed by a transcoder.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-
-
-/* Forward declarations */
-LOCAL(void) transencode_master_selection
-	JPP((j_compress_ptr cinfo, jvirt_barray_ptr * coef_arrays));
-LOCAL(void) transencode_coef_controller
-	JPP((j_compress_ptr cinfo, jvirt_barray_ptr * coef_arrays));
-
-
-/*
- * Compression initialization for writing raw-coefficient data.
- * Before calling this, all parameters and a data destination must be set up.
- * Call jpeg_finish_compress() to actually write the data.
- *
- * The number of passed virtual arrays must match cinfo->num_components.
- * Note that the virtual arrays need not be filled or even realized at
- * the time write_coefficients is called; indeed, if the virtual arrays
- * were requested from this compression object's memory manager, they
- * typically will be realized during this routine and filled afterwards.
- */
-
-GLOBAL(void)
-jpeg_write_coefficients (j_compress_ptr cinfo, jvirt_barray_ptr * coef_arrays)
-{
-  if (cinfo->global_state != CSTATE_START)
-    ERREXIT1(cinfo, JERR_BAD_STATE, cinfo->global_state);
-  /* Mark all tables to be written */
-  jpeg_suppress_tables(cinfo, FALSE);
-  /* (Re)initialize error mgr and destination modules */
-  (*cinfo->err->reset_error_mgr) ((j_common_ptr) cinfo);
-  (*cinfo->dest->init_destination) (cinfo);
-  /* Perform master selection of active modules */
-  transencode_master_selection(cinfo, coef_arrays);
-  /* Wait for jpeg_finish_compress() call */
-  cinfo->next_scanline = 0;	/* so jpeg_write_marker works */
-  cinfo->global_state = CSTATE_WRCOEFS;
-}
-
-
-/*
- * Initialize the compression object with default parameters,
- * then copy from the source object all parameters needed for lossless
- * transcoding.  Parameters that can be varied without loss (such as
- * scan script and Huffman optimization) are left in their default states.
- */
-
-GLOBAL(void)
-jpeg_copy_critical_parameters (j_decompress_ptr srcinfo,
-			       j_compress_ptr dstinfo)
-{
-  JQUANT_TBL ** qtblptr;
-  jpeg_component_info *incomp, *outcomp;
-  JQUANT_TBL *c_quant, *slot_quant;
-  int tblno, ci, coefi;
-
-  /* Safety check to ensure start_compress not called yet. */
-  if (dstinfo->global_state != CSTATE_START)
-    ERREXIT1(dstinfo, JERR_BAD_STATE, dstinfo->global_state);
-  /* Copy fundamental image dimensions */
-  dstinfo->image_width = srcinfo->image_width;
-  dstinfo->image_height = srcinfo->image_height;
-  dstinfo->input_components = srcinfo->num_components;
-  dstinfo->in_color_space = srcinfo->jpeg_color_space;
-  /* Initialize all parameters to default values */
-  jpeg_set_defaults(dstinfo);
-  /* jpeg_set_defaults may choose wrong colorspace, eg YCbCr if input is RGB.
-   * Fix it to get the right header markers for the image colorspace.
-   */
-  jpeg_set_colorspace(dstinfo, srcinfo->jpeg_color_space);
-  dstinfo->data_precision = srcinfo->data_precision;
-  dstinfo->CCIR601_sampling = srcinfo->CCIR601_sampling;
-  /* Copy the source's quantization tables. */
-  for (tblno = 0; tblno < NUM_QUANT_TBLS; tblno++) {
-    if (srcinfo->quant_tbl_ptrs[tblno] != NULL) {
-      qtblptr = & dstinfo->quant_tbl_ptrs[tblno];
-      if (*qtblptr == NULL)
-	*qtblptr = jpeg_alloc_quant_table((j_common_ptr) dstinfo);
-      MEMCOPY((*qtblptr)->quantval,
-	      srcinfo->quant_tbl_ptrs[tblno]->quantval,
-	      SIZEOF((*qtblptr)->quantval));
-      (*qtblptr)->sent_table = FALSE;
-    }
-  }
-  /* Copy the source's per-component info.
-   * Note we assume jpeg_set_defaults has allocated the dest comp_info array.
-   */
-  dstinfo->num_components = srcinfo->num_components;
-  if (dstinfo->num_components < 1 || dstinfo->num_components > MAX_COMPONENTS)
-    ERREXIT2(dstinfo, JERR_COMPONENT_COUNT, dstinfo->num_components,
-	     MAX_COMPONENTS);
-  for (ci = 0, incomp = srcinfo->comp_info, outcomp = dstinfo->comp_info;
-       ci < dstinfo->num_components; ci++, incomp++, outcomp++) {
-    outcomp->component_id = incomp->component_id;
-    outcomp->h_samp_factor = incomp->h_samp_factor;
-    outcomp->v_samp_factor = incomp->v_samp_factor;
-    outcomp->quant_tbl_no = incomp->quant_tbl_no;
-    /* Make sure saved quantization table for component matches the qtable
-     * slot.  If not, the input file re-used this qtable slot.
-     * IJG encoder currently cannot duplicate this.
-     */
-    tblno = outcomp->quant_tbl_no;
-    if (tblno < 0 || tblno >= NUM_QUANT_TBLS ||
-	srcinfo->quant_tbl_ptrs[tblno] == NULL)
-      ERREXIT1(dstinfo, JERR_NO_QUANT_TABLE, tblno);
-    slot_quant = srcinfo->quant_tbl_ptrs[tblno];
-    c_quant = incomp->quant_table;
-    if (c_quant != NULL) {
-      for (coefi = 0; coefi < DCTSIZE2; coefi++) {
-	if (c_quant->quantval[coefi] != slot_quant->quantval[coefi])
-	  ERREXIT1(dstinfo, JERR_MISMATCHED_QUANT_TABLE, tblno);
-      }
-    }
-    /* Note: we do not copy the source's Huffman table assignments;
-     * instead we rely on jpeg_set_colorspace to have made a suitable choice.
-     */
-  }
-  /* Also copy JFIF version and resolution information, if available.
-   * Strictly speaking this isn't "critical" info, but it's nearly
-   * always appropriate to copy it if available.  In particular,
-   * if the application chooses to copy JFIF 1.02 extension markers from
-   * the source file, we need to copy the version to make sure we don't
-   * emit a file that has 1.02 extensions but a claimed version of 1.01.
-   * We will *not*, however, copy version info from mislabeled "2.01" files.
-   */
-  if (srcinfo->saw_JFIF_marker) {
-    if (srcinfo->JFIF_major_version == 1) {
-      dstinfo->JFIF_major_version = srcinfo->JFIF_major_version;
-      dstinfo->JFIF_minor_version = srcinfo->JFIF_minor_version;
-    }
-    dstinfo->density_unit = srcinfo->density_unit;
-    dstinfo->X_density = srcinfo->X_density;
-    dstinfo->Y_density = srcinfo->Y_density;
-  }
-}
-
-
-/*
- * Master selection of compression modules for transcoding.
- * This substitutes for jcinit.c's initialization of the full compressor.
- */
-
-LOCAL(void)
-transencode_master_selection (j_compress_ptr cinfo,
-			      jvirt_barray_ptr * coef_arrays)
-{
-  /* Although we don't actually use input_components for transcoding,
-   * jcmaster.c's initial_setup will complain if input_components is 0.
-   */
-  cinfo->input_components = 1;
-  /* Initialize master control (includes parameter checking/processing) */
-  jinit_c_master_control(cinfo, TRUE /* transcode only */);
-
-  /* Entropy encoding: either Huffman or arithmetic coding. */
-  if (cinfo->arith_code) {
-    ERREXIT(cinfo, JERR_ARITH_NOTIMPL);
-  } else {
-    if (cinfo->progressive_mode) {
-#ifdef C_PROGRESSIVE_SUPPORTED
-      jinit_phuff_encoder(cinfo);
-#else
-      ERREXIT(cinfo, JERR_NOT_COMPILED);
-#endif
-    } else
-      jinit_huff_encoder(cinfo);
-  }
-
-  /* We need a special coefficient buffer controller. */
-  transencode_coef_controller(cinfo, coef_arrays);
-
-  jinit_marker_writer(cinfo);
-
-  /* We can now tell the memory manager to allocate virtual arrays. */
-  (*cinfo->mem->realize_virt_arrays) ((j_common_ptr) cinfo);
-
-  /* Write the datastream header (SOI, JFIF) immediately.
-   * Frame and scan headers are postponed till later.
-   * This lets application insert special markers after the SOI.
-   */
-  (*cinfo->marker->write_file_header) (cinfo);
-}
-
-
-/*
- * The rest of this file is a special implementation of the coefficient
- * buffer controller.  This is similar to jccoefct.c, but it handles only
- * output from presupplied virtual arrays.  Furthermore, we generate any
- * dummy padding blocks on-the-fly rather than expecting them to be present
- * in the arrays.
- */
-
-/* Private buffer controller object */
-
-typedef struct {
-  struct jpeg_c_coef_controller pub; /* public fields */
-
-  JDIMENSION iMCU_row_num;	/* iMCU row # within image */
-  JDIMENSION mcu_ctr;		/* counts MCUs processed in current row */
-  int MCU_vert_offset;		/* counts MCU rows within iMCU row */
-  int MCU_rows_per_iMCU_row;	/* number of such rows needed */
-
-  /* Virtual block array for each component. */
-  jvirt_barray_ptr * whole_image;
-
-  /* Workspace for constructing dummy blocks at right/bottom edges. */
-  JBLOCKROW dummy_buffer[C_MAX_BLOCKS_IN_MCU];
-} my_coef_controller;
-
-typedef my_coef_controller * my_coef_ptr;
-
-
-LOCAL(void)
-start_iMCU_row (j_compress_ptr cinfo)
-/* Reset within-iMCU-row counters for a new row */
-{
-  my_coef_ptr coef = (my_coef_ptr) cinfo->coef;
-
-  /* In an interleaved scan, an MCU row is the same as an iMCU row.
-   * In a noninterleaved scan, an iMCU row has v_samp_factor MCU rows.
-   * But at the bottom of the image, process only what's left.
-   */
-  if (cinfo->comps_in_scan > 1) {
-    coef->MCU_rows_per_iMCU_row = 1;
-  } else {
-    if (coef->iMCU_row_num < (cinfo->total_iMCU_rows-1))
-      coef->MCU_rows_per_iMCU_row = cinfo->cur_comp_info[0]->v_samp_factor;
-    else
-      coef->MCU_rows_per_iMCU_row = cinfo->cur_comp_info[0]->last_row_height;
-  }
-
-  coef->mcu_ctr = 0;
-  coef->MCU_vert_offset = 0;
-}
-
-
-/*
- * Initialize for a processing pass.
- */
-
-METHODDEF(void)
-start_pass_coef (j_compress_ptr cinfo, J_BUF_MODE pass_mode)
-{
-  my_coef_ptr coef = (my_coef_ptr) cinfo->coef;
-
-  if (pass_mode != JBUF_CRANK_DEST)
-    ERREXIT(cinfo, JERR_BAD_BUFFER_MODE);
-
-  coef->iMCU_row_num = 0;
-  start_iMCU_row(cinfo);
-}
-
-
-/*
- * Process some data.
- * We process the equivalent of one fully interleaved MCU row ("iMCU" row)
- * per call, ie, v_samp_factor block rows for each component in the scan.
- * The data is obtained from the virtual arrays and fed to the entropy coder.
- * Returns TRUE if the iMCU row is completed, FALSE if suspended.
- *
- * NB: input_buf is ignored; it is likely to be a NULL pointer.
- */
-
-METHODDEF(boolean)
-compress_output (j_compress_ptr cinfo, JSAMPIMAGE input_buf)
-{
-  my_coef_ptr coef = (my_coef_ptr) cinfo->coef;
-  JDIMENSION MCU_col_num;	/* index of current MCU within row */
-  JDIMENSION last_MCU_col = cinfo->MCUs_per_row - 1;
-  JDIMENSION last_iMCU_row = cinfo->total_iMCU_rows - 1;
-  int blkn, ci, xindex, yindex, yoffset, blockcnt;
-  JDIMENSION start_col;
-  JBLOCKARRAY buffer[MAX_COMPS_IN_SCAN];
-  JBLOCKROW MCU_buffer[C_MAX_BLOCKS_IN_MCU];
-  JBLOCKROW buffer_ptr;
-  jpeg_component_info *compptr;
-
-  /* Align the virtual buffers for the components used in this scan. */
-  for (ci = 0; ci < cinfo->comps_in_scan; ci++) {
-    compptr = cinfo->cur_comp_info[ci];
-    buffer[ci] = (*cinfo->mem->access_virt_barray)
-      ((j_common_ptr) cinfo, coef->whole_image[compptr->component_index],
-       coef->iMCU_row_num * compptr->v_samp_factor,
-       (JDIMENSION) compptr->v_samp_factor, FALSE);
-  }
-
-  /* Loop to process one whole iMCU row */
-  for (yoffset = coef->MCU_vert_offset; yoffset < coef->MCU_rows_per_iMCU_row;
-       yoffset++) {
-    for (MCU_col_num = coef->mcu_ctr; MCU_col_num < cinfo->MCUs_per_row;
-	 MCU_col_num++) {
-      /* Construct list of pointers to DCT blocks belonging to this MCU */
-      blkn = 0;			/* index of current DCT block within MCU */
-      for (ci = 0; ci < cinfo->comps_in_scan; ci++) {
-	compptr = cinfo->cur_comp_info[ci];
-	start_col = MCU_col_num * compptr->MCU_width;
-	blockcnt = (MCU_col_num < last_MCU_col) ? compptr->MCU_width
-						: compptr->last_col_width;
-	for (yindex = 0; yindex < compptr->MCU_height; yindex++) {
-	  if (coef->iMCU_row_num < last_iMCU_row ||
-	      yindex+yoffset < compptr->last_row_height) {
-	    /* Fill in pointers to real blocks in this row */
-	    buffer_ptr = buffer[ci][yindex+yoffset] + start_col;
-	    for (xindex = 0; xindex < blockcnt; xindex++)
-	      MCU_buffer[blkn++] = buffer_ptr++;
-	  } else {
-	    /* At bottom of image, need a whole row of dummy blocks */
-	    xindex = 0;
-	  }
-	  /* Fill in any dummy blocks needed in this row.
-	   * Dummy blocks are filled in the same way as in jccoefct.c:
-	   * all zeroes in the AC entries, DC entries equal to previous
-	   * block's DC value.  The init routine has already zeroed the
-	   * AC entries, so we need only set the DC entries correctly.
-	   */
-	  for (; xindex < compptr->MCU_width; xindex++) {
-	    MCU_buffer[blkn] = coef->dummy_buffer[blkn];
-	    MCU_buffer[blkn][0][0] = MCU_buffer[blkn-1][0][0];
-	    blkn++;
-	  }
-	}
-      }
-      /* Try to write the MCU. */
-      if (! (*cinfo->entropy->encode_mcu) (cinfo, MCU_buffer)) {
-	/* Suspension forced; update state counters and exit */
-	coef->MCU_vert_offset = yoffset;
-	coef->mcu_ctr = MCU_col_num;
-	return FALSE;
-      }
-    }
-    /* Completed an MCU row, but perhaps not an iMCU row */
-    coef->mcu_ctr = 0;
-  }
-  /* Completed the iMCU row, advance counters for next one */
-  coef->iMCU_row_num++;
-  start_iMCU_row(cinfo);
-  return TRUE;
-}
-
-
-/*
- * Initialize coefficient buffer controller.
- *
- * Each passed coefficient array must be the right size for that
- * coefficient: width_in_blocks wide and height_in_blocks high,
- * with unitheight at least v_samp_factor.
- */
-
-LOCAL(void)
-transencode_coef_controller (j_compress_ptr cinfo,
-			     jvirt_barray_ptr * coef_arrays)
-{
-  my_coef_ptr coef;
-  JBLOCKROW buffer;
-  int i;
-
-  coef = (my_coef_ptr)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				SIZEOF(my_coef_controller));
-  cinfo->coef = (struct jpeg_c_coef_controller *) coef;
-  coef->pub.start_pass = start_pass_coef;
-  coef->pub.compress_data = compress_output;
-
-  /* Save pointer to virtual arrays */
-  coef->whole_image = coef_arrays;
-
-  /* Allocate and pre-zero space for dummy DCT blocks. */
-  buffer = (JBLOCKROW)
-    (*cinfo->mem->alloc_large) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				C_MAX_BLOCKS_IN_MCU * SIZEOF(JBLOCK));
-  jzero_far((void FAR *) buffer, C_MAX_BLOCKS_IN_MCU * SIZEOF(JBLOCK));
-  for (i = 0; i < C_MAX_BLOCKS_IN_MCU; i++) {
-    coef->dummy_buffer[i] = buffer + i;
-  }
-}
@@ -1,401 +0,0 @@
-/*
- * jdapimin.c
- *
- * Copyright (C) 1994-1998, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains application interface code for the decompression half
- * of the JPEG library.  These are the "minimum" API routines that may be
- * needed in either the normal full-decompression case or the
- * transcoding-only case.
- *
- * Most of the routines intended to be called directly by an application
- * are in this file or in jdapistd.c.  But also see jcomapi.c for routines
- * shared by compression and decompression, and jdtrans.c for the transcoding
- * case.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-
-
-/*
- * Initialization of a JPEG decompression object.
- * The error manager must already be set up (in case memory manager fails).
- */
-
-GLOBAL(void)
-jpeg_CreateDecompress (j_decompress_ptr cinfo, int version, size_t structsize)
-{
-  int i;
-
-  /* Guard against version mismatches between library and caller. */
-  cinfo->mem = NULL;		/* so jpeg_destroy knows mem mgr not called */
-  if (version != JPEG_LIB_VERSION)
-    ERREXIT2(cinfo, JERR_BAD_LIB_VERSION, JPEG_LIB_VERSION, version);
-  if (structsize != SIZEOF(struct jpeg_decompress_struct))
-    ERREXIT2(cinfo, JERR_BAD_STRUCT_SIZE, 
-	     (int) SIZEOF(struct jpeg_decompress_struct), (int) structsize);
-
-  /* For debugging purposes, we zero the whole master structure.
-   * But the application has already set the err pointer, and may have set
-   * client_data, so we have to save and restore those fields.
-   * Note: if application hasn't set client_data, tools like Purify may
-   * complain here.
-   */
-  {
-    struct jpeg_error_mgr * err = cinfo->err;
-    void * client_data = cinfo->client_data; /* ignore Purify complaint here */
-    MEMZERO(cinfo, SIZEOF(struct jpeg_decompress_struct));
-    cinfo->err = err;
-    cinfo->client_data = client_data;
-  }
-  cinfo->is_decompressor = TRUE;
-  cinfo->tile_decode = FALSE;
-
-  /* Initialize a memory manager instance for this object */
-  jinit_memory_mgr((j_common_ptr) cinfo);
-
-  /* Zero out pointers to permanent structures. */
-  cinfo->progress = NULL;
-  cinfo->src = NULL;
-
-  for (i = 0; i < NUM_QUANT_TBLS; i++)
-    cinfo->quant_tbl_ptrs[i] = NULL;
-
-  for (i = 0; i < NUM_HUFF_TBLS; i++) {
-    cinfo->dc_huff_tbl_ptrs[i] = NULL;
-    cinfo->ac_huff_tbl_ptrs[i] = NULL;
-  }
-
-  /* Initialize marker processor so application can override methods
-   * for COM, APPn markers before calling jpeg_read_header.
-   */
-  cinfo->marker_list = NULL;
-  jinit_marker_reader(cinfo);
-
-  /* And initialize the overall input controller. */
-  jinit_input_controller(cinfo);
-
-  /* OK, I'm ready */
-  cinfo->global_state = DSTATE_START;
-}
-
-
-/*
- * Destruction of a JPEG decompression object
- */
-
-GLOBAL(void)
-jpeg_destroy_decompress (j_decompress_ptr cinfo)
-{
-  jpeg_destroy((j_common_ptr) cinfo); /* use common routine */
-}
-
-
-/*
- * Abort processing of a JPEG decompression operation,
- * but don't destroy the object itself.
- */
-
-GLOBAL(void)
-jpeg_abort_decompress (j_decompress_ptr cinfo)
-{
-  jpeg_abort((j_common_ptr) cinfo); /* use common routine */
-}
-
-
-/*
- * Set default decompression parameters.
- */
-
-LOCAL(void)
-default_decompress_parms (j_decompress_ptr cinfo)
-{
-  /* Guess the input colorspace, and set output colorspace accordingly. */
-  /* (Wish JPEG committee had provided a real way to specify this...) */
-  /* Note application may override our guesses. */
-  switch (cinfo->num_components) {
-  case 1:
-    cinfo->jpeg_color_space = JCS_GRAYSCALE;
-    cinfo->out_color_space = JCS_GRAYSCALE;
-    break;
-    
-  case 3:
-    if (cinfo->saw_JFIF_marker) {
-      cinfo->jpeg_color_space = JCS_YCbCr; /* JFIF implies YCbCr */
-    } else if (cinfo->saw_Adobe_marker) {
-      switch (cinfo->Adobe_transform) {
-      case 0:
-	cinfo->jpeg_color_space = JCS_RGB;
-	break;
-      case 1:
-	cinfo->jpeg_color_space = JCS_YCbCr;
-	break;
-      default:
-	WARNMS1(cinfo, JWRN_ADOBE_XFORM, cinfo->Adobe_transform);
-	cinfo->jpeg_color_space = JCS_YCbCr; /* assume it's YCbCr */
-	break;
-      }
-    } else {
-      /* Saw no special markers, try to guess from the component IDs */
-      int cid0 = cinfo->comp_info[0].component_id;
-      int cid1 = cinfo->comp_info[1].component_id;
-      int cid2 = cinfo->comp_info[2].component_id;
-
-      if (cid0 == 1 && cid1 == 2 && cid2 == 3)
-	cinfo->jpeg_color_space = JCS_YCbCr; /* assume JFIF w/out marker */
-      else if (cid0 == 82 && cid1 == 71 && cid2 == 66)
-	cinfo->jpeg_color_space = JCS_RGB; /* ASCII 'R', 'G', 'B' */
-      else {
-	TRACEMS3(cinfo, 1, JTRC_UNKNOWN_IDS, cid0, cid1, cid2);
-	cinfo->jpeg_color_space = JCS_YCbCr; /* assume it's YCbCr */
-      }
-    }
-    /* Always guess RGB is proper output colorspace. */
-    cinfo->out_color_space = JCS_RGB;
-    break;
-    
-  case 4:
-    if (cinfo->saw_Adobe_marker) {
-      switch (cinfo->Adobe_transform) {
-      case 0:
-	cinfo->jpeg_color_space = JCS_CMYK;
-	break;
-      case 2:
-	cinfo->jpeg_color_space = JCS_YCCK;
-	break;
-      default:
-	WARNMS1(cinfo, JWRN_ADOBE_XFORM, cinfo->Adobe_transform);
-	cinfo->jpeg_color_space = JCS_YCCK; /* assume it's YCCK */
-	break;
-      }
-    } else {
-      /* No special markers, assume straight CMYK. */
-      cinfo->jpeg_color_space = JCS_CMYK;
-    }
-    cinfo->out_color_space = JCS_CMYK;
-    break;
-    
-  default:
-    cinfo->jpeg_color_space = JCS_UNKNOWN;
-    cinfo->out_color_space = JCS_UNKNOWN;
-    break;
-  }
-
-  /* Set defaults for other decompression parameters. */
-  cinfo->scale_num = 1;		/* 1:1 scaling */
-  cinfo->scale_denom = 1;
-  cinfo->output_gamma = 1.0;
-  cinfo->buffered_image = FALSE;
-  cinfo->raw_data_out = FALSE;
-  cinfo->dct_method = JDCT_DEFAULT;
-  cinfo->do_fancy_upsampling = TRUE;
-  cinfo->do_block_smoothing = TRUE;
-  cinfo->quantize_colors = FALSE;
-  /* We set these in case application only sets quantize_colors. */
-  cinfo->dither_mode = JDITHER_FS;
-#ifdef QUANT_2PASS_SUPPORTED
-  cinfo->two_pass_quantize = TRUE;
-#else
-  cinfo->two_pass_quantize = FALSE;
-#endif
-  cinfo->desired_number_of_colors = 256;
-  cinfo->colormap = NULL;
-  /* Initialize for no mode change in buffered-image mode. */
-  cinfo->enable_1pass_quant = FALSE;
-  cinfo->enable_external_quant = FALSE;
-  cinfo->enable_2pass_quant = FALSE;
-}
-
-
-/*
- * Decompression startup: read start of JPEG datastream to see what's there.
- * Need only initialize JPEG object and supply a data source before calling.
- *
- * This routine will read as far as the first SOS marker (ie, actual start of
- * compressed data), and will save all tables and parameters in the JPEG
- * object.  It will also initialize the decompression parameters to default
- * values, and finally return JPEG_HEADER_OK.  On return, the application may
- * adjust the decompression parameters and then call jpeg_start_decompress.
- * (Or, if the application only wanted to determine the image parameters,
- * the data need not be decompressed.  In that case, call jpeg_abort or
- * jpeg_destroy to release any temporary space.)
- * If an abbreviated (tables only) datastream is presented, the routine will
- * return JPEG_HEADER_TABLES_ONLY upon reaching EOI.  The application may then
- * re-use the JPEG object to read the abbreviated image datastream(s).
- * It is unnecessary (but OK) to call jpeg_abort in this case.
- * The JPEG_SUSPENDED return code only occurs if the data source module
- * requests suspension of the decompressor.  In this case the application
- * should load more source data and then re-call jpeg_read_header to resume
- * processing.
- * If a non-suspending data source is used and require_image is TRUE, then the
- * return code need not be inspected since only JPEG_HEADER_OK is possible.
- *
- * This routine is now just a front end to jpeg_consume_input, with some
- * extra error checking.
- */
-
-GLOBAL(int)
-jpeg_read_header (j_decompress_ptr cinfo, boolean require_image)
-{
-  int retcode;
-
-  if (cinfo->global_state != DSTATE_START &&
-      cinfo->global_state != DSTATE_INHEADER)
-    ERREXIT1(cinfo, JERR_BAD_STATE, cinfo->global_state);
-
-  retcode = jpeg_consume_input(cinfo);
-
-  switch (retcode) {
-  case JPEG_REACHED_SOS:
-    retcode = JPEG_HEADER_OK;
-    break;
-  case JPEG_REACHED_EOI:
-    if (require_image)		/* Complain if application wanted an image */
-      ERREXIT(cinfo, JERR_NO_IMAGE);
-    /* Reset to start state; it would be safer to require the application to
-     * call jpeg_abort, but we can't change it now for compatibility reasons.
-     * A side effect is to free any temporary memory (there shouldn't be any).
-     */
-    jpeg_abort((j_common_ptr) cinfo); /* sets state = DSTATE_START */
-    retcode = JPEG_HEADER_TABLES_ONLY;
-    break;
-  case JPEG_SUSPENDED:
-    /* no work */
-    break;
-  }
-
-  return retcode;
-}
-
-
-/*
- * Consume data in advance of what the decompressor requires.
- * This can be called at any time once the decompressor object has
- * been created and a data source has been set up.
- *
- * This routine is essentially a state machine that handles a couple
- * of critical state-transition actions, namely initial setup and
- * transition from header scanning to ready-for-start_decompress.
- * All the actual input is done via the input controller's consume_input
- * method.
- */
-
-GLOBAL(int)
-jpeg_consume_input (j_decompress_ptr cinfo)
-{
-  int retcode = JPEG_SUSPENDED;
-
-  /* NB: every possible DSTATE value should be listed in this switch */
-  switch (cinfo->global_state) {
-  case DSTATE_START:
-    /* Start-of-datastream actions: reset appropriate modules */
-    (*cinfo->inputctl->reset_input_controller) (cinfo);
-    /* Initialize application's data source module */
-    (*cinfo->src->init_source) (cinfo);
-    cinfo->global_state = DSTATE_INHEADER;
-    /*FALLTHROUGH*/
-  case DSTATE_INHEADER:
-    retcode = (*cinfo->inputctl->consume_input) (cinfo);
-    if (retcode == JPEG_REACHED_SOS) { /* Found SOS, prepare to decompress */
-      /* Set up default parameters based on header data */
-      default_decompress_parms(cinfo);
-      /* Set global state: ready for start_decompress */
-      cinfo->global_state = DSTATE_READY;
-    }
-    break;
-  case DSTATE_READY:
-    /* Can't advance past first SOS until start_decompress is called */
-    retcode = JPEG_REACHED_SOS;
-    break;
-  case DSTATE_PRELOAD:
-  case DSTATE_PRESCAN:
-  case DSTATE_SCANNING:
-  case DSTATE_RAW_OK:
-  case DSTATE_BUFIMAGE:
-  case DSTATE_BUFPOST:
-  case DSTATE_STOPPING:
-    retcode = (*cinfo->inputctl->consume_input) (cinfo);
-    break;
-  default:
-    ERREXIT1(cinfo, JERR_BAD_STATE, cinfo->global_state);
-  }
-  return retcode;
-}
-
-
-/*
- * Have we finished reading the input file?
- */
-
-GLOBAL(boolean)
-jpeg_input_complete (j_decompress_ptr cinfo)
-{
-  /* Check for valid jpeg object */
-  if (cinfo->global_state < DSTATE_START ||
-      cinfo->global_state > DSTATE_STOPPING)
-    ERREXIT1(cinfo, JERR_BAD_STATE, cinfo->global_state);
-  return cinfo->inputctl->eoi_reached;
-}
-
-
-/*
- * Is there more than one scan?
- */
-
-GLOBAL(boolean)
-jpeg_has_multiple_scans (j_decompress_ptr cinfo)
-{
-  /* Only valid after jpeg_read_header completes */
-  if (cinfo->global_state < DSTATE_READY ||
-      cinfo->global_state > DSTATE_STOPPING)
-    ERREXIT1(cinfo, JERR_BAD_STATE, cinfo->global_state);
-  return cinfo->inputctl->has_multiple_scans;
-}
-
-
-/*
- * Finish JPEG decompression.
- *
- * This will normally just verify the file trailer and release temp storage.
- *
- * Returns FALSE if suspended.  The return value need be inspected only if
- * a suspending data source is used.
- */
-
-GLOBAL(boolean)
-jpeg_finish_decompress (j_decompress_ptr cinfo)
-{
-  if ((cinfo->global_state == DSTATE_SCANNING ||
-       cinfo->global_state == DSTATE_RAW_OK) && ! cinfo->buffered_image) {
-    /* Terminate final pass of non-buffered mode */
-#ifdef ANDROID_TILE_BASED_DECODE
-    cinfo->output_scanline = cinfo->output_height;
-#endif
-    if (cinfo->output_scanline < cinfo->output_height)
-      ERREXIT(cinfo, JERR_TOO_LITTLE_DATA);
-    (*cinfo->master->finish_output_pass) (cinfo);
-    cinfo->global_state = DSTATE_STOPPING;
-  } else if (cinfo->global_state == DSTATE_BUFIMAGE) {
-    /* Finishing after a buffered-image operation */
-    cinfo->global_state = DSTATE_STOPPING;
-  } else if (cinfo->global_state != DSTATE_STOPPING) {
-    /* STOPPING = repeat call after a suspension, anything else is error */
-    ERREXIT1(cinfo, JERR_BAD_STATE, cinfo->global_state);
-  }
-  /* Read until EOI */
-#ifndef ANDROID_TILE_BASED_DECODE
-  while (! cinfo->inputctl->eoi_reached) {
-    if ((*cinfo->inputctl->consume_input) (cinfo) == JPEG_SUSPENDED)
-      return FALSE;		/* Suspend, come back later */
-  }
-#endif
-  /* Do final cleanup */
-  (*cinfo->src->term_source) (cinfo);
-  /* We can use jpeg_abort to release memory and reset global_state */
-  jpeg_abort((j_common_ptr) cinfo);
-  return TRUE;
-}
@@ -1,397 +0,0 @@
-/*
- * jdapistd.c
- *
- * Copyright (C) 1994-1996, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains application interface code for the decompression half
- * of the JPEG library.  These are the "standard" API routines that are
- * used in the normal full-decompression case.  They are not used by a
- * transcoding-only application.  Note that if an application links in
- * jpeg_start_decompress, it will end up linking in the entire decompressor.
- * We thus must separate this file from jdapimin.c to avoid linking the
- * whole decompression library into a transcoder.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-
-
-/* Forward declarations */
-LOCAL(boolean) output_pass_setup JPP((j_decompress_ptr cinfo));
-
-
-/*
- * Decompression initialization.
- * jpeg_read_header must be completed before calling this.
- *
- * If a multipass operating mode was selected, this will do all but the
- * last pass, and thus may take a great deal of time.
- *
- * Returns FALSE if suspended.  The return value need be inspected only if
- * a suspending data source is used.
- */
-
-GLOBAL(boolean)
-jpeg_start_decompress (j_decompress_ptr cinfo)
-{
-  if (cinfo->global_state == DSTATE_READY) {
-    /* First call: initialize master control, select active modules */
-    jinit_master_decompress(cinfo);
-    if (cinfo->buffered_image) {
-      /* No more work here; expecting jpeg_start_output next */
-      cinfo->global_state = DSTATE_BUFIMAGE;
-      return TRUE;
-    }
-    cinfo->global_state = DSTATE_PRELOAD;
-  }
-  if (cinfo->global_state == DSTATE_PRELOAD) {
-    /* If file has multiple scans, absorb them all into the coef buffer */
-    if (cinfo->inputctl->has_multiple_scans) {
-#ifdef D_MULTISCAN_FILES_SUPPORTED
-      for (;;) {
-	int retcode;
-	/* Call progress monitor hook if present */
-	if (cinfo->progress != NULL)
-	  (*cinfo->progress->progress_monitor) ((j_common_ptr) cinfo);
-	/* Absorb some more input */
-	retcode = (*cinfo->inputctl->consume_input) (cinfo);
-	if (retcode == JPEG_SUSPENDED)
-	  return FALSE;
-	if (retcode == JPEG_REACHED_EOI)
-	  break;
-	/* Advance progress counter if appropriate */
-	if (cinfo->progress != NULL &&
-	    (retcode == JPEG_ROW_COMPLETED || retcode == JPEG_REACHED_SOS)) {
-	  if (++cinfo->progress->pass_counter >= cinfo->progress->pass_limit) {
-	    /* jdmaster underestimated number of scans; ratchet up one scan */
-	    cinfo->progress->pass_limit += (long) cinfo->total_iMCU_rows;
-	  }
-	}
-      }
-#else
-      ERREXIT(cinfo, JERR_NOT_COMPILED);
-#endif /* D_MULTISCAN_FILES_SUPPORTED */
-    }
-    cinfo->output_scan_number = cinfo->input_scan_number;
-  } else if (cinfo->global_state != DSTATE_PRESCAN)
-    ERREXIT1(cinfo, JERR_BAD_STATE, cinfo->global_state);
-  /* Perform any dummy output passes, and set up for the final pass */
-  return output_pass_setup(cinfo);
-}
-
-/*
- * Tile decompression initialization.
- * jpeg_read_header must be completed before calling this.
- */
-
-GLOBAL(boolean)
-jpeg_start_tile_decompress (j_decompress_ptr cinfo)
-{
-  if (cinfo->global_state == DSTATE_READY) {
-    /* First call: initialize master control, select active modules */
-    cinfo->tile_decode = TRUE;
-    jinit_master_decompress(cinfo);
-    if (cinfo->buffered_image) {
-      cinfo->global_state = DSTATE_BUFIMAGE;
-      return TRUE;
-    }
-    cinfo->global_state = DSTATE_PRELOAD;
-  }
-  if (cinfo->global_state == DSTATE_PRELOAD) {
-    cinfo->output_scan_number = cinfo->input_scan_number;
-  } else if (cinfo->global_state != DSTATE_PRESCAN)
-    ERREXIT1(cinfo, JERR_BAD_STATE, cinfo->global_state);
-  /* Perform any dummy output passes, and set up for the final pass */
-  return output_pass_setup(cinfo);
-}
-
-
-/*
- * Set up for an output pass, and perform any dummy pass(es) needed.
- * Common subroutine for jpeg_start_decompress and jpeg_start_output.
- * Entry: global_state = DSTATE_PRESCAN only if previously suspended.
- * Exit: If done, returns TRUE and sets global_state for proper output mode.
- *       If suspended, returns FALSE and sets global_state = DSTATE_PRESCAN.
- */
-
-LOCAL(boolean)
-output_pass_setup (j_decompress_ptr cinfo)
-{
-  if (cinfo->global_state != DSTATE_PRESCAN) {
-    /* First call: do pass setup */
-    (*cinfo->master->prepare_for_output_pass) (cinfo);
-    cinfo->output_scanline = 0;
-    cinfo->global_state = DSTATE_PRESCAN;
-  }
-  /* Loop over any required dummy passes */
-  while (cinfo->master->is_dummy_pass) {
-#ifdef QUANT_2PASS_SUPPORTED
-    /* Crank through the dummy pass */
-    while (cinfo->output_scanline < cinfo->output_height) {
-      JDIMENSION last_scanline;
-      /* Call progress monitor hook if present */
-      if (cinfo->progress != NULL) {
-	cinfo->progress->pass_counter = (long) cinfo->output_scanline;
-	cinfo->progress->pass_limit = (long) cinfo->output_height;
-	(*cinfo->progress->progress_monitor) ((j_common_ptr) cinfo);
-      }
-      /* Process some data */
-      last_scanline = cinfo->output_scanline;
-      (*cinfo->main->process_data) (cinfo, (JSAMPARRAY) NULL,
-				    &cinfo->output_scanline, (JDIMENSION) 0);
-      if (cinfo->output_scanline == last_scanline)
-	return FALSE;		/* No progress made, must suspend */
-    }
-    /* Finish up dummy pass, and set up for another one */
-    (*cinfo->master->finish_output_pass) (cinfo);
-    (*cinfo->master->prepare_for_output_pass) (cinfo);
-    cinfo->output_scanline = 0;
-#else
-    ERREXIT(cinfo, JERR_NOT_COMPILED);
-#endif /* QUANT_2PASS_SUPPORTED */
-  }
-  /* Ready for application to drive output pass through
-   * jpeg_read_scanlines or jpeg_read_raw_data.
-   */
-  cinfo->global_state = cinfo->raw_data_out ? DSTATE_RAW_OK : DSTATE_SCANNING;
-  return TRUE;
-}
-
-
-/*
- * Read some scanlines of data from the JPEG decompressor.
- *
- * The return value will be the number of lines actually read.
- * This may be less than the number requested in several cases,
- * including bottom of image, data source suspension, and operating
- * modes that emit multiple scanlines at a time.
- *
- * Note: we warn about excess calls to jpeg_read_scanlines() since
- * this likely signals an application programmer error.  However,
- * an oversize buffer (max_lines > scanlines remaining) is not an error.
- */
-
-GLOBAL(JDIMENSION)
-jpeg_read_scanlines (j_decompress_ptr cinfo, JSAMPARRAY scanlines,
-		     JDIMENSION max_lines)
-{
-  JDIMENSION row_ctr;
-
-  if (cinfo->global_state != DSTATE_SCANNING)
-    ERREXIT1(cinfo, JERR_BAD_STATE, cinfo->global_state);
-  if (cinfo->output_scanline >= cinfo->output_height) {
-    WARNMS(cinfo, JWRN_TOO_MUCH_DATA);
-    return 0;
-  }
-
-  /* Call progress monitor hook if present */
-  if (cinfo->progress != NULL) {
-    cinfo->progress->pass_counter = (long) cinfo->output_scanline;
-    cinfo->progress->pass_limit = (long) cinfo->output_height;
-    (*cinfo->progress->progress_monitor) ((j_common_ptr) cinfo);
-  }
-
-  /* Process some data */
-  row_ctr = 0;
-  (*cinfo->main->process_data) (cinfo, scanlines, &row_ctr, max_lines);
-  cinfo->output_scanline += row_ctr;
-  return row_ctr;
-}
-/*
- * Initialize the jpeg decoder to decompressing a rectangle with size of (width, height)
- * and its upper-left corner located at (start_x, start_y).
- * Align start_x and start_y to multiplies of iMCU width and height, respectively.
- * Also, the new reader position and sampled image size will be returned in
- * (start_x, start_y) and (width, height), respectively.
- */
-
-GLOBAL(void)
-jpeg_init_read_tile_scanline(j_decompress_ptr cinfo, huffman_index *index,
-		     int *start_x, int *start_y, int *width, int *height)
-{
-  // Calculates the boundary of iMCU
-  int lines_per_iMCU_row = cinfo->max_v_samp_factor * DCTSIZE;
-  int lines_per_iMCU_col = cinfo->max_h_samp_factor * DCTSIZE;
-  int row_offset = *start_y / lines_per_iMCU_row;
-  int col_left_boundary = ((*start_x / lines_per_iMCU_col)
-            / index->MCU_sample_size) * index->MCU_sample_size;
-  int col_right_boundary =
-                  jdiv_round_up(*start_x + *width, lines_per_iMCU_col);
-
-  cinfo->coef->MCU_columns_to_skip =
-      *start_x / lines_per_iMCU_col - col_left_boundary;
-
-  *height = (*start_y - row_offset * lines_per_iMCU_row) + *height;
-  *start_x = col_left_boundary * lines_per_iMCU_col;
-  *start_y = row_offset * lines_per_iMCU_row;
-  cinfo->image_width = jmin(cinfo->original_image_width,
-          col_right_boundary * lines_per_iMCU_col) -
-          col_left_boundary * lines_per_iMCU_col;
-  cinfo->input_iMCU_row = row_offset;
-  cinfo->output_iMCU_row = row_offset;
-
-  // Updates JPEG decoder parameter
-  jinit_color_deconverter(cinfo);
-  jpeg_calc_output_dimensions(cinfo);
-  jinit_upsampler(cinfo);
-  (*cinfo->master->prepare_for_output_pass) (cinfo);
-  if (cinfo->progressive_mode)
-    (*cinfo->entropy->start_pass) (cinfo);
-  else
-    jpeg_decompress_per_scan_setup(cinfo);
-
-  int sample_size = DCTSIZE / cinfo->min_DCT_scaled_size;
-
-  *height = jdiv_round_up(*height, sample_size);
-  *width = cinfo->output_width;
-  cinfo->output_scanline = lines_per_iMCU_row * row_offset / sample_size;
-  cinfo->inputctl->consume_input = cinfo->coef->consume_data;
-  cinfo->inputctl->consume_input_build_huffman_index =
-      cinfo->coef->consume_data_build_huffman_index;
-  cinfo->entropy->index = index;
-  cinfo->input_iMCU_row = row_offset;
-  cinfo->output_iMCU_row = row_offset;
-  cinfo->coef->MCU_column_left_boundary = col_left_boundary;
-  cinfo->coef->MCU_column_right_boundary = col_right_boundary;
-  cinfo->coef->column_left_boundary =
-      col_left_boundary / index->MCU_sample_size;
-  cinfo->coef->column_right_boundary =
-      jdiv_round_up(col_right_boundary, index->MCU_sample_size);
-}
-
-/*
- * Read a scanline from the current position.
- *
- * Return the number of lines actually read.
- */
-
-GLOBAL(JDIMENSION)
-jpeg_read_tile_scanline (j_decompress_ptr cinfo, huffman_index *index,
-        JSAMPARRAY scanlines)
-{
-  // Calculates the boundary of iMCU
-  int lines_per_iMCU_row = cinfo->max_v_samp_factor * DCTSIZE;
-  int lines_per_iMCU_col = cinfo->max_h_samp_factor * DCTSIZE;
-  int sample_size = DCTSIZE / cinfo->min_DCT_scaled_size;
-  JDIMENSION row_ctr = 0;
-
-  if (cinfo->progressive_mode) {
-    (*cinfo->main->process_data) (cinfo, scanlines, &row_ctr, 1);
-  } else {
-    if (cinfo->output_scanline % (lines_per_iMCU_row / sample_size) == 0) {
-      // Set the read head to the next iMCU row
-      int iMCU_row_offset = cinfo->output_scanline /
-            (lines_per_iMCU_row / sample_size);
-      int offset_data_col_position = cinfo->coef->MCU_column_left_boundary /
-            index->MCU_sample_size;
-      huffman_offset_data offset_data =
-          index->scan[0].offset[iMCU_row_offset][offset_data_col_position];
-      (*cinfo->entropy->configure_huffman_decoder) (cinfo, offset_data);
-    }
-    (*cinfo->main->process_data) (cinfo, scanlines, &row_ctr, 1);
-  }
-
-  cinfo->output_scanline += row_ctr;
-  return row_ctr;
-}
-
-/*
- * Alternate entry point to read raw data.
- * Processes exactly one iMCU row per call, unless suspended.
- */
-
-GLOBAL(JDIMENSION)
-jpeg_read_raw_data (j_decompress_ptr cinfo, JSAMPIMAGE data,
-		    JDIMENSION max_lines)
-{
-  JDIMENSION lines_per_iMCU_row;
-
-  if (cinfo->global_state != DSTATE_RAW_OK)
-    ERREXIT1(cinfo, JERR_BAD_STATE, cinfo->global_state);
-  if (cinfo->output_scanline >= cinfo->output_height) {
-    WARNMS(cinfo, JWRN_TOO_MUCH_DATA);
-    return 0;
-  }
-
-  /* Call progress monitor hook if present */
-  if (cinfo->progress != NULL) {
-    cinfo->progress->pass_counter = (long) cinfo->output_scanline;
-    cinfo->progress->pass_limit = (long) cinfo->output_height;
-    (*cinfo->progress->progress_monitor) ((j_common_ptr) cinfo);
-  }
-
-  /* Verify that at least one iMCU row can be returned. */
-  lines_per_iMCU_row = cinfo->max_v_samp_factor * cinfo->min_DCT_scaled_size;
-  if (max_lines < lines_per_iMCU_row)
-    ERREXIT(cinfo, JERR_BUFFER_SIZE);
-
-  /* Decompress directly into user's buffer. */
-  if (! (*cinfo->coef->decompress_data) (cinfo, data))
-    return 0;			/* suspension forced, can do nothing more */
-
-  /* OK, we processed one iMCU row. */
-  cinfo->output_scanline += lines_per_iMCU_row;
-  return lines_per_iMCU_row;
-}
-
-
-/* Additional entry points for buffered-image mode. */
-
-#ifdef D_MULTISCAN_FILES_SUPPORTED
-
-/*
- * Initialize for an output pass in buffered-image mode.
- */
-
-GLOBAL(boolean)
-jpeg_start_output (j_decompress_ptr cinfo, int scan_number)
-{
-  if (cinfo->global_state != DSTATE_BUFIMAGE &&
-      cinfo->global_state != DSTATE_PRESCAN)
-    ERREXIT1(cinfo, JERR_BAD_STATE, cinfo->global_state);
-  /* Limit scan number to valid range */
-  if (scan_number <= 0)
-    scan_number = 1;
-  if (cinfo->inputctl->eoi_reached &&
-      scan_number > cinfo->input_scan_number)
-    scan_number = cinfo->input_scan_number;
-  cinfo->output_scan_number = scan_number;
-  /* Perform any dummy output passes, and set up for the real pass */
-  return output_pass_setup(cinfo);
-}
-
-
-/*
- * Finish up after an output pass in buffered-image mode.
- *
- * Returns FALSE if suspended.  The return value need be inspected only if
- * a suspending data source is used.
- */
-
-GLOBAL(boolean)
-jpeg_finish_output (j_decompress_ptr cinfo)
-{
-  if ((cinfo->global_state == DSTATE_SCANNING ||
-       cinfo->global_state == DSTATE_RAW_OK) && cinfo->buffered_image) {
-    /* Terminate this pass. */
-    /* We do not require the whole pass to have been completed. */
-    (*cinfo->master->finish_output_pass) (cinfo);
-    cinfo->global_state = DSTATE_BUFPOST;
-  } else if (cinfo->global_state != DSTATE_BUFPOST) {
-    /* BUFPOST = repeat call after a suspension, anything else is error */
-    ERREXIT1(cinfo, JERR_BAD_STATE, cinfo->global_state);
-  }
-  /* Read markers looking for SOS or EOI */
-  while (cinfo->input_scan_number <= cinfo->output_scan_number &&
-	 ! cinfo->inputctl->eoi_reached) {
-    if ((*cinfo->inputctl->consume_input) (cinfo) == JPEG_SUSPENDED)
-      return FALSE;		/* Suspend, come back later */
-  }
-  cinfo->global_state = DSTATE_BUFIMAGE;
-  return TRUE;
-}
-
-#endif /* D_MULTISCAN_FILES_SUPPORTED */
@@ -1,151 +0,0 @@
-/*
- * jdatadst.c
- *
- * Copyright (C) 1994-1996, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains compression data destination routines for the case of
- * emitting JPEG data to a file (or any stdio stream).  While these routines
- * are sufficient for most applications, some will want to use a different
- * destination manager.
- * IMPORTANT: we assume that fwrite() will correctly transcribe an array of
- * JOCTETs into 8-bit-wide elements on external storage.  If char is wider
- * than 8 bits on your machine, you may need to do some tweaking.
- */
-
-/* this is not a core library module, so it doesn't define JPEG_INTERNALS */
-#include "jinclude.h"
-#include "jpeglib.h"
-#include "jerror.h"
-
-
-/* Expanded data destination object for stdio output */
-
-typedef struct {
-  struct jpeg_destination_mgr pub; /* public fields */
-
-  FILE * outfile;		/* target stream */
-  JOCTET * buffer;		/* start of buffer */
-} my_destination_mgr;
-
-typedef my_destination_mgr * my_dest_ptr;
-
-#define OUTPUT_BUF_SIZE  4096	/* choose an efficiently fwrite'able size */
-
-
-/*
- * Initialize destination --- called by jpeg_start_compress
- * before any data is actually written.
- */
-
-METHODDEF(void)
-init_destination (j_compress_ptr cinfo)
-{
-  my_dest_ptr dest = (my_dest_ptr) cinfo->dest;
-
-  /* Allocate the output buffer --- it will be released when done with image */
-  dest->buffer = (JOCTET *)
-      (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				  OUTPUT_BUF_SIZE * SIZEOF(JOCTET));
-
-  dest->pub.next_output_byte = dest->buffer;
-  dest->pub.free_in_buffer = OUTPUT_BUF_SIZE;
-}
-
-
-/*
- * Empty the output buffer --- called whenever buffer fills up.
- *
- * In typical applications, this should write the entire output buffer
- * (ignoring the current state of next_output_byte & free_in_buffer),
- * reset the pointer & count to the start of the buffer, and return TRUE
- * indicating that the buffer has been dumped.
- *
- * In applications that need to be able to suspend compression due to output
- * overrun, a FALSE return indicates that the buffer cannot be emptied now.
- * In this situation, the compressor will return to its caller (possibly with
- * an indication that it has not accepted all the supplied scanlines).  The
- * application should resume compression after it has made more room in the
- * output buffer.  Note that there are substantial restrictions on the use of
- * suspension --- see the documentation.
- *
- * When suspending, the compressor will back up to a convenient restart point
- * (typically the start of the current MCU). next_output_byte & free_in_buffer
- * indicate where the restart point will be if the current call returns FALSE.
- * Data beyond this point will be regenerated after resumption, so do not
- * write it out when emptying the buffer externally.
- */
-
-METHODDEF(boolean)
-empty_output_buffer (j_compress_ptr cinfo)
-{
-  my_dest_ptr dest = (my_dest_ptr) cinfo->dest;
-
-  if (JFWRITE(dest->outfile, dest->buffer, OUTPUT_BUF_SIZE) !=
-      (size_t) OUTPUT_BUF_SIZE)
-    ERREXIT(cinfo, JERR_FILE_WRITE);
-
-  dest->pub.next_output_byte = dest->buffer;
-  dest->pub.free_in_buffer = OUTPUT_BUF_SIZE;
-
-  return TRUE;
-}
-
-
-/*
- * Terminate destination --- called by jpeg_finish_compress
- * after all data has been written.  Usually needs to flush buffer.
- *
- * NB: *not* called by jpeg_abort or jpeg_destroy; surrounding
- * application must deal with any cleanup that should happen even
- * for error exit.
- */
-
-METHODDEF(void)
-term_destination (j_compress_ptr cinfo)
-{
-  my_dest_ptr dest = (my_dest_ptr) cinfo->dest;
-  size_t datacount = OUTPUT_BUF_SIZE - dest->pub.free_in_buffer;
-
-  /* Write any data remaining in the buffer */
-  if (datacount > 0) {
-    if (JFWRITE(dest->outfile, dest->buffer, datacount) != datacount)
-      ERREXIT(cinfo, JERR_FILE_WRITE);
-  }
-  fflush(dest->outfile);
-  /* Make sure we wrote the output file OK */
-  if (ferror(dest->outfile))
-    ERREXIT(cinfo, JERR_FILE_WRITE);
-}
-
-
-/*
- * Prepare for output to a stdio stream.
- * The caller must have already opened the stream, and is responsible
- * for closing it after finishing compression.
- */
-
-GLOBAL(void)
-jpeg_stdio_dest (j_compress_ptr cinfo, FILE * outfile)
-{
-  my_dest_ptr dest;
-
-  /* The destination object is made permanent so that multiple JPEG images
-   * can be written to the same file without re-executing jpeg_stdio_dest.
-   * This makes it dangerous to use this manager and a different destination
-   * manager serially with the same JPEG object, because their private object
-   * sizes may be different.  Caveat programmer.
-   */
-  if (cinfo->dest == NULL) {	/* first time for this JPEG object? */
-    cinfo->dest = (struct jpeg_destination_mgr *)
-      (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_PERMANENT,
-				  SIZEOF(my_destination_mgr));
-  }
-
-  dest = (my_dest_ptr) cinfo->dest;
-  dest->pub.init_destination = init_destination;
-  dest->pub.empty_output_buffer = empty_output_buffer;
-  dest->pub.term_destination = term_destination;
-  dest->outfile = outfile;
-}
@@ -1,212 +0,0 @@
-/*
- * jdatasrc.c
- *
- * Copyright (C) 1994-1996, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains decompression data source routines for the case of
- * reading JPEG data from a file (or any stdio stream).  While these routines
- * are sufficient for most applications, some will want to use a different
- * source manager.
- * IMPORTANT: we assume that fread() will correctly transcribe an array of
- * JOCTETs from 8-bit-wide elements on external storage.  If char is wider
- * than 8 bits on your machine, you may need to do some tweaking.
- */
-
-/* this is not a core library module, so it doesn't define JPEG_INTERNALS */
-#include "jinclude.h"
-#include "jpeglib.h"
-#include "jerror.h"
-
-
-/* Expanded data source object for stdio input */
-
-typedef struct {
-  struct jpeg_source_mgr pub;	/* public fields */
-
-  FILE * infile;		/* source stream */
-  JOCTET * buffer;		/* start of buffer */
-  boolean start_of_file;	/* have we gotten any data yet? */
-} my_source_mgr;
-
-typedef my_source_mgr * my_src_ptr;
-
-#define INPUT_BUF_SIZE  4096	/* choose an efficiently fread'able size */
-
-
-/*
- * Initialize source --- called by jpeg_read_header
- * before any data is actually read.
- */
-
-METHODDEF(void)
-init_source (j_decompress_ptr cinfo)
-{
-  my_src_ptr src = (my_src_ptr) cinfo->src;
-
-  /* We reset the empty-input-file flag for each image,
-   * but we don't clear the input buffer.
-   * This is correct behavior for reading a series of images from one source.
-   */
-  src->start_of_file = TRUE;
-}
-
-
-/*
- * Fill the input buffer --- called whenever buffer is emptied.
- *
- * In typical applications, this should read fresh data into the buffer
- * (ignoring the current state of next_input_byte & bytes_in_buffer),
- * reset the pointer & count to the start of the buffer, and return TRUE
- * indicating that the buffer has been reloaded.  It is not necessary to
- * fill the buffer entirely, only to obtain at least one more byte.
- *
- * There is no such thing as an EOF return.  If the end of the file has been
- * reached, the routine has a choice of ERREXIT() or inserting fake data into
- * the buffer.  In most cases, generating a warning message and inserting a
- * fake EOI marker is the best course of action --- this will allow the
- * decompressor to output however much of the image is there.  However,
- * the resulting error message is misleading if the real problem is an empty
- * input file, so we handle that case specially.
- *
- * In applications that need to be able to suspend compression due to input
- * not being available yet, a FALSE return indicates that no more data can be
- * obtained right now, but more may be forthcoming later.  In this situation,
- * the decompressor will return to its caller (with an indication of the
- * number of scanlines it has read, if any).  The application should resume
- * decompression after it has loaded more data into the input buffer.  Note
- * that there are substantial restrictions on the use of suspension --- see
- * the documentation.
- *
- * When suspending, the decompressor will back up to a convenient restart point
- * (typically the start of the current MCU). next_input_byte & bytes_in_buffer
- * indicate where the restart point will be if the current call returns FALSE.
- * Data beyond this point must be rescanned after resumption, so move it to
- * the front of the buffer rather than discarding it.
- */
-
-METHODDEF(boolean)
-fill_input_buffer (j_decompress_ptr cinfo)
-{
-  my_src_ptr src = (my_src_ptr) cinfo->src;
-  size_t nbytes;
-
-  nbytes = JFREAD(src->infile, src->buffer, INPUT_BUF_SIZE);
-
-  if (nbytes <= 0) {
-    if (src->start_of_file)	/* Treat empty input file as fatal error */
-      ERREXIT(cinfo, JERR_INPUT_EMPTY);
-    WARNMS(cinfo, JWRN_JPEG_EOF);
-    /* Insert a fake EOI marker */
-    src->buffer[0] = (JOCTET) 0xFF;
-    src->buffer[1] = (JOCTET) JPEG_EOI;
-    nbytes = 2;
-  }
-
-  src->pub.next_input_byte = src->buffer;
-  src->pub.bytes_in_buffer = nbytes;
-  src->start_of_file = FALSE;
-
-  return TRUE;
-}
-
-
-/*
- * Skip data --- used to skip over a potentially large amount of
- * uninteresting data (such as an APPn marker).
- *
- * Writers of suspendable-input applications must note that skip_input_data
- * is not granted the right to give a suspension return.  If the skip extends
- * beyond the data currently in the buffer, the buffer can be marked empty so
- * that the next read will cause a fill_input_buffer call that can suspend.
- * Arranging for additional bytes to be discarded before reloading the input
- * buffer is the application writer's problem.
- */
-
-METHODDEF(void)
-skip_input_data (j_decompress_ptr cinfo, long num_bytes)
-{
-  my_src_ptr src = (my_src_ptr) cinfo->src;
-
-  /* Just a dumb implementation for now.  Could use fseek() except
-   * it doesn't work on pipes.  Not clear that being smart is worth
-   * any trouble anyway --- large skips are infrequent.
-   */
-  if (num_bytes > 0) {
-    while (num_bytes > (long) src->pub.bytes_in_buffer) {
-      num_bytes -= (long) src->pub.bytes_in_buffer;
-      (void) fill_input_buffer(cinfo);
-      /* note we assume that fill_input_buffer will never return FALSE,
-       * so suspension need not be handled.
-       */
-    }
-    src->pub.next_input_byte += (size_t) num_bytes;
-    src->pub.bytes_in_buffer -= (size_t) num_bytes;
-  }
-}
-
-
-/*
- * An additional method that can be provided by data source modules is the
- * resync_to_restart method for error recovery in the presence of RST markers.
- * For the moment, this source module just uses the default resync method
- * provided by the JPEG library.  That method assumes that no backtracking
- * is possible.
- */
-
-
-/*
- * Terminate source --- called by jpeg_finish_decompress
- * after all data has been read.  Often a no-op.
- *
- * NB: *not* called by jpeg_abort or jpeg_destroy; surrounding
- * application must deal with any cleanup that should happen even
- * for error exit.
- */
-
-METHODDEF(void)
-term_source (j_decompress_ptr cinfo)
-{
-  /* no work necessary here */
-}
-
-
-/*
- * Prepare for input from a stdio stream.
- * The caller must have already opened the stream, and is responsible
- * for closing it after finishing decompression.
- */
-
-GLOBAL(void)
-jpeg_stdio_src (j_decompress_ptr cinfo, FILE * infile)
-{
-  my_src_ptr src;
-
-  /* The source object and input buffer are made permanent so that a series
-   * of JPEG images can be read from the same file by calling jpeg_stdio_src
-   * only before the first one.  (If we discarded the buffer at the end of
-   * one image, we'd likely lose the start of the next one.)
-   * This makes it unsafe to use this manager and a different source
-   * manager serially with the same JPEG object.  Caveat programmer.
-   */
-  if (cinfo->src == NULL) {	/* first time for this JPEG object? */
-    cinfo->src = (struct jpeg_source_mgr *)
-      (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_PERMANENT,
-				  SIZEOF(my_source_mgr));
-    src = (my_src_ptr) cinfo->src;
-    src->buffer = (JOCTET *)
-      (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_PERMANENT,
-				  INPUT_BUF_SIZE * SIZEOF(JOCTET));
-  }
-
-  src = (my_src_ptr) cinfo->src;
-  src->pub.init_source = init_source;
-  src->pub.fill_input_buffer = fill_input_buffer;
-  src->pub.skip_input_data = skip_input_data;
-  src->pub.resync_to_restart = jpeg_resync_to_restart; /* use default method */
-  src->pub.term_source = term_source;
-  src->infile = infile;
-  src->pub.bytes_in_buffer = 0; /* forces fill_input_buffer on first read */
-  src->pub.next_input_byte = NULL; /* until buffer loaded */
-}
@@ -1,899 +0,0 @@
-/*
- * jdcolor.c
- *
- * Copyright (C) 1991-1997, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains output colorspace conversion routines.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-#ifdef NV_ARM_NEON
-#include "jsimd_neon.h"
-#endif
-
-/* Private subobject */
-
-typedef struct {
-  struct jpeg_color_deconverter pub; /* public fields */
-
-  /* Private state for YCC->RGB conversion */
-  int * Cr_r_tab;		/* => table for Cr to R conversion */
-  int * Cb_b_tab;		/* => table for Cb to B conversion */
-  INT32 * Cr_g_tab;		/* => table for Cr to G conversion */
-  INT32 * Cb_g_tab;		/* => table for Cb to G conversion */
-} my_color_deconverter;
-
-typedef my_color_deconverter * my_cconvert_ptr;
-
-
-#ifdef ANDROID_RGB
-
-/* Declarations for ordered dithering.
- * 
- * We use 4x4 ordered dither array packed into 32 bits. This array is
- * sufficent for dithering RGB_888 to RGB_565.
- */
-
-#define DITHER_MASK         0x3
-#define DITHER_ROTATE(x)    (((x)<<24) | (((x)>>8)&0x00FFFFFF))
-static const INT32 dither_matrix[4] = {
-  0x0008020A,
-  0x0C040E06,
-  0x030B0109,
-  0x0F070D05
-};
-
-#endif
-
-
-/**************** YCbCr -> RGB conversion: most common case **************/
-
-/*
- * YCbCr is defined per CCIR 601-1, except that Cb and Cr are
- * normalized to the range 0..MAXJSAMPLE rather than -0.5 .. 0.5.
- * The conversion equations to be implemented are therefore
- *	R = Y                + 1.40200 * Cr
- *	G = Y - 0.34414 * Cb - 0.71414 * Cr
- *	B = Y + 1.77200 * Cb
- * where Cb and Cr represent the incoming values less CENTERJSAMPLE.
- * (These numbers are derived from TIFF 6.0 section 21, dated 3-June-92.)
- *
- * To avoid floating-point arithmetic, we represent the fractional constants
- * as integers scaled up by 2^16 (about 4 digits precision); we have to divide
- * the products by 2^16, with appropriate rounding, to get the correct answer.
- * Notice that Y, being an integral input, does not contribute any fraction
- * so it need not participate in the rounding.
- *
- * For even more speed, we avoid doing any multiplications in the inner loop
- * by precalculating the constants times Cb and Cr for all possible values.
- * For 8-bit JSAMPLEs this is very reasonable (only 256 entries per table);
- * for 12-bit samples it is still acceptable.  It's not very reasonable for
- * 16-bit samples, but if you want lossless storage you shouldn't be changing
- * colorspace anyway.
- * The Cr=>R and Cb=>B values can be rounded to integers in advance; the
- * values for the G calculation are left scaled up, since we must add them
- * together before rounding.
- */
-
-#define SCALEBITS	16	/* speediest right-shift on some machines */
-#define ONE_HALF	((INT32) 1 << (SCALEBITS-1))
-#define FIX(x)		((INT32) ((x) * (1L<<SCALEBITS) + 0.5))
-
-
-/*
- * Initialize tables for YCC->RGB colorspace conversion.
- */
-
-LOCAL(void)
-build_ycc_rgb_table (j_decompress_ptr cinfo)
-{
-  my_cconvert_ptr cconvert = (my_cconvert_ptr) cinfo->cconvert;
-  int i;
-  INT32 x;
-  SHIFT_TEMPS
-
-  cconvert->Cr_r_tab = (int *)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-                                (MAXJSAMPLE+1) * SIZEOF(int));
-  cconvert->Cb_b_tab = (int *)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-                                (MAXJSAMPLE+1) * SIZEOF(int));
-  cconvert->Cr_g_tab = (INT32 *)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-                                (MAXJSAMPLE+1) * SIZEOF(INT32));
-  cconvert->Cb_g_tab = (INT32 *)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-                                (MAXJSAMPLE+1) * SIZEOF(INT32));
-
-  for (i = 0, x = -CENTERJSAMPLE; i <= MAXJSAMPLE; i++, x++) {
-    /* i is the actual input pixel value, in the range 0..MAXJSAMPLE */
-    /* The Cb or Cr value we are thinking of is x = i - CENTERJSAMPLE */
-    /* Cr=>R value is nearest int to 1.40200 * x */
-    cconvert->Cr_r_tab[i] = (int)
-                    RIGHT_SHIFT(FIX(1.40200) * x + ONE_HALF, SCALEBITS);
-    /* Cb=>B value is nearest int to 1.77200 * x */
-    cconvert->Cb_b_tab[i] = (int)
-                    RIGHT_SHIFT(FIX(1.77200) * x + ONE_HALF, SCALEBITS);
-    /* Cr=>G value is scaled-up -0.71414 * x */
-    cconvert->Cr_g_tab[i] = (- FIX(0.71414)) * x;
-    /* Cb=>G value is scaled-up -0.34414 * x */
-    /* We also add in ONE_HALF so that need not do it in inner loop */
-    cconvert->Cb_g_tab[i] = (- FIX(0.34414)) * x + ONE_HALF;
-  }
-}
-
-/*
- * Convert some rows of samples to the output colorspace.
- *
- * Note that we change from noninterleaved, one-plane-per-component format
- * to interleaved-pixel format.  The output buffer is therefore three times
- * as wide as the input buffer.
- * A starting row offset is provided only for the input buffer.  The caller
- * can easily adjust the passed output_buf value to accommodate any row
- * offset required on that side.
- */
-
-METHODDEF(void)
-ycc_rgb_convert (j_decompress_ptr cinfo,
-		 JSAMPIMAGE input_buf, JDIMENSION input_row,
-		 JSAMPARRAY output_buf, int num_rows)
-{
-  my_cconvert_ptr cconvert = (my_cconvert_ptr) cinfo->cconvert;
-  register int y, cb, cr;
-  register JSAMPROW outptr;
-  register JSAMPROW inptr0, inptr1, inptr2;
-  register JDIMENSION col;
-  JDIMENSION num_cols = cinfo->output_width;
-  /* copy these pointers into registers if possible */
-  register JSAMPLE * range_limit = cinfo->sample_range_limit;
-  register int * Crrtab = cconvert->Cr_r_tab;
-  register int * Cbbtab = cconvert->Cb_b_tab;
-  register INT32 * Crgtab = cconvert->Cr_g_tab;
-  register INT32 * Cbgtab = cconvert->Cb_g_tab;
-  SHIFT_TEMPS
-
-  while (--num_rows >= 0) {
-    inptr0 = input_buf[0][input_row];
-    inptr1 = input_buf[1][input_row];
-    inptr2 = input_buf[2][input_row];
-    input_row++;
-    outptr = *output_buf++;
-    for (col = 0; col < num_cols; col++) {
-      y  = GETJSAMPLE(inptr0[col]);
-      cb = GETJSAMPLE(inptr1[col]);
-      cr = GETJSAMPLE(inptr2[col]);
-      /* Range-limiting is essential due to noise introduced by DCT losses. */
-      outptr[RGB_RED] =   range_limit[y + Crrtab[cr]];
-      outptr[RGB_GREEN] = range_limit[y +
-                              ((int) RIGHT_SHIFT(Cbgtab[cb] + Crgtab[cr],
-                                                 SCALEBITS))];
-      outptr[RGB_BLUE] =  range_limit[y + Cbbtab[cb]];
-      outptr += RGB_PIXELSIZE;
-    }
-  }
-}
-
-#ifdef ANDROID_RGB
-METHODDEF(void)
-ycc_rgba_8888_convert (j_decompress_ptr cinfo,
-         JSAMPIMAGE input_buf, JDIMENSION input_row,
-         JSAMPARRAY output_buf, int num_rows)
-{
-  my_cconvert_ptr cconvert = (my_cconvert_ptr) cinfo->cconvert;
-  register int y, cb, cr;
-  register JSAMPROW outptr;
-  register JSAMPROW inptr0, inptr1, inptr2;
-  register JDIMENSION col;
-  JDIMENSION num_cols = cinfo->output_width;
-  /* copy these pointers into registers if possible */
-  register JSAMPLE * range_limit = cinfo->sample_range_limit;
-  register int * Crrtab = cconvert->Cr_r_tab;
-  register int * Cbbtab = cconvert->Cb_b_tab;
-  register INT32 * Crgtab = cconvert->Cr_g_tab;
-  register INT32 * Cbgtab = cconvert->Cb_g_tab;
-  SHIFT_TEMPS
-
-  while (--num_rows >= 0) {
-    inptr0 = input_buf[0][input_row];
-    inptr1 = input_buf[1][input_row];
-    inptr2 = input_buf[2][input_row];
-    input_row++;
-    outptr = *output_buf++;
-    for (col = 0; col < num_cols; col++) {
-      y  = GETJSAMPLE(inptr0[col]);
-      cb = GETJSAMPLE(inptr1[col]);
-      cr = GETJSAMPLE(inptr2[col]);
-      /* Range-limiting is essential due to noise introduced by DCT losses. */
-      outptr[RGB_RED] =   range_limit[y + Crrtab[cr]];
-      outptr[RGB_GREEN] = range_limit[y +
-                              ((int) RIGHT_SHIFT(Cbgtab[cb] + Crgtab[cr],
-                                                 SCALEBITS))];
-      outptr[RGB_BLUE] =  range_limit[y + Cbbtab[cb]];
-      outptr[RGB_ALPHA] =  0xFF;
-      outptr += 4;
-    }
-  }
-}
-
-METHODDEF(void)
-ycc_rgb_565_convert (j_decompress_ptr cinfo,
-         JSAMPIMAGE input_buf, JDIMENSION input_row,
-         JSAMPARRAY output_buf, int num_rows)
-{
-  my_cconvert_ptr cconvert = (my_cconvert_ptr) cinfo->cconvert;
-  register int y, cb, cr;
-  register JSAMPROW outptr;
-  register JSAMPROW inptr0, inptr1, inptr2;
-  register JDIMENSION col;
-  JDIMENSION num_cols = cinfo->output_width;
-  /* copy these pointers into registers if possible */
-  register JSAMPLE * range_limit = cinfo->sample_range_limit;
-  register int * Crrtab = cconvert->Cr_r_tab;
-  register int * Cbbtab = cconvert->Cb_b_tab;
-  register INT32 * Crgtab = cconvert->Cr_g_tab;
-  register INT32 * Cbgtab = cconvert->Cb_g_tab;
-  SHIFT_TEMPS
-
-  while (--num_rows >= 0) {
-    INT32 rgb;
-    unsigned int r, g, b;
-    inptr0 = input_buf[0][input_row];
-    inptr1 = input_buf[1][input_row];
-    inptr2 = input_buf[2][input_row];
-    input_row++;
-    outptr = *output_buf++;
-    
-    if (PACK_NEED_ALIGNMENT(outptr)) {
-        y  = GETJSAMPLE(*inptr0++);
-        cb = GETJSAMPLE(*inptr1++);
-        cr = GETJSAMPLE(*inptr2++);
-        r = range_limit[y + Crrtab[cr]];
-        g = range_limit[y + ((int)RIGHT_SHIFT(Cbgtab[cb]+Crgtab[cr], SCALEBITS))];
-        b = range_limit[y + Cbbtab[cb]];
-        rgb = PACK_SHORT_565(r,g,b);
-        *(INT16*)outptr = rgb;
-        outptr += 2;
-        num_cols--;
-    }
-    for (col = 0; col < (num_cols>>1); col++) {
-      y  = GETJSAMPLE(*inptr0++);
-      cb = GETJSAMPLE(*inptr1++);
-      cr = GETJSAMPLE(*inptr2++);
-      r = range_limit[y + Crrtab[cr]];
-      g = range_limit[y + ((int)RIGHT_SHIFT(Cbgtab[cb]+Crgtab[cr], SCALEBITS))];
-      b = range_limit[y + Cbbtab[cb]];
-      rgb = PACK_SHORT_565(r,g,b);
-
-      y  = GETJSAMPLE(*inptr0++);
-      cb = GETJSAMPLE(*inptr1++);
-      cr = GETJSAMPLE(*inptr2++);
-      r = range_limit[y + Crrtab[cr]];
-      g = range_limit[y + ((int)RIGHT_SHIFT(Cbgtab[cb]+Crgtab[cr], SCALEBITS))];
-      b = range_limit[y + Cbbtab[cb]];
-      rgb = PACK_TWO_PIXELS(rgb, PACK_SHORT_565(r,g,b));
-      WRITE_TWO_ALIGNED_PIXELS(outptr, rgb);
-      outptr += 4;
-    }
-    if (num_cols&1) {
-      y  = GETJSAMPLE(*inptr0);
-      cb = GETJSAMPLE(*inptr1);
-      cr = GETJSAMPLE(*inptr2);
-      r = range_limit[y + Crrtab[cr]];
-      g = range_limit[y + ((int)RIGHT_SHIFT(Cbgtab[cb]+Crgtab[cr], SCALEBITS))];
-      b = range_limit[y + Cbbtab[cb]];
-      rgb = PACK_SHORT_565(r,g,b);
-      *(INT16*)outptr = rgb;
-    }
-  }
-}
-
-METHODDEF(void)
-ycc_rgb_565D_convert (j_decompress_ptr cinfo,
-         JSAMPIMAGE input_buf, JDIMENSION input_row,
-         JSAMPARRAY output_buf, int num_rows)
-{
-  my_cconvert_ptr cconvert = (my_cconvert_ptr) cinfo->cconvert;
-  register int y, cb, cr;
-  register JSAMPROW outptr;
-  register JSAMPROW inptr0, inptr1, inptr2;
-  register JDIMENSION col;
-  JDIMENSION num_cols = cinfo->output_width;
-  /* copy these pointers into registers if possible */
-  register JSAMPLE * range_limit = cinfo->sample_range_limit;
-  register int * Crrtab = cconvert->Cr_r_tab;
-  register int * Cbbtab = cconvert->Cb_b_tab;
-  register INT32 * Crgtab = cconvert->Cr_g_tab;
-  register INT32 * Cbgtab = cconvert->Cb_g_tab;
-  INT32 d0 = dither_matrix[cinfo->output_scanline & DITHER_MASK];
-  SHIFT_TEMPS
-
-  while (--num_rows >= 0) {
-    INT32 rgb;
-    unsigned int r, g, b;
-    inptr0 = input_buf[0][input_row];
-    inptr1 = input_buf[1][input_row];
-    inptr2 = input_buf[2][input_row];
-    input_row++;
-    outptr = *output_buf++;
-    if (PACK_NEED_ALIGNMENT(outptr)) {
-        y  = GETJSAMPLE(*inptr0++);
-        cb = GETJSAMPLE(*inptr1++);
-        cr = GETJSAMPLE(*inptr2++);
-        r = range_limit[DITHER_565_R(y + Crrtab[cr], d0)];
-        g = range_limit[DITHER_565_G(y + ((int)RIGHT_SHIFT(Cbgtab[cb]+Crgtab[cr], SCALEBITS)), d0)];
-        b = range_limit[DITHER_565_B(y + Cbbtab[cb], d0)];
-        rgb = PACK_SHORT_565(r,g,b);
-        *(INT16*)outptr = rgb;
-        outptr += 2;
-        num_cols--;
-    }
-    for (col = 0; col < (num_cols>>1); col++) {
-      y  = GETJSAMPLE(*inptr0++);
-      cb = GETJSAMPLE(*inptr1++);
-      cr = GETJSAMPLE(*inptr2++);
-      r = range_limit[DITHER_565_R(y + Crrtab[cr], d0)];
-      g = range_limit[DITHER_565_G(y + ((int)RIGHT_SHIFT(Cbgtab[cb]+Crgtab[cr], SCALEBITS)), d0)];
-      b = range_limit[DITHER_565_B(y + Cbbtab[cb], d0)];
-      d0 = DITHER_ROTATE(d0);
-      rgb = PACK_SHORT_565(r,g,b);
-      y  = GETJSAMPLE(*inptr0++);
-      cb = GETJSAMPLE(*inptr1++);
-      cr = GETJSAMPLE(*inptr2++);
-      r = range_limit[DITHER_565_R(y + Crrtab[cr], d0)];
-      g = range_limit[DITHER_565_G(y + ((int)RIGHT_SHIFT(Cbgtab[cb]+Crgtab[cr], SCALEBITS)), d0)];
-      b = range_limit[DITHER_565_B(y + Cbbtab[cb], d0)];
-      d0 = DITHER_ROTATE(d0);
-      rgb = PACK_TWO_PIXELS(rgb, PACK_SHORT_565(r,g,b));
-      WRITE_TWO_ALIGNED_PIXELS(outptr, rgb);
-      outptr += 4;
-    }
-    if (num_cols&1) {
-      y  = GETJSAMPLE(*inptr0);
-      cb = GETJSAMPLE(*inptr1);
-      cr = GETJSAMPLE(*inptr2);
-      r = range_limit[DITHER_565_R(y + Crrtab[cr], d0)];
-      g = range_limit[DITHER_565_G(y + ((int)RIGHT_SHIFT(Cbgtab[cb]+Crgtab[cr], SCALEBITS)), d0)];
-      b = range_limit[DITHER_565_B(y + Cbbtab[cb], d0)];
-      rgb = PACK_SHORT_565(r,g,b);
-      *(INT16*)outptr = rgb;
-    }
-  }
-}
-
-#endif
-
-/**************** Cases other than YCbCr -> RGB(A) **************/
-
-#ifdef ANDROID_RGB
-METHODDEF(void)
-rgb_rgba_8888_convert (j_decompress_ptr cinfo,
-         JSAMPIMAGE input_buf, JDIMENSION input_row,
-         JSAMPARRAY output_buf, int num_rows)
-{
-  register JSAMPROW outptr;
-  register JSAMPROW inptr0, inptr1, inptr2;
-  register JDIMENSION col;
-  JDIMENSION num_cols = cinfo->output_width;
-  SHIFT_TEMPS
-
-  while (--num_rows >= 0) {
-    inptr0 = input_buf[0][input_row];
-    inptr1 = input_buf[1][input_row];
-    inptr2 = input_buf[2][input_row];
-    input_row++;
-    outptr = *output_buf++;
-    for (col = 0; col < num_cols; col++) {
-      *outptr++ = *inptr0++;
-      *outptr++ = *inptr1++;
-      *outptr++ = *inptr2++;
-      *outptr++ = 0xFF;
-    }
-  }
-}
-
-METHODDEF(void)
-rgb_rgb_565_convert (j_decompress_ptr cinfo,
-         JSAMPIMAGE input_buf, JDIMENSION input_row,
-         JSAMPARRAY output_buf, int num_rows)
-{
-  register JSAMPROW outptr;
-  register JSAMPROW inptr0, inptr1, inptr2;
-  register JDIMENSION col;
-  JDIMENSION num_cols = cinfo->output_width;
-  SHIFT_TEMPS
-
-  while (--num_rows >= 0) {
-    INT32 rgb;
-    unsigned int r, g, b;
-    inptr0 = input_buf[0][input_row];
-    inptr1 = input_buf[1][input_row];
-    inptr2 = input_buf[2][input_row];
-    input_row++;
-    outptr = *output_buf++;
-    if (PACK_NEED_ALIGNMENT(outptr)) {
-        r = GETJSAMPLE(*inptr0++);
-        g = GETJSAMPLE(*inptr1++);
-        b = GETJSAMPLE(*inptr2++);
-        rgb = PACK_SHORT_565(r,g,b);
-        *(INT16*)outptr = rgb;
-        outptr += 2;
-        num_cols--;
-    }
-    for (col = 0; col < (num_cols>>1); col++) {
-      r = GETJSAMPLE(*inptr0++);
-      g = GETJSAMPLE(*inptr1++);
-      b = GETJSAMPLE(*inptr2++);
-      rgb = PACK_SHORT_565(r,g,b);
-      r = GETJSAMPLE(*inptr0++);
-      g = GETJSAMPLE(*inptr1++);
-      b = GETJSAMPLE(*inptr2++);
-      rgb = PACK_TWO_PIXELS(rgb, PACK_SHORT_565(r,g,b));
-      WRITE_TWO_ALIGNED_PIXELS(outptr, rgb);
-      outptr += 4;
-    }
-    if (num_cols&1) {
-      r = GETJSAMPLE(*inptr0);
-      g = GETJSAMPLE(*inptr1);
-      b = GETJSAMPLE(*inptr2);
-      rgb = PACK_SHORT_565(r,g,b);
-      *(INT16*)outptr = rgb;
-    }
-  }
-}
-
-
-METHODDEF(void)
-rgb_rgb_565D_convert (j_decompress_ptr cinfo,
-         JSAMPIMAGE input_buf, JDIMENSION input_row,
-         JSAMPARRAY output_buf, int num_rows)
-{
-  register JSAMPROW outptr;
-  register JSAMPROW inptr0, inptr1, inptr2;
-  register JDIMENSION col;
-  register JSAMPLE * range_limit = cinfo->sample_range_limit;
-  JDIMENSION num_cols = cinfo->output_width;
-  INT32 d0 = dither_matrix[cinfo->output_scanline & DITHER_MASK];
-  SHIFT_TEMPS
-
-  while (--num_rows >= 0) {
-    INT32 rgb;
-    unsigned int r, g, b;
-    inptr0 = input_buf[0][input_row];
-    inptr1 = input_buf[1][input_row];
-    inptr2 = input_buf[2][input_row];
-    input_row++;
-    outptr = *output_buf++;
-    if (PACK_NEED_ALIGNMENT(outptr)) {
-        r = range_limit[DITHER_565_R(GETJSAMPLE(*inptr0++), d0)];
-        g = range_limit[DITHER_565_G(GETJSAMPLE(*inptr1++), d0)];
-        b = range_limit[DITHER_565_B(GETJSAMPLE(*inptr2++), d0)];
-        rgb = PACK_SHORT_565(r,g,b);
-        *(INT16*)outptr = rgb;
-        outptr += 2;
-        num_cols--;
-    }
-    for (col = 0; col < (num_cols>>1); col++) {
-      r = range_limit[DITHER_565_R(GETJSAMPLE(*inptr0++), d0)];
-      g = range_limit[DITHER_565_G(GETJSAMPLE(*inptr1++), d0)];
-      b = range_limit[DITHER_565_B(GETJSAMPLE(*inptr2++), d0)];
-      d0 = DITHER_ROTATE(d0);
-      rgb = PACK_SHORT_565(r,g,b);
-      r = range_limit[DITHER_565_R(GETJSAMPLE(*inptr0++), d0)];
-      g = range_limit[DITHER_565_G(GETJSAMPLE(*inptr1++), d0)];
-      b = range_limit[DITHER_565_B(GETJSAMPLE(*inptr2++), d0)];
-      d0 = DITHER_ROTATE(d0);
-      rgb = PACK_TWO_PIXELS(rgb, PACK_SHORT_565(r,g,b));
-      WRITE_TWO_ALIGNED_PIXELS(outptr, rgb);
-      outptr += 4;
-    }
-    if (num_cols&1) {
-      r = range_limit[DITHER_565_R(GETJSAMPLE(*inptr0), d0)];
-      g = range_limit[DITHER_565_G(GETJSAMPLE(*inptr1), d0)];
-      b = range_limit[DITHER_565_B(GETJSAMPLE(*inptr2), d0)];
-      rgb = PACK_SHORT_565(r,g,b);
-      *(INT16*)outptr = rgb;
-    }
-  }
-}
-
-#endif
-
-/*
- * Color conversion for no colorspace change: just copy the data,
- * converting from separate-planes to interleaved representation.
- */
-
-METHODDEF(void)
-null_convert (j_decompress_ptr cinfo,
-	      JSAMPIMAGE input_buf, JDIMENSION input_row,
-	      JSAMPARRAY output_buf, int num_rows)
-{
-  register JSAMPROW inptr, outptr;
-  register JDIMENSION count;
-  register int num_components = cinfo->num_components;
-  JDIMENSION num_cols = cinfo->output_width;
-  int ci;
-
-  while (--num_rows >= 0) {
-    for (ci = 0; ci < num_components; ci++) {
-      inptr = input_buf[ci][input_row];
-      outptr = output_buf[0] + ci;
-      for (count = num_cols; count > 0; count--) {
-	*outptr = *inptr++;	/* needn't bother with GETJSAMPLE() here */
-	outptr += num_components;
-      }
-    }
-    input_row++;
-    output_buf++;
-  }
-}
-
-
-/*
- * Color conversion for grayscale: just copy the data.
- * This also works for YCbCr -> grayscale conversion, in which
- * we just copy the Y (luminance) component and ignore chrominance.
- */
-
-METHODDEF(void)
-grayscale_convert (j_decompress_ptr cinfo,
-		   JSAMPIMAGE input_buf, JDIMENSION input_row,
-		   JSAMPARRAY output_buf, int num_rows)
-{
-  jcopy_sample_rows(input_buf[0], (int) input_row, output_buf, 0,
-		    num_rows, cinfo->output_width);
-}
-
-
-/*
- * Convert grayscale to RGB: just duplicate the graylevel three times.
- * This is provided to support applications that don't want to cope
- * with grayscale as a separate case.
- */
-
-METHODDEF(void)
-gray_rgb_convert (j_decompress_ptr cinfo,
-		  JSAMPIMAGE input_buf, JDIMENSION input_row,
-		  JSAMPARRAY output_buf, int num_rows)
-{
-  register JSAMPROW inptr, outptr;
-  register JDIMENSION col;
-  JDIMENSION num_cols = cinfo->output_width;
-
-  while (--num_rows >= 0) {
-    inptr = input_buf[0][input_row++];
-    outptr = *output_buf++;
-    for (col = 0; col < num_cols; col++) {
-      /* We can dispense with GETJSAMPLE() here */
-      outptr[RGB_RED] = outptr[RGB_GREEN] = outptr[RGB_BLUE] = inptr[col];
-      outptr += RGB_PIXELSIZE;
-    }
-  }
-}
-
-#ifdef ANDROID_RGB
-METHODDEF(void)
-gray_rgba_8888_convert (j_decompress_ptr cinfo,
-          JSAMPIMAGE input_buf, JDIMENSION input_row,
-          JSAMPARRAY output_buf, int num_rows)
-{
-  register JSAMPROW inptr, outptr;
-  register JDIMENSION col;
-  JDIMENSION num_cols = cinfo->output_width;
-
-  while (--num_rows >= 0) {
-    inptr = input_buf[0][input_row++];
-    outptr = *output_buf++;
-    for (col = 0; col < num_cols; col++) {
-      /* We can dispense with GETJSAMPLE() here */
-      outptr[RGB_RED] = outptr[RGB_GREEN] = outptr[RGB_BLUE] = inptr[col];
-      outptr[RGB_ALPHA] = 0xff;
-      outptr += 4;
-    }
-  }
-}
-
-METHODDEF(void)
-gray_rgb_565_convert (j_decompress_ptr cinfo,
-          JSAMPIMAGE input_buf, JDIMENSION input_row,
-          JSAMPARRAY output_buf, int num_rows)
-{
-  register JSAMPROW inptr, outptr;
-  register JDIMENSION col;
-  JDIMENSION num_cols = cinfo->output_width;
-
-  while (--num_rows >= 0) {
-    INT32 rgb;
-    unsigned int g;
-    inptr = input_buf[0][input_row++];
-    outptr = *output_buf++;
-    if (PACK_NEED_ALIGNMENT(outptr)) {
-        g = *inptr++;
-        rgb = PACK_SHORT_565(g, g, g);
-        *(INT16*)outptr = rgb;
-        outptr += 2;
-        num_cols--;
-    }
-    for (col = 0; col < (num_cols>>1); col++) {
-      g = *inptr++;
-      rgb = PACK_SHORT_565(g, g, g);
-      g = *inptr++;
-      rgb = PACK_TWO_PIXELS(rgb, PACK_SHORT_565(g, g, g));
-      WRITE_TWO_ALIGNED_PIXELS(outptr, rgb);
-      outptr += 4;
-    }
-    if (num_cols&1) {
-      g = *inptr;
-      rgb = PACK_SHORT_565(g, g, g);
-      *(INT16*)outptr = rgb;
-    }
-  }
-}
-
-METHODDEF(void)
-gray_rgb_565D_convert (j_decompress_ptr cinfo,
-          JSAMPIMAGE input_buf, JDIMENSION input_row,
-          JSAMPARRAY output_buf, int num_rows)
-{
-  register JSAMPROW inptr, outptr;
-  register JDIMENSION col;
-  register JSAMPLE * range_limit = cinfo->sample_range_limit;
-  JDIMENSION num_cols = cinfo->output_width;
-  INT32 d0 = dither_matrix[cinfo->output_scanline & DITHER_MASK];
-
-  while (--num_rows >= 0) {
-    INT32 rgb;
-    unsigned int g;
-    inptr = input_buf[0][input_row++];
-    outptr = *output_buf++;
-    if (PACK_NEED_ALIGNMENT(outptr)) {
-        g = *inptr++;
-        g = range_limit[DITHER_565_R(g, d0)];
-        rgb = PACK_SHORT_565(g, g, g);
-        *(INT16*)outptr = rgb;
-        outptr += 2;
-        num_cols--;
-    }
-    for (col = 0; col < (num_cols>>1); col++) {
-      g = *inptr++;
-      g = range_limit[DITHER_565_R(g, d0)];
-      rgb = PACK_SHORT_565(g, g, g);
-      d0 = DITHER_ROTATE(d0);
-      g = *inptr++;
-      g = range_limit[DITHER_565_R(g, d0)];
-      rgb = PACK_TWO_PIXELS(rgb, PACK_SHORT_565(g, g, g));
-      d0 = DITHER_ROTATE(d0);
-      WRITE_TWO_ALIGNED_PIXELS(outptr, rgb);
-      outptr += 4;
-    }
-    if (num_cols&1) {
-      g = *inptr;
-      g = range_limit[DITHER_565_R(g, d0)];
-      rgb = PACK_SHORT_565(g, g, g);
-      *(INT16*)outptr = rgb;
-    }
-  }
-}
-#endif
-
-/*
- * Adobe-style YCCK->CMYK conversion.
- * We convert YCbCr to R=1-C, G=1-M, and B=1-Y using the same
- * conversion as above, while passing K (black) unchanged.
- * We assume build_ycc_rgb_table has been called.
- */
-
-METHODDEF(void)
-ycck_cmyk_convert (j_decompress_ptr cinfo,
-		   JSAMPIMAGE input_buf, JDIMENSION input_row,
-		   JSAMPARRAY output_buf, int num_rows)
-{
-  my_cconvert_ptr cconvert = (my_cconvert_ptr) cinfo->cconvert;
-  register int y, cb, cr;
-  register JSAMPROW outptr;
-  register JSAMPROW inptr0, inptr1, inptr2, inptr3;
-  register JDIMENSION col;
-  JDIMENSION num_cols = cinfo->output_width;
-  /* copy these pointers into registers if possible */
-  register JSAMPLE * range_limit = cinfo->sample_range_limit;
-  register int * Crrtab = cconvert->Cr_r_tab;
-  register int * Cbbtab = cconvert->Cb_b_tab;
-  register INT32 * Crgtab = cconvert->Cr_g_tab;
-  register INT32 * Cbgtab = cconvert->Cb_g_tab;
-  SHIFT_TEMPS
-
-  while (--num_rows >= 0) {
-    inptr0 = input_buf[0][input_row];
-    inptr1 = input_buf[1][input_row];
-    inptr2 = input_buf[2][input_row];
-    inptr3 = input_buf[3][input_row];
-    input_row++;
-    outptr = *output_buf++;
-    for (col = 0; col < num_cols; col++) {
-      y  = GETJSAMPLE(inptr0[col]);
-      cb = GETJSAMPLE(inptr1[col]);
-      cr = GETJSAMPLE(inptr2[col]);
-      /* Range-limiting is essential due to noise introduced by DCT losses. */
-      outptr[0] = range_limit[MAXJSAMPLE - (y + Crrtab[cr])];   /* red */
-      outptr[1] = range_limit[MAXJSAMPLE - (y +                 /* green */
-                              ((int) RIGHT_SHIFT(Cbgtab[cb] + Crgtab[cr],
-                                                 SCALEBITS)))];
-      outptr[2] = range_limit[MAXJSAMPLE - (y + Cbbtab[cb])];   /* blue */
-      /* K passes through unchanged */
-      outptr[3] = inptr3[col];	/* don't need GETJSAMPLE here */
-      outptr += 4;
-    }
-  }
-}
-
-
-/*
- * Empty method for start_pass.
- */
-
-METHODDEF(void)
-start_pass_dcolor (j_decompress_ptr cinfo)
-{
-  /* no work needed */
-}
-
-
-/*
- * Module initialization routine for output colorspace conversion.
- */
-
-GLOBAL(void)
-jinit_color_deconverter (j_decompress_ptr cinfo)
-{
-  my_cconvert_ptr cconvert;
-  int ci;
-
-  cconvert = (my_cconvert_ptr)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				SIZEOF(my_color_deconverter));
-  cinfo->cconvert = (struct jpeg_color_deconverter *) cconvert;
-  cconvert->pub.start_pass = start_pass_dcolor;
-
-  /* Make sure num_components agrees with jpeg_color_space */
-  switch (cinfo->jpeg_color_space) {
-  case JCS_GRAYSCALE:
-    if (cinfo->num_components != 1)
-      ERREXIT(cinfo, JERR_BAD_J_COLORSPACE);
-    break;
-
-  case JCS_RGB:
-  case JCS_YCbCr:
-    if (cinfo->num_components != 3)
-      ERREXIT(cinfo, JERR_BAD_J_COLORSPACE);
-    break;
-
-  case JCS_CMYK:
-  case JCS_YCCK:
-    if (cinfo->num_components != 4)
-      ERREXIT(cinfo, JERR_BAD_J_COLORSPACE);
-    break;
-
-  default:			/* JCS_UNKNOWN can be anything */
-    if (cinfo->num_components < 1)
-      ERREXIT(cinfo, JERR_BAD_J_COLORSPACE);
-    break;
-  }
-
-  /* Set out_color_components and conversion method based on requested space.
-   * Also clear the component_needed flags for any unused components,
-   * so that earlier pipeline stages can avoid useless computation.
-   */
-
-  switch (cinfo->out_color_space) {
-  case JCS_GRAYSCALE:
-    cinfo->out_color_components = 1;
-    if (cinfo->jpeg_color_space == JCS_GRAYSCALE ||
-	cinfo->jpeg_color_space == JCS_YCbCr) {
-      cconvert->pub.color_convert = grayscale_convert;
-      /* For color->grayscale conversion, only the Y (0) component is needed */
-      for (ci = 1; ci < cinfo->num_components; ci++)
-	cinfo->comp_info[ci].component_needed = FALSE;
-    } else
-      ERREXIT(cinfo, JERR_CONVERSION_NOTIMPL);
-    break;
-
-  case JCS_RGB:
-    cinfo->out_color_components = RGB_PIXELSIZE;
-    if (cinfo->jpeg_color_space == JCS_YCbCr) {
-      cconvert->pub.color_convert = ycc_rgb_convert;
-      build_ycc_rgb_table(cinfo);
-    } else if (cinfo->jpeg_color_space == JCS_GRAYSCALE) {
-      cconvert->pub.color_convert = gray_rgb_convert;
-    } else if (cinfo->jpeg_color_space == JCS_RGB && RGB_PIXELSIZE == 3) {
-      cconvert->pub.color_convert = null_convert;
-    } else
-      ERREXIT(cinfo, JERR_CONVERSION_NOTIMPL);
-    break;
-
-#ifdef ANDROID_RGB
-  case JCS_RGBA_8888:
-    cinfo->out_color_components = 4;
-    if (cinfo->jpeg_color_space == JCS_YCbCr) {
-#if defined(NV_ARM_NEON) && defined(__ARM_HAVE_NEON)
-      if (cap_neon_ycc_rgb()) {
-        cconvert->pub.color_convert = jsimd_ycc_rgba8888_convert;
-      } else {
-        cconvert->pub.color_convert = ycc_rgba_8888_convert;
-      }
-#else
-      cconvert->pub.color_convert = ycc_rgba_8888_convert;
-#endif
-      build_ycc_rgb_table(cinfo);
-    } else if (cinfo->jpeg_color_space == JCS_GRAYSCALE) {
-      cconvert->pub.color_convert = gray_rgba_8888_convert;
-    } else if (cinfo->jpeg_color_space == JCS_RGB) {
-      cconvert->pub.color_convert = rgb_rgba_8888_convert;
-    } else
-      ERREXIT(cinfo, JERR_CONVERSION_NOTIMPL);
-    break;
-
-  case JCS_RGB_565:
-    cinfo->out_color_components = RGB_PIXELSIZE;
-    if (cinfo->dither_mode == JDITHER_NONE) {
-      if (cinfo->jpeg_color_space == JCS_YCbCr) {
-#if defined(NV_ARM_NEON) && defined(__ARM_HAVE_NEON)
-        if (cap_neon_ycc_rgb())  {
-          cconvert->pub.color_convert = jsimd_ycc_rgb565_convert;
-        } else {
-          cconvert->pub.color_convert = ycc_rgb_565_convert;
-        }
-#else
-        cconvert->pub.color_convert = ycc_rgb_565_convert;
-#endif
-        build_ycc_rgb_table(cinfo);
-      } else if (cinfo->jpeg_color_space == JCS_GRAYSCALE) {
-        cconvert->pub.color_convert = gray_rgb_565_convert;
-      } else if (cinfo->jpeg_color_space == JCS_RGB) {
-        cconvert->pub.color_convert = rgb_rgb_565_convert;
-      } else
-        ERREXIT(cinfo, JERR_CONVERSION_NOTIMPL);
-    } else {
-      /* only ordered dither is supported */
-      if (cinfo->jpeg_color_space == JCS_YCbCr) {
-        cconvert->pub.color_convert = ycc_rgb_565D_convert;
-        build_ycc_rgb_table(cinfo);
-      } else if (cinfo->jpeg_color_space == JCS_GRAYSCALE) {
-        cconvert->pub.color_convert = gray_rgb_565D_convert;
-      } else if (cinfo->jpeg_color_space == JCS_RGB) {
-        cconvert->pub.color_convert = rgb_rgb_565D_convert;
-      } else
-        ERREXIT(cinfo, JERR_CONVERSION_NOTIMPL);
-    }
-    break;
-#endif
-    
-  case JCS_CMYK:
-    cinfo->out_color_components = 4;
-    if (cinfo->jpeg_color_space == JCS_YCCK) {
-      cconvert->pub.color_convert = ycck_cmyk_convert;
-      build_ycc_rgb_table(cinfo);
-    } else if (cinfo->jpeg_color_space == JCS_CMYK) {
-      cconvert->pub.color_convert = null_convert;
-    } else
-      ERREXIT(cinfo, JERR_CONVERSION_NOTIMPL);
-    break;
-
-  default:
-    /* Permit null conversion to same output space */
-    if (cinfo->out_color_space == cinfo->jpeg_color_space) {
-      cinfo->out_color_components = cinfo->num_components;
-      cconvert->pub.color_convert = null_convert;
-    } else			/* unsupported non-null conversion */
-      ERREXIT(cinfo, JERR_CONVERSION_NOTIMPL);
-    break;
-  }
-
-  if (cinfo->quantize_colors)
-    cinfo->output_components = 1; /* single colormapped output component */
-  else
-    cinfo->output_components = cinfo->out_color_components;
-}
@@ -1,180 +0,0 @@
-/*
- * jdct.h
- *
- * Copyright (C) 1994-1996, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This include file contains common declarations for the forward and
- * inverse DCT modules.  These declarations are private to the DCT managers
- * (jcdctmgr.c, jddctmgr.c) and the individual DCT algorithms.
- * The individual DCT algorithms are kept in separate files to ease 
- * machine-dependent tuning (e.g., assembly coding).
- */
-
-
-/*
- * A forward DCT routine is given a pointer to a work area of type DCTELEM[];
- * the DCT is to be performed in-place in that buffer.  Type DCTELEM is int
- * for 8-bit samples, INT32 for 12-bit samples.  (NOTE: Floating-point DCT
- * implementations use an array of type FAST_FLOAT, instead.)
- * The DCT inputs are expected to be signed (range +-CENTERJSAMPLE).
- * The DCT outputs are returned scaled up by a factor of 8; they therefore
- * have a range of +-8K for 8-bit data, +-128K for 12-bit data.  This
- * convention improves accuracy in integer implementations and saves some
- * work in floating-point ones.
- * Quantization of the output coefficients is done by jcdctmgr.c.
- */
-
-#if BITS_IN_JSAMPLE == 8
-#ifdef ANDROID_MIPS_IDCT
-typedef short DCTELEM;		/* 16 or 32 bits is fine */
-#else
-typedef int DCTELEM;		/* 16 or 32 bits is fine */
-#endif
-#else
-typedef INT32 DCTELEM;		/* must have 32 bits */
-#endif
-
-typedef JMETHOD(void, forward_DCT_method_ptr, (DCTELEM * data));
-typedef JMETHOD(void, float_DCT_method_ptr, (FAST_FLOAT * data));
-
-
-/*
- * An inverse DCT routine is given a pointer to the input JBLOCK and a pointer
- * to an output sample array.  The routine must dequantize the input data as
- * well as perform the IDCT; for dequantization, it uses the multiplier table
- * pointed to by compptr->dct_table.  The output data is to be placed into the
- * sample array starting at a specified column.  (Any row offset needed will
- * be applied to the array pointer before it is passed to the IDCT code.)
- * Note that the number of samples emitted by the IDCT routine is
- * DCT_scaled_size * DCT_scaled_size.
- */
-
-/* typedef inverse_DCT_method_ptr is declared in jpegint.h */
-
-/*
- * Each IDCT routine has its own ideas about the best dct_table element type.
- */
-
-typedef MULTIPLIER ISLOW_MULT_TYPE; /* short or int, whichever is faster */
-#if BITS_IN_JSAMPLE == 8
-typedef MULTIPLIER IFAST_MULT_TYPE; /* 16 bits is OK, use short if faster */
-#define IFAST_SCALE_BITS  2	/* fractional bits in scale factors */
-#else
-typedef INT32 IFAST_MULT_TYPE;	/* need 32 bits for scaled quantizers */
-#define IFAST_SCALE_BITS  13	/* fractional bits in scale factors */
-#endif
-typedef FAST_FLOAT FLOAT_MULT_TYPE; /* preferred floating type */
-
-
-/*
- * Each IDCT routine is responsible for range-limiting its results and
- * converting them to unsigned form (0..MAXJSAMPLE).  The raw outputs could
- * be quite far out of range if the input data is corrupt, so a bulletproof
- * range-limiting step is required.  We use a mask-and-table-lookup method
- * to do the combined operations quickly.  See the comments with
- * prepare_range_limit_table (in jdmaster.c) for more info.
- */
-
-#define IDCT_range_limit(cinfo)  ((cinfo)->sample_range_limit + CENTERJSAMPLE)
-
-#define RANGE_MASK  (MAXJSAMPLE * 4 + 3) /* 2 bits wider than legal samples */
-
-
-/* Short forms of external names for systems with brain-damaged linkers. */
-
-#ifdef NEED_SHORT_EXTERNAL_NAMES
-#define jpeg_fdct_islow		jFDislow
-#define jpeg_fdct_ifast		jFDifast
-#define jpeg_fdct_float		jFDfloat
-#define jpeg_idct_islow		jRDislow
-#define jpeg_idct_ifast		jRDifast
-#define jpeg_idct_float		jRDfloat
-#define jpeg_idct_4x4		jRD4x4
-#define jpeg_idct_2x2		jRD2x2
-#define jpeg_idct_1x1		jRD1x1
-#endif /* NEED_SHORT_EXTERNAL_NAMES */
-
-/* Extern declarations for the forward and inverse DCT routines. */
-
-EXTERN(void) jpeg_fdct_islow JPP((DCTELEM * data));
-EXTERN(void) jpeg_fdct_ifast JPP((DCTELEM * data));
-EXTERN(void) jpeg_fdct_float JPP((FAST_FLOAT * data));
-
-EXTERN(void) jpeg_idct_islow
-    JPP((j_decompress_ptr cinfo, jpeg_component_info * compptr,
-	 JCOEFPTR coef_block, JSAMPARRAY output_buf, JDIMENSION output_col));
-EXTERN(void) jpeg_idct_ifast
-    JPP((j_decompress_ptr cinfo, jpeg_component_info * compptr,
-	 JCOEFPTR coef_block, JSAMPARRAY output_buf, JDIMENSION output_col));
-EXTERN(void) jpeg_idct_float
-    JPP((j_decompress_ptr cinfo, jpeg_component_info * compptr,
-	 JCOEFPTR coef_block, JSAMPARRAY output_buf, JDIMENSION output_col));
-EXTERN(void) jpeg_idct_4x4
-    JPP((j_decompress_ptr cinfo, jpeg_component_info * compptr,
-	 JCOEFPTR coef_block, JSAMPARRAY output_buf, JDIMENSION output_col));
-EXTERN(void) jpeg_idct_2x2
-    JPP((j_decompress_ptr cinfo, jpeg_component_info * compptr,
-	 JCOEFPTR coef_block, JSAMPARRAY output_buf, JDIMENSION output_col));
-EXTERN(void) jpeg_idct_1x1
-    JPP((j_decompress_ptr cinfo, jpeg_component_info * compptr,
-	 JCOEFPTR coef_block, JSAMPARRAY output_buf, JDIMENSION output_col));
-
-
-/*
- * Macros for handling fixed-point arithmetic; these are used by many
- * but not all of the DCT/IDCT modules.
- *
- * All values are expected to be of type INT32.
- * Fractional constants are scaled left by CONST_BITS bits.
- * CONST_BITS is defined within each module using these macros,
- * and may differ from one module to the next.
- */
-
-#define ONE	((INT32) 1)
-#define CONST_SCALE (ONE << CONST_BITS)
-
-/* Convert a positive real constant to an integer scaled by CONST_SCALE.
- * Caution: some C compilers fail to reduce "FIX(constant)" at compile time,
- * thus causing a lot of useless floating-point operations at run time.
- */
-
-#define FIX(x)	((INT32) ((x) * CONST_SCALE + 0.5))
-
-/* Descale and correctly round an INT32 value that's scaled by N bits.
- * We assume RIGHT_SHIFT rounds towards minus infinity, so adding
- * the fudge factor is correct for either sign of X.
- */
-
-#define DESCALE(x,n)  RIGHT_SHIFT((x) + (ONE << ((n)-1)), n)
-
-/* Multiply an INT32 variable by an INT32 constant to yield an INT32 result.
- * This macro is used only when the two inputs will actually be no more than
- * 16 bits wide, so that a 16x16->32 bit multiply can be used instead of a
- * full 32x32 multiply.  This provides a useful speedup on many machines.
- * Unfortunately there is no way to specify a 16x16->32 multiply portably
- * in C, but some C compilers will do the right thing if you provide the
- * correct combination of casts.
- */
-
-#ifdef SHORTxSHORT_32		/* may work if 'int' is 32 bits */
-#define MULTIPLY16C16(var,const)  (((INT16) (var)) * ((INT16) (const)))
-#endif
-#ifdef SHORTxLCONST_32		/* known to work with Microsoft C 6.0 */
-#define MULTIPLY16C16(var,const)  (((INT16) (var)) * ((INT32) (const)))
-#endif
-
-#ifndef MULTIPLY16C16		/* default definition */
-#define MULTIPLY16C16(var,const)  ((var) * (const))
-#endif
-
-/* Same except both inputs are variables. */
-
-#ifdef SHORTxSHORT_32		/* may work if 'int' is 32 bits */
-#define MULTIPLY16V16(var1,var2)  (((INT16) (var1)) * ((INT16) (var2)))
-#endif
-
-#ifndef MULTIPLY16V16		/* default definition */
-#define MULTIPLY16V16(var1,var2)  ((var1) * (var2))
-#endif
@@ -1,383 +0,0 @@
-/*
- * jddctmgr.c
- *
- * Copyright (C) 1994-1996, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains the inverse-DCT management logic.
- * This code selects a particular IDCT implementation to be used,
- * and it performs related housekeeping chores.  No code in this file
- * is executed per IDCT step, only during output pass setup.
- *
- * Note that the IDCT routines are responsible for performing coefficient
- * dequantization as well as the IDCT proper.  This module sets up the
- * dequantization multiplier table needed by the IDCT routine.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-#include "jdct.h"		/* Private declarations for DCT subsystem */
-
-#ifdef ANDROID_ARMV6_IDCT
-  #undef ANDROID_ARMV6_IDCT
-  #ifdef __arm__
-    #include <machine/cpu-features.h>
-    #if __ARM_ARCH__ >= 6
-      #define ANDROID_ARMV6_IDCT
-    #else
-      #warning "ANDROID_ARMV6_IDCT is disabled"
-    #endif
-  #endif
-#endif
-
-#ifdef NV_ARM_NEON
-#include "jsimd_neon.h"
-#endif
-
-#ifdef ANDROID_ARMV6_IDCT
-
-/* Intentionally declare the prototype with arguments of primitive types instead
- * of type-defined ones. This will at least generate some warnings if jmorecfg.h
- * is changed and becomes incompatible with the assembly code.
- */
-extern void armv6_idct(short *coefs, int *quans, unsigned char **rows, int col);
-
-void jpeg_idct_armv6 (j_decompress_ptr cinfo, jpeg_component_info * compptr,
-		 JCOEFPTR coef_block,
-		 JSAMPARRAY output_buf, JDIMENSION output_col)
-{
-  IFAST_MULT_TYPE *dct_table = (IFAST_MULT_TYPE *)compptr->dct_table;
-  armv6_idct(coef_block, dct_table, output_buf, output_col);
-}
-
-#endif
-
-#ifdef ANDROID_INTELSSE2_IDCT
-extern short __attribute__((aligned(16))) quantptrSSE[DCTSIZE2];
-extern void jpeg_idct_intelsse (j_decompress_ptr cinfo, jpeg_component_info * compptr,
-		JCOEFPTR coef_block,
-		JSAMPARRAY output_buf, JDIMENSION output_col);
-#endif
-
-#ifdef ANDROID_MIPS_IDCT
-extern void jpeg_idct_mips(j_decompress_ptr, jpeg_component_info *, JCOEFPTR, JSAMPARRAY, JDIMENSION);
-#endif
-
-/*
- * The decompressor input side (jdinput.c) saves away the appropriate
- * quantization table for each component at the start of the first scan
- * involving that component.  (This is necessary in order to correctly
- * decode files that reuse Q-table slots.)
- * When we are ready to make an output pass, the saved Q-table is converted
- * to a multiplier table that will actually be used by the IDCT routine.
- * The multiplier table contents are IDCT-method-dependent.  To support
- * application changes in IDCT method between scans, we can remake the
- * multiplier tables if necessary.
- * In buffered-image mode, the first output pass may occur before any data
- * has been seen for some components, and thus before their Q-tables have
- * been saved away.  To handle this case, multiplier tables are preset
- * to zeroes; the result of the IDCT will be a neutral gray level.
- */
-
-
-/* Private subobject for this module */
-
-typedef struct {
-  struct jpeg_inverse_dct pub;	/* public fields */
-
-  /* This array contains the IDCT method code that each multiplier table
-   * is currently set up for, or -1 if it's not yet set up.
-   * The actual multiplier tables are pointed to by dct_table in the
-   * per-component comp_info structures.
-   */
-  int cur_method[MAX_COMPONENTS];
-} my_idct_controller;
-
-typedef my_idct_controller * my_idct_ptr;
-
-
-/* Allocated multiplier tables: big enough for any supported variant */
-
-typedef union {
-  ISLOW_MULT_TYPE islow_array[DCTSIZE2];
-#ifdef DCT_IFAST_SUPPORTED
-  IFAST_MULT_TYPE ifast_array[DCTSIZE2];
-#endif
-#ifdef DCT_FLOAT_SUPPORTED
-  FLOAT_MULT_TYPE float_array[DCTSIZE2];
-#endif
-} multiplier_table;
-
-
-/* The current scaled-IDCT routines require ISLOW-style multiplier tables,
- * so be sure to compile that code if either ISLOW or SCALING is requested.
- */
-#ifdef DCT_ISLOW_SUPPORTED
-#define PROVIDE_ISLOW_TABLES
-#else
-#ifdef IDCT_SCALING_SUPPORTED
-#define PROVIDE_ISLOW_TABLES
-#endif
-#endif
-
-
-/*
- * Prepare for an output pass.
- * Here we select the proper IDCT routine for each component and build
- * a matching multiplier table.
- */
-
-METHODDEF(void)
-start_pass (j_decompress_ptr cinfo)
-{
-  my_idct_ptr idct = (my_idct_ptr) cinfo->idct;
-  int ci, i;
-  jpeg_component_info *compptr;
-  int method = 0;
-  inverse_DCT_method_ptr method_ptr = NULL;
-  JQUANT_TBL * qtbl;
-
-  for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
-       ci++, compptr++) {
-    /* Select the proper IDCT routine for this component's scaling */
-    switch (compptr->DCT_scaled_size) {
-#ifdef IDCT_SCALING_SUPPORTED
-    case 1:
-      method_ptr = jpeg_idct_1x1;
-      method = JDCT_ISLOW;	/* jidctred uses islow-style table */
-      break;
-    case 2:
-#if defined(NV_ARM_NEON) && defined(__ARM_HAVE_NEON)
-      if (cap_neon_idct_2x2()) {
-        method_ptr = jsimd_idct_2x2;
-      } else {
-        method_ptr = jpeg_idct_2x2;
-      }
-#else
-      method_ptr = jpeg_idct_2x2;
-#endif
-      method = JDCT_ISLOW;	/* jidctred uses islow-style table */
-      break;
-    case 4:
-#if defined(NV_ARM_NEON) && defined(__ARM_HAVE_NEON)
-	  if (cap_neon_idct_4x4()) {
-        method_ptr = jsimd_idct_4x4;
-      } else {
-        method_ptr = jpeg_idct_4x4;
-      }
-#else
-      method_ptr = jpeg_idct_4x4;
-#endif
-      method = JDCT_ISLOW;	/* jidctred uses islow-style table */
-      break;
-#endif
-    case DCTSIZE:
-      switch (cinfo->dct_method) {
-#ifdef ANDROID_ARMV6_IDCT
-      case JDCT_ISLOW:
-      case JDCT_IFAST:
-	method_ptr = jpeg_idct_armv6;
-	method = JDCT_IFAST;
-	break;
-#else /* ANDROID_ARMV6_IDCT */
-#ifdef ANDROID_INTELSSE2_IDCT
-      case JDCT_ISLOW:
-      case JDCT_IFAST:
-	method_ptr = jpeg_idct_intelsse;
-	method = JDCT_ISLOW; /* Use quant table of ISLOW.*/
-	break;
-#else /* ANDROID_INTELSSE2_IDCT */
-#ifdef ANDROID_MIPS_IDCT
-      case JDCT_ISLOW:
-      case JDCT_IFAST:
-	method_ptr = jpeg_idct_mips;
-	method = JDCT_IFAST;
-	break;
-#else /* ANDROID_MIPS_IDCT */
-#ifdef DCT_ISLOW_SUPPORTED
-      case JDCT_ISLOW:
-	method_ptr = jpeg_idct_islow;
-	method = JDCT_ISLOW;
-	break;
-#endif
-#ifdef DCT_IFAST_SUPPORTED
-      case JDCT_IFAST:
-#if defined(NV_ARM_NEON) && defined(__ARM_HAVE_NEON)
-        if (cap_neon_idct_ifast()) {
-          method_ptr = jsimd_idct_ifast;
-        } else {
-          method_ptr = jpeg_idct_ifast;
-        }
-#else
-        method_ptr = jpeg_idct_ifast;
-#endif
-	method = JDCT_IFAST;
-	break;
-#endif
-#endif /* ANDROID_MIPS_IDCT */
-#endif /* ANDROID_INTELSSE2_IDCT*/
-#endif /* ANDROID_ARMV6_IDCT */
-#ifdef DCT_FLOAT_SUPPORTED
-      case JDCT_FLOAT:
-	method_ptr = jpeg_idct_float;
-	method = JDCT_FLOAT;
-	break;
-#endif
-      default:
-	ERREXIT(cinfo, JERR_NOT_COMPILED);
-	break;
-      }
-      break;
-    default:
-      ERREXIT1(cinfo, JERR_BAD_DCTSIZE, compptr->DCT_scaled_size);
-      break;
-    }
-    idct->pub.inverse_DCT[ci] = method_ptr;
-    /* Create multiplier table from quant table.
-     * However, we can skip this if the component is uninteresting
-     * or if we already built the table.  Also, if no quant table
-     * has yet been saved for the component, we leave the
-     * multiplier table all-zero; we'll be reading zeroes from the
-     * coefficient controller's buffer anyway.
-     */
-    if (! compptr->component_needed || idct->cur_method[ci] == method)
-      continue;
-    qtbl = compptr->quant_table;
-    if (qtbl == NULL)		/* happens if no data yet for component */
-      continue;
-    idct->cur_method[ci] = method;
-    switch (method) {
-#ifdef PROVIDE_ISLOW_TABLES
-    case JDCT_ISLOW:
-      {
-	/* For LL&M IDCT method, multipliers are equal to raw quantization
-	 * coefficients, but are stored as ints to ensure access efficiency.
-	 */
-	ISLOW_MULT_TYPE * ismtbl = (ISLOW_MULT_TYPE *) compptr->dct_table;
-	for (i = 0; i < DCTSIZE2; i++) {
-	  ismtbl[i] = (ISLOW_MULT_TYPE) qtbl->quantval[i];
-	}
-      }
-      break;
-#endif
-#ifdef DCT_IFAST_SUPPORTED
-    case JDCT_IFAST:
-      {
-	/* For AA&N IDCT method, multipliers are equal to quantization
-	 * coefficients scaled by scalefactor[row]*scalefactor[col], where
-	 *   scalefactor[0] = 1
-	 *   scalefactor[k] = cos(k*PI/16) * sqrt(2)    for k=1..7
-	 * For integer operation, the multiplier table is to be scaled by
-	 * IFAST_SCALE_BITS.
-	 */
-	IFAST_MULT_TYPE * ifmtbl = (IFAST_MULT_TYPE *) compptr->dct_table;
-#ifdef ANDROID_ARMV6_IDCT
-	/* Precomputed values scaled up by 15 bits. */
-	static const unsigned short scales[DCTSIZE2] = {
-	  32768, 45451, 42813, 38531, 32768, 25746, 17734,  9041,
-	  45451, 63042, 59384, 53444, 45451, 35710, 24598, 12540,
-	  42813, 59384, 55938, 50343, 42813, 33638, 23170, 11812,
-	  38531, 53444, 50343, 45308, 38531, 30274, 20853, 10631,
-	  32768, 45451, 42813, 38531, 32768, 25746, 17734,  9041,
-	  25746, 35710, 33638, 30274, 25746, 20228, 13933,  7103,
-	  17734, 24598, 23170, 20853, 17734, 13933,  9598,  4893,
-	   9041, 12540, 11812, 10631,  9041,  7103,  4893,  2494,
-	};
-	/* Inverse map of [7, 5, 1, 3, 0, 2, 4, 6]. */
-	static const char orders[DCTSIZE] = {4, 2, 5, 3, 6, 1, 7, 0};
-	/* Reorder the columns after transposing. */
-	for (i = 0; i < DCTSIZE2; ++i) {
-	  int j = ((i & 7) << 3) + orders[i >> 3];
-	  ifmtbl[j] = (qtbl->quantval[i] * scales[i] + 2) >> 2;
-	}
-#else /* ANDROID_ARMV6_IDCT */
-
-#define CONST_BITS 14
-	static const INT16 aanscales[DCTSIZE2] = {
-	  /* precomputed values scaled up by 14 bits */
-	  16384, 22725, 21407, 19266, 16384, 12873,  8867,  4520,
-	  22725, 31521, 29692, 26722, 22725, 17855, 12299,  6270,
-	  21407, 29692, 27969, 25172, 21407, 16819, 11585,  5906,
-	  19266, 26722, 25172, 22654, 19266, 15137, 10426,  5315,
-	  16384, 22725, 21407, 19266, 16384, 12873,  8867,  4520,
-	  12873, 17855, 16819, 15137, 12873, 10114,  6967,  3552,
-	   8867, 12299, 11585, 10426,  8867,  6967,  4799,  2446,
-	   4520,  6270,  5906,  5315,  4520,  3552,  2446,  1247
-	};
-	SHIFT_TEMPS
-
-	for (i = 0; i < DCTSIZE2; i++) {
-	  ifmtbl[i] = (IFAST_MULT_TYPE)
-	    DESCALE(MULTIPLY16V16((INT32) qtbl->quantval[i],
-				  (INT32) aanscales[i]),
-		    CONST_BITS-IFAST_SCALE_BITS);
-	}
-#endif /* ANDROID_ARMV6_IDCT */
-      }
-      break;
-#endif
-#ifdef DCT_FLOAT_SUPPORTED
-    case JDCT_FLOAT:
-      {
-	/* For float AA&N IDCT method, multipliers are equal to quantization
-	 * coefficients scaled by scalefactor[row]*scalefactor[col], where
-	 *   scalefactor[0] = 1
-	 *   scalefactor[k] = cos(k*PI/16) * sqrt(2)    for k=1..7
-	 */
-	FLOAT_MULT_TYPE * fmtbl = (FLOAT_MULT_TYPE *) compptr->dct_table;
-	int row, col;
-	static const double aanscalefactor[DCTSIZE] = {
-	  1.0, 1.387039845, 1.306562965, 1.175875602,
-	  1.0, 0.785694958, 0.541196100, 0.275899379
-	};
-
-	i = 0;
-	for (row = 0; row < DCTSIZE; row++) {
-	  for (col = 0; col < DCTSIZE; col++) {
-	    fmtbl[i] = (FLOAT_MULT_TYPE)
-	      ((double) qtbl->quantval[i] *
-	       aanscalefactor[row] * aanscalefactor[col]);
-	    i++;
-	  }
-	}
-      }
-      break;
-#endif
-    default:
-      ERREXIT(cinfo, JERR_NOT_COMPILED);
-      break;
-    }
-  }
-}
-
-
-/*
- * Initialize IDCT manager.
- */
-
-GLOBAL(void)
-jinit_inverse_dct (j_decompress_ptr cinfo)
-{
-  my_idct_ptr idct;
-  int ci;
-  jpeg_component_info *compptr;
-
-  idct = (my_idct_ptr)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				SIZEOF(my_idct_controller));
-  cinfo->idct = (struct jpeg_inverse_dct *) idct;
-  idct->pub.start_pass = start_pass;
-
-  for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
-       ci++, compptr++) {
-    /* Allocate and pre-zero a multiplier table for each component */
-    compptr->dct_table =
-      (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				  SIZEOF(multiplier_table));
-    MEMZERO(compptr->dct_table, SIZEOF(multiplier_table));
-    /* Mark multiplier table not yet set up for any method */
-    idct->cur_method[ci] = -1;
-  }
-}
@@ -1,894 +0,0 @@
-/*
- * jdhuff.c
- *
- * Copyright (C) 1991-1997, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains Huffman entropy decoding routines.
- *
- * Much of the complexity here has to do with supporting input suspension.
- * If the data source module demands suspension, we want to be able to back
- * up to the start of the current MCU.  To do this, we copy state variables
- * into local working storage, and update them back to the permanent
- * storage only upon successful completion of an MCU.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-#include "jdhuff.h"		/* Declarations shared with jdphuff.c */
-
-LOCAL(boolean) process_restart (j_decompress_ptr cinfo);
-
-
-/*
- * Expanded entropy decoder object for Huffman decoding.
- *
- * The savable_state subrecord contains fields that change within an MCU,
- * but must not be updated permanently until we complete the MCU.
- */
-
-typedef struct {
-  int last_dc_val[MAX_COMPS_IN_SCAN]; /* last DC coef for each component */
-} savable_state;
-
-/* This macro is to work around compilers with missing or broken
- * structure assignment.  You'll need to fix this code if you have
- * such a compiler and you change MAX_COMPS_IN_SCAN.
- */
-
-#ifndef NO_STRUCT_ASSIGN
-#define ASSIGN_STATE(dest,src)  ((dest) = (src))
-#else
-#if MAX_COMPS_IN_SCAN == 4
-#define ASSIGN_STATE(dest,src)  \
-	((dest).last_dc_val[0] = (src).last_dc_val[0], \
-	 (dest).last_dc_val[1] = (src).last_dc_val[1], \
-	 (dest).last_dc_val[2] = (src).last_dc_val[2], \
-	 (dest).last_dc_val[3] = (src).last_dc_val[3])
-#endif
-#endif
-
-
-typedef struct {
-  struct jpeg_entropy_decoder pub; /* public fields */
-
-  /* These fields are loaded into local variables at start of each MCU.
-   * In case of suspension, we exit WITHOUT updating them.
-   */
-  bitread_perm_state bitstate;	/* Bit buffer at start of MCU */
-  savable_state saved;		/* Other state at start of MCU */
-
-  /* These fields are NOT loaded into local working state. */
-  unsigned int restarts_to_go;	/* MCUs left in this restart interval */
-
-  /* Pointers to derived tables (these workspaces have image lifespan) */
-  d_derived_tbl * dc_derived_tbls[NUM_HUFF_TBLS];
-  d_derived_tbl * ac_derived_tbls[NUM_HUFF_TBLS];
-
-  /* Precalculated info set up by start_pass for use in decode_mcu: */
-
-  /* Pointers to derived tables to be used for each block within an MCU */
-  d_derived_tbl * dc_cur_tbls[D_MAX_BLOCKS_IN_MCU];
-  d_derived_tbl * ac_cur_tbls[D_MAX_BLOCKS_IN_MCU];
-  /* Whether we care about the DC and AC coefficient values for each block */
-  boolean dc_needed[D_MAX_BLOCKS_IN_MCU];
-  boolean ac_needed[D_MAX_BLOCKS_IN_MCU];
-} huff_entropy_decoder;
-
-typedef huff_entropy_decoder * huff_entropy_ptr;
-
-/*
- * Initialize for a Huffman-compressed scan.
- */
-
-METHODDEF(void)
-start_pass_huff_decoder (j_decompress_ptr cinfo)
-{
-  huff_entropy_ptr entropy = (huff_entropy_ptr) cinfo->entropy;
-  int ci, blkn, dctbl, actbl;
-  jpeg_component_info * compptr;
-
-  /* Check that the scan parameters Ss, Se, Ah/Al are OK for sequential JPEG.
-   * This ought to be an error condition, but we make it a warning because
-   * there are some baseline files out there with all zeroes in these bytes.
-   */
-  if (cinfo->Ss != 0 || cinfo->Se != DCTSIZE2-1 ||
-      cinfo->Ah != 0 || cinfo->Al != 0)
-    WARNMS(cinfo, JWRN_NOT_SEQUENTIAL);
-
-  for (ci = 0; ci < cinfo->comps_in_scan; ci++) {
-    compptr = cinfo->cur_comp_info[ci];
-    dctbl = compptr->dc_tbl_no;
-    actbl = compptr->ac_tbl_no;
-    /* Compute derived values for Huffman tables */
-    /* We may do this more than once for a table, but it's not expensive */
-    jpeg_make_d_derived_tbl(cinfo, TRUE, dctbl,
-			    & entropy->dc_derived_tbls[dctbl]);
-    jpeg_make_d_derived_tbl(cinfo, FALSE, actbl,
-			    & entropy->ac_derived_tbls[actbl]);
-    /* Initialize DC predictions to 0 */
-    entropy->saved.last_dc_val[ci] = 0;
-  }
-
-  /* Precalculate decoding info for each block in an MCU of this scan */
-  for (blkn = 0; blkn < cinfo->blocks_in_MCU; blkn++) {
-    ci = cinfo->MCU_membership[blkn];
-    compptr = cinfo->cur_comp_info[ci];
-    /* Precalculate which table to use for each block */
-    entropy->dc_cur_tbls[blkn] = entropy->dc_derived_tbls[compptr->dc_tbl_no];
-    entropy->ac_cur_tbls[blkn] = entropy->ac_derived_tbls[compptr->ac_tbl_no];
-    /* Decide whether we really care about the coefficient values */
-    if (compptr->component_needed) {
-      entropy->dc_needed[blkn] = TRUE;
-      /* we don't need the ACs if producing a 1/8th-size image */
-      entropy->ac_needed[blkn] = (compptr->DCT_scaled_size > 1);
-    } else {
-      entropy->dc_needed[blkn] = entropy->ac_needed[blkn] = FALSE;
-    }
-  }
-
-  /* Initialize bitread state variables */
-  entropy->bitstate.bits_left = 0;
-  entropy->bitstate.get_buffer = 0; /* unnecessary, but keeps Purify quiet */
-  entropy->pub.insufficient_data = FALSE;
-
-  /* Initialize restart counter */
-  entropy->restarts_to_go = cinfo->restart_interval;
-}
-
-
-/*
- * Compute the derived values for a Huffman table.
- * This routine also performs some validation checks on the table.
- *
- * Note this is also used by jdphuff.c.
- */
-
-GLOBAL(void)
-jpeg_make_d_derived_tbl (j_decompress_ptr cinfo, boolean isDC, int tblno,
-			 d_derived_tbl ** pdtbl)
-{
-  JHUFF_TBL *htbl;
-  d_derived_tbl *dtbl;
-  int p, i, l, si, numsymbols;
-  int lookbits, ctr;
-  char huffsize[257];
-  unsigned int huffcode[257];
-  unsigned int code;
-
-  /* Note that huffsize[] and huffcode[] are filled in code-length order,
-   * paralleling the order of the symbols themselves in htbl->huffval[].
-   */
-
-  /* Find the input Huffman table */
-  if (tblno < 0 || tblno >= NUM_HUFF_TBLS)
-    ERREXIT1(cinfo, JERR_NO_HUFF_TABLE, tblno);
-  htbl =
-    isDC ? cinfo->dc_huff_tbl_ptrs[tblno] : cinfo->ac_huff_tbl_ptrs[tblno];
-  if (htbl == NULL)
-    ERREXIT1(cinfo, JERR_NO_HUFF_TABLE, tblno);
-
-  /* Allocate a workspace if we haven't already done so. */
-  if (*pdtbl == NULL)
-    *pdtbl = (d_derived_tbl *)
-      (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				  SIZEOF(d_derived_tbl));
-  dtbl = *pdtbl;
-  dtbl->pub = htbl;		/* fill in back link */
-  
-  /* Figure C.1: make table of Huffman code length for each symbol */
-
-  p = 0;
-  for (l = 1; l <= 16; l++) {
-    i = (int) htbl->bits[l];
-    if (i < 0 || p + i > 256)	/* protect against table overrun */
-      ERREXIT(cinfo, JERR_BAD_HUFF_TABLE);
-    while (i--)
-      huffsize[p++] = (char) l;
-  }
-  huffsize[p] = 0;
-  numsymbols = p;
-  
-  /* Figure C.2: generate the codes themselves */
-  /* We also validate that the counts represent a legal Huffman code tree. */
-  
-  code = 0;
-  si = huffsize[0];
-  p = 0;
-  while (huffsize[p]) {
-    while (((int) huffsize[p]) == si) {
-      huffcode[p++] = code;
-      code++;
-    }
-    /* code is now 1 more than the last code used for codelength si; but
-     * it must still fit in si bits, since no code is allowed to be all ones.
-     */
-    if (((INT32) code) >= (((INT32) 1) << si))
-      ERREXIT(cinfo, JERR_BAD_HUFF_TABLE);
-    code <<= 1;
-    si++;
-  }
-
-  /* Figure F.15: generate decoding tables for bit-sequential decoding */
-
-  p = 0;
-  for (l = 1; l <= 16; l++) {
-    if (htbl->bits[l]) {
-      /* valoffset[l] = huffval[] index of 1st symbol of code length l,
-       * minus the minimum code of length l
-       */
-      dtbl->valoffset[l] = (INT32) p - (INT32) huffcode[p];
-      p += htbl->bits[l];
-      dtbl->maxcode[l] = huffcode[p-1]; /* maximum code of length l */
-    } else {
-      dtbl->maxcode[l] = -1;	/* -1 if no codes of this length */
-    }
-  }
-  dtbl->maxcode[17] = 0xFFFFFL; /* ensures jpeg_huff_decode terminates */
-
-  /* Compute lookahead tables to speed up decoding.
-   * First we set all the table entries to 0, indicating "too long";
-   * then we iterate through the Huffman codes that are short enough and
-   * fill in all the entries that correspond to bit sequences starting
-   * with that code.
-   */
-
-  MEMZERO(dtbl->look_nbits, SIZEOF(dtbl->look_nbits));
-
-  p = 0;
-  for (l = 1; l <= HUFF_LOOKAHEAD; l++) {
-    for (i = 1; i <= (int) htbl->bits[l]; i++, p++) {
-      /* l = current code's length, p = its index in huffcode[] & huffval[]. */
-      /* Generate left-justified code followed by all possible bit sequences */
-      lookbits = huffcode[p] << (HUFF_LOOKAHEAD-l);
-      for (ctr = 1 << (HUFF_LOOKAHEAD-l); ctr > 0; ctr--) {
-	dtbl->look_nbits[lookbits] = l;
-	dtbl->look_sym[lookbits] = htbl->huffval[p];
-	lookbits++;
-      }
-    }
-  }
-
-  /* Validate symbols as being reasonable.
-   * For AC tables, we make no check, but accept all byte values 0..255.
-   * For DC tables, we require the symbols to be in range 0..15.
-   * (Tighter bounds could be applied depending on the data depth and mode,
-   * but this is sufficient to ensure safe decoding.)
-   */
-  if (isDC) {
-    for (i = 0; i < numsymbols; i++) {
-      int sym = htbl->huffval[i];
-      if (sym < 0 || sym > 15)
-	ERREXIT(cinfo, JERR_BAD_HUFF_TABLE);
-    }
-  }
-}
-
-
-/*
- * Out-of-line code for bit fetching (shared with jdphuff.c).
- * See jdhuff.h for info about usage.
- * Note: current values of get_buffer and bits_left are passed as parameters,
- * but are returned in the corresponding fields of the state struct.
- *
- * On most machines MIN_GET_BITS should be 25 to allow the full 32-bit width
- * of get_buffer to be used.  (On machines with wider words, an even larger
- * buffer could be used.)  However, on some machines 32-bit shifts are
- * quite slow and take time proportional to the number of places shifted.
- * (This is true with most PC compilers, for instance.)  In this case it may
- * be a win to set MIN_GET_BITS to the minimum value of 15.  This reduces the
- * average shift distance at the cost of more calls to jpeg_fill_bit_buffer.
- */
-
-#ifdef SLOW_SHIFT_32
-#define MIN_GET_BITS  15	/* minimum allowable value */
-#else
-#define MIN_GET_BITS  (BIT_BUF_SIZE-7)
-#endif
-
-
-GLOBAL(boolean)
-jpeg_fill_bit_buffer (bitread_working_state * state,
-		      register bit_buf_type get_buffer, register int bits_left,
-		      int nbits)
-/* Load up the bit buffer to a depth of at least nbits */
-{
-  /* Copy heavily used state fields into locals (hopefully registers) */
-  register const JOCTET * next_input_byte = state->next_input_byte;
-  register size_t bytes_in_buffer = state->bytes_in_buffer;
-  j_decompress_ptr cinfo = state->cinfo;
-
-  /* Attempt to load at least MIN_GET_BITS bits into get_buffer. */
-  /* (It is assumed that no request will be for more than that many bits.) */
-  /* We fail to do so only if we hit a marker or are forced to suspend. */
-
-  if (cinfo->unread_marker == 0) {	/* cannot advance past a marker */
-    while (bits_left < MIN_GET_BITS) {
-      register int c;
-
-      /* Attempt to read a byte */
-      if (bytes_in_buffer == 0) {
-	if (! (*cinfo->src->fill_input_buffer) (cinfo))
-	  return FALSE;
-	next_input_byte = cinfo->src->next_input_byte;
-	bytes_in_buffer = cinfo->src->bytes_in_buffer;
-      }
-      bytes_in_buffer--;
-      c = GETJOCTET(*next_input_byte++);
-
-      /* If it's 0xFF, check and discard stuffed zero byte */
-      if (c == 0xFF) {
-	/* Loop here to discard any padding FF's on terminating marker,
-	 * so that we can save a valid unread_marker value.  NOTE: we will
-	 * accept multiple FF's followed by a 0 as meaning a single FF data
-	 * byte.  This data pattern is not valid according to the standard.
-	 */
-	do {
-	  if (bytes_in_buffer == 0) {
-	    if (! (*cinfo->src->fill_input_buffer) (cinfo))
-	      return FALSE;
-	    next_input_byte = cinfo->src->next_input_byte;
-	    bytes_in_buffer = cinfo->src->bytes_in_buffer;
-	  }
-	  bytes_in_buffer--;
-	  c = GETJOCTET(*next_input_byte++);
-	} while (c == 0xFF);
-
-	if (c == 0) {
-	  /* Found FF/00, which represents an FF data byte */
-	  c = 0xFF;
-	} else {
-	  /* Oops, it's actually a marker indicating end of compressed data.
-	   * Save the marker code for later use.
-	   * Fine point: it might appear that we should save the marker into
-	   * bitread working state, not straight into permanent state.  But
-	   * once we have hit a marker, we cannot need to suspend within the
-	   * current MCU, because we will read no more bytes from the data
-	   * source.  So it is OK to update permanent state right away.
-	   */
-	  cinfo->unread_marker = c;
-	  /* See if we need to insert some fake zero bits. */
-	  goto no_more_bytes;
-	}
-      }
-
-      /* OK, load c into get_buffer */
-      get_buffer = (get_buffer << 8) | c;
-      bits_left += 8;
-    } /* end while */
-  } else {
-  no_more_bytes:
-    /* We get here if we've read the marker that terminates the compressed
-     * data segment.  There should be enough bits in the buffer register
-     * to satisfy the request; if so, no problem.
-     */
-    if (nbits > bits_left) {
-      /* Uh-oh.  Report corrupted data to user and stuff zeroes into
-       * the data stream, so that we can produce some kind of image.
-       * We use a nonvolatile flag to ensure that only one warning message
-       * appears per data segment.
-       */
-      if (! cinfo->entropy->insufficient_data) {
-	WARNMS(cinfo, JWRN_HIT_MARKER);
-	cinfo->entropy->insufficient_data = TRUE;
-      }
-      /* Fill the buffer with zero bits */
-      get_buffer <<= MIN_GET_BITS - bits_left;
-      bits_left = MIN_GET_BITS;
-    }
-  }
-
-  /* Unload the local registers */
-  state->next_input_byte = next_input_byte;
-  state->bytes_in_buffer = bytes_in_buffer;
-  state->get_buffer = get_buffer;
-  state->bits_left = bits_left;
-
-  return TRUE;
-}
-
-
-/*
- * Out-of-line code for Huffman code decoding.
- * See jdhuff.h for info about usage.
- */
-
-GLOBAL(int)
-jpeg_huff_decode (bitread_working_state * state,
-		  register bit_buf_type get_buffer, register int bits_left,
-		  d_derived_tbl * htbl, int min_bits)
-{
-  register int l = min_bits;
-  register INT32 code;
-
-  /* HUFF_DECODE has determined that the code is at least min_bits */
-  /* bits long, so fetch that many bits in one swoop. */
-
-  CHECK_BIT_BUFFER(*state, l, return -1);
-  code = GET_BITS(l);
-
-  /* Collect the rest of the Huffman code one bit at a time. */
-  /* This is per Figure F.16 in the JPEG spec. */
-
-  while (code > htbl->maxcode[l]) {
-    code <<= 1;
-    CHECK_BIT_BUFFER(*state, 1, return -1);
-    code |= GET_BITS(1);
-    l++;
-  }
-
-  /* Unload the local registers */
-  state->get_buffer = get_buffer;
-  state->bits_left = bits_left;
-
-  /* With garbage input we may reach the sentinel value l = 17. */
-
-  if (l > 16) {
-    WARNMS(state->cinfo, JWRN_HUFF_BAD_CODE);
-    return 0;			/* fake a zero as the safest result */
-  }
-
-  return htbl->pub->huffval[ (int) (code + htbl->valoffset[l]) ];
-}
-
-
-/*
- * Figure F.12: extend sign bit.
- * On some machines, a shift and add will be faster than a table lookup.
- */
-
-#ifdef AVOID_TABLES
-
-#define HUFF_EXTEND(x,s)  ((x) < (1<<((s)-1)) ? (x) + (((-1)<<(s)) + 1) : (x))
-
-#else
-
-#define HUFF_EXTEND(x,s)  ((x) < extend_test[s] ? (x) + extend_offset[s] : (x))
-
-static const int extend_test[16] =   /* entry n is 2**(n-1) */
-  { 0, 0x0001, 0x0002, 0x0004, 0x0008, 0x0010, 0x0020, 0x0040, 0x0080,
-    0x0100, 0x0200, 0x0400, 0x0800, 0x1000, 0x2000, 0x4000 };
-
-static const int extend_offset[16] = /* entry n is (-1 << n) + 1 */
-  { 0, ((-1)<<1) + 1, ((-1)<<2) + 1, ((-1)<<3) + 1, ((-1)<<4) + 1,
-    ((-1)<<5) + 1, ((-1)<<6) + 1, ((-1)<<7) + 1, ((-1)<<8) + 1,
-    ((-1)<<9) + 1, ((-1)<<10) + 1, ((-1)<<11) + 1, ((-1)<<12) + 1,
-    ((-1)<<13) + 1, ((-1)<<14) + 1, ((-1)<<15) + 1 };
-
-#endif /* AVOID_TABLES */
-
-
-/*
- * Check for a restart marker & resynchronize decoder.
- * Returns FALSE if must suspend.
- */
-
-LOCAL(boolean)
-process_restart (j_decompress_ptr cinfo)
-{
-  huff_entropy_ptr entropy = (huff_entropy_ptr) cinfo->entropy;
-  int ci;
-
-  /* Throw away any unused bits remaining in bit buffer; */
-  /* include any full bytes in next_marker's count of discarded bytes */
-  cinfo->marker->discarded_bytes += entropy->bitstate.bits_left / 8;
-  entropy->bitstate.bits_left = 0;
-
-  /* Advance past the RSTn marker */
-  if (! (*cinfo->marker->read_restart_marker) (cinfo))
-    return FALSE;
-
-  /* Re-initialize DC predictions to 0 */
-  for (ci = 0; ci < cinfo->comps_in_scan; ci++)
-    entropy->saved.last_dc_val[ci] = 0;
-
-  /* Reset restart counter */
-  entropy->restarts_to_go = cinfo->restart_interval;
-
-  /* Reset out-of-data flag, unless read_restart_marker left us smack up
-   * against a marker.  In that case we will end up treating the next data
-   * segment as empty, and we can avoid producing bogus output pixels by
-   * leaving the flag set.
-   */
-  if (cinfo->unread_marker == 0)
-    entropy->pub.insufficient_data = FALSE;
-
-  return TRUE;
-}
-
-/*
- * Save the current Huffman deocde position and the DC coefficients
- * for each component into bitstream_offset and dc_info[], respectively.
- */
-METHODDEF(void)
-get_huffman_decoder_configuration(j_decompress_ptr cinfo,
-        huffman_offset_data *offset)
-{
-  huff_entropy_ptr entropy = (huff_entropy_ptr) cinfo->entropy;
-  short int *dc_info = offset->prev_dc;
-  int i;
-  jpeg_get_huffman_decoder_configuration(cinfo, offset);
-  for (i = 0; i < cinfo->comps_in_scan; i++) {
-    dc_info[i] = entropy->saved.last_dc_val[i];
-  }
-}
-
-/*
- * Save the current Huffman decoder position and the bit buffer
- * into bitstream_offset and get_buffer, respectively.
- */
-GLOBAL(void)
-jpeg_get_huffman_decoder_configuration(j_decompress_ptr cinfo,
-        huffman_offset_data *offset)
-{
-  huff_entropy_ptr entropy = (huff_entropy_ptr) cinfo->entropy;
-
-  if (cinfo->restart_interval) {
-    // We are at the end of a data segment
-    if (entropy->restarts_to_go == 0)
-      if (! process_restart(cinfo))
-	return;
-  }
-
-  // Save restarts_to_go and next_restart_num
-  offset->restarts_to_go = (unsigned short) entropy->restarts_to_go;
-  offset->next_restart_num = cinfo->marker->next_restart_num;
-
-  offset->bitstream_offset =
-      (jget_input_stream_position(cinfo) << LOG_TWO_BIT_BUF_SIZE)
-      + entropy->bitstate.bits_left;
-
-  offset->get_buffer = entropy->bitstate.get_buffer;
-}
-
-/*
- * Configure the Huffman decoder to decode the image
- * starting from the bitstream position recorded in offset.
- */
-METHODDEF(void)
-configure_huffman_decoder(j_decompress_ptr cinfo, huffman_offset_data offset)
-{
-  huff_entropy_ptr entropy = (huff_entropy_ptr) cinfo->entropy;
-  short int *dc_info = offset.prev_dc;
-  int i;
-  jpeg_configure_huffman_decoder(cinfo, offset);
-  for (i = 0; i < cinfo->comps_in_scan; i++) {
-    entropy->saved.last_dc_val[i] = dc_info[i];
-  }
-}
-
-/*
- * Configure the Huffman decoder reader position and bit buffer.
- */
-GLOBAL(void)
-jpeg_configure_huffman_decoder(j_decompress_ptr cinfo,
-        huffman_offset_data offset)
-{
-  huff_entropy_ptr entropy = (huff_entropy_ptr) cinfo->entropy;
-
-  // Restore restarts_to_go and next_restart_num
-  cinfo->unread_marker = 0;
-  entropy->restarts_to_go = offset.restarts_to_go;
-  cinfo->marker->next_restart_num = offset.next_restart_num;
-
-  unsigned int bitstream_offset = offset.bitstream_offset;
-  int blkn, i;
-
-  unsigned int byte_offset = bitstream_offset >> LOG_TWO_BIT_BUF_SIZE;
-  unsigned int bit_in_bit_buffer =
-      bitstream_offset & ((1 << LOG_TWO_BIT_BUF_SIZE) - 1);
-
-  jset_input_stream_position_bit(cinfo, byte_offset,
-          bit_in_bit_buffer, offset.get_buffer);
-}
-
-/*
- * Decode and return one MCU's worth of Huffman-compressed coefficients.
- * The coefficients are reordered from zigzag order into natural array order,
- * but are not dequantized.
- *
- * The i'th block of the MCU is stored into the block pointed to by
- * MCU_data[i].  WE ASSUME THIS AREA HAS BEEN ZEROED BY THE CALLER.
- * (Wholesale zeroing is usually a little faster than retail...)
- *
- * Returns FALSE if data source requested suspension.  In that case no
- * changes have been made to permanent state.  (Exception: some output
- * coefficients may already have been assigned.  This is harmless for
- * this module, since we'll just re-assign them on the next call.)
- */
-
-METHODDEF(boolean)
-decode_mcu (j_decompress_ptr cinfo, JBLOCKROW *MCU_data)
-{
-  huff_entropy_ptr entropy = (huff_entropy_ptr) cinfo->entropy;
-  int blkn;
-  BITREAD_STATE_VARS;
-  savable_state state;
-
-  /* Process restart marker if needed; may have to suspend */
-  if (cinfo->restart_interval) {
-    if (entropy->restarts_to_go == 0)
-      if (! process_restart(cinfo))
-	return FALSE;
-  }
-
-  /* If we've run out of data, just leave the MCU set to zeroes.
-   * This way, we return uniform gray for the remainder of the segment.
-   */
-  if (! entropy->pub.insufficient_data) {
-    /* Load up working state */
-    BITREAD_LOAD_STATE(cinfo,entropy->bitstate);
-    ASSIGN_STATE(state, entropy->saved);
-
-    /* Outer loop handles each block in the MCU */
-
-    for (blkn = 0; blkn < cinfo->blocks_in_MCU; blkn++) {
-      JBLOCKROW block = MCU_data[blkn];
-      d_derived_tbl * dctbl = entropy->dc_cur_tbls[blkn];
-      d_derived_tbl * actbl = entropy->ac_cur_tbls[blkn];
-      register int s, k, r;
-
-      /* Decode a single block's worth of coefficients */
-
-      /* Section F.2.2.1: decode the DC coefficient difference */
-      HUFF_DECODE(s, br_state, dctbl, return FALSE, label1);
-      if (s) {
-	CHECK_BIT_BUFFER(br_state, s, return FALSE);
-	r = GET_BITS(s);
-	s = HUFF_EXTEND(r, s);
-      }
-
-      if (entropy->dc_needed[blkn]) {
-	/* Convert DC difference to actual value, update last_dc_val */
-	int ci = cinfo->MCU_membership[blkn];
-	s += state.last_dc_val[ci];
-	state.last_dc_val[ci] = s;
-	/* Output the DC coefficient (assumes jpeg_natural_order[0] = 0) */
-	(*block)[0] = (JCOEF) s;
-      }
-
-      if (entropy->ac_needed[blkn]) {
-
-	/* Section F.2.2.2: decode the AC coefficients */
-	/* Since zeroes are skipped, output area must be cleared beforehand */
-	for (k = 1; k < DCTSIZE2; k++) {
-	  HUFF_DECODE(s, br_state, actbl, return FALSE, label2);
-      
-	  r = s >> 4;
-	  s &= 15;
-      
-	  if (s) {
-	    k += r;
-	    CHECK_BIT_BUFFER(br_state, s, return FALSE);
-	    r = GET_BITS(s);
-	    s = HUFF_EXTEND(r, s);
-	    /* Output coefficient in natural (dezigzagged) order.
-	     * Note: the extra entries in jpeg_natural_order[] will save us
-	     * if k >= DCTSIZE2, which could happen if the data is corrupted.
-	     */
-	    (*block)[jpeg_natural_order[k]] = (JCOEF) s;
-	  } else {
-	    if (r != 15)
-	      break;
-	    k += 15;
-	  }
-	}
-
-      } else {
-
-	/* Section F.2.2.2: decode the AC coefficients */
-	/* In this path we just discard the values */
-	for (k = 1; k < DCTSIZE2; k++) {
-	  HUFF_DECODE(s, br_state, actbl, return FALSE, label3);
-      
-	  r = s >> 4;
-	  s &= 15;
-      
-	  if (s) {
-	    k += r;
-	    CHECK_BIT_BUFFER(br_state, s, return FALSE);
-	    DROP_BITS(s);
-	  } else {
-	    if (r != 15)
-	      break;
-	    k += 15;
-	  }
-	}
-
-      }
-    }
-
-    /* Completed MCU, so update state */
-    BITREAD_SAVE_STATE(cinfo,entropy->bitstate);
-    ASSIGN_STATE(entropy->saved, state);
-  }
-
-  /* Account for restart interval (no-op if not using restarts) */
-  entropy->restarts_to_go--;
-
-  return TRUE;
-}
-
-/*
- * Decode one MCU's worth of Huffman-compressed coefficients.
- * The propose of this method is to calculate the
- * data length of one MCU in Huffman-coded format.
- * Therefore, all coefficients are discarded.
- */
-
-METHODDEF(boolean)
-decode_mcu_discard_coef (j_decompress_ptr cinfo)
-{
-  huff_entropy_ptr entropy = (huff_entropy_ptr) cinfo->entropy;
-  int blkn;
-  BITREAD_STATE_VARS;
-  savable_state state;
-
-  /* Process restart marker if needed; may have to suspend */
-  if (cinfo->restart_interval) {
-    if (entropy->restarts_to_go == 0)
-      if (! process_restart(cinfo))
-	return FALSE;
-  }
-
-  if (! entropy->pub.insufficient_data) {
-
-    /* Load up working state */
-    BITREAD_LOAD_STATE(cinfo,entropy->bitstate);
-    ASSIGN_STATE(state, entropy->saved);
-
-    /* Outer loop handles each block in the MCU */
-
-    for (blkn = 0; blkn < cinfo->blocks_in_MCU; blkn++) {
-      d_derived_tbl * dctbl = entropy->dc_cur_tbls[blkn];
-      d_derived_tbl * actbl = entropy->ac_cur_tbls[blkn];
-      register int s, k, r;
-
-      /* Decode a single block's worth of coefficients */
-
-      /* Section F.2.2.1: decode the DC coefficient difference */
-      HUFF_DECODE(s, br_state, dctbl, return FALSE, label1);
-      if (s) {
-	CHECK_BIT_BUFFER(br_state, s, return FALSE);
-	r = GET_BITS(s);
-	s = HUFF_EXTEND(r, s);
-      }
-
-      /* discard all coefficients */
-      if (entropy->dc_needed[blkn]) {
-	/* Convert DC difference to actual value, update last_dc_val */
-	int ci = cinfo->MCU_membership[blkn];
-	s += state.last_dc_val[ci];
-	state.last_dc_val[ci] = s;
-      }
-      for (k = 1; k < DCTSIZE2; k++) {
-        HUFF_DECODE(s, br_state, actbl, return FALSE, label3);
-
-        r = s >> 4;
-        s &= 15;
-
-        if (s) {
-          k += r;
-          CHECK_BIT_BUFFER(br_state, s, return FALSE);
-          DROP_BITS(s);
-        } else {
-          if (r != 15)
-            break;
-          k += 15;
-        }
-      }
-    }
-
-    /* Completed MCU, so update state */
-    BITREAD_SAVE_STATE(cinfo,entropy->bitstate);
-    ASSIGN_STATE(entropy->saved, state);
-  }
-
-  /* Account for restart interval (no-op if not using restarts) */
-  entropy->restarts_to_go--;
-
-  return TRUE;
-}
-
-
-/*
- * Module initialization routine for Huffman entropy decoding.
- */
-
-GLOBAL(void)
-jinit_huff_decoder (j_decompress_ptr cinfo)
-{
-  huff_entropy_ptr entropy;
-  int i;
-
-  entropy = (huff_entropy_ptr)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				SIZEOF(huff_entropy_decoder));
-  cinfo->entropy = (struct jpeg_entropy_decoder *) entropy;
-  entropy->pub.start_pass = start_pass_huff_decoder;
-  entropy->pub.decode_mcu = decode_mcu;
-  entropy->pub.decode_mcu_discard_coef = decode_mcu_discard_coef;
-  entropy->pub.configure_huffman_decoder = configure_huffman_decoder;
-  entropy->pub.get_huffman_decoder_configuration =
-        get_huffman_decoder_configuration;
-  entropy->pub.index = NULL;
-
-  /* Mark tables unallocated */
-  for (i = 0; i < NUM_HUFF_TBLS; i++) {
-    entropy->dc_derived_tbls[i] = entropy->ac_derived_tbls[i] = NULL;
-  }
-}
-
-/*
- * Call after jpeg_read_header
- */
-GLOBAL(void)
-jpeg_create_huffman_index(j_decompress_ptr cinfo, huffman_index *index)
-{
-  int i, s;
-  index->scan_count = 1;
-  index->total_iMCU_rows = cinfo->total_iMCU_rows;
-  index->scan = (huffman_scan_header*)malloc(index->scan_count
-          * sizeof(huffman_scan_header));
-  index->scan[0].offset = (huffman_offset_data**)malloc(cinfo->total_iMCU_rows
-          * sizeof(huffman_offset_data*));
-  index->scan[0].prev_MCU_offset.bitstream_offset = 0;
-  index->MCU_sample_size = DEFAULT_MCU_SAMPLE_SIZE;
-
-  index->mem_used = sizeof(huffman_scan_header)
-      + cinfo->total_iMCU_rows * sizeof(huffman_offset_data*);
-}
-
-GLOBAL(void)
-jpeg_destroy_huffman_index(huffman_index *index)
-{
-    int i, j;
-    for (i = 0; i < index->scan_count; i++) {
-        for(j = 0; j < index->total_iMCU_rows; j++) {
-            free(index->scan[i].offset[j]);
-        }
-        free(index->scan[i].offset);
-    }
-    free(index->scan);
-}
-
-/*
- * Set the reader byte position to offset
- */
-GLOBAL(void)
-jset_input_stream_position(j_decompress_ptr cinfo, int offset)
-{
-  if (cinfo->src->seek_input_data) {
-    cinfo->src->seek_input_data(cinfo, offset);
-  } else {
-    cinfo->src->bytes_in_buffer = cinfo->src->current_offset - offset;
-    cinfo->src->next_input_byte = cinfo->src->start_input_byte + offset;
-  }
-}
-
-/*
- * Set the reader byte position to offset and bit position to bit_left
- * with bit buffer set to buf.
- */
-GLOBAL(void)
-jset_input_stream_position_bit(j_decompress_ptr cinfo,
-        int byte_offset, int bit_left, INT32 buf)
-{
-  huff_entropy_ptr entropy = (huff_entropy_ptr) cinfo->entropy;
-
-  entropy->bitstate.bits_left = bit_left;
-  entropy->bitstate.get_buffer = buf;
-
-  jset_input_stream_position(cinfo, byte_offset);
-}
-
-/*
- * Get the current reader byte position.
- */
-GLOBAL(int)
-jget_input_stream_position(j_decompress_ptr cinfo)
-{
-  return cinfo->src->current_offset - cinfo->src->bytes_in_buffer;
-}
@@ -1,202 +0,0 @@
-/*
- * jdhuff.h
- *
- * Copyright (C) 1991-1997, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains declarations for Huffman entropy decoding routines
- * that are shared between the sequential decoder (jdhuff.c) and the
- * progressive decoder (jdphuff.c).  No other modules need to see these.
- */
-
-/* Short forms of external names for systems with brain-damaged linkers. */
-
-#ifdef NEED_SHORT_EXTERNAL_NAMES
-#define jpeg_make_d_derived_tbl	jMkDDerived
-#define jpeg_fill_bit_buffer	jFilBitBuf
-#define jpeg_huff_decode	jHufDecode
-#endif /* NEED_SHORT_EXTERNAL_NAMES */
-
-
-/* Derived data constructed for each Huffman table */
-
-#define HUFF_LOOKAHEAD	8	/* # of bits of lookahead */
-
-typedef struct {
-  /* Basic tables: (element [0] of each array is unused) */
-  INT32 maxcode[18];		/* largest code of length k (-1 if none) */
-  /* (maxcode[17] is a sentinel to ensure jpeg_huff_decode terminates) */
-  INT32 valoffset[17];		/* huffval[] offset for codes of length k */
-  /* valoffset[k] = huffval[] index of 1st symbol of code length k, less
-   * the smallest code of length k; so given a code of length k, the
-   * corresponding symbol is huffval[code + valoffset[k]]
-   */
-
-  /* Link to public Huffman table (needed only in jpeg_huff_decode) */
-  JHUFF_TBL *pub;
-
-  /* Lookahead tables: indexed by the next HUFF_LOOKAHEAD bits of
-   * the input data stream.  If the next Huffman code is no more
-   * than HUFF_LOOKAHEAD bits long, we can obtain its length and
-   * the corresponding symbol directly from these tables.
-   */
-  int look_nbits[1<<HUFF_LOOKAHEAD]; /* # bits, or 0 if too long */
-  UINT8 look_sym[1<<HUFF_LOOKAHEAD]; /* symbol, or unused */
-} d_derived_tbl;
-
-/* Expand a Huffman table definition into the derived format */
-EXTERN(void) jpeg_make_d_derived_tbl
-	JPP((j_decompress_ptr cinfo, boolean isDC, int tblno,
-	     d_derived_tbl ** pdtbl));
-
-
-/*
- * Fetching the next N bits from the input stream is a time-critical operation
- * for the Huffman decoders.  We implement it with a combination of inline
- * macros and out-of-line subroutines.  Note that N (the number of bits
- * demanded at one time) never exceeds 15 for JPEG use.
- *
- * We read source bytes into get_buffer and dole out bits as needed.
- * If get_buffer already contains enough bits, they are fetched in-line
- * by the macros CHECK_BIT_BUFFER and GET_BITS.  When there aren't enough
- * bits, jpeg_fill_bit_buffer is called; it will attempt to fill get_buffer
- * as full as possible (not just to the number of bits needed; this
- * prefetching reduces the overhead cost of calling jpeg_fill_bit_buffer).
- * Note that jpeg_fill_bit_buffer may return FALSE to indicate suspension.
- * On TRUE return, jpeg_fill_bit_buffer guarantees that get_buffer contains
- * at least the requested number of bits --- dummy zeroes are inserted if
- * necessary.
- */
-
-typedef INT32 bit_buf_type;	/* type of bit-extraction buffer */
-#define BIT_BUF_SIZE  32	/* size of buffer in bits */
-#define LOG_TWO_BIT_BUF_SIZE  5	/* log_2(BIT_BUF_SIZE) */
-
-/* If long is > 32 bits on your machine, and shifting/masking longs is
- * reasonably fast, making bit_buf_type be long and setting BIT_BUF_SIZE
- * appropriately should be a win.  Unfortunately we can't define the size
- * with something like  #define BIT_BUF_SIZE (sizeof(bit_buf_type)*8)
- * because not all machines measure sizeof in 8-bit bytes.
- */
-
-typedef struct {		/* Bitreading state saved across MCUs */
-  bit_buf_type get_buffer;	/* current bit-extraction buffer */
-  int bits_left;		/* # of unused bits in it */
-} bitread_perm_state;
-
-typedef struct {		/* Bitreading working state within an MCU */
-  /* Current data source location */
-  /* We need a copy, rather than munging the original, in case of suspension */
-  const JOCTET * next_input_byte; /* => next byte to read from source */
-  size_t bytes_in_buffer;	/* # of bytes remaining in source buffer */
-  /* Bit input buffer --- note these values are kept in register variables,
-   * not in this struct, inside the inner loops.
-   */
-  bit_buf_type get_buffer;	/* current bit-extraction buffer */
-  int bits_left;		/* # of unused bits in it */
-  /* Pointer needed by jpeg_fill_bit_buffer. */
-  j_decompress_ptr cinfo;	/* back link to decompress master record */
-} bitread_working_state;
-
-/* Macros to declare and load/save bitread local variables. */
-#define BITREAD_STATE_VARS  \
-	register bit_buf_type get_buffer;  \
-	register int bits_left;  \
-	bitread_working_state br_state
-
-#define BITREAD_LOAD_STATE(cinfop,permstate)  \
-	br_state.cinfo = cinfop; \
-	br_state.next_input_byte = cinfop->src->next_input_byte; \
-	br_state.bytes_in_buffer = cinfop->src->bytes_in_buffer; \
-	get_buffer = permstate.get_buffer; \
-	bits_left = permstate.bits_left;
-
-#define BITREAD_SAVE_STATE(cinfop,permstate)  \
-	cinfop->src->next_input_byte = br_state.next_input_byte; \
-	cinfop->src->bytes_in_buffer = br_state.bytes_in_buffer; \
-	permstate.get_buffer = get_buffer; \
-	permstate.bits_left = bits_left
-
-/*
- * These macros provide the in-line portion of bit fetching.
- * Use CHECK_BIT_BUFFER to ensure there are N bits in get_buffer
- * before using GET_BITS, PEEK_BITS, or DROP_BITS.
- * The variables get_buffer and bits_left are assumed to be locals,
- * but the state struct might not be (jpeg_huff_decode needs this).
- *	CHECK_BIT_BUFFER(state,n,action);
- *		Ensure there are N bits in get_buffer; if suspend, take action.
- *      val = GET_BITS(n);
- *		Fetch next N bits.
- *      val = PEEK_BITS(n);
- *		Fetch next N bits without removing them from the buffer.
- *	DROP_BITS(n);
- *		Discard next N bits.
- * The value N should be a simple variable, not an expression, because it
- * is evaluated multiple times.
- */
-
-#define CHECK_BIT_BUFFER(state,nbits,action) \
-	{ if (bits_left < (nbits)) {  \
-	    if (! jpeg_fill_bit_buffer(&(state),get_buffer,bits_left,nbits))  \
-	      { action; }  \
-	    get_buffer = (state).get_buffer; bits_left = (state).bits_left; } }
-
-#define GET_BITS(nbits) \
-	(((int) (get_buffer >> (bits_left -= (nbits)))) & ((1<<(nbits))-1))
-
-#define PEEK_BITS(nbits) \
-	(((int) (get_buffer >> (bits_left -  (nbits)))) & ((1<<(nbits))-1))
-
-#define DROP_BITS(nbits) \
-	(bits_left -= (nbits))
-
-/* Load up the bit buffer to a depth of at least nbits */
-EXTERN(boolean) jpeg_fill_bit_buffer
-	JPP((bitread_working_state * state, register bit_buf_type get_buffer,
-	     register int bits_left, int nbits));
-
-
-/*
- * Code for extracting next Huffman-coded symbol from input bit stream.
- * Again, this is time-critical and we make the main paths be macros.
- *
- * We use a lookahead table to process codes of up to HUFF_LOOKAHEAD bits
- * without looping.  Usually, more than 95% of the Huffman codes will be 8
- * or fewer bits long.  The few overlength codes are handled with a loop,
- * which need not be inline code.
- *
- * Notes about the HUFF_DECODE macro:
- * 1. Near the end of the data segment, we may fail to get enough bits
- *    for a lookahead.  In that case, we do it the hard way.
- * 2. If the lookahead table contains no entry, the next code must be
- *    more than HUFF_LOOKAHEAD bits long.
- * 3. jpeg_huff_decode returns -1 if forced to suspend.
- */
-
-#define HUFF_DECODE(result,state,htbl,failaction,slowlabel) \
-{ register int nb, look; \
-  if (bits_left < HUFF_LOOKAHEAD) { \
-    if (! jpeg_fill_bit_buffer(&state,get_buffer,bits_left, 0)) {failaction;} \
-    get_buffer = state.get_buffer; bits_left = state.bits_left; \
-    if (bits_left < HUFF_LOOKAHEAD) { \
-      nb = 1; goto slowlabel; \
-    } \
-  } \
-  look = PEEK_BITS(HUFF_LOOKAHEAD); \
-  if ((nb = htbl->look_nbits[look]) != 0) { \
-    DROP_BITS(nb); \
-    result = htbl->look_sym[look]; \
-  } else { \
-    nb = HUFF_LOOKAHEAD+1; \
-slowlabel: \
-    if ((result=jpeg_huff_decode(&state,get_buffer,bits_left,htbl,nb)) < 0) \
-	{ failaction; } \
-    get_buffer = state.get_buffer; bits_left = state.bits_left; \
-  } \
-}
-
-/* Out-of-line case for Huffman code fetching */
-EXTERN(int) jpeg_huff_decode
-	JPP((bitread_working_state * state, register bit_buf_type get_buffer,
-	     register int bits_left, d_derived_tbl * htbl, int min_bits));
@@ -1,415 +0,0 @@
-/*
- * jdinput.c
- *
- * Copyright (C) 1991-1997, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains input control logic for the JPEG decompressor.
- * These routines are concerned with controlling the decompressor's input
- * processing (marker reading and coefficient decoding).  The actual input
- * reading is done in jdmarker.c, jdhuff.c, and jdphuff.c.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-
-
-/* Private state */
-
-typedef struct {
-  struct jpeg_input_controller pub; /* public fields */
-
-  boolean inheaders;		/* TRUE until first SOS is reached */
-} my_input_controller;
-
-typedef my_input_controller * my_inputctl_ptr;
-
-
-/* Forward declarations */
-METHODDEF(int) consume_markers JPP((j_decompress_ptr cinfo));
-METHODDEF(int) consume_markers_with_huffman_index JPP((j_decompress_ptr cinfo,
-                    huffman_index *index, int current_scan));
-
-
-/*
- * Routines to calculate various quantities related to the size of the image.
- */
-
-LOCAL(void)
-initial_setup (j_decompress_ptr cinfo)
-/* Called once, when first SOS marker is reached */
-{
-  int ci;
-  jpeg_component_info *compptr;
-
-  /* Make sure image isn't bigger than I can handle */
-  if ((long) cinfo->image_height > (long) JPEG_MAX_DIMENSION ||
-      (long) cinfo->image_width > (long) JPEG_MAX_DIMENSION)
-    ERREXIT1(cinfo, JERR_IMAGE_TOO_BIG, (unsigned int) JPEG_MAX_DIMENSION);
-
-  /* For now, precision must match compiled-in value... */
-  if (cinfo->data_precision != BITS_IN_JSAMPLE)
-    ERREXIT1(cinfo, JERR_BAD_PRECISION, cinfo->data_precision);
-
-  /* Check that number of components won't exceed internal array sizes */
-  if (cinfo->num_components > MAX_COMPONENTS)
-    ERREXIT2(cinfo, JERR_COMPONENT_COUNT, cinfo->num_components,
-	     MAX_COMPONENTS);
-
-  /* Compute maximum sampling factors; check factor validity */
-  cinfo->max_h_samp_factor = 1;
-  cinfo->max_v_samp_factor = 1;
-  for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
-       ci++, compptr++) {
-    if (compptr->h_samp_factor<=0 || compptr->h_samp_factor>MAX_SAMP_FACTOR ||
-	compptr->v_samp_factor<=0 || compptr->v_samp_factor>MAX_SAMP_FACTOR)
-      ERREXIT(cinfo, JERR_BAD_SAMPLING);
-    cinfo->max_h_samp_factor = MAX(cinfo->max_h_samp_factor,
-				   compptr->h_samp_factor);
-    cinfo->max_v_samp_factor = MAX(cinfo->max_v_samp_factor,
-				   compptr->v_samp_factor);
-  }
-
-  /* We initialize DCT_scaled_size and min_DCT_scaled_size to DCTSIZE.
-   * In the full decompressor, this will be overridden by jdmaster.c;
-   * but in the transcoder, jdmaster.c is not used, so we must do it here.
-   */
-  cinfo->min_DCT_scaled_size = DCTSIZE;
-
-  /* Compute dimensions of components */
-  for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
-       ci++, compptr++) {
-    compptr->DCT_scaled_size = DCTSIZE;
-    /* Size in DCT blocks */
-    compptr->width_in_blocks = (JDIMENSION)
-      jdiv_round_up((long) cinfo->image_width * (long) compptr->h_samp_factor,
-		    (long) (cinfo->max_h_samp_factor * DCTSIZE));
-    compptr->height_in_blocks = (JDIMENSION)
-      jdiv_round_up((long) cinfo->image_height * (long) compptr->v_samp_factor,
-		    (long) (cinfo->max_v_samp_factor * DCTSIZE));
-    /* downsampled_width and downsampled_height will also be overridden by
-     * jdmaster.c if we are doing full decompression.  The transcoder library
-     * doesn't use these values, but the calling application might.
-     */
-    /* Size in samples */
-    compptr->downsampled_width = (JDIMENSION)
-      jdiv_round_up((long) cinfo->image_width * (long) compptr->h_samp_factor,
-		    (long) cinfo->max_h_samp_factor);
-    compptr->downsampled_height = (JDIMENSION)
-      jdiv_round_up((long) cinfo->image_height * (long) compptr->v_samp_factor,
-		    (long) cinfo->max_v_samp_factor);
-    /* Mark component needed, until color conversion says otherwise */
-    compptr->component_needed = TRUE;
-    /* Mark no quantization table yet saved for component */
-    compptr->quant_table = NULL;
-  }
-
-  /* Compute number of fully interleaved MCU rows. */
-  cinfo->total_iMCU_rows = (JDIMENSION)
-    jdiv_round_up((long) cinfo->image_height,
-		  (long) (cinfo->max_v_samp_factor*DCTSIZE));
-
-  /* Decide whether file contains multiple scans */
-  if (cinfo->comps_in_scan < cinfo->num_components || cinfo->progressive_mode)
-    cinfo->inputctl->has_multiple_scans = TRUE;
-  else
-    cinfo->inputctl->has_multiple_scans = FALSE;
-  cinfo->original_image_width = cinfo->image_width;
-}
-
-LOCAL(void)
-per_scan_setup (j_decompress_ptr cinfo)
-/* Do computations that are needed before processing a JPEG scan */
-/* cinfo->comps_in_scan and cinfo->cur_comp_info[] were set from SOS marker */
-{
-  int ci, mcublks, tmp;
-  jpeg_component_info *compptr;
-  
-  if (cinfo->comps_in_scan == 1) {
-    
-    /* Noninterleaved (single-component) scan */
-    compptr = cinfo->cur_comp_info[0];
-    
-    /* Overall image size in MCUs */
-    cinfo->MCUs_per_row = compptr->width_in_blocks;
-    cinfo->MCU_rows_in_scan = compptr->height_in_blocks;
-    
-    /* For noninterleaved scan, always one block per MCU */
-    compptr->MCU_width = 1;
-    compptr->MCU_height = 1;
-    compptr->MCU_blocks = 1;
-    compptr->MCU_sample_width = compptr->DCT_scaled_size;
-    compptr->last_col_width = 1;
-    /* For noninterleaved scans, it is convenient to define last_row_height
-     * as the number of block rows present in the last iMCU row.
-     */
-    tmp = (int) (compptr->height_in_blocks % compptr->v_samp_factor);
-    if (tmp == 0) tmp = compptr->v_samp_factor;
-    compptr->last_row_height = tmp;
-    
-    /* Prepare array describing MCU composition */
-    cinfo->blocks_in_MCU = 1;
-    cinfo->MCU_membership[0] = 0;
-    
-  } else {
-    
-    /* Interleaved (multi-component) scan */
-    if (cinfo->comps_in_scan <= 0 || cinfo->comps_in_scan > MAX_COMPS_IN_SCAN)
-      ERREXIT2(cinfo, JERR_COMPONENT_COUNT, cinfo->comps_in_scan,
-	       MAX_COMPS_IN_SCAN);
-    
-    /* Overall image size in MCUs */
-    cinfo->MCUs_per_row = (JDIMENSION)
-      jdiv_round_up((long) cinfo->image_width,
-		    (long) (cinfo->max_h_samp_factor*DCTSIZE));
-    cinfo->MCU_rows_in_scan = (JDIMENSION)
-      jdiv_round_up((long) cinfo->image_height,
-		    (long) (cinfo->max_v_samp_factor*DCTSIZE));
-    
-    cinfo->blocks_in_MCU = 0;
-    
-    for (ci = 0; ci < cinfo->comps_in_scan; ci++) {
-      compptr = cinfo->cur_comp_info[ci];
-      /* Sampling factors give # of blocks of component in each MCU */
-      compptr->MCU_width = compptr->h_samp_factor;
-      compptr->MCU_height = compptr->v_samp_factor;
-      compptr->MCU_blocks = compptr->MCU_width * compptr->MCU_height;
-      compptr->MCU_sample_width = compptr->MCU_width * compptr->DCT_scaled_size;
-      /* Figure number of non-dummy blocks in last MCU column & row */
-      tmp = (int) (compptr->width_in_blocks % compptr->MCU_width);
-      if (tmp == 0) tmp = compptr->MCU_width;
-      compptr->last_col_width = tmp;
-#ifdef ANDROID_TILE_BASED_DECODE
-      if (cinfo->tile_decode) {
-        tmp = (int) (jdiv_round_up(cinfo->image_width, 8)
-                % compptr->MCU_width);
-        if (tmp == 0) tmp = compptr->MCU_width;
-        compptr->last_col_width = tmp;
-      }
-#endif
-
-      tmp = (int) (compptr->height_in_blocks % compptr->MCU_height);
-      if (tmp == 0) tmp = compptr->MCU_height;
-      compptr->last_row_height = tmp;
-      /* Prepare array describing MCU composition */
-      mcublks = compptr->MCU_blocks;
-      if (cinfo->blocks_in_MCU + mcublks > D_MAX_BLOCKS_IN_MCU)
-	ERREXIT(cinfo, JERR_BAD_MCU_SIZE);
-      while (mcublks-- > 0) {
-	cinfo->MCU_membership[cinfo->blocks_in_MCU++] = ci;
-      }
-    }
-    
-  }
-}
-
-GLOBAL(void)
-jpeg_decompress_per_scan_setup(j_decompress_ptr cinfo)
-{
-    per_scan_setup(cinfo);
-}
-
-
-
-/*
- * Save away a copy of the Q-table referenced by each component present
- * in the current scan, unless already saved during a prior scan.
- *
- * In a multiple-scan JPEG file, the encoder could assign different components
- * the same Q-table slot number, but change table definitions between scans
- * so that each component uses a different Q-table.  (The IJG encoder is not
- * currently capable of doing this, but other encoders might.)  Since we want
- * to be able to dequantize all the components at the end of the file, this
- * means that we have to save away the table actually used for each component.
- * We do this by copying the table at the start of the first scan containing
- * the component.
- * The JPEG spec prohibits the encoder from changing the contents of a Q-table
- * slot between scans of a component using that slot.  If the encoder does so
- * anyway, this decoder will simply use the Q-table values that were current
- * at the start of the first scan for the component.
- *
- * The decompressor output side looks only at the saved quant tables,
- * not at the current Q-table slots.
- */
-
-LOCAL(void)
-latch_quant_tables (j_decompress_ptr cinfo)
-{
-  int ci, qtblno;
-  jpeg_component_info *compptr;
-  JQUANT_TBL * qtbl;
-
-  for (ci = 0; ci < cinfo->comps_in_scan; ci++) {
-    compptr = cinfo->cur_comp_info[ci];
-    /* No work if we already saved Q-table for this component */
-    if (compptr->quant_table != NULL)
-      continue;
-    /* Make sure specified quantization table is present */
-    qtblno = compptr->quant_tbl_no;
-    if (qtblno < 0 || qtblno >= NUM_QUANT_TBLS ||
-	cinfo->quant_tbl_ptrs[qtblno] == NULL)
-      ERREXIT1(cinfo, JERR_NO_QUANT_TABLE, qtblno);
-    /* OK, save away the quantization table */
-    qtbl = (JQUANT_TBL *)
-      (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				  SIZEOF(JQUANT_TBL));
-    MEMCOPY(qtbl, cinfo->quant_tbl_ptrs[qtblno], SIZEOF(JQUANT_TBL));
-    compptr->quant_table = qtbl;
-  }
-}
-
-
-/*
- * Initialize the input modules to read a scan of compressed data.
- * The first call to this is done by jdmaster.c after initializing
- * the entire decompressor (during jpeg_start_decompress).
- * Subsequent calls come from consume_markers, below.
- */
-
-METHODDEF(void)
-start_input_pass (j_decompress_ptr cinfo)
-{
-  per_scan_setup(cinfo);
-  latch_quant_tables(cinfo);
-  (*cinfo->entropy->start_pass) (cinfo);
-  (*cinfo->coef->start_input_pass) (cinfo);
-  cinfo->inputctl->consume_input = cinfo->coef->consume_data;
-  cinfo->inputctl->consume_input_build_huffman_index =
-        cinfo->coef->consume_data_build_huffman_index;
-}
-
-
-/*
- * Finish up after inputting a compressed-data scan.
- * This is called by the coefficient controller after it's read all
- * the expected data of the scan.
- */
-
-METHODDEF(void)
-finish_input_pass (j_decompress_ptr cinfo)
-{
-  cinfo->inputctl->consume_input = consume_markers;
-  cinfo->inputctl->consume_input_build_huffman_index =
-        consume_markers_with_huffman_index;
-}
-
-
-METHODDEF(int)
-consume_markers_with_huffman_index (j_decompress_ptr cinfo,
-        huffman_index *index, int current_scan)
-{
-    return consume_markers(cinfo);
-}
-/*
- * Read JPEG markers before, between, or after compressed-data scans.
- * Change state as necessary when a new scan is reached.
- * Return value is JPEG_SUSPENDED, JPEG_REACHED_SOS, or JPEG_REACHED_EOI.
- *
- * The consume_input method pointer points either here or to the
- * coefficient controller's consume_data routine, depending on whether
- * we are reading a compressed data segment or inter-segment markers.
- */
-
-METHODDEF(int)
-consume_markers (j_decompress_ptr cinfo)
-{
-  my_inputctl_ptr inputctl = (my_inputctl_ptr) cinfo->inputctl;
-  int val;
-
-  if (inputctl->pub.eoi_reached) /* After hitting EOI, read no further */
-    return JPEG_REACHED_EOI;
-
-  val = (*cinfo->marker->read_markers) (cinfo);
-
-  switch (val) {
-  case JPEG_REACHED_SOS:	/* Found SOS */
-    if (inputctl->inheaders) {	/* 1st SOS */
-      initial_setup(cinfo);
-      inputctl->inheaders = FALSE;
-      /* Note: start_input_pass must be called by jdmaster.c
-       * before any more input can be consumed.  jdapimin.c is
-       * responsible for enforcing this sequencing.
-       */
-    } else {			/* 2nd or later SOS marker */
-      if (! inputctl->pub.has_multiple_scans)
-	ERREXIT(cinfo, JERR_EOI_EXPECTED); /* Oops, I wasn't expecting this! */
-      start_input_pass(cinfo);
-    }
-    break;
-  case JPEG_REACHED_EOI:	/* Found EOI */
-    inputctl->pub.eoi_reached = TRUE;
-    if (inputctl->inheaders) {	/* Tables-only datastream, apparently */
-      if (cinfo->marker->saw_SOF)
-	ERREXIT(cinfo, JERR_SOF_NO_SOS);
-    } else {
-      /* Prevent infinite loop in coef ctlr's decompress_data routine
-       * if user set output_scan_number larger than number of scans.
-       */
-      if (cinfo->output_scan_number > cinfo->input_scan_number)
-	cinfo->output_scan_number = cinfo->input_scan_number;
-    }
-    break;
-  case JPEG_SUSPENDED:
-    break;
-  }
-
-  return val;
-}
-
-
-/*
- * Reset state to begin a fresh datastream.
- */
-
-METHODDEF(void)
-reset_input_controller (j_decompress_ptr cinfo)
-{
-  my_inputctl_ptr inputctl = (my_inputctl_ptr) cinfo->inputctl;
-
-  inputctl->pub.consume_input = consume_markers;
-  inputctl->pub.consume_input_build_huffman_index =
-        consume_markers_with_huffman_index;
-  inputctl->pub.has_multiple_scans = FALSE; /* "unknown" would be better */
-  inputctl->pub.eoi_reached = FALSE;
-  inputctl->inheaders = TRUE;
-  /* Reset other modules */
-  (*cinfo->err->reset_error_mgr) ((j_common_ptr) cinfo);
-  (*cinfo->marker->reset_marker_reader) (cinfo);
-  /* Reset progression state -- would be cleaner if entropy decoder did this */
-  cinfo->coef_bits = NULL;
-}
-
-
-/*
- * Initialize the input controller module.
- * This is called only once, when the decompression object is created.
- */
-
-GLOBAL(void)
-jinit_input_controller (j_decompress_ptr cinfo)
-{
-  my_inputctl_ptr inputctl;
-
-  /* Create subobject in permanent pool */
-  inputctl = (my_inputctl_ptr)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_PERMANENT,
-				SIZEOF(my_input_controller));
-  cinfo->inputctl = (struct jpeg_input_controller *) inputctl;
-  /* Initialize method pointers */
-  inputctl->pub.consume_input = consume_markers;
-  inputctl->pub.reset_input_controller = reset_input_controller;
-  inputctl->pub.start_input_pass = start_input_pass;
-  inputctl->pub.finish_input_pass = finish_input_pass;
-
-  inputctl->pub.consume_markers = consume_markers_with_huffman_index;
-  inputctl->pub.consume_input_build_huffman_index =
-        consume_markers_with_huffman_index;
-  /* Initialize state: can't use reset_input_controller since we don't
-   * want to try to reset other modules yet.
-   */
-  inputctl->pub.has_multiple_scans = FALSE; /* "unknown" would be better */
-  inputctl->pub.eoi_reached = FALSE;
-  inputctl->inheaders = TRUE;
-}
@@ -1,512 +0,0 @@
-/*
- * jdmainct.c
- *
- * Copyright (C) 1994-1996, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains the main buffer controller for decompression.
- * The main buffer lies between the JPEG decompressor proper and the
- * post-processor; it holds downsampled data in the JPEG colorspace.
- *
- * Note that this code is bypassed in raw-data mode, since the application
- * supplies the equivalent of the main buffer in that case.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-
-
-/*
- * In the current system design, the main buffer need never be a full-image
- * buffer; any full-height buffers will be found inside the coefficient or
- * postprocessing controllers.  Nonetheless, the main controller is not
- * trivial.  Its responsibility is to provide context rows for upsampling/
- * rescaling, and doing this in an efficient fashion is a bit tricky.
- *
- * Postprocessor input data is counted in "row groups".  A row group
- * is defined to be (v_samp_factor * DCT_scaled_size / min_DCT_scaled_size)
- * sample rows of each component.  (We require DCT_scaled_size values to be
- * chosen such that these numbers are integers.  In practice DCT_scaled_size
- * values will likely be powers of two, so we actually have the stronger
- * condition that DCT_scaled_size / min_DCT_scaled_size is an integer.)
- * Upsampling will typically produce max_v_samp_factor pixel rows from each
- * row group (times any additional scale factor that the upsampler is
- * applying).
- *
- * The coefficient controller will deliver data to us one iMCU row at a time;
- * each iMCU row contains v_samp_factor * DCT_scaled_size sample rows, or
- * exactly min_DCT_scaled_size row groups.  (This amount of data corresponds
- * to one row of MCUs when the image is fully interleaved.)  Note that the
- * number of sample rows varies across components, but the number of row
- * groups does not.  Some garbage sample rows may be included in the last iMCU
- * row at the bottom of the image.
- *
- * Depending on the vertical scaling algorithm used, the upsampler may need
- * access to the sample row(s) above and below its current input row group.
- * The upsampler is required to set need_context_rows TRUE at global selection
- * time if so.  When need_context_rows is FALSE, this controller can simply
- * obtain one iMCU row at a time from the coefficient controller and dole it
- * out as row groups to the postprocessor.
- *
- * When need_context_rows is TRUE, this controller guarantees that the buffer
- * passed to postprocessing contains at least one row group's worth of samples
- * above and below the row group(s) being processed.  Note that the context
- * rows "above" the first passed row group appear at negative row offsets in
- * the passed buffer.  At the top and bottom of the image, the required
- * context rows are manufactured by duplicating the first or last real sample
- * row; this avoids having special cases in the upsampling inner loops.
- *
- * The amount of context is fixed at one row group just because that's a
- * convenient number for this controller to work with.  The existing
- * upsamplers really only need one sample row of context.  An upsampler
- * supporting arbitrary output rescaling might wish for more than one row
- * group of context when shrinking the image; tough, we don't handle that.
- * (This is justified by the assumption that downsizing will be handled mostly
- * by adjusting the DCT_scaled_size values, so that the actual scale factor at
- * the upsample step needn't be much less than one.)
- *
- * To provide the desired context, we have to retain the last two row groups
- * of one iMCU row while reading in the next iMCU row.  (The last row group
- * can't be processed until we have another row group for its below-context,
- * and so we have to save the next-to-last group too for its above-context.)
- * We could do this most simply by copying data around in our buffer, but
- * that'd be very slow.  We can avoid copying any data by creating a rather
- * strange pointer structure.  Here's how it works.  We allocate a workspace
- * consisting of M+2 row groups (where M = min_DCT_scaled_size is the number
- * of row groups per iMCU row).  We create two sets of redundant pointers to
- * the workspace.  Labeling the physical row groups 0 to M+1, the synthesized
- * pointer lists look like this:
- *                   M+1                          M-1
- * master pointer --> 0         master pointer --> 0
- *                    1                            1
- *                   ...                          ...
- *                   M-3                          M-3
- *                   M-2                           M
- *                   M-1                          M+1
- *                    M                           M-2
- *                   M+1                          M-1
- *                    0                            0
- * We read alternate iMCU rows using each master pointer; thus the last two
- * row groups of the previous iMCU row remain un-overwritten in the workspace.
- * The pointer lists are set up so that the required context rows appear to
- * be adjacent to the proper places when we pass the pointer lists to the
- * upsampler.
- *
- * The above pictures describe the normal state of the pointer lists.
- * At top and bottom of the image, we diddle the pointer lists to duplicate
- * the first or last sample row as necessary (this is cheaper than copying
- * sample rows around).
- *
- * This scheme breaks down if M < 2, ie, min_DCT_scaled_size is 1.  In that
- * situation each iMCU row provides only one row group so the buffering logic
- * must be different (eg, we must read two iMCU rows before we can emit the
- * first row group).  For now, we simply do not support providing context
- * rows when min_DCT_scaled_size is 1.  That combination seems unlikely to
- * be worth providing --- if someone wants a 1/8th-size preview, they probably
- * want it quick and dirty, so a context-free upsampler is sufficient.
- */
-
-
-/* Private buffer controller object */
-
-typedef struct {
-  struct jpeg_d_main_controller pub; /* public fields */
-
-  /* Pointer to allocated workspace (M or M+2 row groups). */
-  JSAMPARRAY buffer[MAX_COMPONENTS];
-
-  boolean buffer_full;		/* Have we gotten an iMCU row from decoder? */
-  JDIMENSION rowgroup_ctr;	/* counts row groups output to postprocessor */
-
-  /* Remaining fields are only used in the context case. */
-
-  /* These are the master pointers to the funny-order pointer lists. */
-  JSAMPIMAGE xbuffer[2];	/* pointers to weird pointer lists */
-
-  int whichptr;			/* indicates which pointer set is now in use */
-  int context_state;		/* process_data state machine status */
-  JDIMENSION rowgroups_avail;	/* row groups available to postprocessor */
-  JDIMENSION iMCU_row_ctr;	/* counts iMCU rows to detect image top/bot */
-} my_main_controller;
-
-typedef my_main_controller * my_main_ptr;
-
-/* context_state values: */
-#define CTX_PREPARE_FOR_IMCU	0	/* need to prepare for MCU row */
-#define CTX_PROCESS_IMCU	1	/* feeding iMCU to postprocessor */
-#define CTX_POSTPONED_ROW	2	/* feeding postponed row group */
-
-
-/* Forward declarations */
-METHODDEF(void) process_data_simple_main
-	JPP((j_decompress_ptr cinfo, JSAMPARRAY output_buf,
-	     JDIMENSION *out_row_ctr, JDIMENSION out_rows_avail));
-METHODDEF(void) process_data_context_main
-	JPP((j_decompress_ptr cinfo, JSAMPARRAY output_buf,
-	     JDIMENSION *out_row_ctr, JDIMENSION out_rows_avail));
-#ifdef QUANT_2PASS_SUPPORTED
-METHODDEF(void) process_data_crank_post
-	JPP((j_decompress_ptr cinfo, JSAMPARRAY output_buf,
-	     JDIMENSION *out_row_ctr, JDIMENSION out_rows_avail));
-#endif
-
-
-LOCAL(void)
-alloc_funny_pointers (j_decompress_ptr cinfo)
-/* Allocate space for the funny pointer lists.
- * This is done only once, not once per pass.
- */
-{
-  my_main_ptr main = (my_main_ptr) cinfo->main;
-  int ci, rgroup;
-  int M = cinfo->min_DCT_scaled_size;
-  jpeg_component_info *compptr;
-  JSAMPARRAY xbuf;
-
-  /* Get top-level space for component array pointers.
-   * We alloc both arrays with one call to save a few cycles.
-   */
-  main->xbuffer[0] = (JSAMPIMAGE)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				cinfo->num_components * 2 * SIZEOF(JSAMPARRAY));
-  main->xbuffer[1] = main->xbuffer[0] + cinfo->num_components;
-
-  for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
-       ci++, compptr++) {
-    rgroup = (compptr->v_samp_factor * compptr->DCT_scaled_size) /
-      cinfo->min_DCT_scaled_size; /* height of a row group of component */
-    /* Get space for pointer lists --- M+4 row groups in each list.
-     * We alloc both pointer lists with one call to save a few cycles.
-     */
-    xbuf = (JSAMPARRAY)
-      (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				  2 * (rgroup * (M + 4)) * SIZEOF(JSAMPROW));
-    xbuf += rgroup;		/* want one row group at negative offsets */
-    main->xbuffer[0][ci] = xbuf;
-    xbuf += rgroup * (M + 4);
-    main->xbuffer[1][ci] = xbuf;
-  }
-}
-
-
-LOCAL(void)
-make_funny_pointers (j_decompress_ptr cinfo)
-/* Create the funny pointer lists discussed in the comments above.
- * The actual workspace is already allocated (in main->buffer),
- * and the space for the pointer lists is allocated too.
- * This routine just fills in the curiously ordered lists.
- * This will be repeated at the beginning of each pass.
- */
-{
-  my_main_ptr main = (my_main_ptr) cinfo->main;
-  int ci, i, rgroup;
-  int M = cinfo->min_DCT_scaled_size;
-  jpeg_component_info *compptr;
-  JSAMPARRAY buf, xbuf0, xbuf1;
-
-  for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
-       ci++, compptr++) {
-    rgroup = (compptr->v_samp_factor * compptr->DCT_scaled_size) /
-      cinfo->min_DCT_scaled_size; /* height of a row group of component */
-    xbuf0 = main->xbuffer[0][ci];
-    xbuf1 = main->xbuffer[1][ci];
-    /* First copy the workspace pointers as-is */
-    buf = main->buffer[ci];
-    for (i = 0; i < rgroup * (M + 2); i++) {
-      xbuf0[i] = xbuf1[i] = buf[i];
-    }
-    /* In the second list, put the last four row groups in swapped order */
-    for (i = 0; i < rgroup * 2; i++) {
-      xbuf1[rgroup*(M-2) + i] = buf[rgroup*M + i];
-      xbuf1[rgroup*M + i] = buf[rgroup*(M-2) + i];
-    }
-    /* The wraparound pointers at top and bottom will be filled later
-     * (see set_wraparound_pointers, below).  Initially we want the "above"
-     * pointers to duplicate the first actual data line.  This only needs
-     * to happen in xbuffer[0].
-     */
-    for (i = 0; i < rgroup; i++) {
-      xbuf0[i - rgroup] = xbuf0[0];
-    }
-  }
-}
-
-
-LOCAL(void)
-set_wraparound_pointers (j_decompress_ptr cinfo)
-/* Set up the "wraparound" pointers at top and bottom of the pointer lists.
- * This changes the pointer list state from top-of-image to the normal state.
- */
-{
-  my_main_ptr main = (my_main_ptr) cinfo->main;
-  int ci, i, rgroup;
-  int M = cinfo->min_DCT_scaled_size;
-  jpeg_component_info *compptr;
-  JSAMPARRAY xbuf0, xbuf1;
-
-  for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
-       ci++, compptr++) {
-    rgroup = (compptr->v_samp_factor * compptr->DCT_scaled_size) /
-      cinfo->min_DCT_scaled_size; /* height of a row group of component */
-    xbuf0 = main->xbuffer[0][ci];
-    xbuf1 = main->xbuffer[1][ci];
-    for (i = 0; i < rgroup; i++) {
-      xbuf0[i - rgroup] = xbuf0[rgroup*(M+1) + i];
-      xbuf1[i - rgroup] = xbuf1[rgroup*(M+1) + i];
-      xbuf0[rgroup*(M+2) + i] = xbuf0[i];
-      xbuf1[rgroup*(M+2) + i] = xbuf1[i];
-    }
-  }
-}
-
-
-LOCAL(void)
-set_bottom_pointers (j_decompress_ptr cinfo)
-/* Change the pointer lists to duplicate the last sample row at the bottom
- * of the image.  whichptr indicates which xbuffer holds the final iMCU row.
- * Also sets rowgroups_avail to indicate number of nondummy row groups in row.
- */
-{
-  my_main_ptr main = (my_main_ptr) cinfo->main;
-  int ci, i, rgroup, iMCUheight, rows_left;
-  jpeg_component_info *compptr;
-  JSAMPARRAY xbuf;
-
-  for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
-       ci++, compptr++) {
-    /* Count sample rows in one iMCU row and in one row group */
-    iMCUheight = compptr->v_samp_factor * compptr->DCT_scaled_size;
-    rgroup = iMCUheight / cinfo->min_DCT_scaled_size;
-    /* Count nondummy sample rows remaining for this component */
-    rows_left = (int) (compptr->downsampled_height % (JDIMENSION) iMCUheight);
-    if (rows_left == 0) rows_left = iMCUheight;
-    /* Count nondummy row groups.  Should get same answer for each component,
-     * so we need only do it once.
-     */
-    if (ci == 0) {
-      main->rowgroups_avail = (JDIMENSION) ((rows_left-1) / rgroup + 1);
-    }
-    /* Duplicate the last real sample row rgroup*2 times; this pads out the
-     * last partial rowgroup and ensures at least one full rowgroup of context.
-     */
-    xbuf = main->xbuffer[main->whichptr][ci];
-    for (i = 0; i < rgroup * 2; i++) {
-      xbuf[rows_left + i] = xbuf[rows_left-1];
-    }
-  }
-}
-
-
-/*
- * Initialize for a processing pass.
- */
-
-METHODDEF(void)
-start_pass_main (j_decompress_ptr cinfo, J_BUF_MODE pass_mode)
-{
-  my_main_ptr main = (my_main_ptr) cinfo->main;
-
-  switch (pass_mode) {
-  case JBUF_PASS_THRU:
-    if (cinfo->upsample->need_context_rows) {
-      main->pub.process_data = process_data_context_main;
-      make_funny_pointers(cinfo); /* Create the xbuffer[] lists */
-      main->whichptr = 0;	/* Read first iMCU row into xbuffer[0] */
-      main->context_state = CTX_PREPARE_FOR_IMCU;
-      main->iMCU_row_ctr = 0;
-    } else {
-      /* Simple case with no context needed */
-      main->pub.process_data = process_data_simple_main;
-    }
-    main->buffer_full = FALSE;	/* Mark buffer empty */
-    main->rowgroup_ctr = 0;
-    break;
-#ifdef QUANT_2PASS_SUPPORTED
-  case JBUF_CRANK_DEST:
-    /* For last pass of 2-pass quantization, just crank the postprocessor */
-    main->pub.process_data = process_data_crank_post;
-    break;
-#endif
-  default:
-    ERREXIT(cinfo, JERR_BAD_BUFFER_MODE);
-    break;
-  }
-}
-
-
-/*
- * Process some data.
- * This handles the simple case where no context is required.
- */
-
-METHODDEF(void)
-process_data_simple_main (j_decompress_ptr cinfo,
-			  JSAMPARRAY output_buf, JDIMENSION *out_row_ctr,
-			  JDIMENSION out_rows_avail)
-{
-  my_main_ptr main = (my_main_ptr) cinfo->main;
-  JDIMENSION rowgroups_avail;
-
-  /* Read input data if we haven't filled the main buffer yet */
-  if (! main->buffer_full) {
-    if (! (*cinfo->coef->decompress_data) (cinfo, main->buffer))
-      return;			/* suspension forced, can do nothing more */
-    main->buffer_full = TRUE;	/* OK, we have an iMCU row to work with */
-  }
-
-  /* There are always min_DCT_scaled_size row groups in an iMCU row. */
-  rowgroups_avail = (JDIMENSION) cinfo->min_DCT_scaled_size;
-  /* Note: at the bottom of the image, we may pass extra garbage row groups
-   * to the postprocessor.  The postprocessor has to check for bottom
-   * of image anyway (at row resolution), so no point in us doing it too.
-   */
-
-  /* Feed the postprocessor */
-  (*cinfo->post->post_process_data) (cinfo, main->buffer,
-				     &main->rowgroup_ctr, rowgroups_avail,
-				     output_buf, out_row_ctr, out_rows_avail);
-
-  /* Has postprocessor consumed all the data yet? If so, mark buffer empty */
-  if (main->rowgroup_ctr >= rowgroups_avail) {
-    main->buffer_full = FALSE;
-    main->rowgroup_ctr = 0;
-  }
-}
-
-
-/*
- * Process some data.
- * This handles the case where context rows must be provided.
- */
-
-METHODDEF(void)
-process_data_context_main (j_decompress_ptr cinfo,
-			   JSAMPARRAY output_buf, JDIMENSION *out_row_ctr,
-			   JDIMENSION out_rows_avail)
-{
-  my_main_ptr main = (my_main_ptr) cinfo->main;
-
-  /* Read input data if we haven't filled the main buffer yet */
-  if (! main->buffer_full) {
-    if (! (*cinfo->coef->decompress_data) (cinfo,
-					   main->xbuffer[main->whichptr]))
-      return;			/* suspension forced, can do nothing more */
-    main->buffer_full = TRUE;	/* OK, we have an iMCU row to work with */
-    main->iMCU_row_ctr++;	/* count rows received */
-  }
-
-  /* Postprocessor typically will not swallow all the input data it is handed
-   * in one call (due to filling the output buffer first).  Must be prepared
-   * to exit and restart.  This switch lets us keep track of how far we got.
-   * Note that each case falls through to the next on successful completion.
-   */
-  switch (main->context_state) {
-  case CTX_POSTPONED_ROW:
-    /* Call postprocessor using previously set pointers for postponed row */
-    (*cinfo->post->post_process_data) (cinfo, main->xbuffer[main->whichptr],
-			&main->rowgroup_ctr, main->rowgroups_avail,
-			output_buf, out_row_ctr, out_rows_avail);
-    if (main->rowgroup_ctr < main->rowgroups_avail)
-      return;			/* Need to suspend */
-    main->context_state = CTX_PREPARE_FOR_IMCU;
-    if (*out_row_ctr >= out_rows_avail)
-      return;			/* Postprocessor exactly filled output buf */
-    /*FALLTHROUGH*/
-  case CTX_PREPARE_FOR_IMCU:
-    /* Prepare to process first M-1 row groups of this iMCU row */
-    main->rowgroup_ctr = 0;
-    main->rowgroups_avail = (JDIMENSION) (cinfo->min_DCT_scaled_size - 1);
-    /* Check for bottom of image: if so, tweak pointers to "duplicate"
-     * the last sample row, and adjust rowgroups_avail to ignore padding rows.
-     */
-    if (main->iMCU_row_ctr == cinfo->total_iMCU_rows)
-      set_bottom_pointers(cinfo);
-    main->context_state = CTX_PROCESS_IMCU;
-    /*FALLTHROUGH*/
-  case CTX_PROCESS_IMCU:
-    /* Call postprocessor using previously set pointers */
-    (*cinfo->post->post_process_data) (cinfo, main->xbuffer[main->whichptr],
-			&main->rowgroup_ctr, main->rowgroups_avail,
-			output_buf, out_row_ctr, out_rows_avail);
-    if (main->rowgroup_ctr < main->rowgroups_avail)
-      return;			/* Need to suspend */
-    /* After the first iMCU, change wraparound pointers to normal state */
-    if (main->iMCU_row_ctr == 1)
-      set_wraparound_pointers(cinfo);
-    /* Prepare to load new iMCU row using other xbuffer list */
-    main->whichptr ^= 1;	/* 0=>1 or 1=>0 */
-    main->buffer_full = FALSE;
-    /* Still need to process last row group of this iMCU row, */
-    /* which is saved at index M+1 of the other xbuffer */
-    main->rowgroup_ctr = (JDIMENSION) (cinfo->min_DCT_scaled_size + 1);
-    main->rowgroups_avail = (JDIMENSION) (cinfo->min_DCT_scaled_size + 2);
-    main->context_state = CTX_POSTPONED_ROW;
-  }
-}
-
-
-/*
- * Process some data.
- * Final pass of two-pass quantization: just call the postprocessor.
- * Source data will be the postprocessor controller's internal buffer.
- */
-
-#ifdef QUANT_2PASS_SUPPORTED
-
-METHODDEF(void)
-process_data_crank_post (j_decompress_ptr cinfo,
-			 JSAMPARRAY output_buf, JDIMENSION *out_row_ctr,
-			 JDIMENSION out_rows_avail)
-{
-  (*cinfo->post->post_process_data) (cinfo, (JSAMPIMAGE) NULL,
-				     (JDIMENSION *) NULL, (JDIMENSION) 0,
-				     output_buf, out_row_ctr, out_rows_avail);
-}
-
-#endif /* QUANT_2PASS_SUPPORTED */
-
-
-/*
- * Initialize main buffer controller.
- */
-
-GLOBAL(void)
-jinit_d_main_controller (j_decompress_ptr cinfo, boolean need_full_buffer)
-{
-  my_main_ptr main;
-  int ci, rgroup, ngroups;
-  jpeg_component_info *compptr;
-
-  main = (my_main_ptr)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				SIZEOF(my_main_controller));
-  cinfo->main = (struct jpeg_d_main_controller *) main;
-  main->pub.start_pass = start_pass_main;
-
-  if (need_full_buffer)		/* shouldn't happen */
-    ERREXIT(cinfo, JERR_BAD_BUFFER_MODE);
-
-  /* Allocate the workspace.
-   * ngroups is the number of row groups we need.
-   */
-  if (cinfo->upsample->need_context_rows) {
-    if (cinfo->min_DCT_scaled_size < 2) /* unsupported, see comments above */
-      ERREXIT(cinfo, JERR_NOTIMPL);
-    alloc_funny_pointers(cinfo); /* Alloc space for xbuffer[] lists */
-    ngroups = cinfo->min_DCT_scaled_size + 2;
-  } else {
-    ngroups = cinfo->min_DCT_scaled_size;
-  }
-
-  for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
-       ci++, compptr++) {
-    rgroup = (compptr->v_samp_factor * compptr->DCT_scaled_size) /
-      cinfo->min_DCT_scaled_size; /* height of a row group of component */
-    main->buffer[ci] = (*cinfo->mem->alloc_sarray)
-			((j_common_ptr) cinfo, JPOOL_IMAGE,
-			 compptr->width_in_blocks * compptr->DCT_scaled_size,
-			 (JDIMENSION) (rgroup * ngroups));
-  }
-}
@@ -1,580 +0,0 @@
-/*
- * jdmaster.c
- *
- * Copyright (C) 1991-1997, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains master control logic for the JPEG decompressor.
- * These routines are concerned with selecting the modules to be executed
- * and with determining the number of passes and the work to be done in each
- * pass.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-
-
-/* Private state */
-
-typedef struct {
-  struct jpeg_decomp_master pub; /* public fields */
-
-  int pass_number;		/* # of passes completed */
-
-  boolean using_merged_upsample; /* TRUE if using merged upsample/cconvert */
-
-  /* Saved references to initialized quantizer modules,
-   * in case we need to switch modes.
-   */
-  struct jpeg_color_quantizer * quantizer_1pass;
-  struct jpeg_color_quantizer * quantizer_2pass;
-} my_decomp_master;
-
-typedef my_decomp_master * my_master_ptr;
-
-
-/*
- * Determine whether merged upsample/color conversion should be used.
- * CRUCIAL: this must match the actual capabilities of jdmerge.c!
- */
-
-LOCAL(boolean)
-use_merged_upsample (j_decompress_ptr cinfo)
-{
-#ifdef UPSAMPLE_MERGING_SUPPORTED
-  /* Merging is the equivalent of plain box-filter upsampling */
-  if (cinfo->do_fancy_upsampling || cinfo->CCIR601_sampling)
-    return FALSE;
-
-#ifdef ANDROID_RGB
-  /* jdmerge.c only supports YCC=>RGB565 and YCC=>RGB color conversion */
-  if (cinfo->jpeg_color_space != JCS_YCbCr || 
-      cinfo->num_components != 3 ||
-      cinfo->out_color_components != 3 ||
-      (cinfo->out_color_space != JCS_RGB_565 && 
-         cinfo->out_color_space != JCS_RGB)) {
-    return FALSE;
-  }
-#else
-  /* jdmerge.c only supports YCC=>RGB color conversion */
-  if (cinfo->jpeg_color_space != JCS_YCbCr || cinfo->num_components != 3 ||
-      cinfo->out_color_space != JCS_RGB ||
-      cinfo->out_color_components != RGB_PIXELSIZE)
-    return FALSE;
-#endif
-
-  /* and it only handles 2h1v or 2h2v sampling ratios */
-  if (cinfo->comp_info[0].h_samp_factor != 2 ||
-      cinfo->comp_info[1].h_samp_factor != 1 ||
-      cinfo->comp_info[2].h_samp_factor != 1 ||
-      cinfo->comp_info[0].v_samp_factor >  2 ||
-      cinfo->comp_info[1].v_samp_factor != 1 ||
-      cinfo->comp_info[2].v_samp_factor != 1)
-    return FALSE;
-  /* furthermore, it doesn't work if we've scaled the IDCTs differently */
-  if (cinfo->comp_info[0].DCT_scaled_size != cinfo->min_DCT_scaled_size ||
-      cinfo->comp_info[1].DCT_scaled_size != cinfo->min_DCT_scaled_size ||
-      cinfo->comp_info[2].DCT_scaled_size != cinfo->min_DCT_scaled_size)
-    return FALSE;
-  /* ??? also need to test for upsample-time rescaling, when & if supported */
-  return TRUE;			/* by golly, it'll work... */
-#else
-  return FALSE;
-#endif
-}
-
-
-/*
- * Compute output image dimensions and related values.
- * NOTE: this is exported for possible use by application.
- * Hence it mustn't do anything that can't be done twice.
- * Also note that it may be called before the master module is initialized!
- */
-
-GLOBAL(void)
-jpeg_calc_output_dimensions (j_decompress_ptr cinfo)
-/* Do computations that are needed before master selection phase */
-{
-#ifdef IDCT_SCALING_SUPPORTED
-  int ci;
-  jpeg_component_info *compptr;
-#endif
-
-  /* Prevent application from calling me at wrong times */
-#if ANDROID_TILE_BASED_DECODE
-  // Tile based decoding may call this function several times.
-  if (!cinfo->tile_decode)
-#endif
-    if (cinfo->global_state != DSTATE_READY)
-      ERREXIT1(cinfo, JERR_BAD_STATE, cinfo->global_state);
-
-#ifdef IDCT_SCALING_SUPPORTED
-
-  /* Compute actual output image dimensions and DCT scaling choices. */
-  if (cinfo->scale_num * 8 <= cinfo->scale_denom) {
-    /* Provide 1/8 scaling */
-    cinfo->output_width = (JDIMENSION)
-      jdiv_round_up((long) cinfo->image_width, 8L);
-    cinfo->output_height = (JDIMENSION)
-      jdiv_round_up((long) cinfo->image_height, 8L);
-    cinfo->min_DCT_scaled_size = 1;
-  } else if (cinfo->scale_num * 4 <= cinfo->scale_denom) {
-    /* Provide 1/4 scaling */
-    cinfo->output_width = (JDIMENSION)
-      jdiv_round_up((long) cinfo->image_width, 4L);
-    cinfo->output_height = (JDIMENSION)
-      jdiv_round_up((long) cinfo->image_height, 4L);
-    cinfo->min_DCT_scaled_size = 2;
-  } else if (cinfo->scale_num * 2 <= cinfo->scale_denom) {
-    /* Provide 1/2 scaling */
-    cinfo->output_width = (JDIMENSION)
-      jdiv_round_up((long) cinfo->image_width, 2L);
-    cinfo->output_height = (JDIMENSION)
-      jdiv_round_up((long) cinfo->image_height, 2L);
-    cinfo->min_DCT_scaled_size = 4;
-  } else {
-    /* Provide 1/1 scaling */
-    cinfo->output_width = cinfo->image_width;
-    cinfo->output_height = cinfo->image_height;
-    cinfo->min_DCT_scaled_size = DCTSIZE;
-  }
-  /* In selecting the actual DCT scaling for each component, we try to
-   * scale up the chroma components via IDCT scaling rather than upsampling.
-   * This saves time if the upsampler gets to use 1:1 scaling.
-   * Note this code assumes that the supported DCT scalings are powers of 2.
-   */
-  for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
-       ci++, compptr++) {
-    int ssize = cinfo->min_DCT_scaled_size;
-    while (ssize < DCTSIZE &&
-	   (compptr->h_samp_factor * ssize * 2 <=
-	    cinfo->max_h_samp_factor * cinfo->min_DCT_scaled_size) &&
-	   (compptr->v_samp_factor * ssize * 2 <=
-	    cinfo->max_v_samp_factor * cinfo->min_DCT_scaled_size)) {
-      ssize = ssize * 2;
-    }
-    compptr->DCT_scaled_size = ssize;
-  }
-
-  /* Recompute downsampled dimensions of components;
-   * application needs to know these if using raw downsampled data.
-   */
-  for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
-       ci++, compptr++) {
-    /* Size in samples, after IDCT scaling */
-    compptr->downsampled_width = (JDIMENSION)
-      jdiv_round_up((long) cinfo->image_width *
-		    (long) (compptr->h_samp_factor * compptr->DCT_scaled_size),
-		    (long) (cinfo->max_h_samp_factor * DCTSIZE));
-    compptr->downsampled_height = (JDIMENSION)
-      jdiv_round_up((long) cinfo->image_height *
-		    (long) (compptr->v_samp_factor * compptr->DCT_scaled_size),
-		    (long) (cinfo->max_v_samp_factor * DCTSIZE));
-  }
-
-#else /* !IDCT_SCALING_SUPPORTED */
-
-  /* Hardwire it to "no scaling" */
-  cinfo->output_width = cinfo->image_width;
-  cinfo->output_height = cinfo->image_height;
-  /* jdinput.c has already initialized DCT_scaled_size to DCTSIZE,
-   * and has computed unscaled downsampled_width and downsampled_height.
-   */
-
-#endif /* IDCT_SCALING_SUPPORTED */
-
-  /* Report number of components in selected colorspace. */
-  /* Probably this should be in the color conversion module... */
-  switch (cinfo->out_color_space) {
-  case JCS_GRAYSCALE:
-    cinfo->out_color_components = 1;
-    break;
-  case JCS_RGB:
-#if RGB_PIXELSIZE != 3
-    cinfo->out_color_components = RGB_PIXELSIZE;
-    break;
-#endif /* else share code with YCbCr */
-#ifdef ANDROID_RGB
-  case JCS_RGB_565:
-#endif
-  case JCS_YCbCr:
-    cinfo->out_color_components = 3;
-    break;
-  case JCS_CMYK:
-  case JCS_YCCK:
-#ifdef ANDROID_RGB
-  case JCS_RGBA_8888:
-#endif
-    cinfo->out_color_components = 4;
-    break;
-  default:			/* else must be same colorspace as in file */
-    cinfo->out_color_components = cinfo->num_components;
-    break;
-  }
-  cinfo->output_components = (cinfo->quantize_colors ? 1 :
-			      cinfo->out_color_components);
-
-  /* See if upsampler will want to emit more than one row at a time */
-  if (use_merged_upsample(cinfo))
-    cinfo->rec_outbuf_height = cinfo->max_v_samp_factor;
-  else
-    cinfo->rec_outbuf_height = 1;
-}
-
-
-/*
- * Several decompression processes need to range-limit values to the range
- * 0..MAXJSAMPLE; the input value may fall somewhat outside this range
- * due to noise introduced by quantization, roundoff error, etc.  These
- * processes are inner loops and need to be as fast as possible.  On most
- * machines, particularly CPUs with pipelines or instruction prefetch,
- * a (subscript-check-less) C table lookup
- *		x = sample_range_limit[x];
- * is faster than explicit tests
- *		if (x < 0)  x = 0;
- *		else if (x > MAXJSAMPLE)  x = MAXJSAMPLE;
- * These processes all use a common table prepared by the routine below.
- *
- * For most steps we can mathematically guarantee that the initial value
- * of x is within MAXJSAMPLE+1 of the legal range, so a table running from
- * -(MAXJSAMPLE+1) to 2*MAXJSAMPLE+1 is sufficient.  But for the initial
- * limiting step (just after the IDCT), a wildly out-of-range value is
- * possible if the input data is corrupt.  To avoid any chance of indexing
- * off the end of memory and getting a bad-pointer trap, we perform the
- * post-IDCT limiting thus:
- *		x = range_limit[x & MASK];
- * where MASK is 2 bits wider than legal sample data, ie 10 bits for 8-bit
- * samples.  Under normal circumstances this is more than enough range and
- * a correct output will be generated; with bogus input data the mask will
- * cause wraparound, and we will safely generate a bogus-but-in-range output.
- * For the post-IDCT step, we want to convert the data from signed to unsigned
- * representation by adding CENTERJSAMPLE at the same time that we limit it.
- * So the post-IDCT limiting table ends up looking like this:
- *   CENTERJSAMPLE,CENTERJSAMPLE+1,...,MAXJSAMPLE,
- *   MAXJSAMPLE (repeat 2*(MAXJSAMPLE+1)-CENTERJSAMPLE times),
- *   0          (repeat 2*(MAXJSAMPLE+1)-CENTERJSAMPLE times),
- *   0,1,...,CENTERJSAMPLE-1
- * Negative inputs select values from the upper half of the table after
- * masking.
- *
- * We can save some space by overlapping the start of the post-IDCT table
- * with the simpler range limiting table.  The post-IDCT table begins at
- * sample_range_limit + CENTERJSAMPLE.
- *
- * Note that the table is allocated in near data space on PCs; it's small
- * enough and used often enough to justify this.
- */
-
-LOCAL(void)
-prepare_range_limit_table (j_decompress_ptr cinfo)
-/* Allocate and fill in the sample_range_limit table */
-{
-  JSAMPLE * table;
-  int i;
-
-  table = (JSAMPLE *)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-		(5 * (MAXJSAMPLE+1) + CENTERJSAMPLE) * SIZEOF(JSAMPLE));
-  table += (MAXJSAMPLE+1);	/* allow negative subscripts of simple table */
-  cinfo->sample_range_limit = table;
-  /* First segment of "simple" table: limit[x] = 0 for x < 0 */
-  MEMZERO(table - (MAXJSAMPLE+1), (MAXJSAMPLE+1) * SIZEOF(JSAMPLE));
-  /* Main part of "simple" table: limit[x] = x */
-  for (i = 0; i <= MAXJSAMPLE; i++)
-    table[i] = (JSAMPLE) i;
-  table += CENTERJSAMPLE;	/* Point to where post-IDCT table starts */
-  /* End of simple table, rest of first half of post-IDCT table */
-  for (i = CENTERJSAMPLE; i < 2*(MAXJSAMPLE+1); i++)
-    table[i] = MAXJSAMPLE;
-  /* Second half of post-IDCT table */
-  MEMZERO(table + (2 * (MAXJSAMPLE+1)),
-	  (2 * (MAXJSAMPLE+1) - CENTERJSAMPLE) * SIZEOF(JSAMPLE));
-  MEMCOPY(table + (4 * (MAXJSAMPLE+1) - CENTERJSAMPLE),
-	  cinfo->sample_range_limit, CENTERJSAMPLE * SIZEOF(JSAMPLE));
-}
-
-
-/*
- * Master selection of decompression modules.
- * This is done once at jpeg_start_decompress time.  We determine
- * which modules will be used and give them appropriate initialization calls.
- * We also initialize the decompressor input side to begin consuming data.
- *
- * Since jpeg_read_header has finished, we know what is in the SOF
- * and (first) SOS markers.  We also have all the application parameter
- * settings.
- */
-
-LOCAL(void)
-master_selection (j_decompress_ptr cinfo)
-{
-  my_master_ptr master = (my_master_ptr) cinfo->master;
-  boolean use_c_buffer;
-  long samplesperrow;
-  JDIMENSION jd_samplesperrow;
-
-  /* Initialize dimensions and other stuff */
-  jpeg_calc_output_dimensions(cinfo);
-  prepare_range_limit_table(cinfo);
-
-  /* Width of an output scanline must be representable as JDIMENSION. */
-  samplesperrow = (long) cinfo->output_width * (long) cinfo->out_color_components;
-  jd_samplesperrow = (JDIMENSION) samplesperrow;
-  if ((long) jd_samplesperrow != samplesperrow)
-    ERREXIT(cinfo, JERR_WIDTH_OVERFLOW);
-
-  /* Initialize my private state */
-  master->pass_number = 0;
-  master->using_merged_upsample = use_merged_upsample(cinfo);
-
-  /* Color quantizer selection */
-  master->quantizer_1pass = NULL;
-  master->quantizer_2pass = NULL;
-  /* No mode changes if not using buffered-image mode. */
-  if (! cinfo->quantize_colors || ! cinfo->buffered_image) {
-    cinfo->enable_1pass_quant = FALSE;
-    cinfo->enable_external_quant = FALSE;
-    cinfo->enable_2pass_quant = FALSE;
-  }
-  if (cinfo->quantize_colors) {
-    if (cinfo->raw_data_out)
-      ERREXIT(cinfo, JERR_NOTIMPL);
-    /* 2-pass quantizer only works in 3-component color space. */
-    if (cinfo->out_color_components != 3) {
-      cinfo->enable_1pass_quant = TRUE;
-      cinfo->enable_external_quant = FALSE;
-      cinfo->enable_2pass_quant = FALSE;
-      cinfo->colormap = NULL;
-    } else if (cinfo->colormap != NULL) {
-      cinfo->enable_external_quant = TRUE;
-    } else if (cinfo->two_pass_quantize) {
-      cinfo->enable_2pass_quant = TRUE;
-    } else {
-      cinfo->enable_1pass_quant = TRUE;
-    }
-
-    if (cinfo->enable_1pass_quant) {
-#ifdef QUANT_1PASS_SUPPORTED
-      jinit_1pass_quantizer(cinfo);
-      master->quantizer_1pass = cinfo->cquantize;
-#else
-      ERREXIT(cinfo, JERR_NOT_COMPILED);
-#endif
-    }
-
-    /* We use the 2-pass code to map to external colormaps. */
-    if (cinfo->enable_2pass_quant || cinfo->enable_external_quant) {
-#ifdef QUANT_2PASS_SUPPORTED
-      jinit_2pass_quantizer(cinfo);
-      master->quantizer_2pass = cinfo->cquantize;
-#else
-      ERREXIT(cinfo, JERR_NOT_COMPILED);
-#endif
-    }
-    /* If both quantizers are initialized, the 2-pass one is left active;
-     * this is necessary for starting with quantization to an external map.
-     */
-  }
-
-  /* Post-processing: in particular, color conversion first */
-  if (! cinfo->raw_data_out) {
-    if (master->using_merged_upsample) {
-#ifdef UPSAMPLE_MERGING_SUPPORTED
-      jinit_merged_upsampler(cinfo); /* does color conversion too */
-#else
-      ERREXIT(cinfo, JERR_NOT_COMPILED);
-#endif
-    } else {
-      jinit_color_deconverter(cinfo);
-      jinit_upsampler(cinfo);
-    }
-    jinit_d_post_controller(cinfo, cinfo->enable_2pass_quant);
-  }
-  /* Inverse DCT */
-  jinit_inverse_dct(cinfo);
-  /* Entropy decoding: either Huffman or arithmetic coding. */
-  if (cinfo->arith_code) {
-    ERREXIT(cinfo, JERR_ARITH_NOTIMPL);
-  } else {
-    if (cinfo->progressive_mode) {
-#ifdef D_PROGRESSIVE_SUPPORTED
-      jinit_phuff_decoder(cinfo);
-#else
-      ERREXIT(cinfo, JERR_NOT_COMPILED);
-#endif
-    } else
-      jinit_huff_decoder(cinfo);
-  }
-
-  /* Initialize principal buffer controllers. */
-  use_c_buffer = cinfo->inputctl->has_multiple_scans || cinfo->buffered_image;
-  jinit_d_coef_controller(cinfo, use_c_buffer);
-
-  if (! cinfo->raw_data_out)
-    jinit_d_main_controller(cinfo, FALSE /* never need full buffer here */);
-
-  /* We can now tell the memory manager to allocate virtual arrays. */
-  (*cinfo->mem->realize_virt_arrays) ((j_common_ptr) cinfo);
-
-  /* Initialize input side of decompressor to consume first scan. */
-  (*cinfo->inputctl->start_input_pass) (cinfo);
-
-#ifdef D_MULTISCAN_FILES_SUPPORTED
-  /* If jpeg_start_decompress will read the whole file, initialize
-   * progress monitoring appropriately.  The input step is counted
-   * as one pass.
-   */
-  if (cinfo->progress != NULL && ! cinfo->buffered_image &&
-      cinfo->inputctl->has_multiple_scans) {
-    int nscans;
-    /* Estimate number of scans to set pass_limit. */
-    if (cinfo->progressive_mode) {
-      /* Arbitrarily estimate 2 interleaved DC scans + 3 AC scans/component. */
-      nscans = 2 + 3 * cinfo->num_components;
-    } else {
-      /* For a nonprogressive multiscan file, estimate 1 scan per component. */
-      nscans = cinfo->num_components;
-    }
-    cinfo->progress->pass_counter = 0L;
-    cinfo->progress->pass_limit = (long) cinfo->total_iMCU_rows * nscans;
-    cinfo->progress->completed_passes = 0;
-    cinfo->progress->total_passes = (cinfo->enable_2pass_quant ? 3 : 2);
-    /* Count the input pass as done */
-    master->pass_number++;
-  }
-#endif /* D_MULTISCAN_FILES_SUPPORTED */
-}
-
-
-/*
- * Per-pass setup.
- * This is called at the beginning of each output pass.  We determine which
- * modules will be active during this pass and give them appropriate
- * start_pass calls.  We also set is_dummy_pass to indicate whether this
- * is a "real" output pass or a dummy pass for color quantization.
- * (In the latter case, jdapistd.c will crank the pass to completion.)
- */
-
-METHODDEF(void)
-prepare_for_output_pass (j_decompress_ptr cinfo)
-{
-  my_master_ptr master = (my_master_ptr) cinfo->master;
-
-  if (master->pub.is_dummy_pass) {
-#ifdef QUANT_2PASS_SUPPORTED
-    /* Final pass of 2-pass quantization */
-    master->pub.is_dummy_pass = FALSE;
-    (*cinfo->cquantize->start_pass) (cinfo, FALSE);
-    (*cinfo->post->start_pass) (cinfo, JBUF_CRANK_DEST);
-    (*cinfo->main->start_pass) (cinfo, JBUF_CRANK_DEST);
-#else
-    ERREXIT(cinfo, JERR_NOT_COMPILED);
-#endif /* QUANT_2PASS_SUPPORTED */
-  } else {
-    if (cinfo->quantize_colors && cinfo->colormap == NULL) {
-      /* Select new quantization method */
-      if (cinfo->two_pass_quantize && cinfo->enable_2pass_quant) {
-	cinfo->cquantize = master->quantizer_2pass;
-	master->pub.is_dummy_pass = TRUE;
-      } else if (cinfo->enable_1pass_quant) {
-	cinfo->cquantize = master->quantizer_1pass;
-      } else {
-	ERREXIT(cinfo, JERR_MODE_CHANGE);
-      }
-    }
-    (*cinfo->idct->start_pass) (cinfo);
-    (*cinfo->coef->start_output_pass) (cinfo);
-    if (! cinfo->raw_data_out) {
-      if (! master->using_merged_upsample)
-	(*cinfo->cconvert->start_pass) (cinfo);
-      (*cinfo->upsample->start_pass) (cinfo);
-      if (cinfo->quantize_colors)
-	(*cinfo->cquantize->start_pass) (cinfo, master->pub.is_dummy_pass);
-      (*cinfo->post->start_pass) (cinfo,
-	    (master->pub.is_dummy_pass ? JBUF_SAVE_AND_PASS : JBUF_PASS_THRU));
-      (*cinfo->main->start_pass) (cinfo, JBUF_PASS_THRU);
-    }
-  }
-
-  /* Set up progress monitor's pass info if present */
-  if (cinfo->progress != NULL) {
-    cinfo->progress->completed_passes = master->pass_number;
-    cinfo->progress->total_passes = master->pass_number +
-				    (master->pub.is_dummy_pass ? 2 : 1);
-    /* In buffered-image mode, we assume one more output pass if EOI not
-     * yet reached, but no more passes if EOI has been reached.
-     */
-    if (cinfo->buffered_image && ! cinfo->inputctl->eoi_reached) {
-      cinfo->progress->total_passes += (cinfo->enable_2pass_quant ? 2 : 1);
-    }
-  }
-}
-
-
-/*
- * Finish up at end of an output pass.
- */
-
-METHODDEF(void)
-finish_output_pass (j_decompress_ptr cinfo)
-{
-  my_master_ptr master = (my_master_ptr) cinfo->master;
-
-  if (cinfo->quantize_colors)
-    (*cinfo->cquantize->finish_pass) (cinfo);
-  master->pass_number++;
-}
-
-
-#ifdef D_MULTISCAN_FILES_SUPPORTED
-
-/*
- * Switch to a new external colormap between output passes.
- */
-
-GLOBAL(void)
-jpeg_new_colormap (j_decompress_ptr cinfo)
-{
-  my_master_ptr master = (my_master_ptr) cinfo->master;
-
-  /* Prevent application from calling me at wrong times */
-  if (cinfo->global_state != DSTATE_BUFIMAGE)
-    ERREXIT1(cinfo, JERR_BAD_STATE, cinfo->global_state);
-
-  if (cinfo->quantize_colors && cinfo->enable_external_quant &&
-      cinfo->colormap != NULL) {
-    /* Select 2-pass quantizer for external colormap use */
-    cinfo->cquantize = master->quantizer_2pass;
-    /* Notify quantizer of colormap change */
-    (*cinfo->cquantize->new_color_map) (cinfo);
-    master->pub.is_dummy_pass = FALSE; /* just in case */
-  } else
-    ERREXIT(cinfo, JERR_MODE_CHANGE);
-}
-
-#endif /* D_MULTISCAN_FILES_SUPPORTED */
-
-
-/*
- * Initialize master decompression control and select active modules.
- * This is performed at the start of jpeg_start_decompress.
- */
-
-GLOBAL(void)
-jinit_master_decompress (j_decompress_ptr cinfo)
-{
-  my_master_ptr master;
-
-  master = (my_master_ptr)
-      (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				  SIZEOF(my_decomp_master));
-  cinfo->master = (struct jpeg_decomp_master *) master;
-  master->pub.prepare_for_output_pass = prepare_for_output_pass;
-  master->pub.finish_output_pass = finish_output_pass;
-
-  master->pub.is_dummy_pass = FALSE;
-
-  master_selection(cinfo);
-}
@@ -1,757 +0,0 @@
-/*
- * jdmerge.c
- *
- * Copyright (C) 1994-1996, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains code for merged upsampling/color conversion.
- *
- * This file combines functions from jdsample.c and jdcolor.c;
- * read those files first to understand what's going on.
- *
- * When the chroma components are to be upsampled by simple replication
- * (ie, box filtering), we can save some work in color conversion by
- * calculating all the output pixels corresponding to a pair of chroma
- * samples at one time.  In the conversion equations
- *	R = Y           + K1 * Cr
- *	G = Y + K2 * Cb + K3 * Cr
- *	B = Y + K4 * Cb
- * only the Y term varies among the group of pixels corresponding to a pair
- * of chroma samples, so the rest of the terms can be calculated just once.
- * At typical sampling ratios, this eliminates half or three-quarters of the
- * multiplications needed for color conversion.
- *
- * This file currently provides implementations for the following cases:
- *	YCbCr => RGB color conversion only.
- *	Sampling ratios of 2h1v or 2h2v.
- *	No scaling needed at upsample time.
- *	Corner-aligned (non-CCIR601) sampling alignment.
- * Other special cases could be added, but in most applications these are
- * the only common cases.  (For uncommon cases we fall back on the more
- * general code in jdsample.c and jdcolor.c.)
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-
-#ifdef UPSAMPLE_MERGING_SUPPORTED
-
-#ifdef ANDROID_RGB
-
-/* Declarations for ordered dithering.
- * 
- * We use 4x4 ordered dither array packed into 32 bits. This array is
- * sufficent for dithering RGB_888 to RGB_565.
- */
-
-#define DITHER_MASK         0x3
-#define DITHER_ROTATE(x)    (((x)<<24) | (((x)>>8)&0x00FFFFFF))
-static const INT32 dither_matrix[4] = {
-  0x0008020A,
-  0x0C040E06,
-  0x030B0109,
-  0x0F070D05
-};
-
-#endif
-
-/* Private subobject */
-
-typedef struct {
-  struct jpeg_upsampler pub;	/* public fields */
-
-  /* Pointer to routine to do actual upsampling/conversion of one row group */
-  JMETHOD(void, upmethod, (j_decompress_ptr cinfo,
-			   JSAMPIMAGE input_buf, JDIMENSION in_row_group_ctr,
-			   JSAMPARRAY output_buf));
-
-  /* Private state for YCC->RGB conversion */
-  int * Cr_r_tab;		/* => table for Cr to R conversion */
-  int * Cb_b_tab;		/* => table for Cb to B conversion */
-  INT32 * Cr_g_tab;		/* => table for Cr to G conversion */
-  INT32 * Cb_g_tab;		/* => table for Cb to G conversion */
-
-  /* For 2:1 vertical sampling, we produce two output rows at a time.
-   * We need a "spare" row buffer to hold the second output row if the
-   * application provides just a one-row buffer; we also use the spare
-   * to discard the dummy last row if the image height is odd.
-   */
-  JSAMPROW spare_row;
-  boolean spare_full;		/* T if spare buffer is occupied */
-
-  JDIMENSION out_row_width;	/* samples per output row */
-  JDIMENSION rows_to_go;	/* counts rows remaining in image */
-} my_upsampler;
-
-typedef my_upsampler * my_upsample_ptr;
-
-#define SCALEBITS	16	/* speediest right-shift on some machines */
-#define ONE_HALF	((INT32) 1 << (SCALEBITS-1))
-#define FIX(x)		((INT32) ((x) * (1L<<SCALEBITS) + 0.5))
-
-
-/*
- * Initialize tables for YCC->RGB colorspace conversion.
- * This is taken directly from jdcolor.c; see that file for more info.
- */
-
-LOCAL(void)
-build_ycc_rgb_table (j_decompress_ptr cinfo)
-{
-  my_upsample_ptr upsample = (my_upsample_ptr) cinfo->upsample;
-  int i;
-  INT32 x;
-  SHIFT_TEMPS
-
-  upsample->Cr_r_tab = (int *)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				(MAXJSAMPLE+1) * SIZEOF(int));
-  upsample->Cb_b_tab = (int *)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				(MAXJSAMPLE+1) * SIZEOF(int));
-  upsample->Cr_g_tab = (INT32 *)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				(MAXJSAMPLE+1) * SIZEOF(INT32));
-  upsample->Cb_g_tab = (INT32 *)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				(MAXJSAMPLE+1) * SIZEOF(INT32));
-
-  for (i = 0, x = -CENTERJSAMPLE; i <= MAXJSAMPLE; i++, x++) {
-    /* i is the actual input pixel value, in the range 0..MAXJSAMPLE */
-    /* The Cb or Cr value we are thinking of is x = i - CENTERJSAMPLE */
-    /* Cr=>R value is nearest int to 1.40200 * x */
-    upsample->Cr_r_tab[i] = (int)
-		    RIGHT_SHIFT(FIX(1.40200) * x + ONE_HALF, SCALEBITS);
-    /* Cb=>B value is nearest int to 1.77200 * x */
-    upsample->Cb_b_tab[i] = (int)
-		    RIGHT_SHIFT(FIX(1.77200) * x + ONE_HALF, SCALEBITS);
-    /* Cr=>G value is scaled-up -0.71414 * x */
-    upsample->Cr_g_tab[i] = (- FIX(0.71414)) * x;
-    /* Cb=>G value is scaled-up -0.34414 * x */
-    /* We also add in ONE_HALF so that need not do it in inner loop */
-    upsample->Cb_g_tab[i] = (- FIX(0.34414)) * x + ONE_HALF;
-  }
-}
-
-
-/*
- * Initialize for an upsampling pass.
- */
-
-METHODDEF(void)
-start_pass_merged_upsample (j_decompress_ptr cinfo)
-{
-  my_upsample_ptr upsample = (my_upsample_ptr) cinfo->upsample;
-
-  /* Mark the spare buffer empty */
-  upsample->spare_full = FALSE;
-  /* Initialize total-height counter for detecting bottom of image */
-  upsample->rows_to_go = cinfo->output_height;
-}
-
-
-/*
- * Control routine to do upsampling (and color conversion).
- *
- * The control routine just handles the row buffering considerations.
- */
-
-METHODDEF(void)
-merged_2v_upsample (j_decompress_ptr cinfo,
-		    JSAMPIMAGE input_buf, JDIMENSION *in_row_group_ctr,
-		    JDIMENSION in_row_groups_avail,
-		    JSAMPARRAY output_buf, JDIMENSION *out_row_ctr,
-		    JDIMENSION out_rows_avail)
-/* 2:1 vertical sampling case: may need a spare row. */
-{
-  my_upsample_ptr upsample = (my_upsample_ptr) cinfo->upsample;
-  JSAMPROW work_ptrs[2];
-  JDIMENSION num_rows;		/* number of rows returned to caller */
-
-  if (upsample->spare_full) {
-    /* If we have a spare row saved from a previous cycle, just return it. */
-      JDIMENSION size = upsample->out_row_width;
-#ifdef ANDROID_RGB
-    if (cinfo->out_color_space == JCS_RGB_565)
-      size = cinfo->output_width*2;
-#endif
-    jcopy_sample_rows(& upsample->spare_row, 0, output_buf + *out_row_ctr, 0,
-		      1, size);
-
-    num_rows = 1;
-    upsample->spare_full = FALSE;
-  } else {
-    /* Figure number of rows to return to caller. */
-    num_rows = 2;
-    /* Not more than the distance to the end of the image. */
-    if (num_rows > upsample->rows_to_go)
-      num_rows = upsample->rows_to_go;
-    /* And not more than what the client can accept: */
-    out_rows_avail -= *out_row_ctr;
-    if (num_rows > out_rows_avail)
-      num_rows = out_rows_avail;
-    /* Create output pointer array for upsampler. */
-    work_ptrs[0] = output_buf[*out_row_ctr];
-    if (num_rows > 1) {
-      work_ptrs[1] = output_buf[*out_row_ctr + 1];
-    } else {
-      work_ptrs[1] = upsample->spare_row;
-      upsample->spare_full = TRUE;
-    }
-    /* Now do the upsampling. */
-    (*upsample->upmethod) (cinfo, input_buf, *in_row_group_ctr, work_ptrs);
-  }
-
-  /* Adjust counts */
-  *out_row_ctr += num_rows;
-  upsample->rows_to_go -= num_rows;
-  /* When the buffer is emptied, declare this input row group consumed */
-  if (! upsample->spare_full)
-    (*in_row_group_ctr)++;
-}
-
-
-METHODDEF(void)
-merged_1v_upsample (j_decompress_ptr cinfo,
-		    JSAMPIMAGE input_buf, JDIMENSION *in_row_group_ctr,
-		    JDIMENSION in_row_groups_avail,
-		    JSAMPARRAY output_buf, JDIMENSION *out_row_ctr,
-		    JDIMENSION out_rows_avail)
-/* 1:1 vertical sampling case: much easier, never need a spare row. */
-{
-  my_upsample_ptr upsample = (my_upsample_ptr) cinfo->upsample;
-
-  /* Just do the upsampling. */
-  (*upsample->upmethod) (cinfo, input_buf, *in_row_group_ctr,
-			 output_buf + *out_row_ctr);
-  /* Adjust counts */
-  (*out_row_ctr)++;
-  (*in_row_group_ctr)++;
-}
-
-
-/*
- * These are the routines invoked by the control routines to do
- * the actual upsampling/conversion.  One row group is processed per call.
- *
- * Note: since we may be writing directly into application-supplied buffers,
- * we have to be honest about the output width; we can't assume the buffer
- * has been rounded up to an even width.
- */
-
-
-/*
- * Upsample and color convert for the case of 2:1 horizontal and 1:1 vertical.
- */
-
-METHODDEF(void)
-h2v1_merged_upsample (j_decompress_ptr cinfo,
-		      JSAMPIMAGE input_buf, JDIMENSION in_row_group_ctr,
-		      JSAMPARRAY output_buf)
-{
-  my_upsample_ptr upsample = (my_upsample_ptr) cinfo->upsample;
-  register int y, cred, cgreen, cblue;
-  int cb, cr;
-  register JSAMPROW outptr;
-  JSAMPROW inptr0, inptr1, inptr2;
-  JDIMENSION col;
-  /* copy these pointers into registers if possible */
-  register JSAMPLE * range_limit = cinfo->sample_range_limit;
-  int * Crrtab = upsample->Cr_r_tab;
-  int * Cbbtab = upsample->Cb_b_tab;
-  INT32 * Crgtab = upsample->Cr_g_tab;
-  INT32 * Cbgtab = upsample->Cb_g_tab;
-  SHIFT_TEMPS
-
-  inptr0 = input_buf[0][in_row_group_ctr];
-  inptr1 = input_buf[1][in_row_group_ctr];
-  inptr2 = input_buf[2][in_row_group_ctr];
-  outptr = output_buf[0];
-  /* Loop for each pair of output pixels */
-  for (col = cinfo->output_width >> 1; col > 0; col--) {
-    /* Do the chroma part of the calculation */
-    cb = GETJSAMPLE(*inptr1++);
-    cr = GETJSAMPLE(*inptr2++);
-    cred = Crrtab[cr];
-    cgreen = (int) RIGHT_SHIFT(Cbgtab[cb] + Crgtab[cr], SCALEBITS);
-    cblue = Cbbtab[cb];
-    /* Fetch 2 Y values and emit 2 pixels */
-    y  = GETJSAMPLE(*inptr0++);
-    outptr[RGB_RED] = range_limit[y + cred];
-    outptr[RGB_GREEN] = range_limit[y + cgreen];
-    outptr[RGB_BLUE] = range_limit[y + cblue];
-    outptr += RGB_PIXELSIZE;
-    y  = GETJSAMPLE(*inptr0++);
-    outptr[RGB_RED] = range_limit[y + cred];
-    outptr[RGB_GREEN] = range_limit[y + cgreen];
-    outptr[RGB_BLUE] = range_limit[y + cblue];
-    outptr += RGB_PIXELSIZE;
-  }
-  /* If image width is odd, do the last output column separately */
-  if (cinfo->output_width & 1) {
-    cb = GETJSAMPLE(*inptr1);
-    cr = GETJSAMPLE(*inptr2);
-    cred = Crrtab[cr];
-    cgreen = (int) RIGHT_SHIFT(Cbgtab[cb] + Crgtab[cr], SCALEBITS);
-    cblue = Cbbtab[cb];
-    y  = GETJSAMPLE(*inptr0);
-    outptr[RGB_RED] = range_limit[y + cred];
-    outptr[RGB_GREEN] = range_limit[y + cgreen];
-    outptr[RGB_BLUE] = range_limit[y + cblue];
-  }
-}
-
-
-#ifdef ANDROID_RGB
-METHODDEF(void)
-h2v1_merged_upsample_565 (j_decompress_ptr cinfo,
-              JSAMPIMAGE input_buf, JDIMENSION in_row_group_ctr,
-              JSAMPARRAY output_buf)
-{
-  my_upsample_ptr upsample = (my_upsample_ptr) cinfo->upsample;
-  register int y, cred, cgreen, cblue;
-  int cb, cr;
-  register JSAMPROW outptr;
-  JSAMPROW inptr0, inptr1, inptr2;
-  JDIMENSION col;
-  /* copy these pointers into registers if possible */
-  register JSAMPLE * range_limit = cinfo->sample_range_limit;
-  int * Crrtab = upsample->Cr_r_tab;
-  int * Cbbtab = upsample->Cb_b_tab;
-  INT32 * Crgtab = upsample->Cr_g_tab;
-  INT32 * Cbgtab = upsample->Cb_g_tab;
-  unsigned int r, g, b;
-  INT32 rgb;
-  SHIFT_TEMPS
-
-  inptr0 = input_buf[0][in_row_group_ctr];
-  inptr1 = input_buf[1][in_row_group_ctr];
-  inptr2 = input_buf[2][in_row_group_ctr];
-  outptr = output_buf[0];
-  /* Loop for each pair of output pixels */
-  for (col = cinfo->output_width >> 1; col > 0; col--) {
-    /* Do the chroma part of the calculation */
-    cb = GETJSAMPLE(*inptr1++);
-    cr = GETJSAMPLE(*inptr2++);
-    cred = Crrtab[cr];
-    cgreen = (int) RIGHT_SHIFT(Cbgtab[cb] + Crgtab[cr], SCALEBITS);
-    cblue = Cbbtab[cb];
-    /* Fetch 2 Y values and emit 2 pixels */
-    y  = GETJSAMPLE(*inptr0++);
-    r = range_limit[y + cred];
-    g = range_limit[y + cgreen];
-    b = range_limit[y + cblue];
-    rgb = PACK_SHORT_565(r,g,b);
-    y  = GETJSAMPLE(*inptr0++);
-    r = range_limit[y + cred];
-    g = range_limit[y + cgreen];
-    b = range_limit[y + cblue];
-    rgb = PACK_TWO_PIXELS(rgb, PACK_SHORT_565(r,g,b));
-    WRITE_TWO_PIXELS(outptr, rgb);
-    outptr += 4;
-  }
-  /* If image width is odd, do the last output column separately */
-  if (cinfo->output_width & 1) {
-    cb = GETJSAMPLE(*inptr1);
-    cr = GETJSAMPLE(*inptr2);
-    cred = Crrtab[cr];
-    cgreen = (int) RIGHT_SHIFT(Cbgtab[cb] + Crgtab[cr], SCALEBITS);
-    cblue = Cbbtab[cb];
-    y  = GETJSAMPLE(*inptr0);
-    r = range_limit[y + cred];
-    g = range_limit[y + cgreen];
-    b = range_limit[y + cblue];
-    rgb = PACK_SHORT_565(r,g,b);
-    *(INT16*)outptr = rgb;
-  }
-}
-
-
-METHODDEF(void)
-h2v1_merged_upsample_565D (j_decompress_ptr cinfo,
-              JSAMPIMAGE input_buf, JDIMENSION in_row_group_ctr,
-              JSAMPARRAY output_buf)
-{
-  my_upsample_ptr upsample = (my_upsample_ptr) cinfo->upsample;
-  register int y, cred, cgreen, cblue;
-  int cb, cr;
-  register JSAMPROW outptr;
-  JSAMPROW inptr0, inptr1, inptr2;
-  JDIMENSION col;
-  /* copy these pointers into registers if possible */
-  register JSAMPLE * range_limit = cinfo->sample_range_limit;
-  int * Crrtab = upsample->Cr_r_tab;
-  int * Cbbtab = upsample->Cb_b_tab;
-  INT32 * Crgtab = upsample->Cr_g_tab;
-  INT32 * Cbgtab = upsample->Cb_g_tab;
-  JDIMENSION col_index = 0;
-  INT32 d0 = dither_matrix[cinfo->output_scanline & DITHER_MASK];
-  unsigned int r, g, b;
-  INT32 rgb;
-  SHIFT_TEMPS
-
-  inptr0 = input_buf[0][in_row_group_ctr];
-  inptr1 = input_buf[1][in_row_group_ctr];
-  inptr2 = input_buf[2][in_row_group_ctr];
-  outptr = output_buf[0];
-  /* Loop for each pair of output pixels */
-  for (col = cinfo->output_width >> 1; col > 0; col--) {
-    /* Do the chroma part of the calculation */
-    cb = GETJSAMPLE(*inptr1++);
-    cr = GETJSAMPLE(*inptr2++);
-    cred = Crrtab[cr];
-    cgreen = (int) RIGHT_SHIFT(Cbgtab[cb] + Crgtab[cr], SCALEBITS);
-    cblue = Cbbtab[cb];
-    /* Fetch 2 Y values and emit 2 pixels */
-    y  = GETJSAMPLE(*inptr0++);
-    r = range_limit[DITHER_565_R(y + cred, d0)];
-    g = range_limit[DITHER_565_G(y + cgreen, d0)];
-    b = range_limit[DITHER_565_B(y + cblue, d0)];
-    d0 = DITHER_ROTATE(d0);
-    rgb = PACK_SHORT_565(r,g,b);
-    y  = GETJSAMPLE(*inptr0++);
-    r = range_limit[DITHER_565_R(y + cred, d0)];
-    g = range_limit[DITHER_565_G(y + cgreen, d0)];
-    b = range_limit[DITHER_565_B(y + cblue, d0)];
-    d0 = DITHER_ROTATE(d0);
-    rgb = PACK_TWO_PIXELS(rgb, PACK_SHORT_565(r,g,b));
-    WRITE_TWO_PIXELS(outptr, rgb);
-    outptr += 4;
-  }
-  /* If image width is odd, do the last output column separately */
-  if (cinfo->output_width & 1) {
-    cb = GETJSAMPLE(*inptr1);
-    cr = GETJSAMPLE(*inptr2);
-    cred = Crrtab[cr];
-    cgreen = (int) RIGHT_SHIFT(Cbgtab[cb] + Crgtab[cr], SCALEBITS);
-    cblue = Cbbtab[cb];
-    y  = GETJSAMPLE(*inptr0);
-    r = range_limit[DITHER_565_R(y + cred, d0)];
-    g = range_limit[DITHER_565_G(y + cgreen, d0)];
-    b = range_limit[DITHER_565_B(y + cblue, d0)];
-    rgb = PACK_SHORT_565(r,g,b);
-    *(INT16*)outptr = rgb;
-  }
-}
-
-
-#endif
-
-/*
- * Upsample and color convert for the case of 2:1 horizontal and 2:1 vertical.
- */
-
-METHODDEF(void)
-h2v2_merged_upsample (j_decompress_ptr cinfo,
-		      JSAMPIMAGE input_buf, JDIMENSION in_row_group_ctr,
-		      JSAMPARRAY output_buf)
-{
-  my_upsample_ptr upsample = (my_upsample_ptr) cinfo->upsample;
-  register int y, cred, cgreen, cblue;
-  int cb, cr;
-  register JSAMPROW outptr0, outptr1;
-  JSAMPROW inptr00, inptr01, inptr1, inptr2;
-  JDIMENSION col;
-  /* copy these pointers into registers if possible */
-  register JSAMPLE * range_limit = cinfo->sample_range_limit;
-  int * Crrtab = upsample->Cr_r_tab;
-  int * Cbbtab = upsample->Cb_b_tab;
-  INT32 * Crgtab = upsample->Cr_g_tab;
-  INT32 * Cbgtab = upsample->Cb_g_tab;
-  SHIFT_TEMPS
-
-  inptr00 = input_buf[0][in_row_group_ctr*2];
-  inptr01 = input_buf[0][in_row_group_ctr*2 + 1];
-  inptr1 = input_buf[1][in_row_group_ctr];
-  inptr2 = input_buf[2][in_row_group_ctr];
-  outptr0 = output_buf[0];
-  outptr1 = output_buf[1];
-  /* Loop for each group of output pixels */
-  for (col = cinfo->output_width >> 1; col > 0; col--) {
-    /* Do the chroma part of the calculation */
-    cb = GETJSAMPLE(*inptr1++);
-    cr = GETJSAMPLE(*inptr2++);
-    cred = Crrtab[cr];
-    cgreen = (int) RIGHT_SHIFT(Cbgtab[cb] + Crgtab[cr], SCALEBITS);
-    cblue = Cbbtab[cb];
-    /* Fetch 4 Y values and emit 4 pixels */
-    y  = GETJSAMPLE(*inptr00++);
-    outptr0[RGB_RED] = range_limit[y + cred];
-    outptr0[RGB_GREEN] = range_limit[y + cgreen];
-    outptr0[RGB_BLUE] = range_limit[y + cblue];
-    outptr0 += RGB_PIXELSIZE;
-    y  = GETJSAMPLE(*inptr00++);
-    outptr0[RGB_RED] = range_limit[y + cred];
-    outptr0[RGB_GREEN] = range_limit[y + cgreen];
-    outptr0[RGB_BLUE] = range_limit[y + cblue];
-    outptr0 += RGB_PIXELSIZE;
-    y  = GETJSAMPLE(*inptr01++);
-    outptr1[RGB_RED] = range_limit[y + cred];
-    outptr1[RGB_GREEN] = range_limit[y + cgreen];
-    outptr1[RGB_BLUE] = range_limit[y + cblue];
-    outptr1 += RGB_PIXELSIZE;
-    y  = GETJSAMPLE(*inptr01++);
-    outptr1[RGB_RED] = range_limit[y + cred];
-    outptr1[RGB_GREEN] = range_limit[y + cgreen];
-    outptr1[RGB_BLUE] = range_limit[y + cblue];
-    outptr1 += RGB_PIXELSIZE;
-  }
-  /* If image width is odd, do the last output column separately */
-  if (cinfo->output_width & 1) {
-    cb = GETJSAMPLE(*inptr1);
-    cr = GETJSAMPLE(*inptr2);
-    cred = Crrtab[cr];
-    cgreen = (int) RIGHT_SHIFT(Cbgtab[cb] + Crgtab[cr], SCALEBITS);
-    cblue = Cbbtab[cb];
-    y  = GETJSAMPLE(*inptr00);
-    outptr0[RGB_RED] = range_limit[y + cred];
-    outptr0[RGB_GREEN] = range_limit[y + cgreen];
-    outptr0[RGB_BLUE] = range_limit[y + cblue];
-    y  = GETJSAMPLE(*inptr01);
-    outptr1[RGB_RED] = range_limit[y + cred];
-    outptr1[RGB_GREEN] = range_limit[y + cgreen];
-    outptr1[RGB_BLUE] = range_limit[y + cblue];
-  }
-}
-
-
-#ifdef ANDROID_RGB
-
-METHODDEF(void)
-h2v2_merged_upsample_565 (j_decompress_ptr cinfo,
-              JSAMPIMAGE input_buf, JDIMENSION in_row_group_ctr,
-              JSAMPARRAY output_buf)
-{
-  my_upsample_ptr upsample = (my_upsample_ptr) cinfo->upsample;
-  register int y, cred, cgreen, cblue;
-  int cb, cr;
-  register JSAMPROW outptr0, outptr1;
-  JSAMPROW inptr00, inptr01, inptr1, inptr2;
-  JDIMENSION col;
-  /* copy these pointers into registers if possible */
-  register JSAMPLE * range_limit = cinfo->sample_range_limit;
-  int * Crrtab = upsample->Cr_r_tab;
-  int * Cbbtab = upsample->Cb_b_tab;
-  INT32 * Crgtab = upsample->Cr_g_tab;
-  INT32 * Cbgtab = upsample->Cb_g_tab;
-  unsigned int r, g, b;
-  INT32 rgb;
-  SHIFT_TEMPS
-
-  inptr00 = input_buf[0][in_row_group_ctr*2];
-  inptr01 = input_buf[0][in_row_group_ctr*2 + 1];
-  inptr1 = input_buf[1][in_row_group_ctr];
-  inptr2 = input_buf[2][in_row_group_ctr];
-  outptr0 = output_buf[0];
-  outptr1 = output_buf[1];
-  /* Loop for each group of output pixels */
-  for (col = cinfo->output_width >> 1; col > 0; col--) {
-    /* Do the chroma part of the calculation */
-    cb = GETJSAMPLE(*inptr1++);
-    cr = GETJSAMPLE(*inptr2++);
-    cred = Crrtab[cr];
-    cgreen = (int) RIGHT_SHIFT(Cbgtab[cb] + Crgtab[cr], SCALEBITS);
-    cblue = Cbbtab[cb];
-    /* Fetch 4 Y values and emit 4 pixels */
-    y  = GETJSAMPLE(*inptr00++);
-    r = range_limit[y + cred];
-    g = range_limit[y + cgreen];
-    b = range_limit[y + cblue];
-    rgb = PACK_SHORT_565(r,g,b);
-    y  = GETJSAMPLE(*inptr00++);
-    r = range_limit[y + cred];
-    g = range_limit[y + cgreen];
-    b = range_limit[y + cblue];
-    rgb = PACK_TWO_PIXELS(rgb, PACK_SHORT_565(r,g,b));
-    WRITE_TWO_PIXELS(outptr0, rgb);
-    outptr0 += 4;
-    y  = GETJSAMPLE(*inptr01++);
-    r = range_limit[y + cred];
-    g = range_limit[y + cgreen];
-    b = range_limit[y + cblue];
-    rgb = PACK_SHORT_565(r,g,b);
-    y  = GETJSAMPLE(*inptr01++);
-    r = range_limit[y + cred];
-    g = range_limit[y + cgreen];
-    b = range_limit[y + cblue];
-    rgb = PACK_TWO_PIXELS(rgb, PACK_SHORT_565(r,g,b));
-    WRITE_TWO_PIXELS(outptr1, rgb);
-    outptr1 += 4;
-  }
-  /* If image width is odd, do the last output column separately */
-  if (cinfo->output_width & 1) {
-    cb = GETJSAMPLE(*inptr1);
-    cr = GETJSAMPLE(*inptr2);
-    cred = Crrtab[cr];
-    cgreen = (int) RIGHT_SHIFT(Cbgtab[cb] + Crgtab[cr], SCALEBITS);
-    cblue = Cbbtab[cb];
-    y  = GETJSAMPLE(*inptr00);
-    r = range_limit[y + cred];
-    g = range_limit[y + cgreen];
-    b = range_limit[y + cblue];
-    rgb = PACK_SHORT_565(r,g,b);
-    *(INT16*)outptr0 = rgb;
-   y  = GETJSAMPLE(*inptr01);
-   r = range_limit[y + cred];
-   g = range_limit[y + cgreen];
-   b = range_limit[y + cblue];
-   rgb = PACK_SHORT_565(r,g,b);
-   *(INT16*)outptr1 = rgb;
-  }
-}
-
-
-
-METHODDEF(void)
-h2v2_merged_upsample_565D (j_decompress_ptr cinfo,
-              JSAMPIMAGE input_buf, JDIMENSION in_row_group_ctr,
-              JSAMPARRAY output_buf)
-{
-  my_upsample_ptr upsample = (my_upsample_ptr) cinfo->upsample;
-  register int y, cred, cgreen, cblue;
-  int cb, cr;
-  register JSAMPROW outptr0, outptr1;
-  JSAMPROW inptr00, inptr01, inptr1, inptr2;
-  JDIMENSION col;
-  /* copy these pointers into registers if possible */
-  register JSAMPLE * range_limit = cinfo->sample_range_limit;
-  int * Crrtab = upsample->Cr_r_tab;
-  int * Cbbtab = upsample->Cb_b_tab;
-  INT32 * Crgtab = upsample->Cr_g_tab;
-  INT32 * Cbgtab = upsample->Cb_g_tab;
-  JDIMENSION col_index = 0;
-  INT32 d0 = dither_matrix[cinfo->output_scanline & DITHER_MASK];
-  INT32 d1 = dither_matrix[(cinfo->output_scanline+1) & DITHER_MASK];
-  unsigned int r, g, b;
-  INT32 rgb;
-  SHIFT_TEMPS
-
-  inptr00 = input_buf[0][in_row_group_ctr*2];
-  inptr01 = input_buf[0][in_row_group_ctr*2 + 1];
-  inptr1 = input_buf[1][in_row_group_ctr];
-  inptr2 = input_buf[2][in_row_group_ctr];
-  outptr0 = output_buf[0];
-  outptr1 = output_buf[1];
-  /* Loop for each group of output pixels */
-  for (col = cinfo->output_width >> 1; col > 0; col--) {
-    
-    /* Do the chroma part of the calculation */
-    cb = GETJSAMPLE(*inptr1++);
-    cr = GETJSAMPLE(*inptr2++);
-    cred = Crrtab[cr];
-    cgreen = (int) RIGHT_SHIFT(Cbgtab[cb] + Crgtab[cr], SCALEBITS);
-    cblue = Cbbtab[cb];
-    /* Fetch 4 Y values and emit 4 pixels */    
-    y  = GETJSAMPLE(*inptr00++);
-    r = range_limit[DITHER_565_R(y + cred, d0)];
-    g = range_limit[DITHER_565_G(y + cgreen, d0)];
-    b = range_limit[DITHER_565_B(y + cblue, d0)];
-    d0 = DITHER_ROTATE(d0);
-    rgb = PACK_SHORT_565(r,g,b);
-    y  = GETJSAMPLE(*inptr00++);
-    r = range_limit[DITHER_565_R(y + cred, d1)];
-    g = range_limit[DITHER_565_G(y + cgreen, d1)];
-    b = range_limit[DITHER_565_B(y + cblue, d1)];
-    d1 = DITHER_ROTATE(d1);
-    rgb = PACK_TWO_PIXELS(rgb, PACK_SHORT_565(r,g,b));
-    WRITE_TWO_PIXELS(outptr0, rgb);
-    outptr0 += 4;
-    y  = GETJSAMPLE(*inptr01++);
-    r = range_limit[DITHER_565_R(y + cred, d0)];
-    g = range_limit[DITHER_565_G(y + cgreen, d0)];
-    b = range_limit[DITHER_565_B(y + cblue, d0)];
-    d0 = DITHER_ROTATE(d0);
-    rgb = PACK_SHORT_565(r,g,b);
-    y  = GETJSAMPLE(*inptr01++);
-    r = range_limit[DITHER_565_R(y + cred, d1)];
-    g = range_limit[DITHER_565_G(y + cgreen, d1)];
-    b = range_limit[DITHER_565_B(y + cblue, d1)];
-    d1 = DITHER_ROTATE(d1);
-    rgb = PACK_TWO_PIXELS(rgb, PACK_SHORT_565(r,g,b));
-    WRITE_TWO_PIXELS(outptr1, rgb);
-    outptr1 += 4;
-  }
-  /* If image width is odd, do the last output column separately */
-  if (cinfo->output_width & 1) {
-    cb = GETJSAMPLE(*inptr1);
-    cr = GETJSAMPLE(*inptr2);
-    cred = Crrtab[cr];
-    cgreen = (int) RIGHT_SHIFT(Cbgtab[cb] + Crgtab[cr], SCALEBITS);
-    cblue = Cbbtab[cb];
-    y  = GETJSAMPLE(*inptr00);
-    r = range_limit[DITHER_565_R(y + cred, d0)];
-    g = range_limit[DITHER_565_G(y + cgreen, d0)];
-    b = range_limit[DITHER_565_B(y + cblue, d0)];
-    rgb = PACK_SHORT_565(r,g,b);
-    *(INT16*)outptr0 = rgb;
-   y  = GETJSAMPLE(*inptr01);
-   r = range_limit[DITHER_565_R(y + cred, d1)];
-   g = range_limit[DITHER_565_G(y + cgreen, d1)];
-   b = range_limit[DITHER_565_B(y + cblue, d1)];
-   rgb = PACK_SHORT_565(r,g,b);
-   *(INT16*)outptr1 = rgb;
-  }
-}
-
-#endif
-
-/*
- * Module initialization routine for merged upsampling/color conversion.
- *
- * NB: this is called under the conditions determined by use_merged_upsample()
- * in jdmaster.c.  That routine MUST correspond to the actual capabilities
- * of this module; no safety checks are made here.
- */
-
-GLOBAL(void)
-jinit_merged_upsampler (j_decompress_ptr cinfo)
-{
-  my_upsample_ptr upsample;
-
-  upsample = (my_upsample_ptr)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				SIZEOF(my_upsampler));
-  cinfo->upsample = (struct jpeg_upsampler *) upsample;
-  upsample->pub.start_pass = start_pass_merged_upsample;
-  upsample->pub.need_context_rows = FALSE;
-
-  upsample->out_row_width = cinfo->output_width * cinfo->out_color_components;
-  
-  if (cinfo->max_v_samp_factor == 2) {
-    upsample->pub.upsample = merged_2v_upsample;
-    upsample->upmethod = h2v2_merged_upsample;
-#ifdef ANDROID_RGB
-    if (cinfo->out_color_space == JCS_RGB_565) {
-        if (cinfo->dither_mode == JDITHER_NONE) {
-            upsample->upmethod = h2v2_merged_upsample_565;
-        } else {
-            upsample->upmethod = h2v2_merged_upsample_565D;
-        }
-    }
-#endif
-    /* Allocate a spare row buffer */
-    upsample->spare_row = (JSAMPROW)
-      (*cinfo->mem->alloc_large) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-		(size_t) (upsample->out_row_width * SIZEOF(JSAMPLE)));
-  } else {
-    upsample->pub.upsample = merged_1v_upsample;
-    upsample->upmethod = h2v1_merged_upsample;
-#ifdef ANDROID_RGB
-    if (cinfo->out_color_space == JCS_RGB_565) {
-        if (cinfo->dither_mode == JDITHER_NONE) {
-            upsample->upmethod = h2v1_merged_upsample_565;
-        } else {
-            upsample->upmethod = h2v1_merged_upsample_565D;
-        }
-    }
-#endif
-    /* No spare row needed */
-    upsample->spare_row = NULL;
-  }
-
-  build_ycc_rgb_table(cinfo);
-}
-
-#endif /* UPSAMPLE_MERGING_SUPPORTED */
@@ -1,770 +0,0 @@
-/*
- * jdphuff.c
- *
- * Copyright (C) 1995-1997, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains Huffman entropy decoding routines for progressive JPEG.
- *
- * Much of the complexity here has to do with supporting input suspension.
- * If the data source module demands suspension, we want to be able to back
- * up to the start of the current MCU.  To do this, we copy state variables
- * into local working storage, and update them back to the permanent
- * storage only upon successful completion of an MCU.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-#include "jdhuff.h"		/* Declarations shared with jdhuff.c */
-
-
-#ifdef D_PROGRESSIVE_SUPPORTED
-
-/*
- * Expanded entropy decoder object for progressive Huffman decoding.
- *
- * The savable_state subrecord contains fields that change within an MCU,
- * but must not be updated permanently until we complete the MCU.
- */
-
-typedef struct {
-  unsigned int EOBRUN;			/* remaining EOBs in EOBRUN */
-  int last_dc_val[MAX_COMPS_IN_SCAN];	/* last DC coef for each component */
-} savable_state;
-
-/* This macro is to work around compilers with missing or broken
- * structure assignment.  You'll need to fix this code if you have
- * such a compiler and you change MAX_COMPS_IN_SCAN.
- */
-
-#ifndef NO_STRUCT_ASSIGN
-#define ASSIGN_STATE(dest,src)  ((dest) = (src))
-#else
-#if MAX_COMPS_IN_SCAN == 4
-#define ASSIGN_STATE(dest,src)  \
-	((dest).EOBRUN = (src).EOBRUN, \
-	 (dest).last_dc_val[0] = (src).last_dc_val[0], \
-	 (dest).last_dc_val[1] = (src).last_dc_val[1], \
-	 (dest).last_dc_val[2] = (src).last_dc_val[2], \
-	 (dest).last_dc_val[3] = (src).last_dc_val[3])
-#endif
-#endif
-
-
-typedef struct {
-  struct jpeg_entropy_decoder pub; /* public fields */
-
-  /* These fields are loaded into local variables at start of each MCU.
-   * In case of suspension, we exit WITHOUT updating them.
-   */
-  bitread_perm_state bitstate;	/* Bit buffer at start of MCU */
-  savable_state saved;		/* Other state at start of MCU */
-
-  /* These fields are NOT loaded into local working state. */
-  unsigned int restarts_to_go;	/* MCUs left in this restart interval */
-
-  /* Pointers to derived tables (these workspaces have image lifespan) */
-  d_derived_tbl * derived_tbls[NUM_HUFF_TBLS];
-
-  d_derived_tbl * ac_derived_tbl; /* active table during an AC scan */
-} phuff_entropy_decoder;
-
-typedef phuff_entropy_decoder * phuff_entropy_ptr;
-
-/* Forward declarations */
-METHODDEF(boolean) decode_mcu_DC_first JPP((j_decompress_ptr cinfo,
-					    JBLOCKROW *MCU_data));
-METHODDEF(boolean) decode_mcu_AC_first JPP((j_decompress_ptr cinfo,
-					    JBLOCKROW *MCU_data));
-METHODDEF(boolean) decode_mcu_DC_refine JPP((j_decompress_ptr cinfo,
-					     JBLOCKROW *MCU_data));
-METHODDEF(boolean) decode_mcu_AC_refine JPP((j_decompress_ptr cinfo,
-					     JBLOCKROW *MCU_data));
-
-/*
- * Initialize for a Huffman-compressed scan.
- */
-
-METHODDEF(void)
-start_pass_phuff_decoder (j_decompress_ptr cinfo)
-{
-  phuff_entropy_ptr entropy = (phuff_entropy_ptr) cinfo->entropy;
-  boolean is_DC_band, bad;
-  int ci, coefi, tbl;
-  int *coef_bit_ptr;
-  jpeg_component_info * compptr;
-
-  is_DC_band = (cinfo->Ss == 0);
-
-  /* Validate scan parameters */
-  bad = FALSE;
-  if (is_DC_band) {
-    if (cinfo->Se != 0)
-      bad = TRUE;
-  } else {
-    /* need not check Ss/Se < 0 since they came from unsigned bytes */
-    if (cinfo->Ss > cinfo->Se || cinfo->Se >= DCTSIZE2)
-      bad = TRUE;
-    /* AC scans may have only one component */
-    if (cinfo->comps_in_scan != 1)
-      bad = TRUE;
-  }
-  if (cinfo->Ah != 0) {
-    /* Successive approximation refinement scan: must have Al = Ah-1. */
-    if (cinfo->Al != cinfo->Ah-1)
-      bad = TRUE;
-  }
-  if (cinfo->Al > 13)		/* need not check for < 0 */
-    bad = TRUE;
-  /* Arguably the maximum Al value should be less than 13 for 8-bit precision,
-   * but the spec doesn't say so, and we try to be liberal about what we
-   * accept.  Note: large Al values could result in out-of-range DC
-   * coefficients during early scans, leading to bizarre displays due to
-   * overflows in the IDCT math.  But we won't crash.
-   */
-  if (bad)
-    ERREXIT4(cinfo, JERR_BAD_PROGRESSION,
-	     cinfo->Ss, cinfo->Se, cinfo->Ah, cinfo->Al);
-  /* Update progression status, and verify that scan order is legal.
-   * Note that inter-scan inconsistencies are treated as warnings
-   * not fatal errors ... not clear if this is right way to behave.
-   */
-  for (ci = 0; ci < cinfo->comps_in_scan; ci++) {
-    int cindex = cinfo->cur_comp_info[ci]->component_index;
-    coef_bit_ptr = & cinfo->coef_bits[cindex][0];
-    if (!is_DC_band && coef_bit_ptr[0] < 0) /* AC without prior DC scan */
-      WARNMS2(cinfo, JWRN_BOGUS_PROGRESSION, cindex, 0);
-    for (coefi = cinfo->Ss; coefi <= cinfo->Se; coefi++) {
-      int expected = (coef_bit_ptr[coefi] < 0) ? 0 : coef_bit_ptr[coefi];
-      if (cinfo->Ah != expected)
-	WARNMS2(cinfo, JWRN_BOGUS_PROGRESSION, cindex, coefi);
-      coef_bit_ptr[coefi] = cinfo->Al;
-    }
-  }
-
-  /* Select MCU decoding routine */
-  if (cinfo->Ah == 0) {
-    if (is_DC_band)
-      entropy->pub.decode_mcu = decode_mcu_DC_first;
-    else
-      entropy->pub.decode_mcu = decode_mcu_AC_first;
-  } else {
-    if (is_DC_band)
-      entropy->pub.decode_mcu = decode_mcu_DC_refine;
-    else
-      entropy->pub.decode_mcu = decode_mcu_AC_refine;
-  }
-
-  for (ci = 0; ci < cinfo->comps_in_scan; ci++) {
-    compptr = cinfo->cur_comp_info[ci];
-    /* Make sure requested tables are present, and compute derived tables.
-     * We may build same derived table more than once, but it's not expensive.
-     */
-    if (is_DC_band) {
-      if (cinfo->Ah == 0) {	/* DC refinement needs no table */
-	tbl = compptr->dc_tbl_no;
-	jpeg_make_d_derived_tbl(cinfo, TRUE, tbl,
-				& entropy->derived_tbls[tbl]);
-      }
-    } else {
-      tbl = compptr->ac_tbl_no;
-      jpeg_make_d_derived_tbl(cinfo, FALSE, tbl,
-			      & entropy->derived_tbls[tbl]);
-      /* remember the single active table */
-      entropy->ac_derived_tbl = entropy->derived_tbls[tbl];
-    }
-    /* Initialize DC predictions to 0 */
-    entropy->saved.last_dc_val[ci] = 0;
-  }
-
-  /* Initialize bitread state variables */
-  entropy->bitstate.bits_left = 0;
-  entropy->bitstate.get_buffer = 0; /* unnecessary, but keeps Purify quiet */
-  entropy->pub.insufficient_data = FALSE;
-
-  /* Initialize private state variables */
-  entropy->saved.EOBRUN = 0;
-
-  /* Initialize restart counter */
-  entropy->restarts_to_go = cinfo->restart_interval;
-}
-
-
-/*
- * Figure F.12: extend sign bit.
- * On some machines, a shift and add will be faster than a table lookup.
- */
-
-#ifdef AVOID_TABLES
-
-#define HUFF_EXTEND(x,s)  ((x) < (1<<((s)-1)) ? (x) + (((-1)<<(s)) + 1) : (x))
-
-#else
-
-#define HUFF_EXTEND(x,s)  ((x) < extend_test[s] ? (x) + extend_offset[s] : (x))
-
-static const int extend_test[16] =   /* entry n is 2**(n-1) */
-  { 0, 0x0001, 0x0002, 0x0004, 0x0008, 0x0010, 0x0020, 0x0040, 0x0080,
-    0x0100, 0x0200, 0x0400, 0x0800, 0x1000, 0x2000, 0x4000 };
-
-static const int extend_offset[16] = /* entry n is (-1 << n) + 1 */
-  { 0, ((-1)<<1) + 1, ((-1)<<2) + 1, ((-1)<<3) + 1, ((-1)<<4) + 1,
-    ((-1)<<5) + 1, ((-1)<<6) + 1, ((-1)<<7) + 1, ((-1)<<8) + 1,
-    ((-1)<<9) + 1, ((-1)<<10) + 1, ((-1)<<11) + 1, ((-1)<<12) + 1,
-    ((-1)<<13) + 1, ((-1)<<14) + 1, ((-1)<<15) + 1 };
-
-#endif /* AVOID_TABLES */
-
-
-/*
- * Check for a restart marker & resynchronize decoder.
- * Returns FALSE if must suspend.
- */
-
-LOCAL(boolean)
-process_restart (j_decompress_ptr cinfo)
-{
-  phuff_entropy_ptr entropy = (phuff_entropy_ptr) cinfo->entropy;
-  int ci;
-
-  /* Throw away any unused bits remaining in bit buffer; */
-  /* include any full bytes in next_marker's count of discarded bytes */
-  cinfo->marker->discarded_bytes += entropy->bitstate.bits_left / 8;
-  entropy->bitstate.bits_left = 0;
-
-  /* Advance past the RSTn marker */
-  if (! (*cinfo->marker->read_restart_marker) (cinfo))
-    return FALSE;
-
-  /* Re-initialize DC predictions to 0 */
-  for (ci = 0; ci < cinfo->comps_in_scan; ci++)
-    entropy->saved.last_dc_val[ci] = 0;
-  /* Re-init EOB run count, too */
-  entropy->saved.EOBRUN = 0;
-
-  /* Reset restart counter */
-  entropy->restarts_to_go = cinfo->restart_interval;
-
-  /* Reset out-of-data flag, unless read_restart_marker left us smack up
-   * against a marker.  In that case we will end up treating the next data
-   * segment as empty, and we can avoid producing bogus output pixels by
-   * leaving the flag set.
-   */
-  if (cinfo->unread_marker == 0)
-    entropy->pub.insufficient_data = FALSE;
-
-  return TRUE;
-}
-
-
-/*
- * Huffman MCU decoding.
- * Each of these routines decodes and returns one MCU's worth of
- * Huffman-compressed coefficients. 
- * The coefficients are reordered from zigzag order into natural array order,
- * but are not dequantized.
- *
- * The i'th block of the MCU is stored into the block pointed to by
- * MCU_data[i].  WE ASSUME THIS AREA IS INITIALLY ZEROED BY THE CALLER.
- *
- * We return FALSE if data source requested suspension.  In that case no
- * changes have been made to permanent state.  (Exception: some output
- * coefficients may already have been assigned.  This is harmless for
- * spectral selection, since we'll just re-assign them on the next call.
- * Successive approximation AC refinement has to be more careful, however.)
- */
-
-/*
- * MCU decoding for DC initial scan (either spectral selection,
- * or first pass of successive approximation).
- */
-
-METHODDEF(boolean)
-decode_mcu_DC_first (j_decompress_ptr cinfo, JBLOCKROW *MCU_data)
-{   
-  phuff_entropy_ptr entropy = (phuff_entropy_ptr) cinfo->entropy;
-  int Al = cinfo->Al;
-  register int s, r;
-  int blkn, ci;
-  JBLOCKROW block;
-  BITREAD_STATE_VARS;
-  savable_state state;
-  d_derived_tbl * tbl;
-  jpeg_component_info * compptr;
-
-  /* Process restart marker if needed; may have to suspend */
-  if (cinfo->restart_interval) {
-    if (entropy->restarts_to_go == 0)
-      if (! process_restart(cinfo))
-	return FALSE;
-  }
-
-  /* If we've run out of data, just leave the MCU set to zeroes.
-   * This way, we return uniform gray for the remainder of the segment.
-   */
-  if (! entropy->pub.insufficient_data) {
-
-    /* Load up working state */
-    BITREAD_LOAD_STATE(cinfo,entropy->bitstate);
-    ASSIGN_STATE(state, entropy->saved);
-
-    /* Outer loop handles each block in the MCU */
-
-    for (blkn = 0; blkn < cinfo->blocks_in_MCU; blkn++) {
-      block = MCU_data[blkn];
-      ci = cinfo->MCU_membership[blkn];
-      compptr = cinfo->cur_comp_info[ci];
-      tbl = entropy->derived_tbls[compptr->dc_tbl_no];
-
-      /* Decode a single block's worth of coefficients */
-
-      /* Section F.2.2.1: decode the DC coefficient difference */
-      HUFF_DECODE(s, br_state, tbl, return FALSE, label1);
-      if (s) {
-	CHECK_BIT_BUFFER(br_state, s, return FALSE);
-	r = GET_BITS(s);
-	s = HUFF_EXTEND(r, s);
-      }
-
-      /* Convert DC difference to actual value, update last_dc_val */
-      s += state.last_dc_val[ci];
-      state.last_dc_val[ci] = s;
-      /* Scale and output the coefficient (assumes jpeg_natural_order[0]=0) */
-      (*block)[0] = (JCOEF) (s << Al);
-    }
-
-    /* Completed MCU, so update state */
-    BITREAD_SAVE_STATE(cinfo,entropy->bitstate);
-    ASSIGN_STATE(entropy->saved, state);
-  }
-
-  /* Account for restart interval (no-op if not using restarts) */
-  entropy->restarts_to_go--;
-
-  return TRUE;
-}
-
-
-/*
- * MCU decoding for AC initial scan (either spectral selection,
- * or first pass of successive approximation).
- */
-
-METHODDEF(boolean)
-decode_mcu_AC_first (j_decompress_ptr cinfo, JBLOCKROW *MCU_data)
-{   
-  phuff_entropy_ptr entropy = (phuff_entropy_ptr) cinfo->entropy;
-  int Se = cinfo->Se;
-  int Al = cinfo->Al;
-  register int s, k, r;
-  unsigned int EOBRUN;
-  JBLOCKROW block;
-  BITREAD_STATE_VARS;
-  d_derived_tbl * tbl;
-
-  /* Process restart marker if needed; may have to suspend */
-  if (cinfo->restart_interval) {
-    if (entropy->restarts_to_go == 0)
-      if (! process_restart(cinfo))
-	return FALSE;
-  }
-
-  /* If we've run out of data, just leave the MCU set to zeroes.
-   * This way, we return uniform gray for the remainder of the segment.
-   */
-  if (! entropy->pub.insufficient_data) {
-
-    /* Load up working state.
-     * We can avoid loading/saving bitread state if in an EOB run.
-     */
-    EOBRUN = entropy->saved.EOBRUN;	/* only part of saved state we need */
-
-    /* There is always only one block per MCU */
-
-    if (EOBRUN > 0)		/* if it's a band of zeroes... */
-      EOBRUN--;			/* ...process it now (we do nothing) */
-    else {
-      BITREAD_LOAD_STATE(cinfo,entropy->bitstate);
-      block = MCU_data[0];
-      tbl = entropy->ac_derived_tbl;
-
-      for (k = cinfo->Ss; k <= Se; k++) {
-	HUFF_DECODE(s, br_state, tbl, return FALSE, label2);
-	r = s >> 4;
-	s &= 15;
-	if (s) {
-	  k += r;
-	  CHECK_BIT_BUFFER(br_state, s, return FALSE);
-	  r = GET_BITS(s);
-	  s = HUFF_EXTEND(r, s);
-	  /* Scale and output coefficient in natural (dezigzagged) order */
-	  (*block)[jpeg_natural_order[k]] = (JCOEF) (s << Al);
-	} else {
-	  if (r == 15) {	/* ZRL */
-	    k += 15;		/* skip 15 zeroes in band */
-	  } else {		/* EOBr, run length is 2^r + appended bits */
-	    EOBRUN = 1 << r;
-	    if (r) {		/* EOBr, r > 0 */
-	      CHECK_BIT_BUFFER(br_state, r, return FALSE);
-	      r = GET_BITS(r);
-	      EOBRUN += r;
-	    }
-	    EOBRUN--;		/* this band is processed at this moment */
-	    break;		/* force end-of-band */
-	  }
-	}
-      }
-
-      BITREAD_SAVE_STATE(cinfo,entropy->bitstate);
-    }
-
-    /* Completed MCU, so update state */
-    entropy->saved.EOBRUN = EOBRUN;	/* only part of saved state we need */
-  }
-
-  /* Account for restart interval (no-op if not using restarts) */
-  entropy->restarts_to_go--;
-
-  return TRUE;
-}
-
-
-/*
- * MCU decoding for DC successive approximation refinement scan.
- * Note: we assume such scans can be multi-component, although the spec
- * is not very clear on the point.
- */
-
-METHODDEF(boolean)
-decode_mcu_DC_refine (j_decompress_ptr cinfo, JBLOCKROW *MCU_data)
-{   
-  phuff_entropy_ptr entropy = (phuff_entropy_ptr) cinfo->entropy;
-  int p1 = 1 << cinfo->Al;	/* 1 in the bit position being coded */
-  int blkn;
-  JBLOCKROW block;
-  BITREAD_STATE_VARS;
-
-  /* Process restart marker if needed; may have to suspend */
-  if (cinfo->restart_interval) {
-    if (entropy->restarts_to_go == 0)
-      if (! process_restart(cinfo))
-	return FALSE;
-  }
-
-  /* Not worth the cycles to check insufficient_data here,
-   * since we will not change the data anyway if we read zeroes.
-   */
-
-  /* Load up working state */
-  BITREAD_LOAD_STATE(cinfo,entropy->bitstate);
-
-  /* Outer loop handles each block in the MCU */
-
-  for (blkn = 0; blkn < cinfo->blocks_in_MCU; blkn++) {
-    block = MCU_data[blkn];
-
-    /* Encoded data is simply the next bit of the two's-complement DC value */
-    CHECK_BIT_BUFFER(br_state, 1, return FALSE);
-    if (GET_BITS(1))
-      (*block)[0] |= p1;
-    /* Note: since we use |=, repeating the assignment later is safe */
-  }
-
-  /* Completed MCU, so update state */
-  BITREAD_SAVE_STATE(cinfo,entropy->bitstate);
-
-  /* Account for restart interval (no-op if not using restarts) */
-  entropy->restarts_to_go--;
-
-  return TRUE;
-}
-
-
-/*
- * MCU decoding for AC successive approximation refinement scan.
- */
-
-METHODDEF(boolean)
-decode_mcu_AC_refine (j_decompress_ptr cinfo, JBLOCKROW *MCU_data)
-{   
-  phuff_entropy_ptr entropy = (phuff_entropy_ptr) cinfo->entropy;
-  int Se = cinfo->Se;
-  int p1 = 1 << cinfo->Al;	/* 1 in the bit position being coded */
-  int m1 = (-1) << cinfo->Al;	/* -1 in the bit position being coded */
-  register int s, k, r;
-  unsigned int EOBRUN;
-  JBLOCKROW block;
-  JCOEFPTR thiscoef;
-  BITREAD_STATE_VARS;
-  d_derived_tbl * tbl;
-  int num_newnz;
-  int newnz_pos[DCTSIZE2];
-
-  /* Process restart marker if needed; may have to suspend */
-  if (cinfo->restart_interval) {
-    if (entropy->restarts_to_go == 0)
-      if (! process_restart(cinfo))
-	return FALSE;
-  }
-
-  /* If we've run out of data, don't modify the MCU.
-   */
-  if (! entropy->pub.insufficient_data) {
-
-    /* Load up working state */
-    BITREAD_LOAD_STATE(cinfo,entropy->bitstate);
-    EOBRUN = entropy->saved.EOBRUN; /* only part of saved state we need */
-
-    /* There is always only one block per MCU */
-    block = MCU_data[0];
-    tbl = entropy->ac_derived_tbl;
-
-    /* If we are forced to suspend, we must undo the assignments to any newly
-     * nonzero coefficients in the block, because otherwise we'd get confused
-     * next time about which coefficients were already nonzero.
-     * But we need not undo addition of bits to already-nonzero coefficients;
-     * instead, we can test the current bit to see if we already did it.
-     */
-    num_newnz = 0;
-
-    /* initialize coefficient loop counter to start of band */
-    k = cinfo->Ss;
-
-    if (EOBRUN == 0) {
-      for (; k <= Se; k++) {
-	HUFF_DECODE(s, br_state, tbl, goto undoit, label3);
-	r = s >> 4;
-	s &= 15;
-	if (s) {
-	  if (s != 1)		/* size of new coef should always be 1 */
-	    WARNMS(cinfo, JWRN_HUFF_BAD_CODE);
-	  CHECK_BIT_BUFFER(br_state, 1, goto undoit);
-	  if (GET_BITS(1))
-	    s = p1;		/* newly nonzero coef is positive */
-	  else
-	    s = m1;		/* newly nonzero coef is negative */
-	} else {
-	  if (r != 15) {
-	    EOBRUN = 1 << r;	/* EOBr, run length is 2^r + appended bits */
-	    if (r) {
-	      CHECK_BIT_BUFFER(br_state, r, goto undoit);
-	      r = GET_BITS(r);
-	      EOBRUN += r;
-	    }
-	    break;		/* rest of block is handled by EOB logic */
-	  }
-	  /* note s = 0 for processing ZRL */
-	}
-	/* Advance over already-nonzero coefs and r still-zero coefs,
-	 * appending correction bits to the nonzeroes.  A correction bit is 1
-	 * if the absolute value of the coefficient must be increased.
-	 */
-	do {
-	  thiscoef = *block + jpeg_natural_order[k];
-	  if (*thiscoef != 0) {
-	    CHECK_BIT_BUFFER(br_state, 1, goto undoit);
-	    if (GET_BITS(1)) {
-	      if ((*thiscoef & p1) == 0) { /* do nothing if already set it */
-		if (*thiscoef >= 0)
-		  *thiscoef += p1;
-		else
-		  *thiscoef += m1;
-	      }
-	    }
-	  } else {
-	    if (--r < 0)
-	      break;		/* reached target zero coefficient */
-	  }
-	  k++;
-	} while (k <= Se);
-	if (s) {
-	  int pos = jpeg_natural_order[k];
-	  /* Output newly nonzero coefficient */
-	  (*block)[pos] = (JCOEF) s;
-	  /* Remember its position in case we have to suspend */
-	  newnz_pos[num_newnz++] = pos;
-	}
-      }
-    }
-
-    if (EOBRUN > 0) {
-      /* Scan any remaining coefficient positions after the end-of-band
-       * (the last newly nonzero coefficient, if any).  Append a correction
-       * bit to each already-nonzero coefficient.  A correction bit is 1
-       * if the absolute value of the coefficient must be increased.
-       */
-      for (; k <= Se; k++) {
-	thiscoef = *block + jpeg_natural_order[k];
-	if (*thiscoef != 0) {
-	  CHECK_BIT_BUFFER(br_state, 1, goto undoit);
-	  if (GET_BITS(1)) {
-	    if ((*thiscoef & p1) == 0) { /* do nothing if already changed it */
-	      if (*thiscoef >= 0)
-		*thiscoef += p1;
-	      else
-		*thiscoef += m1;
-	    }
-	  }
-	}
-      }
-      /* Count one block completed in EOB run */
-      EOBRUN--;
-    }
-
-    /* Completed MCU, so update state */
-    BITREAD_SAVE_STATE(cinfo,entropy->bitstate);
-    entropy->saved.EOBRUN = EOBRUN; /* only part of saved state we need */
-  }
-
-  /* Account for restart interval (no-op if not using restarts) */
-  entropy->restarts_to_go--;
-
-  return TRUE;
-
-undoit:
-  /* Re-zero any output coefficients that we made newly nonzero */
-  while (num_newnz > 0)
-    (*block)[newnz_pos[--num_newnz]] = 0;
-
-  return FALSE;
-}
-
-/*
- * Save the current Huffman decoder position and the bit buffer
- * into bitstream_offset and get_buffer, respectively.
- */
-GLOBAL(void)
-jpeg_get_huffman_decoder_configuration_progressive(j_decompress_ptr cinfo,
-        huffman_offset_data *offset)
-{
-  phuff_entropy_ptr entropy = (phuff_entropy_ptr) cinfo->entropy;
-
-  if (cinfo->restart_interval) {
-    // We are at the end of a data segment
-    if (entropy->restarts_to_go == 0)
-      if (! process_restart(cinfo))
-	return;
-  }
-
-  // Save restarts_to_go and next_restart_num.
-  offset->restarts_to_go = (unsigned short) entropy->restarts_to_go;
-  offset->next_restart_num = cinfo->marker->next_restart_num;
-
-  offset->bitstream_offset =
-      (jget_input_stream_position(cinfo) << LOG_TWO_BIT_BUF_SIZE)
-      + entropy->bitstate.bits_left;
-
-  offset->get_buffer = entropy->bitstate.get_buffer;
-}
-
-/*
- * Save the current Huffman deocde position and the DC coefficients
- * for each component into bitstream_offset and dc_info[], respectively.
- */
-METHODDEF(void)
-get_huffman_decoder_configuration(j_decompress_ptr cinfo,
-        huffman_offset_data *offset)
-{
-  int i;
-  phuff_entropy_ptr entropy = (phuff_entropy_ptr) cinfo->entropy;
-  jpeg_get_huffman_decoder_configuration_progressive(cinfo, offset);
-  offset->EOBRUN = entropy->saved.EOBRUN;
-  for (i = 0; i < cinfo->comps_in_scan; i++)
-    offset->prev_dc[i] = entropy->saved.last_dc_val[i];
-}
-
-/*
- * Configure the Huffman decoder reader position and bit buffer.
- */
-GLOBAL(void)
-jpeg_configure_huffman_decoder_progressive(j_decompress_ptr cinfo,
-        huffman_offset_data offset)
-{
-	phuff_entropy_ptr entropy = (phuff_entropy_ptr) cinfo->entropy;
-
-  // Restore restarts_to_go and next_restart_num
-  cinfo->unread_marker = 0;
-  entropy->restarts_to_go = offset.restarts_to_go;
-  cinfo->marker->next_restart_num = offset.next_restart_num;
-
-  unsigned int bitstream_offset = offset.bitstream_offset;
-  int blkn, i;
-
-  unsigned int byte_offset = bitstream_offset >> LOG_TWO_BIT_BUF_SIZE;
-  unsigned int bit_in_bit_buffer =
-      bitstream_offset & ((1 << LOG_TWO_BIT_BUF_SIZE) - 1);
-
-  jset_input_stream_position_bit(cinfo, byte_offset,
-          bit_in_bit_buffer, offset.get_buffer);
-}
-
-/*
- * Configure the Huffman decoder to decode the image
- * starting from (iMCU_row_offset, iMCU_col_offset).
- */
-METHODDEF(void)
-configure_huffman_decoder(j_decompress_ptr cinfo, huffman_offset_data offset)
-{
-  int i;
-  phuff_entropy_ptr entropy = (phuff_entropy_ptr) cinfo->entropy;
-  jpeg_configure_huffman_decoder_progressive(cinfo, offset);
-  entropy->saved.EOBRUN = offset.EOBRUN;
-  for (i = 0; i < cinfo->comps_in_scan; i++)
-    entropy->saved.last_dc_val[i] = offset.prev_dc[i];
-}
-
-GLOBAL(void)
-jpeg_configure_huffman_index_scan(j_decompress_ptr cinfo,
-        huffman_index *index, int scan_no, int offset)
-{
-  phuff_entropy_ptr entropy = (phuff_entropy_ptr) cinfo->entropy;
-  if (scan_no >= index->scan_count) {
-    index->scan = realloc(index->scan,
-                    (scan_no + 1) * sizeof(huffman_scan_header));
-    index->mem_used += (scan_no - index->scan_count + 1)
-      * (sizeof(huffman_scan_header) + cinfo->total_iMCU_rows
-      * sizeof(huffman_offset_data*));
-    index->scan_count = scan_no + 1;
-  }
-  index->scan[scan_no].offset = (huffman_offset_data**)malloc(
-          cinfo->total_iMCU_rows * sizeof(huffman_offset_data*));
-  index->scan[scan_no].bitstream_offset = offset;
-}
-
-/*
- * Module initialization routine for progressive Huffman entropy decoding.
- */
-GLOBAL(void)
-jinit_phuff_decoder (j_decompress_ptr cinfo)
-{
-  phuff_entropy_ptr entropy;
-  int *coef_bit_ptr;
-  int ci, i;
-
-  entropy = (phuff_entropy_ptr)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				SIZEOF(phuff_entropy_decoder));
-  cinfo->entropy = (struct jpeg_entropy_decoder *) entropy;
-  entropy->pub.start_pass = start_pass_phuff_decoder;
-  entropy->pub.configure_huffman_decoder = configure_huffman_decoder;
-  entropy->pub.get_huffman_decoder_configuration =
-        get_huffman_decoder_configuration;
-
-  /* Mark derived tables unallocated */
-  for (i = 0; i < NUM_HUFF_TBLS; i++) {
-    entropy->derived_tbls[i] = NULL;
-  }
-
-  /* Create progression status table */
-  cinfo->coef_bits = (int (*)[DCTSIZE2])
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				cinfo->num_components*DCTSIZE2*SIZEOF(int));
-  coef_bit_ptr = & cinfo->coef_bits[0][0];
-  for (ci = 0; ci < cinfo->num_components; ci++) 
-    for (i = 0; i < DCTSIZE2; i++)
-      *coef_bit_ptr++ = -1;
-}
-
-#endif /* D_PROGRESSIVE_SUPPORTED */
@@ -1,290 +0,0 @@
-/*
- * jdpostct.c
- *
- * Copyright (C) 1994-1996, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains the decompression postprocessing controller.
- * This controller manages the upsampling, color conversion, and color
- * quantization/reduction steps; specifically, it controls the buffering
- * between upsample/color conversion and color quantization/reduction.
- *
- * If no color quantization/reduction is required, then this module has no
- * work to do, and it just hands off to the upsample/color conversion code.
- * An integrated upsample/convert/quantize process would replace this module
- * entirely.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-
-
-/* Private buffer controller object */
-
-typedef struct {
-  struct jpeg_d_post_controller pub; /* public fields */
-
-  /* Color quantization source buffer: this holds output data from
-   * the upsample/color conversion step to be passed to the quantizer.
-   * For two-pass color quantization, we need a full-image buffer;
-   * for one-pass operation, a strip buffer is sufficient.
-   */
-  jvirt_sarray_ptr whole_image;	/* virtual array, or NULL if one-pass */
-  JSAMPARRAY buffer;		/* strip buffer, or current strip of virtual */
-  JDIMENSION strip_height;	/* buffer size in rows */
-  /* for two-pass mode only: */
-  JDIMENSION starting_row;	/* row # of first row in current strip */
-  JDIMENSION next_row;		/* index of next row to fill/empty in strip */
-} my_post_controller;
-
-typedef my_post_controller * my_post_ptr;
-
-
-/* Forward declarations */
-METHODDEF(void) post_process_1pass
-	JPP((j_decompress_ptr cinfo,
-	     JSAMPIMAGE input_buf, JDIMENSION *in_row_group_ctr,
-	     JDIMENSION in_row_groups_avail,
-	     JSAMPARRAY output_buf, JDIMENSION *out_row_ctr,
-	     JDIMENSION out_rows_avail));
-#ifdef QUANT_2PASS_SUPPORTED
-METHODDEF(void) post_process_prepass
-	JPP((j_decompress_ptr cinfo,
-	     JSAMPIMAGE input_buf, JDIMENSION *in_row_group_ctr,
-	     JDIMENSION in_row_groups_avail,
-	     JSAMPARRAY output_buf, JDIMENSION *out_row_ctr,
-	     JDIMENSION out_rows_avail));
-METHODDEF(void) post_process_2pass
-	JPP((j_decompress_ptr cinfo,
-	     JSAMPIMAGE input_buf, JDIMENSION *in_row_group_ctr,
-	     JDIMENSION in_row_groups_avail,
-	     JSAMPARRAY output_buf, JDIMENSION *out_row_ctr,
-	     JDIMENSION out_rows_avail));
-#endif
-
-
-/*
- * Initialize for a processing pass.
- */
-
-METHODDEF(void)
-start_pass_dpost (j_decompress_ptr cinfo, J_BUF_MODE pass_mode)
-{
-  my_post_ptr post = (my_post_ptr) cinfo->post;
-
-  switch (pass_mode) {
-  case JBUF_PASS_THRU:
-    if (cinfo->quantize_colors) {
-      /* Single-pass processing with color quantization. */
-      post->pub.post_process_data = post_process_1pass;
-      /* We could be doing buffered-image output before starting a 2-pass
-       * color quantization; in that case, jinit_d_post_controller did not
-       * allocate a strip buffer.  Use the virtual-array buffer as workspace.
-       */
-      if (post->buffer == NULL) {
-	post->buffer = (*cinfo->mem->access_virt_sarray)
-	  ((j_common_ptr) cinfo, post->whole_image,
-	   (JDIMENSION) 0, post->strip_height, TRUE);
-      }
-    } else {
-      /* For single-pass processing without color quantization,
-       * I have no work to do; just call the upsampler directly.
-       */
-      post->pub.post_process_data = cinfo->upsample->upsample;
-    }
-    break;
-#ifdef QUANT_2PASS_SUPPORTED
-  case JBUF_SAVE_AND_PASS:
-    /* First pass of 2-pass quantization */
-    if (post->whole_image == NULL)
-      ERREXIT(cinfo, JERR_BAD_BUFFER_MODE);
-    post->pub.post_process_data = post_process_prepass;
-    break;
-  case JBUF_CRANK_DEST:
-    /* Second pass of 2-pass quantization */
-    if (post->whole_image == NULL)
-      ERREXIT(cinfo, JERR_BAD_BUFFER_MODE);
-    post->pub.post_process_data = post_process_2pass;
-    break;
-#endif /* QUANT_2PASS_SUPPORTED */
-  default:
-    ERREXIT(cinfo, JERR_BAD_BUFFER_MODE);
-    break;
-  }
-  post->starting_row = post->next_row = 0;
-}
-
-
-/*
- * Process some data in the one-pass (strip buffer) case.
- * This is used for color precision reduction as well as one-pass quantization.
- */
-
-METHODDEF(void)
-post_process_1pass (j_decompress_ptr cinfo,
-		    JSAMPIMAGE input_buf, JDIMENSION *in_row_group_ctr,
-		    JDIMENSION in_row_groups_avail,
-		    JSAMPARRAY output_buf, JDIMENSION *out_row_ctr,
-		    JDIMENSION out_rows_avail)
-{
-  my_post_ptr post = (my_post_ptr) cinfo->post;
-  JDIMENSION num_rows, max_rows;
-
-  /* Fill the buffer, but not more than what we can dump out in one go. */
-  /* Note we rely on the upsampler to detect bottom of image. */
-  max_rows = out_rows_avail - *out_row_ctr;
-  if (max_rows > post->strip_height)
-    max_rows = post->strip_height;
-  num_rows = 0;
-  (*cinfo->upsample->upsample) (cinfo,
-		input_buf, in_row_group_ctr, in_row_groups_avail,
-		post->buffer, &num_rows, max_rows);
-  /* Quantize and emit data. */
-  (*cinfo->cquantize->color_quantize) (cinfo,
-		post->buffer, output_buf + *out_row_ctr, (int) num_rows);
-  *out_row_ctr += num_rows;
-}
-
-
-#ifdef QUANT_2PASS_SUPPORTED
-
-/*
- * Process some data in the first pass of 2-pass quantization.
- */
-
-METHODDEF(void)
-post_process_prepass (j_decompress_ptr cinfo,
-		      JSAMPIMAGE input_buf, JDIMENSION *in_row_group_ctr,
-		      JDIMENSION in_row_groups_avail,
-		      JSAMPARRAY output_buf, JDIMENSION *out_row_ctr,
-		      JDIMENSION out_rows_avail)
-{
-  my_post_ptr post = (my_post_ptr) cinfo->post;
-  JDIMENSION old_next_row, num_rows;
-
-  /* Reposition virtual buffer if at start of strip. */
-  if (post->next_row == 0) {
-    post->buffer = (*cinfo->mem->access_virt_sarray)
-	((j_common_ptr) cinfo, post->whole_image,
-	 post->starting_row, post->strip_height, TRUE);
-  }
-
-  /* Upsample some data (up to a strip height's worth). */
-  old_next_row = post->next_row;
-  (*cinfo->upsample->upsample) (cinfo,
-		input_buf, in_row_group_ctr, in_row_groups_avail,
-		post->buffer, &post->next_row, post->strip_height);
-
-  /* Allow quantizer to scan new data.  No data is emitted, */
-  /* but we advance out_row_ctr so outer loop can tell when we're done. */
-  if (post->next_row > old_next_row) {
-    num_rows = post->next_row - old_next_row;
-    (*cinfo->cquantize->color_quantize) (cinfo, post->buffer + old_next_row,
-					 (JSAMPARRAY) NULL, (int) num_rows);
-    *out_row_ctr += num_rows;
-  }
-
-  /* Advance if we filled the strip. */
-  if (post->next_row >= post->strip_height) {
-    post->starting_row += post->strip_height;
-    post->next_row = 0;
-  }
-}
-
-
-/*
- * Process some data in the second pass of 2-pass quantization.
- */
-
-METHODDEF(void)
-post_process_2pass (j_decompress_ptr cinfo,
-		    JSAMPIMAGE input_buf, JDIMENSION *in_row_group_ctr,
-		    JDIMENSION in_row_groups_avail,
-		    JSAMPARRAY output_buf, JDIMENSION *out_row_ctr,
-		    JDIMENSION out_rows_avail)
-{
-  my_post_ptr post = (my_post_ptr) cinfo->post;
-  JDIMENSION num_rows, max_rows;
-
-  /* Reposition virtual buffer if at start of strip. */
-  if (post->next_row == 0) {
-    post->buffer = (*cinfo->mem->access_virt_sarray)
-	((j_common_ptr) cinfo, post->whole_image,
-	 post->starting_row, post->strip_height, FALSE);
-  }
-
-  /* Determine number of rows to emit. */
-  num_rows = post->strip_height - post->next_row; /* available in strip */
-  max_rows = out_rows_avail - *out_row_ctr; /* available in output area */
-  if (num_rows > max_rows)
-    num_rows = max_rows;
-  /* We have to check bottom of image here, can't depend on upsampler. */
-  max_rows = cinfo->output_height - post->starting_row;
-  if (num_rows > max_rows)
-    num_rows = max_rows;
-
-  /* Quantize and emit data. */
-  (*cinfo->cquantize->color_quantize) (cinfo,
-		post->buffer + post->next_row, output_buf + *out_row_ctr,
-		(int) num_rows);
-  *out_row_ctr += num_rows;
-
-  /* Advance if we filled the strip. */
-  post->next_row += num_rows;
-  if (post->next_row >= post->strip_height) {
-    post->starting_row += post->strip_height;
-    post->next_row = 0;
-  }
-}
-
-#endif /* QUANT_2PASS_SUPPORTED */
-
-
-/*
- * Initialize postprocessing controller.
- */
-
-GLOBAL(void)
-jinit_d_post_controller (j_decompress_ptr cinfo, boolean need_full_buffer)
-{
-  my_post_ptr post;
-
-  post = (my_post_ptr)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				SIZEOF(my_post_controller));
-  cinfo->post = (struct jpeg_d_post_controller *) post;
-  post->pub.start_pass = start_pass_dpost;
-  post->whole_image = NULL;	/* flag for no virtual arrays */
-  post->buffer = NULL;		/* flag for no strip buffer */
-
-  /* Create the quantization buffer, if needed */
-  if (cinfo->quantize_colors) {
-    /* The buffer strip height is max_v_samp_factor, which is typically
-     * an efficient number of rows for upsampling to return.
-     * (In the presence of output rescaling, we might want to be smarter?)
-     */
-    post->strip_height = (JDIMENSION) cinfo->max_v_samp_factor;
-    if (need_full_buffer) {
-      /* Two-pass color quantization: need full-image storage. */
-      /* We round up the number of rows to a multiple of the strip height. */
-#ifdef QUANT_2PASS_SUPPORTED
-      post->whole_image = (*cinfo->mem->request_virt_sarray)
-	((j_common_ptr) cinfo, JPOOL_IMAGE, FALSE,
-	 cinfo->output_width * cinfo->out_color_components,
-	 (JDIMENSION) jround_up((long) cinfo->output_height,
-				(long) post->strip_height),
-	 post->strip_height);
-#else
-      ERREXIT(cinfo, JERR_BAD_BUFFER_MODE);
-#endif /* QUANT_2PASS_SUPPORTED */
-    } else {
-      /* One-pass color quantization: just make a strip buffer. */
-      post->buffer = (*cinfo->mem->alloc_sarray)
-	((j_common_ptr) cinfo, JPOOL_IMAGE,
-	 cinfo->output_width * cinfo->out_color_components,
-	 post->strip_height);
-    }
-  }
-}
@@ -1,478 +0,0 @@
-/*
- * jdsample.c
- *
- * Copyright (C) 1991-1996, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains upsampling routines.
- *
- * Upsampling input data is counted in "row groups".  A row group
- * is defined to be (v_samp_factor * DCT_scaled_size / min_DCT_scaled_size)
- * sample rows of each component.  Upsampling will normally produce
- * max_v_samp_factor pixel rows from each row group (but this could vary
- * if the upsampler is applying a scale factor of its own).
- *
- * An excellent reference for image resampling is
- *   Digital Image Warping, George Wolberg, 1990.
- *   Pub. by IEEE Computer Society Press, Los Alamitos, CA. ISBN 0-8186-8944-7.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-
-
-/* Pointer to routine to upsample a single component */
-typedef JMETHOD(void, upsample1_ptr,
-		(j_decompress_ptr cinfo, jpeg_component_info * compptr,
-		 JSAMPARRAY input_data, JSAMPARRAY * output_data_ptr));
-
-/* Private subobject */
-
-typedef struct {
-  struct jpeg_upsampler pub;	/* public fields */
-
-  /* Color conversion buffer.  When using separate upsampling and color
-   * conversion steps, this buffer holds one upsampled row group until it
-   * has been color converted and output.
-   * Note: we do not allocate any storage for component(s) which are full-size,
-   * ie do not need rescaling.  The corresponding entry of color_buf[] is
-   * simply set to point to the input data array, thereby avoiding copying.
-   */
-  JSAMPARRAY color_buf[MAX_COMPONENTS];
-
-  /* Per-component upsampling method pointers */
-  upsample1_ptr methods[MAX_COMPONENTS];
-
-  int next_row_out;		/* counts rows emitted from color_buf */
-  JDIMENSION rows_to_go;	/* counts rows remaining in image */
-
-  /* Height of an input row group for each component. */
-  int rowgroup_height[MAX_COMPONENTS];
-
-  /* These arrays save pixel expansion factors so that int_expand need not
-   * recompute them each time.  They are unused for other upsampling methods.
-   */
-  UINT8 h_expand[MAX_COMPONENTS];
-  UINT8 v_expand[MAX_COMPONENTS];
-} my_upsampler;
-
-typedef my_upsampler * my_upsample_ptr;
-
-
-/*
- * Initialize for an upsampling pass.
- */
-
-METHODDEF(void)
-start_pass_upsample (j_decompress_ptr cinfo)
-{
-  my_upsample_ptr upsample = (my_upsample_ptr) cinfo->upsample;
-
-  /* Mark the conversion buffer empty */
-  upsample->next_row_out = cinfo->max_v_samp_factor;
-  /* Initialize total-height counter for detecting bottom of image */
-  upsample->rows_to_go = cinfo->output_height;
-}
-
-
-/*
- * Control routine to do upsampling (and color conversion).
- *
- * In this version we upsample each component independently.
- * We upsample one row group into the conversion buffer, then apply
- * color conversion a row at a time.
- */
-
-METHODDEF(void)
-sep_upsample (j_decompress_ptr cinfo,
-	      JSAMPIMAGE input_buf, JDIMENSION *in_row_group_ctr,
-	      JDIMENSION in_row_groups_avail,
-	      JSAMPARRAY output_buf, JDIMENSION *out_row_ctr,
-	      JDIMENSION out_rows_avail)
-{
-  my_upsample_ptr upsample = (my_upsample_ptr) cinfo->upsample;
-  int ci;
-  jpeg_component_info * compptr;
-  JDIMENSION num_rows;
-
-  /* Fill the conversion buffer, if it's empty */
-  if (upsample->next_row_out >= cinfo->max_v_samp_factor) {
-    for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
-	 ci++, compptr++) {
-      /* Invoke per-component upsample method.  Notice we pass a POINTER
-       * to color_buf[ci], so that fullsize_upsample can change it.
-       */
-      (*upsample->methods[ci]) (cinfo, compptr,
-	input_buf[ci] + (*in_row_group_ctr * upsample->rowgroup_height[ci]),
-	upsample->color_buf + ci);
-    }
-    upsample->next_row_out = 0;
-  }
-
-  /* Color-convert and emit rows */
-
-  /* How many we have in the buffer: */
-  num_rows = (JDIMENSION) (cinfo->max_v_samp_factor - upsample->next_row_out);
-  /* Not more than the distance to the end of the image.  Need this test
-   * in case the image height is not a multiple of max_v_samp_factor:
-   */
-  if (num_rows > upsample->rows_to_go) 
-    num_rows = upsample->rows_to_go;
-  /* And not more than what the client can accept: */
-  out_rows_avail -= *out_row_ctr;
-  if (num_rows > out_rows_avail)
-    num_rows = out_rows_avail;
-
-  (*cinfo->cconvert->color_convert) (cinfo, upsample->color_buf,
-				     (JDIMENSION) upsample->next_row_out,
-				     output_buf + *out_row_ctr,
-				     (int) num_rows);
-
-  /* Adjust counts */
-  *out_row_ctr += num_rows;
-  upsample->rows_to_go -= num_rows;
-  upsample->next_row_out += num_rows;
-  /* When the buffer is emptied, declare this input row group consumed */
-  if (upsample->next_row_out >= cinfo->max_v_samp_factor)
-    (*in_row_group_ctr)++;
-}
-
-
-/*
- * These are the routines invoked by sep_upsample to upsample pixel values
- * of a single component.  One row group is processed per call.
- */
-
-
-/*
- * For full-size components, we just make color_buf[ci] point at the
- * input buffer, and thus avoid copying any data.  Note that this is
- * safe only because sep_upsample doesn't declare the input row group
- * "consumed" until we are done color converting and emitting it.
- */
-
-METHODDEF(void)
-fullsize_upsample (j_decompress_ptr cinfo, jpeg_component_info * compptr,
-		   JSAMPARRAY input_data, JSAMPARRAY * output_data_ptr)
-{
-  *output_data_ptr = input_data;
-}
-
-
-/*
- * This is a no-op version used for "uninteresting" components.
- * These components will not be referenced by color conversion.
- */
-
-METHODDEF(void)
-noop_upsample (j_decompress_ptr cinfo, jpeg_component_info * compptr,
-	       JSAMPARRAY input_data, JSAMPARRAY * output_data_ptr)
-{
-  *output_data_ptr = NULL;	/* safety check */
-}
-
-
-/*
- * This version handles any integral sampling ratios.
- * This is not used for typical JPEG files, so it need not be fast.
- * Nor, for that matter, is it particularly accurate: the algorithm is
- * simple replication of the input pixel onto the corresponding output
- * pixels.  The hi-falutin sampling literature refers to this as a
- * "box filter".  A box filter tends to introduce visible artifacts,
- * so if you are actually going to use 3:1 or 4:1 sampling ratios
- * you would be well advised to improve this code.
- */
-
-METHODDEF(void)
-int_upsample (j_decompress_ptr cinfo, jpeg_component_info * compptr,
-	      JSAMPARRAY input_data, JSAMPARRAY * output_data_ptr)
-{
-  my_upsample_ptr upsample = (my_upsample_ptr) cinfo->upsample;
-  JSAMPARRAY output_data = *output_data_ptr;
-  register JSAMPROW inptr, outptr;
-  register JSAMPLE invalue;
-  register int h;
-  JSAMPROW outend;
-  int h_expand, v_expand;
-  int inrow, outrow;
-
-  h_expand = upsample->h_expand[compptr->component_index];
-  v_expand = upsample->v_expand[compptr->component_index];
-
-  inrow = outrow = 0;
-  while (outrow < cinfo->max_v_samp_factor) {
-    /* Generate one output row with proper horizontal expansion */
-    inptr = input_data[inrow];
-    outptr = output_data[outrow];
-    outend = outptr + cinfo->output_width;
-    while (outptr < outend) {
-      invalue = *inptr++;	/* don't need GETJSAMPLE() here */
-      for (h = h_expand; h > 0; h--) {
-	*outptr++ = invalue;
-      }
-    }
-    /* Generate any additional output rows by duplicating the first one */
-    if (v_expand > 1) {
-      jcopy_sample_rows(output_data, outrow, output_data, outrow+1,
-			v_expand-1, cinfo->output_width);
-    }
-    inrow++;
-    outrow += v_expand;
-  }
-}
-
-
-/*
- * Fast processing for the common case of 2:1 horizontal and 1:1 vertical.
- * It's still a box filter.
- */
-
-METHODDEF(void)
-h2v1_upsample (j_decompress_ptr cinfo, jpeg_component_info * compptr,
-	       JSAMPARRAY input_data, JSAMPARRAY * output_data_ptr)
-{
-  JSAMPARRAY output_data = *output_data_ptr;
-  register JSAMPROW inptr, outptr;
-  register JSAMPLE invalue;
-  JSAMPROW outend;
-  int inrow;
-
-  for (inrow = 0; inrow < cinfo->max_v_samp_factor; inrow++) {
-    inptr = input_data[inrow];
-    outptr = output_data[inrow];
-    outend = outptr + cinfo->output_width;
-    while (outptr < outend) {
-      invalue = *inptr++;	/* don't need GETJSAMPLE() here */
-      *outptr++ = invalue;
-      *outptr++ = invalue;
-    }
-  }
-}
-
-
-/*
- * Fast processing for the common case of 2:1 horizontal and 2:1 vertical.
- * It's still a box filter.
- */
-
-METHODDEF(void)
-h2v2_upsample (j_decompress_ptr cinfo, jpeg_component_info * compptr,
-	       JSAMPARRAY input_data, JSAMPARRAY * output_data_ptr)
-{
-  JSAMPARRAY output_data = *output_data_ptr;
-  register JSAMPROW inptr, outptr;
-  register JSAMPLE invalue;
-  JSAMPROW outend;
-  int inrow, outrow;
-
-  inrow = outrow = 0;
-  while (outrow < cinfo->max_v_samp_factor) {
-    inptr = input_data[inrow];
-    outptr = output_data[outrow];
-    outend = outptr + cinfo->output_width;
-    while (outptr < outend) {
-      invalue = *inptr++;	/* don't need GETJSAMPLE() here */
-      *outptr++ = invalue;
-      *outptr++ = invalue;
-    }
-    jcopy_sample_rows(output_data, outrow, output_data, outrow+1,
-		      1, cinfo->output_width);
-    inrow++;
-    outrow += 2;
-  }
-}
-
-
-/*
- * Fancy processing for the common case of 2:1 horizontal and 1:1 vertical.
- *
- * The upsampling algorithm is linear interpolation between pixel centers,
- * also known as a "triangle filter".  This is a good compromise between
- * speed and visual quality.  The centers of the output pixels are 1/4 and 3/4
- * of the way between input pixel centers.
- *
- * A note about the "bias" calculations: when rounding fractional values to
- * integer, we do not want to always round 0.5 up to the next integer.
- * If we did that, we'd introduce a noticeable bias towards larger values.
- * Instead, this code is arranged so that 0.5 will be rounded up or down at
- * alternate pixel locations (a simple ordered dither pattern).
- */
-
-METHODDEF(void)
-h2v1_fancy_upsample (j_decompress_ptr cinfo, jpeg_component_info * compptr,
-		     JSAMPARRAY input_data, JSAMPARRAY * output_data_ptr)
-{
-  JSAMPARRAY output_data = *output_data_ptr;
-  register JSAMPROW inptr, outptr;
-  register int invalue;
-  register JDIMENSION colctr;
-  int inrow;
-
-  for (inrow = 0; inrow < cinfo->max_v_samp_factor; inrow++) {
-    inptr = input_data[inrow];
-    outptr = output_data[inrow];
-    /* Special case for first column */
-    invalue = GETJSAMPLE(*inptr++);
-    *outptr++ = (JSAMPLE) invalue;
-    *outptr++ = (JSAMPLE) ((invalue * 3 + GETJSAMPLE(*inptr) + 2) >> 2);
-
-    for (colctr = compptr->downsampled_width - 2; colctr > 0; colctr--) {
-      /* General case: 3/4 * nearer pixel + 1/4 * further pixel */
-      invalue = GETJSAMPLE(*inptr++) * 3;
-      *outptr++ = (JSAMPLE) ((invalue + GETJSAMPLE(inptr[-2]) + 1) >> 2);
-      *outptr++ = (JSAMPLE) ((invalue + GETJSAMPLE(*inptr) + 2) >> 2);
-    }
-
-    /* Special case for last column */
-    invalue = GETJSAMPLE(*inptr);
-    *outptr++ = (JSAMPLE) ((invalue * 3 + GETJSAMPLE(inptr[-1]) + 1) >> 2);
-    *outptr++ = (JSAMPLE) invalue;
-  }
-}
-
-
-/*
- * Fancy processing for the common case of 2:1 horizontal and 2:1 vertical.
- * Again a triangle filter; see comments for h2v1 case, above.
- *
- * It is OK for us to reference the adjacent input rows because we demanded
- * context from the main buffer controller (see initialization code).
- */
-
-METHODDEF(void)
-h2v2_fancy_upsample (j_decompress_ptr cinfo, jpeg_component_info * compptr,
-		     JSAMPARRAY input_data, JSAMPARRAY * output_data_ptr)
-{
-  JSAMPARRAY output_data = *output_data_ptr;
-  register JSAMPROW inptr0, inptr1, outptr;
-#if BITS_IN_JSAMPLE == 8
-  register int thiscolsum, lastcolsum, nextcolsum;
-#else
-  register INT32 thiscolsum, lastcolsum, nextcolsum;
-#endif
-  register JDIMENSION colctr;
-  int inrow, outrow, v;
-
-  inrow = outrow = 0;
-  while (outrow < cinfo->max_v_samp_factor) {
-    for (v = 0; v < 2; v++) {
-      /* inptr0 points to nearest input row, inptr1 points to next nearest */
-      inptr0 = input_data[inrow];
-      if (v == 0)		/* next nearest is row above */
-	inptr1 = input_data[inrow-1];
-      else			/* next nearest is row below */
-	inptr1 = input_data[inrow+1];
-      outptr = output_data[outrow++];
-
-      /* Special case for first column */
-      thiscolsum = GETJSAMPLE(*inptr0++) * 3 + GETJSAMPLE(*inptr1++);
-      nextcolsum = GETJSAMPLE(*inptr0++) * 3 + GETJSAMPLE(*inptr1++);
-      *outptr++ = (JSAMPLE) ((thiscolsum * 4 + 8) >> 4);
-      *outptr++ = (JSAMPLE) ((thiscolsum * 3 + nextcolsum + 7) >> 4);
-      lastcolsum = thiscolsum; thiscolsum = nextcolsum;
-
-      for (colctr = compptr->downsampled_width - 2; colctr > 0; colctr--) {
-	/* General case: 3/4 * nearer pixel + 1/4 * further pixel in each */
-	/* dimension, thus 9/16, 3/16, 3/16, 1/16 overall */
-	nextcolsum = GETJSAMPLE(*inptr0++) * 3 + GETJSAMPLE(*inptr1++);
-	*outptr++ = (JSAMPLE) ((thiscolsum * 3 + lastcolsum + 8) >> 4);
-	*outptr++ = (JSAMPLE) ((thiscolsum * 3 + nextcolsum + 7) >> 4);
-	lastcolsum = thiscolsum; thiscolsum = nextcolsum;
-      }
-
-      /* Special case for last column */
-      *outptr++ = (JSAMPLE) ((thiscolsum * 3 + lastcolsum + 8) >> 4);
-      *outptr++ = (JSAMPLE) ((thiscolsum * 4 + 7) >> 4);
-    }
-    inrow++;
-  }
-}
-
-
-/*
- * Module initialization routine for upsampling.
- */
-
-GLOBAL(void)
-jinit_upsampler (j_decompress_ptr cinfo)
-{
-  my_upsample_ptr upsample;
-  int ci;
-  jpeg_component_info * compptr;
-  boolean need_buffer, do_fancy;
-  int h_in_group, v_in_group, h_out_group, v_out_group;
-
-  upsample = (my_upsample_ptr)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				SIZEOF(my_upsampler));
-  cinfo->upsample = (struct jpeg_upsampler *) upsample;
-  upsample->pub.start_pass = start_pass_upsample;
-  upsample->pub.upsample = sep_upsample;
-  upsample->pub.need_context_rows = FALSE; /* until we find out differently */
-
-  if (cinfo->CCIR601_sampling)	/* this isn't supported */
-    ERREXIT(cinfo, JERR_CCIR601_NOTIMPL);
-
-  /* jdmainct.c doesn't support context rows when min_DCT_scaled_size = 1,
-   * so don't ask for it.
-   */
-  do_fancy = cinfo->do_fancy_upsampling && cinfo->min_DCT_scaled_size > 1;
-
-  /* Verify we can handle the sampling factors, select per-component methods,
-   * and create storage as needed.
-   */
-  for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
-       ci++, compptr++) {
-    /* Compute size of an "input group" after IDCT scaling.  This many samples
-     * are to be converted to max_h_samp_factor * max_v_samp_factor pixels.
-     */
-    h_in_group = (compptr->h_samp_factor * compptr->DCT_scaled_size) /
-		 cinfo->min_DCT_scaled_size;
-    v_in_group = (compptr->v_samp_factor * compptr->DCT_scaled_size) /
-		 cinfo->min_DCT_scaled_size;
-    h_out_group = cinfo->max_h_samp_factor;
-    v_out_group = cinfo->max_v_samp_factor;
-    upsample->rowgroup_height[ci] = v_in_group; /* save for use later */
-    need_buffer = TRUE;
-    if (! compptr->component_needed) {
-      /* Don't bother to upsample an uninteresting component. */
-      upsample->methods[ci] = noop_upsample;
-      need_buffer = FALSE;
-    } else if (h_in_group == h_out_group && v_in_group == v_out_group) {
-      /* Fullsize components can be processed without any work. */
-      upsample->methods[ci] = fullsize_upsample;
-      need_buffer = FALSE;
-    } else if (h_in_group * 2 == h_out_group &&
-	       v_in_group == v_out_group) {
-      /* Special cases for 2h1v upsampling */
-      if (do_fancy && compptr->downsampled_width > 2)
-	upsample->methods[ci] = h2v1_fancy_upsample;
-      else
-	upsample->methods[ci] = h2v1_upsample;
-    } else if (h_in_group * 2 == h_out_group &&
-	       v_in_group * 2 == v_out_group) {
-      /* Special cases for 2h2v upsampling */
-      if (do_fancy && compptr->downsampled_width > 2) {
-	upsample->methods[ci] = h2v2_fancy_upsample;
-	upsample->pub.need_context_rows = TRUE;
-      } else
-	upsample->methods[ci] = h2v2_upsample;
-    } else if ((h_out_group % h_in_group) == 0 &&
-	       (v_out_group % v_in_group) == 0) {
-      /* Generic integral-factors upsampling method */
-      upsample->methods[ci] = int_upsample;
-      upsample->h_expand[ci] = (UINT8) (h_out_group / h_in_group);
-      upsample->v_expand[ci] = (UINT8) (v_out_group / v_in_group);
-    } else
-      ERREXIT(cinfo, JERR_FRACT_SAMPLE_NOTIMPL);
-    if (need_buffer) {
-      upsample->color_buf[ci] = (*cinfo->mem->alloc_sarray)
-	((j_common_ptr) cinfo, JPOOL_IMAGE,
-	 (JDIMENSION) jround_up((long) cinfo->output_width,
-				(long) cinfo->max_h_samp_factor),
-	 (JDIMENSION) cinfo->max_v_samp_factor);
-    }
-  }
-}
@@ -1,270 +0,0 @@
-/*
- * jdtrans.c
- *
- * Copyright (C) 1995-1997, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains library routines for transcoding decompression,
- * that is, reading raw DCT coefficient arrays from an input JPEG file.
- * The routines in jdapimin.c will also be needed by a transcoder.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-
-
-/* Forward declarations */
-LOCAL(void) transdecode_master_selection JPP((j_decompress_ptr cinfo));
-
-
-/*
- * Read the coefficient arrays from a JPEG file.
- * jpeg_read_header must be completed before calling this.
- *
- * The entire image is read into a set of virtual coefficient-block arrays,
- * one per component.  The return value is a pointer to the array of
- * virtual-array descriptors.  These can be manipulated directly via the
- * JPEG memory manager, or handed off to jpeg_write_coefficients().
- * To release the memory occupied by the virtual arrays, call
- * jpeg_finish_decompress() when done with the data.
- *
- * An alternative usage is to simply obtain access to the coefficient arrays
- * during a buffered-image-mode decompression operation.  This is allowed
- * after any jpeg_finish_output() call.  The arrays can be accessed until
- * jpeg_finish_decompress() is called.  (Note that any call to the library
- * may reposition the arrays, so don't rely on access_virt_barray() results
- * to stay valid across library calls.)
- *
- * Returns NULL if suspended.  This case need be checked only if
- * a suspending data source is used.
- */
-
-GLOBAL(jvirt_barray_ptr *)
-jpeg_read_coefficients (j_decompress_ptr cinfo)
-{
-  if (cinfo->global_state == DSTATE_READY) {
-    /* First call: initialize active modules */
-    transdecode_master_selection(cinfo);
-    cinfo->global_state = DSTATE_RDCOEFS;
-  }
-  if (cinfo->global_state == DSTATE_RDCOEFS) {
-    /* Absorb whole file into the coef buffer */
-    for (;;) {
-      int retcode;
-      /* Call progress monitor hook if present */
-      if (cinfo->progress != NULL)
-        (*cinfo->progress->progress_monitor) ((j_common_ptr) cinfo);
-      /* Absorb some more input */
-      retcode = (*cinfo->inputctl->consume_input) (cinfo);
-      if (retcode == JPEG_SUSPENDED)
-        return NULL;
-      if (retcode == JPEG_REACHED_EOI)
-        break;
-      /* Advance progress counter if appropriate */
-      if (cinfo->progress != NULL &&
-	  (retcode == JPEG_ROW_COMPLETED || retcode == JPEG_REACHED_SOS)) {
-        if (++cinfo->progress->pass_counter >= cinfo->progress->pass_limit) {
-	  /* startup underestimated number of scans; ratchet up one scan */
-        cinfo->progress->pass_limit += (long) cinfo->total_iMCU_rows;
-        }
-      }
-    }
-    /* Set state so that jpeg_finish_decompress does the right thing */
-    cinfo->global_state = DSTATE_STOPPING;
-  }
-  /* At this point we should be in state DSTATE_STOPPING if being used
-   * standalone, or in state DSTATE_BUFIMAGE if being invoked to get access
-   * to the coefficients during a full buffered-image-mode decompression.
-   */
-  if ((cinfo->global_state == DSTATE_STOPPING ||
-       cinfo->global_state == DSTATE_BUFIMAGE) && cinfo->buffered_image) {
-    return cinfo->coef->coef_arrays;
-  }
-  /* Oops, improper usage */
-  ERREXIT1(cinfo, JERR_BAD_STATE, cinfo->global_state);
-  return NULL;			/* keep compiler happy */
-}
-
-LOCAL(boolean)
-jpeg_build_huffman_index_progressive(j_decompress_ptr cinfo,
-        huffman_index *index)
-{
-  if (cinfo->global_state == DSTATE_READY) {
-    printf("Progressive Mode\n");
-    /* First call: initialize active modules */
-    transdecode_master_selection(cinfo);
-    cinfo->global_state = DSTATE_RDCOEFS;
-  }
-  if (cinfo->global_state == DSTATE_RDCOEFS) {
-    int mcu, i;
-    cinfo->marker->get_sos_marker_position(cinfo, index);
-
-    /* Absorb whole file into the coef buffer */
-    for (mcu = 0; mcu < cinfo->total_iMCU_rows; mcu++) {
-      int retcode = 0;
-      /* Call progress monitor hook if present */
-      if (cinfo->progress != NULL)
-        (*cinfo->progress->progress_monitor) ((j_common_ptr) cinfo);
-      /* Absorb some more input */
-      jinit_phuff_decoder(cinfo);
-      for (i = 0; i < index->scan_count; i++) {
-        (*cinfo->inputctl->finish_input_pass) (cinfo);
-        jset_input_stream_position(cinfo, index->scan[i].bitstream_offset);
-        cinfo->unread_marker = 0;
-        retcode = (*cinfo->inputctl->consume_input_build_huffman_index)
-                    (cinfo, index, i);
-        if (retcode == JPEG_REACHED_EOI)
-          break;
-        cinfo->input_iMCU_row = mcu;
-        if (mcu != 0)
-          (*cinfo->entropy->configure_huffman_decoder)
-                (cinfo, index->scan[i].prev_MCU_offset);
-        cinfo->input_scan_number = i;
-        retcode = (*cinfo->inputctl->consume_input_build_huffman_index)
-                    (cinfo, index, i);
-      }
-      if (retcode == JPEG_SUSPENDED)
-        return FALSE;
-      if (retcode == JPEG_REACHED_EOI)
-        break;
-      /* Advance progress counter if appropriate */
-      if (cinfo->progress != NULL &&
-	  (retcode == JPEG_ROW_COMPLETED || retcode == JPEG_REACHED_SOS)) {
-        if (++cinfo->progress->pass_counter >= cinfo->progress->pass_limit) {
-	  /* startup underestimated number of scans; ratchet up one scan */
-          cinfo->progress->pass_limit += (long) cinfo->total_iMCU_rows;
-        }
-      }
-    }
-    cinfo->global_state = DSTATE_STOPPING;
-  }
-  /* At this point we should be in state DSTATE_STOPPING if being used
-   * standalone, or in state DSTATE_BUFIMAGE if being invoked to get access
-   * to the coefficients during a full buffered-image-mode decompression.
-   */
-  if ((cinfo->global_state == DSTATE_STOPPING ||
-       cinfo->global_state == DSTATE_BUFIMAGE) && cinfo->buffered_image) {
-    return TRUE;
-  }
-  /* Oops, improper usage */
-  ERREXIT1(cinfo, JERR_BAD_STATE, cinfo->global_state);
-  return FALSE;			/* keep compiler happy */
-}
-
-LOCAL(boolean)
-jpeg_build_huffman_index_baseline(j_decompress_ptr cinfo, huffman_index *index)
-{
-  if (cinfo->global_state == DSTATE_READY) {
-    printf("Baseline Mode\n");
-    /* First call: initialize active modules */
-    transdecode_master_selection(cinfo);
-    cinfo->global_state = DSTATE_RDCOEFS;
-  }
-  if (cinfo->global_state == DSTATE_RDCOEFS) {
-    /* Absorb whole file into the coef buffer */
-    for (;;) {
-      int retcode;
-      /* Call progress monitor hook if present */
-      if (cinfo->progress != NULL)
-        (*cinfo->progress->progress_monitor) ((j_common_ptr) cinfo);
-      /* Absorb some more input */
-      retcode = (*cinfo->inputctl->consume_input_build_huffman_index)
-                    (cinfo, index, 0);
-      if (retcode == JPEG_SUSPENDED)
-        return FALSE;
-      if (retcode == JPEG_REACHED_EOI)
-        break;
-      if (retcode == JPEG_SCAN_COMPLETED)
-        break;
-
-      /* Advance progress counter if appropriate */
-      if (cinfo->progress != NULL &&
-	  (retcode == JPEG_ROW_COMPLETED || retcode == JPEG_REACHED_SOS)) {
-        if (++cinfo->progress->pass_counter >= cinfo->progress->pass_limit) {
-	  /* startup underestimated number of scans; ratchet up one scan */
-        cinfo->progress->pass_limit += (long) cinfo->total_iMCU_rows;
-        }
-      }
-    }
-    /* Set state so that jpeg_finish_decompress does the right thing */
-    cinfo->global_state = DSTATE_STOPPING;
-  }
-  /* At this point we should be in state DSTATE_STOPPING if being used
-   * standalone, or in state DSTATE_BUFIMAGE if being invoked to get access
-   * to the coefficients during a full buffered-image-mode decompression.
-   */
-  if ((cinfo->global_state == DSTATE_STOPPING ||
-       cinfo->global_state == DSTATE_BUFIMAGE) && cinfo->buffered_image) {
-    return TRUE;
-  }
-  /* Oops, improper usage */
-  ERREXIT1(cinfo, JERR_BAD_STATE, cinfo->global_state);
-  return FALSE;			/* keep compiler happy */
-}
-
-GLOBAL(boolean)
-jpeg_build_huffman_index(j_decompress_ptr cinfo, huffman_index *index)
-{
-    cinfo->tile_decode = TRUE;
-    if (cinfo->progressive_mode)
-      return jpeg_build_huffman_index_progressive(cinfo, index);
-    else
-      return jpeg_build_huffman_index_baseline(cinfo, index);
-}
-
-/*
- * Master selection of decompression modules for transcoding.
- * This substitutes for jdmaster.c's initialization of the full decompressor.
- */
-
-LOCAL(void)
-transdecode_master_selection (j_decompress_ptr cinfo)
-{
-  /* This is effectively a buffered-image operation. */
-  cinfo->buffered_image = TRUE;
-
-  /* Entropy decoding: either Huffman or arithmetic coding. */
-  if (cinfo->arith_code) {
-    ERREXIT(cinfo, JERR_ARITH_NOTIMPL);
-  } else {
-    if (cinfo->progressive_mode) {
-#ifdef D_PROGRESSIVE_SUPPORTED
-      jinit_phuff_decoder(cinfo);
-#else
-      ERREXIT(cinfo, JERR_NOT_COMPILED);
-#endif
-    } else {
-      jinit_huff_decoder(cinfo);
-    }
-  }
-
-  /* Always get a full-image coefficient buffer. */
-  jinit_d_coef_controller(cinfo, TRUE);
-
-  /* We can now tell the memory manager to allocate virtual arrays. */
-  (*cinfo->mem->realize_virt_arrays) ((j_common_ptr) cinfo);
-
-  /* Initialize input side of decompressor to consume first scan. */
-  (*cinfo->inputctl->start_input_pass) (cinfo);
-
-  /* Initialize progress monitoring. */
-  if (cinfo->progress != NULL) {
-    int nscans;
-    /* Estimate number of scans to set pass_limit. */
-    if (cinfo->progressive_mode) {
-      /* Arbitrarily estimate 2 interleaved DC scans + 3 AC scans/component. */
-      nscans = 2 + 3 * cinfo->num_components;
-    } else if (cinfo->inputctl->has_multiple_scans) {
-      /* For a nonprogressive multiscan file, estimate 1 scan per component. */
-      nscans = cinfo->num_components;
-    } else {
-      nscans = 1;
-    }
-    cinfo->progress->pass_counter = 0L;
-    cinfo->progress->pass_limit = (long) cinfo->total_iMCU_rows * nscans;
-    cinfo->progress->completed_passes = 0;
-    cinfo->progress->total_passes = 1;
-  }
-}
@@ -1,252 +0,0 @@
-/*
- * jerror.c
- *
- * Copyright (C) 1991-1998, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains simple error-reporting and trace-message routines.
- * These are suitable for Unix-like systems and others where writing to
- * stderr is the right thing to do.  Many applications will want to replace
- * some or all of these routines.
- *
- * If you define USE_WINDOWS_MESSAGEBOX in jconfig.h or in the makefile,
- * you get a Windows-specific hack to display error messages in a dialog box.
- * It ain't much, but it beats dropping error messages into the bit bucket,
- * which is what happens to output to stderr under most Windows C compilers.
- *
- * These routines are used by both the compression and decompression code.
- */
-
-/* this is not a core library module, so it doesn't define JPEG_INTERNALS */
-#include "jinclude.h"
-#include "jpeglib.h"
-#include "jversion.h"
-#include "jerror.h"
-
-#ifdef USE_WINDOWS_MESSAGEBOX
-#include <windows.h>
-#endif
-
-#ifndef EXIT_FAILURE		/* define exit() codes if not provided */
-#define EXIT_FAILURE  1
-#endif
-
-
-/*
- * Create the message string table.
- * We do this from the master message list in jerror.h by re-reading
- * jerror.h with a suitable definition for macro JMESSAGE.
- * The message table is made an external symbol just in case any applications
- * want to refer to it directly.
- */
-
-#ifdef NEED_SHORT_EXTERNAL_NAMES
-#define jpeg_std_message_table	jMsgTable
-#endif
-
-#define JMESSAGE(code,string)	string ,
-
-const char * const jpeg_std_message_table[] = {
-#include "jerror.h"
-  NULL
-};
-
-
-/*
- * Error exit handler: must not return to caller.
- *
- * Applications may override this if they want to get control back after
- * an error.  Typically one would longjmp somewhere instead of exiting.
- * The setjmp buffer can be made a private field within an expanded error
- * handler object.  Note that the info needed to generate an error message
- * is stored in the error object, so you can generate the message now or
- * later, at your convenience.
- * You should make sure that the JPEG object is cleaned up (with jpeg_abort
- * or jpeg_destroy) at some point.
- */
-
-METHODDEF(void)
-error_exit (j_common_ptr cinfo)
-{
-  /* Always display the message */
-  (*cinfo->err->output_message) (cinfo);
-
-  /* Let the memory manager delete any temp files before we die */
-  jpeg_destroy(cinfo);
-
-  exit(EXIT_FAILURE);
-}
-
-
-/*
- * Actual output of an error or trace message.
- * Applications may override this method to send JPEG messages somewhere
- * other than stderr.
- *
- * On Windows, printing to stderr is generally completely useless,
- * so we provide optional code to produce an error-dialog popup.
- * Most Windows applications will still prefer to override this routine,
- * but if they don't, it'll do something at least marginally useful.
- *
- * NOTE: to use the library in an environment that doesn't support the
- * C stdio library, you may have to delete the call to fprintf() entirely,
- * not just not use this routine.
- */
-
-METHODDEF(void)
-output_message (j_common_ptr cinfo)
-{
-  char buffer[JMSG_LENGTH_MAX];
-
-  /* Create the message */
-  (*cinfo->err->format_message) (cinfo, buffer);
-
-#ifdef USE_WINDOWS_MESSAGEBOX
-  /* Display it in a message dialog box */
-  MessageBox(GetActiveWindow(), buffer, "JPEG Library Error",
-	     MB_OK | MB_ICONERROR);
-#else
-  /* Send it to stderr, adding a newline */
-  fprintf(stderr, "%s\n", buffer);
-#endif
-}
-
-
-/*
- * Decide whether to emit a trace or warning message.
- * msg_level is one of:
- *   -1: recoverable corrupt-data warning, may want to abort.
- *    0: important advisory messages (always display to user).
- *    1: first level of tracing detail.
- *    2,3,...: successively more detailed tracing messages.
- * An application might override this method if it wanted to abort on warnings
- * or change the policy about which messages to display.
- */
-
-METHODDEF(void)
-emit_message (j_common_ptr cinfo, int msg_level)
-{
-  struct jpeg_error_mgr * err = cinfo->err;
-
-  if (msg_level < 0) {
-    /* It's a warning message.  Since corrupt files may generate many warnings,
-     * the policy implemented here is to show only the first warning,
-     * unless trace_level >= 3.
-     */
-    if (err->num_warnings == 0 || err->trace_level >= 3)
-      (*err->output_message) (cinfo);
-    /* Always count warnings in num_warnings. */
-    err->num_warnings++;
-  } else {
-    /* It's a trace message.  Show it if trace_level >= msg_level. */
-    if (err->trace_level >= msg_level)
-      (*err->output_message) (cinfo);
-  }
-}
-
-
-/*
- * Format a message string for the most recent JPEG error or message.
- * The message is stored into buffer, which should be at least JMSG_LENGTH_MAX
- * characters.  Note that no '\n' character is added to the string.
- * Few applications should need to override this method.
- */
-
-METHODDEF(void)
-format_message (j_common_ptr cinfo, char * buffer)
-{
-  struct jpeg_error_mgr * err = cinfo->err;
-  int msg_code = err->msg_code;
-  const char * msgtext = NULL;
-  const char * msgptr;
-  char ch;
-  boolean isstring;
-
-  /* Look up message string in proper table */
-  if (msg_code > 0 && msg_code <= err->last_jpeg_message) {
-    msgtext = err->jpeg_message_table[msg_code];
-  } else if (err->addon_message_table != NULL &&
-	     msg_code >= err->first_addon_message &&
-	     msg_code <= err->last_addon_message) {
-    msgtext = err->addon_message_table[msg_code - err->first_addon_message];
-  }
-
-  /* Defend against bogus message number */
-  if (msgtext == NULL) {
-    err->msg_parm.i[0] = msg_code;
-    msgtext = err->jpeg_message_table[0];
-  }
-
-  /* Check for string parameter, as indicated by %s in the message text */
-  isstring = FALSE;
-  msgptr = msgtext;
-  while ((ch = *msgptr++) != '\0') {
-    if (ch == '%') {
-      if (*msgptr == 's') isstring = TRUE;
-      break;
-    }
-  }
-
-  /* Format the message into the passed buffer */
-  if (isstring)
-    sprintf(buffer, msgtext, err->msg_parm.s);
-  else
-    sprintf(buffer, msgtext,
-	    err->msg_parm.i[0], err->msg_parm.i[1],
-	    err->msg_parm.i[2], err->msg_parm.i[3],
-	    err->msg_parm.i[4], err->msg_parm.i[5],
-	    err->msg_parm.i[6], err->msg_parm.i[7]);
-}
-
-
-/*
- * Reset error state variables at start of a new image.
- * This is called during compression startup to reset trace/error
- * processing to default state, without losing any application-specific
- * method pointers.  An application might possibly want to override
- * this method if it has additional error processing state.
- */
-
-METHODDEF(void)
-reset_error_mgr (j_common_ptr cinfo)
-{
-  cinfo->err->num_warnings = 0;
-  /* trace_level is not reset since it is an application-supplied parameter */
-  cinfo->err->msg_code = 0;	/* may be useful as a flag for "no error" */
-}
-
-
-/*
- * Fill in the standard error-handling methods in a jpeg_error_mgr object.
- * Typical call is:
- *	struct jpeg_compress_struct cinfo;
- *	struct jpeg_error_mgr err;
- *
- *	cinfo.err = jpeg_std_error(&err);
- * after which the application may override some of the methods.
- */
-
-GLOBAL(struct jpeg_error_mgr *)
-jpeg_std_error (struct jpeg_error_mgr * err)
-{
-  err->error_exit = error_exit;
-  err->emit_message = emit_message;
-  err->output_message = output_message;
-  err->format_message = format_message;
-  err->reset_error_mgr = reset_error_mgr;
-
-  err->trace_level = 0;		/* default = no tracing */
-  err->num_warnings = 0;	/* no warnings emitted yet */
-  err->msg_code = 0;		/* may be useful as a flag for "no error" */
-
-  /* Initialize message table pointers */
-  err->jpeg_message_table = jpeg_std_message_table;
-  err->last_jpeg_message = (int) JMSG_LASTMSGCODE - 1;
-
-  err->addon_message_table = NULL;
-  err->first_addon_message = 0;	/* for safety */
-  err->last_addon_message = 0;
-
-  return err;
-}
@@ -1,291 +0,0 @@
-/*
- * jerror.h
- *
- * Copyright (C) 1994-1997, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file defines the error and message codes for the JPEG library.
- * Edit this file to add new codes, or to translate the message strings to
- * some other language.
- * A set of error-reporting macros are defined too.  Some applications using
- * the JPEG library may wish to include this file to get the error codes
- * and/or the macros.
- */
-
-/*
- * To define the enum list of message codes, include this file without
- * defining macro JMESSAGE.  To create a message string table, include it
- * again with a suitable JMESSAGE definition (see jerror.c for an example).
- */
-#ifndef JMESSAGE
-#ifndef JERROR_H
-/* First time through, define the enum list */
-#define JMAKE_ENUM_LIST
-#else
-/* Repeated inclusions of this file are no-ops unless JMESSAGE is defined */
-#define JMESSAGE(code,string)
-#endif /* JERROR_H */
-#endif /* JMESSAGE */
-
-#ifdef JMAKE_ENUM_LIST
-
-typedef enum {
-
-#define JMESSAGE(code,string)	code ,
-
-#endif /* JMAKE_ENUM_LIST */
-
-JMESSAGE(JMSG_NOMESSAGE, "Bogus message code %d") /* Must be first entry! */
-
-/* For maintenance convenience, list is alphabetical by message code name */
-JMESSAGE(JERR_ARITH_NOTIMPL,
-	 "Sorry, there are legal restrictions on arithmetic coding")
-JMESSAGE(JERR_BAD_ALIGN_TYPE, "ALIGN_TYPE is wrong, please fix")
-JMESSAGE(JERR_BAD_ALLOC_CHUNK, "MAX_ALLOC_CHUNK is wrong, please fix")
-JMESSAGE(JERR_BAD_BUFFER_MODE, "Bogus buffer control mode")
-JMESSAGE(JERR_BAD_COMPONENT_ID, "Invalid component ID %d in SOS")
-JMESSAGE(JERR_BAD_DCT_COEF, "DCT coefficient out of range")
-JMESSAGE(JERR_BAD_DCTSIZE, "IDCT output block size %d not supported")
-JMESSAGE(JERR_BAD_HUFF_TABLE, "Bogus Huffman table definition")
-JMESSAGE(JERR_BAD_IN_COLORSPACE, "Bogus input colorspace")
-JMESSAGE(JERR_BAD_J_COLORSPACE, "Bogus JPEG colorspace")
-JMESSAGE(JERR_BAD_LENGTH, "Bogus marker length")
-JMESSAGE(JERR_BAD_LIB_VERSION,
-	 "Wrong JPEG library version: library is %d, caller expects %d")
-JMESSAGE(JERR_BAD_MCU_SIZE, "Sampling factors too large for interleaved scan")
-JMESSAGE(JERR_BAD_POOL_ID, "Invalid memory pool code %d")
-JMESSAGE(JERR_BAD_PRECISION, "Unsupported JPEG data precision %d")
-JMESSAGE(JERR_BAD_PROGRESSION,
-	 "Invalid progressive parameters Ss=%d Se=%d Ah=%d Al=%d")
-JMESSAGE(JERR_BAD_PROG_SCRIPT,
-	 "Invalid progressive parameters at scan script entry %d")
-JMESSAGE(JERR_BAD_SAMPLING, "Bogus sampling factors")
-JMESSAGE(JERR_BAD_SCAN_SCRIPT, "Invalid scan script at entry %d")
-JMESSAGE(JERR_BAD_STATE, "Improper call to JPEG library in state %d")
-JMESSAGE(JERR_BAD_STRUCT_SIZE,
-	 "JPEG parameter struct mismatch: library thinks size is %u, caller expects %u")
-JMESSAGE(JERR_BAD_VIRTUAL_ACCESS, "Bogus virtual array access")
-JMESSAGE(JERR_BUFFER_SIZE, "Buffer passed to JPEG library is too small")
-JMESSAGE(JERR_CANT_SUSPEND, "Suspension not allowed here")
-JMESSAGE(JERR_CCIR601_NOTIMPL, "CCIR601 sampling not implemented yet")
-JMESSAGE(JERR_COMPONENT_COUNT, "Too many color components: %d, max %d")
-JMESSAGE(JERR_CONVERSION_NOTIMPL, "Unsupported color conversion request")
-JMESSAGE(JERR_DAC_INDEX, "Bogus DAC index %d")
-JMESSAGE(JERR_DAC_VALUE, "Bogus DAC value 0x%x")
-JMESSAGE(JERR_DHT_INDEX, "Bogus DHT index %d")
-JMESSAGE(JERR_DQT_INDEX, "Bogus DQT index %d")
-JMESSAGE(JERR_EMPTY_IMAGE, "Empty JPEG image (DNL not supported)")
-JMESSAGE(JERR_EMS_READ, "Read from EMS failed")
-JMESSAGE(JERR_EMS_WRITE, "Write to EMS failed")
-JMESSAGE(JERR_EOI_EXPECTED, "Didn't expect more than one scan")
-JMESSAGE(JERR_FILE_READ, "Input file read error")
-JMESSAGE(JERR_FILE_WRITE, "Output file write error --- out of disk space?")
-JMESSAGE(JERR_FRACT_SAMPLE_NOTIMPL, "Fractional sampling not implemented yet")
-JMESSAGE(JERR_HUFF_CLEN_OVERFLOW, "Huffman code size table overflow")
-JMESSAGE(JERR_HUFF_MISSING_CODE, "Missing Huffman code table entry")
-JMESSAGE(JERR_IMAGE_TOO_BIG, "Maximum supported image dimension is %u pixels")
-JMESSAGE(JERR_INPUT_EMPTY, "Empty input file")
-JMESSAGE(JERR_INPUT_EOF, "Premature end of input file")
-JMESSAGE(JERR_MISMATCHED_QUANT_TABLE,
-	 "Cannot transcode due to multiple use of quantization table %d")
-JMESSAGE(JERR_MISSING_DATA, "Scan script does not transmit all data")
-JMESSAGE(JERR_MODE_CHANGE, "Invalid color quantization mode change")
-JMESSAGE(JERR_NOTIMPL, "Not implemented yet")
-JMESSAGE(JERR_NOT_COMPILED, "Requested feature was omitted at compile time")
-JMESSAGE(JERR_NO_BACKING_STORE, "Backing store not supported")
-JMESSAGE(JERR_NO_HUFF_TABLE, "Huffman table 0x%02x was not defined")
-JMESSAGE(JERR_NO_IMAGE, "JPEG datastream contains no image")
-JMESSAGE(JERR_NO_QUANT_TABLE, "Quantization table 0x%02x was not defined")
-JMESSAGE(JERR_NO_SOI, "Not a JPEG file: starts with 0x%02x 0x%02x")
-JMESSAGE(JERR_OUT_OF_MEMORY, "Insufficient memory (case %d)")
-JMESSAGE(JERR_QUANT_COMPONENTS,
-	 "Cannot quantize more than %d color components")
-JMESSAGE(JERR_QUANT_FEW_COLORS, "Cannot quantize to fewer than %d colors")
-JMESSAGE(JERR_QUANT_MANY_COLORS, "Cannot quantize to more than %d colors")
-JMESSAGE(JERR_SOF_DUPLICATE, "Invalid JPEG file structure: two SOF markers")
-JMESSAGE(JERR_SOF_NO_SOS, "Invalid JPEG file structure: missing SOS marker")
-JMESSAGE(JERR_SOF_UNSUPPORTED, "Unsupported JPEG process: SOF type 0x%02x")
-JMESSAGE(JERR_SOI_DUPLICATE, "Invalid JPEG file structure: two SOI markers")
-JMESSAGE(JERR_SOS_NO_SOF, "Invalid JPEG file structure: SOS before SOF")
-JMESSAGE(JERR_TFILE_CREATE, "Failed to create temporary file %s")
-JMESSAGE(JERR_TFILE_READ, "Read failed on temporary file")
-JMESSAGE(JERR_TFILE_SEEK, "Seek failed on temporary file")
-JMESSAGE(JERR_TFILE_WRITE,
-	 "Write failed on temporary file --- out of disk space?")
-JMESSAGE(JERR_TOO_LITTLE_DATA, "Application transferred too few scanlines")
-JMESSAGE(JERR_UNKNOWN_MARKER, "Unsupported marker type 0x%02x")
-JMESSAGE(JERR_VIRTUAL_BUG, "Virtual array controller messed up")
-JMESSAGE(JERR_WIDTH_OVERFLOW, "Image too wide for this implementation")
-JMESSAGE(JERR_XMS_READ, "Read from XMS failed")
-JMESSAGE(JERR_XMS_WRITE, "Write to XMS failed")
-JMESSAGE(JMSG_COPYRIGHT, JCOPYRIGHT)
-JMESSAGE(JMSG_VERSION, JVERSION)
-JMESSAGE(JTRC_16BIT_TABLES,
-	 "Caution: quantization tables are too coarse for baseline JPEG")
-JMESSAGE(JTRC_ADOBE,
-	 "Adobe APP14 marker: version %d, flags 0x%04x 0x%04x, transform %d")
-JMESSAGE(JTRC_APP0, "Unknown APP0 marker (not JFIF), length %u")
-JMESSAGE(JTRC_APP14, "Unknown APP14 marker (not Adobe), length %u")
-JMESSAGE(JTRC_DAC, "Define Arithmetic Table 0x%02x: 0x%02x")
-JMESSAGE(JTRC_DHT, "Define Huffman Table 0x%02x")
-JMESSAGE(JTRC_DQT, "Define Quantization Table %d  precision %d")
-JMESSAGE(JTRC_DRI, "Define Restart Interval %u")
-JMESSAGE(JTRC_EMS_CLOSE, "Freed EMS handle %u")
-JMESSAGE(JTRC_EMS_OPEN, "Obtained EMS handle %u")
-JMESSAGE(JTRC_EOI, "End Of Image")
-JMESSAGE(JTRC_HUFFBITS, "        %3d %3d %3d %3d %3d %3d %3d %3d")
-JMESSAGE(JTRC_JFIF, "JFIF APP0 marker: version %d.%02d, density %dx%d  %d")
-JMESSAGE(JTRC_JFIF_BADTHUMBNAILSIZE,
-	 "Warning: thumbnail image size does not match data length %u")
-JMESSAGE(JTRC_JFIF_EXTENSION,
-	 "JFIF extension marker: type 0x%02x, length %u")
-JMESSAGE(JTRC_JFIF_THUMBNAIL, "    with %d x %d thumbnail image")
-JMESSAGE(JTRC_MISC_MARKER, "Miscellaneous marker 0x%02x, length %u")
-JMESSAGE(JTRC_PARMLESS_MARKER, "Unexpected marker 0x%02x")
-JMESSAGE(JTRC_QUANTVALS, "        %4u %4u %4u %4u %4u %4u %4u %4u")
-JMESSAGE(JTRC_QUANT_3_NCOLORS, "Quantizing to %d = %d*%d*%d colors")
-JMESSAGE(JTRC_QUANT_NCOLORS, "Quantizing to %d colors")
-JMESSAGE(JTRC_QUANT_SELECTED, "Selected %d colors for quantization")
-JMESSAGE(JTRC_RECOVERY_ACTION, "At marker 0x%02x, recovery action %d")
-JMESSAGE(JTRC_RST, "RST%d")
-JMESSAGE(JTRC_SMOOTH_NOTIMPL,
-	 "Smoothing not supported with nonstandard sampling ratios")
-JMESSAGE(JTRC_SOF, "Start Of Frame 0x%02x: width=%u, height=%u, components=%d")
-JMESSAGE(JTRC_SOF_COMPONENT, "    Component %d: %dhx%dv q=%d")
-JMESSAGE(JTRC_SOI, "Start of Image")
-JMESSAGE(JTRC_SOS, "Start Of Scan: %d components")
-JMESSAGE(JTRC_SOS_COMPONENT, "    Component %d: dc=%d ac=%d")
-JMESSAGE(JTRC_SOS_PARAMS, "  Ss=%d, Se=%d, Ah=%d, Al=%d")
-JMESSAGE(JTRC_TFILE_CLOSE, "Closed temporary file %s")
-JMESSAGE(JTRC_TFILE_OPEN, "Opened temporary file %s")
-JMESSAGE(JTRC_THUMB_JPEG,
-	 "JFIF extension marker: JPEG-compressed thumbnail image, length %u")
-JMESSAGE(JTRC_THUMB_PALETTE,
-	 "JFIF extension marker: palette thumbnail image, length %u")
-JMESSAGE(JTRC_THUMB_RGB,
-	 "JFIF extension marker: RGB thumbnail image, length %u")
-JMESSAGE(JTRC_UNKNOWN_IDS,
-	 "Unrecognized component IDs %d %d %d, assuming YCbCr")
-JMESSAGE(JTRC_XMS_CLOSE, "Freed XMS handle %u")
-JMESSAGE(JTRC_XMS_OPEN, "Obtained XMS handle %u")
-JMESSAGE(JWRN_ADOBE_XFORM, "Unknown Adobe color transform code %d")
-JMESSAGE(JWRN_BOGUS_PROGRESSION,
-	 "Inconsistent progression sequence for component %d coefficient %d")
-JMESSAGE(JWRN_EXTRANEOUS_DATA,
-	 "Corrupt JPEG data: %u extraneous bytes before marker 0x%02x")
-JMESSAGE(JWRN_HIT_MARKER, "Corrupt JPEG data: premature end of data segment")
-JMESSAGE(JWRN_HUFF_BAD_CODE, "Corrupt JPEG data: bad Huffman code")
-JMESSAGE(JWRN_JFIF_MAJOR, "Warning: unknown JFIF revision number %d.%02d")
-JMESSAGE(JWRN_JPEG_EOF, "Premature end of JPEG file")
-JMESSAGE(JWRN_MUST_RESYNC,
-	 "Corrupt JPEG data: found marker 0x%02x instead of RST%d")
-JMESSAGE(JWRN_NOT_SEQUENTIAL, "Invalid SOS parameters for sequential JPEG")
-JMESSAGE(JWRN_TOO_MUCH_DATA, "Application transferred too many scanlines")
-
-#ifdef JMAKE_ENUM_LIST
-
-  JMSG_LASTMSGCODE
-} J_MESSAGE_CODE;
-
-#undef JMAKE_ENUM_LIST
-#endif /* JMAKE_ENUM_LIST */
-
-/* Zap JMESSAGE macro so that future re-inclusions do nothing by default */
-#undef JMESSAGE
-
-
-#ifndef JERROR_H
-#define JERROR_H
-
-/* Macros to simplify using the error and trace message stuff */
-/* The first parameter is either type of cinfo pointer */
-
-/* Fatal errors (print message and exit) */
-#define ERREXIT(cinfo,code)  \
-  ((cinfo)->err->msg_code = (code), \
-   (*(cinfo)->err->error_exit) ((j_common_ptr) (cinfo)))
-#define ERREXIT1(cinfo,code,p1)  \
-  ((cinfo)->err->msg_code = (code), \
-   (cinfo)->err->msg_parm.i[0] = (p1), \
-   (*(cinfo)->err->error_exit) ((j_common_ptr) (cinfo)))
-#define ERREXIT2(cinfo,code,p1,p2)  \
-  ((cinfo)->err->msg_code = (code), \
-   (cinfo)->err->msg_parm.i[0] = (p1), \
-   (cinfo)->err->msg_parm.i[1] = (p2), \
-   (*(cinfo)->err->error_exit) ((j_common_ptr) (cinfo)))
-#define ERREXIT3(cinfo,code,p1,p2,p3)  \
-  ((cinfo)->err->msg_code = (code), \
-   (cinfo)->err->msg_parm.i[0] = (p1), \
-   (cinfo)->err->msg_parm.i[1] = (p2), \
-   (cinfo)->err->msg_parm.i[2] = (p3), \
-   (*(cinfo)->err->error_exit) ((j_common_ptr) (cinfo)))
-#define ERREXIT4(cinfo,code,p1,p2,p3,p4)  \
-  ((cinfo)->err->msg_code = (code), \
-   (cinfo)->err->msg_parm.i[0] = (p1), \
-   (cinfo)->err->msg_parm.i[1] = (p2), \
-   (cinfo)->err->msg_parm.i[2] = (p3), \
-   (cinfo)->err->msg_parm.i[3] = (p4), \
-   (*(cinfo)->err->error_exit) ((j_common_ptr) (cinfo)))
-#define ERREXITS(cinfo,code,str)  \
-  ((cinfo)->err->msg_code = (code), \
-   strncpy((cinfo)->err->msg_parm.s, (str), JMSG_STR_PARM_MAX), \
-   (*(cinfo)->err->error_exit) ((j_common_ptr) (cinfo)))
-
-#define MAKESTMT(stuff)		do { stuff } while (0)
-
-/* Nonfatal errors (we can keep going, but the data is probably corrupt) */
-#define WARNMS(cinfo,code)  \
-  ((cinfo)->err->msg_code = (code), \
-   (*(cinfo)->err->emit_message) ((j_common_ptr) (cinfo), -1))
-#define WARNMS1(cinfo,code,p1)  \
-  ((cinfo)->err->msg_code = (code), \
-   (cinfo)->err->msg_parm.i[0] = (p1), \
-   (*(cinfo)->err->emit_message) ((j_common_ptr) (cinfo), -1))
-#define WARNMS2(cinfo,code,p1,p2)  \
-  ((cinfo)->err->msg_code = (code), \
-   (cinfo)->err->msg_parm.i[0] = (p1), \
-   (cinfo)->err->msg_parm.i[1] = (p2), \
-   (*(cinfo)->err->emit_message) ((j_common_ptr) (cinfo), -1))
-
-/* Informational/debugging messages */
-#define TRACEMS(cinfo,lvl,code)  \
-  ((cinfo)->err->msg_code = (code), \
-   (*(cinfo)->err->emit_message) ((j_common_ptr) (cinfo), (lvl)))
-#define TRACEMS1(cinfo,lvl,code,p1)  \
-  ((cinfo)->err->msg_code = (code), \
-   (cinfo)->err->msg_parm.i[0] = (p1), \
-   (*(cinfo)->err->emit_message) ((j_common_ptr) (cinfo), (lvl)))
-#define TRACEMS2(cinfo,lvl,code,p1,p2)  \
-  ((cinfo)->err->msg_code = (code), \
-   (cinfo)->err->msg_parm.i[0] = (p1), \
-   (cinfo)->err->msg_parm.i[1] = (p2), \
-   (*(cinfo)->err->emit_message) ((j_common_ptr) (cinfo), (lvl)))
-#define TRACEMS3(cinfo,lvl,code,p1,p2,p3)  \
-  MAKESTMT(int * _mp = (cinfo)->err->msg_parm.i; \
-	   _mp[0] = (p1); _mp[1] = (p2); _mp[2] = (p3); \
-	   (cinfo)->err->msg_code = (code); \
-	   (*(cinfo)->err->emit_message) ((j_common_ptr) (cinfo), (lvl)); )
-#define TRACEMS4(cinfo,lvl,code,p1,p2,p3,p4)  \
-  MAKESTMT(int * _mp = (cinfo)->err->msg_parm.i; \
-	   _mp[0] = (p1); _mp[1] = (p2); _mp[2] = (p3); _mp[3] = (p4); \
-	   (cinfo)->err->msg_code = (code); \
-	   (*(cinfo)->err->emit_message) ((j_common_ptr) (cinfo), (lvl)); )
-#define TRACEMS5(cinfo,lvl,code,p1,p2,p3,p4,p5)  \
-  MAKESTMT(int * _mp = (cinfo)->err->msg_parm.i; \
-	   _mp[0] = (p1); _mp[1] = (p2); _mp[2] = (p3); _mp[3] = (p4); \
-	   _mp[4] = (p5); \
-	   (cinfo)->err->msg_code = (code); \
-	   (*(cinfo)->err->emit_message) ((j_common_ptr) (cinfo), (lvl)); )
-#define TRACEMS8(cinfo,lvl,code,p1,p2,p3,p4,p5,p6,p7,p8)  \
-  MAKESTMT(int * _mp = (cinfo)->err->msg_parm.i; \
-	   _mp[0] = (p1); _mp[1] = (p2); _mp[2] = (p3); _mp[3] = (p4); \
-	   _mp[4] = (p5); _mp[5] = (p6); _mp[6] = (p7); _mp[7] = (p8); \
-	   (cinfo)->err->msg_code = (code); \
-	   (*(cinfo)->err->emit_message) ((j_common_ptr) (cinfo), (lvl)); )
-#define TRACEMSS(cinfo,lvl,code,str)  \
-  ((cinfo)->err->msg_code = (code), \
-   strncpy((cinfo)->err->msg_parm.s, (str), JMSG_STR_PARM_MAX), \
-   (*(cinfo)->err->emit_message) ((j_common_ptr) (cinfo), (lvl)))
-
-#endif /* JERROR_H */
@@ -1,168 +0,0 @@
-/*
- * jfdctflt.c
- *
- * Copyright (C) 1994-1996, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains a floating-point implementation of the
- * forward DCT (Discrete Cosine Transform).
- *
- * This implementation should be more accurate than either of the integer
- * DCT implementations.  However, it may not give the same results on all
- * machines because of differences in roundoff behavior.  Speed will depend
- * on the hardware's floating point capacity.
- *
- * A 2-D DCT can be done by 1-D DCT on each row followed by 1-D DCT
- * on each column.  Direct algorithms are also available, but they are
- * much more complex and seem not to be any faster when reduced to code.
- *
- * This implementation is based on Arai, Agui, and Nakajima's algorithm for
- * scaled DCT.  Their original paper (Trans. IEICE E-71(11):1095) is in
- * Japanese, but the algorithm is described in the Pennebaker & Mitchell
- * JPEG textbook (see REFERENCES section in file README).  The following code
- * is based directly on figure 4-8 in P&M.
- * While an 8-point DCT cannot be done in less than 11 multiplies, it is
- * possible to arrange the computation so that many of the multiplies are
- * simple scalings of the final outputs.  These multiplies can then be
- * folded into the multiplications or divisions by the JPEG quantization
- * table entries.  The AA&N method leaves only 5 multiplies and 29 adds
- * to be done in the DCT itself.
- * The primary disadvantage of this method is that with a fixed-point
- * implementation, accuracy is lost due to imprecise representation of the
- * scaled quantization values.  However, that problem does not arise if
- * we use floating point arithmetic.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-#include "jdct.h"		/* Private declarations for DCT subsystem */
-
-#ifdef DCT_FLOAT_SUPPORTED
-
-
-/*
- * This module is specialized to the case DCTSIZE = 8.
- */
-
-#if DCTSIZE != 8
-  Sorry, this code only copes with 8x8 DCTs. /* deliberate syntax err */
-#endif
-
-
-/*
- * Perform the forward DCT on one block of samples.
- */
-
-GLOBAL(void)
-jpeg_fdct_float (FAST_FLOAT * data)
-{
-  FAST_FLOAT tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
-  FAST_FLOAT tmp10, tmp11, tmp12, tmp13;
-  FAST_FLOAT z1, z2, z3, z4, z5, z11, z13;
-  FAST_FLOAT *dataptr;
-  int ctr;
-
-  /* Pass 1: process rows. */
-
-  dataptr = data;
-  for (ctr = DCTSIZE-1; ctr >= 0; ctr--) {
-    tmp0 = dataptr[0] + dataptr[7];
-    tmp7 = dataptr[0] - dataptr[7];
-    tmp1 = dataptr[1] + dataptr[6];
-    tmp6 = dataptr[1] - dataptr[6];
-    tmp2 = dataptr[2] + dataptr[5];
-    tmp5 = dataptr[2] - dataptr[5];
-    tmp3 = dataptr[3] + dataptr[4];
-    tmp4 = dataptr[3] - dataptr[4];
-    
-    /* Even part */
-    
-    tmp10 = tmp0 + tmp3;	/* phase 2 */
-    tmp13 = tmp0 - tmp3;
-    tmp11 = tmp1 + tmp2;
-    tmp12 = tmp1 - tmp2;
-    
-    dataptr[0] = tmp10 + tmp11; /* phase 3 */
-    dataptr[4] = tmp10 - tmp11;
-    
-    z1 = (tmp12 + tmp13) * ((FAST_FLOAT) 0.707106781); /* c4 */
-    dataptr[2] = tmp13 + z1;	/* phase 5 */
-    dataptr[6] = tmp13 - z1;
-    
-    /* Odd part */
-
-    tmp10 = tmp4 + tmp5;	/* phase 2 */
-    tmp11 = tmp5 + tmp6;
-    tmp12 = tmp6 + tmp7;
-
-    /* The rotator is modified from fig 4-8 to avoid extra negations. */
-    z5 = (tmp10 - tmp12) * ((FAST_FLOAT) 0.382683433); /* c6 */
-    z2 = ((FAST_FLOAT) 0.541196100) * tmp10 + z5; /* c2-c6 */
-    z4 = ((FAST_FLOAT) 1.306562965) * tmp12 + z5; /* c2+c6 */
-    z3 = tmp11 * ((FAST_FLOAT) 0.707106781); /* c4 */
-
-    z11 = tmp7 + z3;		/* phase 5 */
-    z13 = tmp7 - z3;
-
-    dataptr[5] = z13 + z2;	/* phase 6 */
-    dataptr[3] = z13 - z2;
-    dataptr[1] = z11 + z4;
-    dataptr[7] = z11 - z4;
-
-    dataptr += DCTSIZE;		/* advance pointer to next row */
-  }
-
-  /* Pass 2: process columns. */
-
-  dataptr = data;
-  for (ctr = DCTSIZE-1; ctr >= 0; ctr--) {
-    tmp0 = dataptr[DCTSIZE*0] + dataptr[DCTSIZE*7];
-    tmp7 = dataptr[DCTSIZE*0] - dataptr[DCTSIZE*7];
-    tmp1 = dataptr[DCTSIZE*1] + dataptr[DCTSIZE*6];
-    tmp6 = dataptr[DCTSIZE*1] - dataptr[DCTSIZE*6];
-    tmp2 = dataptr[DCTSIZE*2] + dataptr[DCTSIZE*5];
-    tmp5 = dataptr[DCTSIZE*2] - dataptr[DCTSIZE*5];
-    tmp3 = dataptr[DCTSIZE*3] + dataptr[DCTSIZE*4];
-    tmp4 = dataptr[DCTSIZE*3] - dataptr[DCTSIZE*4];
-    
-    /* Even part */
-    
-    tmp10 = tmp0 + tmp3;	/* phase 2 */
-    tmp13 = tmp0 - tmp3;
-    tmp11 = tmp1 + tmp2;
-    tmp12 = tmp1 - tmp2;
-    
-    dataptr[DCTSIZE*0] = tmp10 + tmp11; /* phase 3 */
-    dataptr[DCTSIZE*4] = tmp10 - tmp11;
-    
-    z1 = (tmp12 + tmp13) * ((FAST_FLOAT) 0.707106781); /* c4 */
-    dataptr[DCTSIZE*2] = tmp13 + z1; /* phase 5 */
-    dataptr[DCTSIZE*6] = tmp13 - z1;
-    
-    /* Odd part */
-
-    tmp10 = tmp4 + tmp5;	/* phase 2 */
-    tmp11 = tmp5 + tmp6;
-    tmp12 = tmp6 + tmp7;
-
-    /* The rotator is modified from fig 4-8 to avoid extra negations. */
-    z5 = (tmp10 - tmp12) * ((FAST_FLOAT) 0.382683433); /* c6 */
-    z2 = ((FAST_FLOAT) 0.541196100) * tmp10 + z5; /* c2-c6 */
-    z4 = ((FAST_FLOAT) 1.306562965) * tmp12 + z5; /* c2+c6 */
-    z3 = tmp11 * ((FAST_FLOAT) 0.707106781); /* c4 */
-
-    z11 = tmp7 + z3;		/* phase 5 */
-    z13 = tmp7 - z3;
-
-    dataptr[DCTSIZE*5] = z13 + z2; /* phase 6 */
-    dataptr[DCTSIZE*3] = z13 - z2;
-    dataptr[DCTSIZE*1] = z11 + z4;
-    dataptr[DCTSIZE*7] = z11 - z4;
-
-    dataptr++;			/* advance pointer to next column */
-  }
-}
-
-#endif /* DCT_FLOAT_SUPPORTED */
@@ -1,224 +0,0 @@
-/*
- * jfdctfst.c
- *
- * Copyright (C) 1994-1996, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains a fast, not so accurate integer implementation of the
- * forward DCT (Discrete Cosine Transform).
- *
- * A 2-D DCT can be done by 1-D DCT on each row followed by 1-D DCT
- * on each column.  Direct algorithms are also available, but they are
- * much more complex and seem not to be any faster when reduced to code.
- *
- * This implementation is based on Arai, Agui, and Nakajima's algorithm for
- * scaled DCT.  Their original paper (Trans. IEICE E-71(11):1095) is in
- * Japanese, but the algorithm is described in the Pennebaker & Mitchell
- * JPEG textbook (see REFERENCES section in file README).  The following code
- * is based directly on figure 4-8 in P&M.
- * While an 8-point DCT cannot be done in less than 11 multiplies, it is
- * possible to arrange the computation so that many of the multiplies are
- * simple scalings of the final outputs.  These multiplies can then be
- * folded into the multiplications or divisions by the JPEG quantization
- * table entries.  The AA&N method leaves only 5 multiplies and 29 adds
- * to be done in the DCT itself.
- * The primary disadvantage of this method is that with fixed-point math,
- * accuracy is lost due to imprecise representation of the scaled
- * quantization values.  The smaller the quantization table entry, the less
- * precise the scaled value, so this implementation does worse with high-
- * quality-setting files than with low-quality ones.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-#include "jdct.h"		/* Private declarations for DCT subsystem */
-
-#ifdef DCT_IFAST_SUPPORTED
-
-
-/*
- * This module is specialized to the case DCTSIZE = 8.
- */
-
-#if DCTSIZE != 8
-  Sorry, this code only copes with 8x8 DCTs. /* deliberate syntax err */
-#endif
-
-
-/* Scaling decisions are generally the same as in the LL&M algorithm;
- * see jfdctint.c for more details.  However, we choose to descale
- * (right shift) multiplication products as soon as they are formed,
- * rather than carrying additional fractional bits into subsequent additions.
- * This compromises accuracy slightly, but it lets us save a few shifts.
- * More importantly, 16-bit arithmetic is then adequate (for 8-bit samples)
- * everywhere except in the multiplications proper; this saves a good deal
- * of work on 16-bit-int machines.
- *
- * Again to save a few shifts, the intermediate results between pass 1 and
- * pass 2 are not upscaled, but are represented only to integral precision.
- *
- * A final compromise is to represent the multiplicative constants to only
- * 8 fractional bits, rather than 13.  This saves some shifting work on some
- * machines, and may also reduce the cost of multiplication (since there
- * are fewer one-bits in the constants).
- */
-
-#define CONST_BITS  8
-
-
-/* Some C compilers fail to reduce "FIX(constant)" at compile time, thus
- * causing a lot of useless floating-point operations at run time.
- * To get around this we use the following pre-calculated constants.
- * If you change CONST_BITS you may want to add appropriate values.
- * (With a reasonable C compiler, you can just rely on the FIX() macro...)
- */
-
-#if CONST_BITS == 8
-#define FIX_0_382683433  ((INT32)   98)		/* FIX(0.382683433) */
-#define FIX_0_541196100  ((INT32)  139)		/* FIX(0.541196100) */
-#define FIX_0_707106781  ((INT32)  181)		/* FIX(0.707106781) */
-#define FIX_1_306562965  ((INT32)  334)		/* FIX(1.306562965) */
-#else
-#define FIX_0_382683433  FIX(0.382683433)
-#define FIX_0_541196100  FIX(0.541196100)
-#define FIX_0_707106781  FIX(0.707106781)
-#define FIX_1_306562965  FIX(1.306562965)
-#endif
-
-
-/* We can gain a little more speed, with a further compromise in accuracy,
- * by omitting the addition in a descaling shift.  This yields an incorrectly
- * rounded result half the time...
- */
-
-#ifndef USE_ACCURATE_ROUNDING
-#undef DESCALE
-#define DESCALE(x,n)  RIGHT_SHIFT(x, n)
-#endif
-
-
-/* Multiply a DCTELEM variable by an INT32 constant, and immediately
- * descale to yield a DCTELEM result.
- */
-
-#define MULTIPLY(var,const)  ((DCTELEM) DESCALE((var) * (const), CONST_BITS))
-
-
-/*
- * Perform the forward DCT on one block of samples.
- */
-
-GLOBAL(void)
-jpeg_fdct_ifast (DCTELEM * data)
-{
-  DCTELEM tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
-  DCTELEM tmp10, tmp11, tmp12, tmp13;
-  DCTELEM z1, z2, z3, z4, z5, z11, z13;
-  DCTELEM *dataptr;
-  int ctr;
-  SHIFT_TEMPS
-
-  /* Pass 1: process rows. */
-
-  dataptr = data;
-  for (ctr = DCTSIZE-1; ctr >= 0; ctr--) {
-    tmp0 = dataptr[0] + dataptr[7];
-    tmp7 = dataptr[0] - dataptr[7];
-    tmp1 = dataptr[1] + dataptr[6];
-    tmp6 = dataptr[1] - dataptr[6];
-    tmp2 = dataptr[2] + dataptr[5];
-    tmp5 = dataptr[2] - dataptr[5];
-    tmp3 = dataptr[3] + dataptr[4];
-    tmp4 = dataptr[3] - dataptr[4];
-    
-    /* Even part */
-    
-    tmp10 = tmp0 + tmp3;	/* phase 2 */
-    tmp13 = tmp0 - tmp3;
-    tmp11 = tmp1 + tmp2;
-    tmp12 = tmp1 - tmp2;
-    
-    dataptr[0] = tmp10 + tmp11; /* phase 3 */
-    dataptr[4] = tmp10 - tmp11;
-    
-    z1 = MULTIPLY(tmp12 + tmp13, FIX_0_707106781); /* c4 */
-    dataptr[2] = tmp13 + z1;	/* phase 5 */
-    dataptr[6] = tmp13 - z1;
-    
-    /* Odd part */
-
-    tmp10 = tmp4 + tmp5;	/* phase 2 */
-    tmp11 = tmp5 + tmp6;
-    tmp12 = tmp6 + tmp7;
-
-    /* The rotator is modified from fig 4-8 to avoid extra negations. */
-    z5 = MULTIPLY(tmp10 - tmp12, FIX_0_382683433); /* c6 */
-    z2 = MULTIPLY(tmp10, FIX_0_541196100) + z5; /* c2-c6 */
-    z4 = MULTIPLY(tmp12, FIX_1_306562965) + z5; /* c2+c6 */
-    z3 = MULTIPLY(tmp11, FIX_0_707106781); /* c4 */
-
-    z11 = tmp7 + z3;		/* phase 5 */
-    z13 = tmp7 - z3;
-
-    dataptr[5] = z13 + z2;	/* phase 6 */
-    dataptr[3] = z13 - z2;
-    dataptr[1] = z11 + z4;
-    dataptr[7] = z11 - z4;
-
-    dataptr += DCTSIZE;		/* advance pointer to next row */
-  }
-
-  /* Pass 2: process columns. */
-
-  dataptr = data;
-  for (ctr = DCTSIZE-1; ctr >= 0; ctr--) {
-    tmp0 = dataptr[DCTSIZE*0] + dataptr[DCTSIZE*7];
-    tmp7 = dataptr[DCTSIZE*0] - dataptr[DCTSIZE*7];
-    tmp1 = dataptr[DCTSIZE*1] + dataptr[DCTSIZE*6];
-    tmp6 = dataptr[DCTSIZE*1] - dataptr[DCTSIZE*6];
-    tmp2 = dataptr[DCTSIZE*2] + dataptr[DCTSIZE*5];
-    tmp5 = dataptr[DCTSIZE*2] - dataptr[DCTSIZE*5];
-    tmp3 = dataptr[DCTSIZE*3] + dataptr[DCTSIZE*4];
-    tmp4 = dataptr[DCTSIZE*3] - dataptr[DCTSIZE*4];
-    
-    /* Even part */
-    
-    tmp10 = tmp0 + tmp3;	/* phase 2 */
-    tmp13 = tmp0 - tmp3;
-    tmp11 = tmp1 + tmp2;
-    tmp12 = tmp1 - tmp2;
-    
-    dataptr[DCTSIZE*0] = tmp10 + tmp11; /* phase 3 */
-    dataptr[DCTSIZE*4] = tmp10 - tmp11;
-    
-    z1 = MULTIPLY(tmp12 + tmp13, FIX_0_707106781); /* c4 */
-    dataptr[DCTSIZE*2] = tmp13 + z1; /* phase 5 */
-    dataptr[DCTSIZE*6] = tmp13 - z1;
-    
-    /* Odd part */
-
-    tmp10 = tmp4 + tmp5;	/* phase 2 */
-    tmp11 = tmp5 + tmp6;
-    tmp12 = tmp6 + tmp7;
-
-    /* The rotator is modified from fig 4-8 to avoid extra negations. */
-    z5 = MULTIPLY(tmp10 - tmp12, FIX_0_382683433); /* c6 */
-    z2 = MULTIPLY(tmp10, FIX_0_541196100) + z5; /* c2-c6 */
-    z4 = MULTIPLY(tmp12, FIX_1_306562965) + z5; /* c2+c6 */
-    z3 = MULTIPLY(tmp11, FIX_0_707106781); /* c4 */
-
-    z11 = tmp7 + z3;		/* phase 5 */
-    z13 = tmp7 - z3;
-
-    dataptr[DCTSIZE*5] = z13 + z2; /* phase 6 */
-    dataptr[DCTSIZE*3] = z13 - z2;
-    dataptr[DCTSIZE*1] = z11 + z4;
-    dataptr[DCTSIZE*7] = z11 - z4;
-
-    dataptr++;			/* advance pointer to next column */
-  }
-}
-
-#endif /* DCT_IFAST_SUPPORTED */
@@ -1,283 +0,0 @@
-/*
- * jfdctint.c
- *
- * Copyright (C) 1991-1996, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains a slow-but-accurate integer implementation of the
- * forward DCT (Discrete Cosine Transform).
- *
- * A 2-D DCT can be done by 1-D DCT on each row followed by 1-D DCT
- * on each column.  Direct algorithms are also available, but they are
- * much more complex and seem not to be any faster when reduced to code.
- *
- * This implementation is based on an algorithm described in
- *   C. Loeffler, A. Ligtenberg and G. Moschytz, "Practical Fast 1-D DCT
- *   Algorithms with 11 Multiplications", Proc. Int'l. Conf. on Acoustics,
- *   Speech, and Signal Processing 1989 (ICASSP '89), pp. 988-991.
- * The primary algorithm described there uses 11 multiplies and 29 adds.
- * We use their alternate method with 12 multiplies and 32 adds.
- * The advantage of this method is that no data path contains more than one
- * multiplication; this allows a very simple and accurate implementation in
- * scaled fixed-point arithmetic, with a minimal number of shifts.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-#include "jdct.h"		/* Private declarations for DCT subsystem */
-
-#ifdef DCT_ISLOW_SUPPORTED
-
-
-/*
- * This module is specialized to the case DCTSIZE = 8.
- */
-
-#if DCTSIZE != 8
-  Sorry, this code only copes with 8x8 DCTs. /* deliberate syntax err */
-#endif
-
-
-/*
- * The poop on this scaling stuff is as follows:
- *
- * Each 1-D DCT step produces outputs which are a factor of sqrt(N)
- * larger than the true DCT outputs.  The final outputs are therefore
- * a factor of N larger than desired; since N=8 this can be cured by
- * a simple right shift at the end of the algorithm.  The advantage of
- * this arrangement is that we save two multiplications per 1-D DCT,
- * because the y0 and y4 outputs need not be divided by sqrt(N).
- * In the IJG code, this factor of 8 is removed by the quantization step
- * (in jcdctmgr.c), NOT in this module.
- *
- * We have to do addition and subtraction of the integer inputs, which
- * is no problem, and multiplication by fractional constants, which is
- * a problem to do in integer arithmetic.  We multiply all the constants
- * by CONST_SCALE and convert them to integer constants (thus retaining
- * CONST_BITS bits of precision in the constants).  After doing a
- * multiplication we have to divide the product by CONST_SCALE, with proper
- * rounding, to produce the correct output.  This division can be done
- * cheaply as a right shift of CONST_BITS bits.  We postpone shifting
- * as long as possible so that partial sums can be added together with
- * full fractional precision.
- *
- * The outputs of the first pass are scaled up by PASS1_BITS bits so that
- * they are represented to better-than-integral precision.  These outputs
- * require BITS_IN_JSAMPLE + PASS1_BITS + 3 bits; this fits in a 16-bit word
- * with the recommended scaling.  (For 12-bit sample data, the intermediate
- * array is INT32 anyway.)
- *
- * To avoid overflow of the 32-bit intermediate results in pass 2, we must
- * have BITS_IN_JSAMPLE + CONST_BITS + PASS1_BITS <= 26.  Error analysis
- * shows that the values given below are the most effective.
- */
-
-#if BITS_IN_JSAMPLE == 8
-#define CONST_BITS  13
-#define PASS1_BITS  2
-#else
-#define CONST_BITS  13
-#define PASS1_BITS  1		/* lose a little precision to avoid overflow */
-#endif
-
-/* Some C compilers fail to reduce "FIX(constant)" at compile time, thus
- * causing a lot of useless floating-point operations at run time.
- * To get around this we use the following pre-calculated constants.
- * If you change CONST_BITS you may want to add appropriate values.
- * (With a reasonable C compiler, you can just rely on the FIX() macro...)
- */
-
-#if CONST_BITS == 13
-#define FIX_0_298631336  ((INT32)  2446)	/* FIX(0.298631336) */
-#define FIX_0_390180644  ((INT32)  3196)	/* FIX(0.390180644) */
-#define FIX_0_541196100  ((INT32)  4433)	/* FIX(0.541196100) */
-#define FIX_0_765366865  ((INT32)  6270)	/* FIX(0.765366865) */
-#define FIX_0_899976223  ((INT32)  7373)	/* FIX(0.899976223) */
-#define FIX_1_175875602  ((INT32)  9633)	/* FIX(1.175875602) */
-#define FIX_1_501321110  ((INT32)  12299)	/* FIX(1.501321110) */
-#define FIX_1_847759065  ((INT32)  15137)	/* FIX(1.847759065) */
-#define FIX_1_961570560  ((INT32)  16069)	/* FIX(1.961570560) */
-#define FIX_2_053119869  ((INT32)  16819)	/* FIX(2.053119869) */
-#define FIX_2_562915447  ((INT32)  20995)	/* FIX(2.562915447) */
-#define FIX_3_072711026  ((INT32)  25172)	/* FIX(3.072711026) */
-#else
-#define FIX_0_298631336  FIX(0.298631336)
-#define FIX_0_390180644  FIX(0.390180644)
-#define FIX_0_541196100  FIX(0.541196100)
-#define FIX_0_765366865  FIX(0.765366865)
-#define FIX_0_899976223  FIX(0.899976223)
-#define FIX_1_175875602  FIX(1.175875602)
-#define FIX_1_501321110  FIX(1.501321110)
-#define FIX_1_847759065  FIX(1.847759065)
-#define FIX_1_961570560  FIX(1.961570560)
-#define FIX_2_053119869  FIX(2.053119869)
-#define FIX_2_562915447  FIX(2.562915447)
-#define FIX_3_072711026  FIX(3.072711026)
-#endif
-
-
-/* Multiply an INT32 variable by an INT32 constant to yield an INT32 result.
- * For 8-bit samples with the recommended scaling, all the variable
- * and constant values involved are no more than 16 bits wide, so a
- * 16x16->32 bit multiply can be used instead of a full 32x32 multiply.
- * For 12-bit samples, a full 32-bit multiplication will be needed.
- */
-
-#if BITS_IN_JSAMPLE == 8
-#define MULTIPLY(var,const)  MULTIPLY16C16(var,const)
-#else
-#define MULTIPLY(var,const)  ((var) * (const))
-#endif
-
-
-/*
- * Perform the forward DCT on one block of samples.
- */
-
-GLOBAL(void)
-jpeg_fdct_islow (DCTELEM * data)
-{
-  INT32 tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
-  INT32 tmp10, tmp11, tmp12, tmp13;
-  INT32 z1, z2, z3, z4, z5;
-  DCTELEM *dataptr;
-  int ctr;
-  SHIFT_TEMPS
-
-  /* Pass 1: process rows. */
-  /* Note results are scaled up by sqrt(8) compared to a true DCT; */
-  /* furthermore, we scale the results by 2**PASS1_BITS. */
-
-  dataptr = data;
-  for (ctr = DCTSIZE-1; ctr >= 0; ctr--) {
-    tmp0 = dataptr[0] + dataptr[7];
-    tmp7 = dataptr[0] - dataptr[7];
-    tmp1 = dataptr[1] + dataptr[6];
-    tmp6 = dataptr[1] - dataptr[6];
-    tmp2 = dataptr[2] + dataptr[5];
-    tmp5 = dataptr[2] - dataptr[5];
-    tmp3 = dataptr[3] + dataptr[4];
-    tmp4 = dataptr[3] - dataptr[4];
-    
-    /* Even part per LL&M figure 1 --- note that published figure is faulty;
-     * rotator "sqrt(2)*c1" should be "sqrt(2)*c6".
-     */
-    
-    tmp10 = tmp0 + tmp3;
-    tmp13 = tmp0 - tmp3;
-    tmp11 = tmp1 + tmp2;
-    tmp12 = tmp1 - tmp2;
-    
-    dataptr[0] = (DCTELEM) ((tmp10 + tmp11) << PASS1_BITS);
-    dataptr[4] = (DCTELEM) ((tmp10 - tmp11) << PASS1_BITS);
-    
-    z1 = MULTIPLY(tmp12 + tmp13, FIX_0_541196100);
-    dataptr[2] = (DCTELEM) DESCALE(z1 + MULTIPLY(tmp13, FIX_0_765366865),
-				   CONST_BITS-PASS1_BITS);
-    dataptr[6] = (DCTELEM) DESCALE(z1 + MULTIPLY(tmp12, - FIX_1_847759065),
-				   CONST_BITS-PASS1_BITS);
-    
-    /* Odd part per figure 8 --- note paper omits factor of sqrt(2).
-     * cK represents cos(K*pi/16).
-     * i0..i3 in the paper are tmp4..tmp7 here.
-     */
-    
-    z1 = tmp4 + tmp7;
-    z2 = tmp5 + tmp6;
-    z3 = tmp4 + tmp6;
-    z4 = tmp5 + tmp7;
-    z5 = MULTIPLY(z3 + z4, FIX_1_175875602); /* sqrt(2) * c3 */
-    
-    tmp4 = MULTIPLY(tmp4, FIX_0_298631336); /* sqrt(2) * (-c1+c3+c5-c7) */
-    tmp5 = MULTIPLY(tmp5, FIX_2_053119869); /* sqrt(2) * ( c1+c3-c5+c7) */
-    tmp6 = MULTIPLY(tmp6, FIX_3_072711026); /* sqrt(2) * ( c1+c3+c5-c7) */
-    tmp7 = MULTIPLY(tmp7, FIX_1_501321110); /* sqrt(2) * ( c1+c3-c5-c7) */
-    z1 = MULTIPLY(z1, - FIX_0_899976223); /* sqrt(2) * (c7-c3) */
-    z2 = MULTIPLY(z2, - FIX_2_562915447); /* sqrt(2) * (-c1-c3) */
-    z3 = MULTIPLY(z3, - FIX_1_961570560); /* sqrt(2) * (-c3-c5) */
-    z4 = MULTIPLY(z4, - FIX_0_390180644); /* sqrt(2) * (c5-c3) */
-    
-    z3 += z5;
-    z4 += z5;
-    
-    dataptr[7] = (DCTELEM) DESCALE(tmp4 + z1 + z3, CONST_BITS-PASS1_BITS);
-    dataptr[5] = (DCTELEM) DESCALE(tmp5 + z2 + z4, CONST_BITS-PASS1_BITS);
-    dataptr[3] = (DCTELEM) DESCALE(tmp6 + z2 + z3, CONST_BITS-PASS1_BITS);
-    dataptr[1] = (DCTELEM) DESCALE(tmp7 + z1 + z4, CONST_BITS-PASS1_BITS);
-    
-    dataptr += DCTSIZE;		/* advance pointer to next row */
-  }
-
-  /* Pass 2: process columns.
-   * We remove the PASS1_BITS scaling, but leave the results scaled up
-   * by an overall factor of 8.
-   */
-
-  dataptr = data;
-  for (ctr = DCTSIZE-1; ctr >= 0; ctr--) {
-    tmp0 = dataptr[DCTSIZE*0] + dataptr[DCTSIZE*7];
-    tmp7 = dataptr[DCTSIZE*0] - dataptr[DCTSIZE*7];
-    tmp1 = dataptr[DCTSIZE*1] + dataptr[DCTSIZE*6];
-    tmp6 = dataptr[DCTSIZE*1] - dataptr[DCTSIZE*6];
-    tmp2 = dataptr[DCTSIZE*2] + dataptr[DCTSIZE*5];
-    tmp5 = dataptr[DCTSIZE*2] - dataptr[DCTSIZE*5];
-    tmp3 = dataptr[DCTSIZE*3] + dataptr[DCTSIZE*4];
-    tmp4 = dataptr[DCTSIZE*3] - dataptr[DCTSIZE*4];
-    
-    /* Even part per LL&M figure 1 --- note that published figure is faulty;
-     * rotator "sqrt(2)*c1" should be "sqrt(2)*c6".
-     */
-    
-    tmp10 = tmp0 + tmp3;
-    tmp13 = tmp0 - tmp3;
-    tmp11 = tmp1 + tmp2;
-    tmp12 = tmp1 - tmp2;
-    
-    dataptr[DCTSIZE*0] = (DCTELEM) DESCALE(tmp10 + tmp11, PASS1_BITS);
-    dataptr[DCTSIZE*4] = (DCTELEM) DESCALE(tmp10 - tmp11, PASS1_BITS);
-    
-    z1 = MULTIPLY(tmp12 + tmp13, FIX_0_541196100);
-    dataptr[DCTSIZE*2] = (DCTELEM) DESCALE(z1 + MULTIPLY(tmp13, FIX_0_765366865),
-					   CONST_BITS+PASS1_BITS);
-    dataptr[DCTSIZE*6] = (DCTELEM) DESCALE(z1 + MULTIPLY(tmp12, - FIX_1_847759065),
-					   CONST_BITS+PASS1_BITS);
-    
-    /* Odd part per figure 8 --- note paper omits factor of sqrt(2).
-     * cK represents cos(K*pi/16).
-     * i0..i3 in the paper are tmp4..tmp7 here.
-     */
-    
-    z1 = tmp4 + tmp7;
-    z2 = tmp5 + tmp6;
-    z3 = tmp4 + tmp6;
-    z4 = tmp5 + tmp7;
-    z5 = MULTIPLY(z3 + z4, FIX_1_175875602); /* sqrt(2) * c3 */
-    
-    tmp4 = MULTIPLY(tmp4, FIX_0_298631336); /* sqrt(2) * (-c1+c3+c5-c7) */
-    tmp5 = MULTIPLY(tmp5, FIX_2_053119869); /* sqrt(2) * ( c1+c3-c5+c7) */
-    tmp6 = MULTIPLY(tmp6, FIX_3_072711026); /* sqrt(2) * ( c1+c3+c5-c7) */
-    tmp7 = MULTIPLY(tmp7, FIX_1_501321110); /* sqrt(2) * ( c1+c3-c5-c7) */
-    z1 = MULTIPLY(z1, - FIX_0_899976223); /* sqrt(2) * (c7-c3) */
-    z2 = MULTIPLY(z2, - FIX_2_562915447); /* sqrt(2) * (-c1-c3) */
-    z3 = MULTIPLY(z3, - FIX_1_961570560); /* sqrt(2) * (-c3-c5) */
-    z4 = MULTIPLY(z4, - FIX_0_390180644); /* sqrt(2) * (c5-c3) */
-    
-    z3 += z5;
-    z4 += z5;
-    
-    dataptr[DCTSIZE*7] = (DCTELEM) DESCALE(tmp4 + z1 + z3,
-					   CONST_BITS+PASS1_BITS);
-    dataptr[DCTSIZE*5] = (DCTELEM) DESCALE(tmp5 + z2 + z4,
-					   CONST_BITS+PASS1_BITS);
-    dataptr[DCTSIZE*3] = (DCTELEM) DESCALE(tmp6 + z2 + z3,
-					   CONST_BITS+PASS1_BITS);
-    dataptr[DCTSIZE*1] = (DCTELEM) DESCALE(tmp7 + z1 + z4,
-					   CONST_BITS+PASS1_BITS);
-    
-    dataptr++;			/* advance pointer to next column */
-  }
-}
-
-#endif /* DCT_ISLOW_SUPPORTED */
@@ -1,242 +0,0 @@
-/*
- * jidctflt.c
- *
- * Copyright (C) 1994-1998, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains a floating-point implementation of the
- * inverse DCT (Discrete Cosine Transform).  In the IJG code, this routine
- * must also perform dequantization of the input coefficients.
- *
- * This implementation should be more accurate than either of the integer
- * IDCT implementations.  However, it may not give the same results on all
- * machines because of differences in roundoff behavior.  Speed will depend
- * on the hardware's floating point capacity.
- *
- * A 2-D IDCT can be done by 1-D IDCT on each column followed by 1-D IDCT
- * on each row (or vice versa, but it's more convenient to emit a row at
- * a time).  Direct algorithms are also available, but they are much more
- * complex and seem not to be any faster when reduced to code.
- *
- * This implementation is based on Arai, Agui, and Nakajima's algorithm for
- * scaled DCT.  Their original paper (Trans. IEICE E-71(11):1095) is in
- * Japanese, but the algorithm is described in the Pennebaker & Mitchell
- * JPEG textbook (see REFERENCES section in file README).  The following code
- * is based directly on figure 4-8 in P&M.
- * While an 8-point DCT cannot be done in less than 11 multiplies, it is
- * possible to arrange the computation so that many of the multiplies are
- * simple scalings of the final outputs.  These multiplies can then be
- * folded into the multiplications or divisions by the JPEG quantization
- * table entries.  The AA&N method leaves only 5 multiplies and 29 adds
- * to be done in the DCT itself.
- * The primary disadvantage of this method is that with a fixed-point
- * implementation, accuracy is lost due to imprecise representation of the
- * scaled quantization values.  However, that problem does not arise if
- * we use floating point arithmetic.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-#include "jdct.h"		/* Private declarations for DCT subsystem */
-
-#ifdef DCT_FLOAT_SUPPORTED
-
-
-/*
- * This module is specialized to the case DCTSIZE = 8.
- */
-
-#if DCTSIZE != 8
-  Sorry, this code only copes with 8x8 DCTs. /* deliberate syntax err */
-#endif
-
-
-/* Dequantize a coefficient by multiplying it by the multiplier-table
- * entry; produce a float result.
- */
-
-#define DEQUANTIZE(coef,quantval)  (((FAST_FLOAT) (coef)) * (quantval))
-
-
-/*
- * Perform dequantization and inverse DCT on one block of coefficients.
- */
-
-GLOBAL(void)
-jpeg_idct_float (j_decompress_ptr cinfo, jpeg_component_info * compptr,
-		 JCOEFPTR coef_block,
-		 JSAMPARRAY output_buf, JDIMENSION output_col)
-{
-  FAST_FLOAT tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
-  FAST_FLOAT tmp10, tmp11, tmp12, tmp13;
-  FAST_FLOAT z5, z10, z11, z12, z13;
-  JCOEFPTR inptr;
-  FLOAT_MULT_TYPE * quantptr;
-  FAST_FLOAT * wsptr;
-  JSAMPROW outptr;
-  JSAMPLE *range_limit = IDCT_range_limit(cinfo);
-  int ctr;
-  FAST_FLOAT workspace[DCTSIZE2]; /* buffers data between passes */
-  SHIFT_TEMPS
-
-  /* Pass 1: process columns from input, store into work array. */
-
-  inptr = coef_block;
-  quantptr = (FLOAT_MULT_TYPE *) compptr->dct_table;
-  wsptr = workspace;
-  for (ctr = DCTSIZE; ctr > 0; ctr--) {
-    /* Due to quantization, we will usually find that many of the input
-     * coefficients are zero, especially the AC terms.  We can exploit this
-     * by short-circuiting the IDCT calculation for any column in which all
-     * the AC terms are zero.  In that case each output is equal to the
-     * DC coefficient (with scale factor as needed).
-     * With typical images and quantization tables, half or more of the
-     * column DCT calculations can be simplified this way.
-     */
-    
-    if (inptr[DCTSIZE*1] == 0 && inptr[DCTSIZE*2] == 0 &&
-	inptr[DCTSIZE*3] == 0 && inptr[DCTSIZE*4] == 0 &&
-	inptr[DCTSIZE*5] == 0 && inptr[DCTSIZE*6] == 0 &&
-	inptr[DCTSIZE*7] == 0) {
-      /* AC terms all zero */
-      FAST_FLOAT dcval = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]);
-      
-      wsptr[DCTSIZE*0] = dcval;
-      wsptr[DCTSIZE*1] = dcval;
-      wsptr[DCTSIZE*2] = dcval;
-      wsptr[DCTSIZE*3] = dcval;
-      wsptr[DCTSIZE*4] = dcval;
-      wsptr[DCTSIZE*5] = dcval;
-      wsptr[DCTSIZE*6] = dcval;
-      wsptr[DCTSIZE*7] = dcval;
-      
-      inptr++;			/* advance pointers to next column */
-      quantptr++;
-      wsptr++;
-      continue;
-    }
-    
-    /* Even part */
-
-    tmp0 = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]);
-    tmp1 = DEQUANTIZE(inptr[DCTSIZE*2], quantptr[DCTSIZE*2]);
-    tmp2 = DEQUANTIZE(inptr[DCTSIZE*4], quantptr[DCTSIZE*4]);
-    tmp3 = DEQUANTIZE(inptr[DCTSIZE*6], quantptr[DCTSIZE*6]);
-
-    tmp10 = tmp0 + tmp2;	/* phase 3 */
-    tmp11 = tmp0 - tmp2;
-
-    tmp13 = tmp1 + tmp3;	/* phases 5-3 */
-    tmp12 = (tmp1 - tmp3) * ((FAST_FLOAT) 1.414213562) - tmp13; /* 2*c4 */
-
-    tmp0 = tmp10 + tmp13;	/* phase 2 */
-    tmp3 = tmp10 - tmp13;
-    tmp1 = tmp11 + tmp12;
-    tmp2 = tmp11 - tmp12;
-    
-    /* Odd part */
-
-    tmp4 = DEQUANTIZE(inptr[DCTSIZE*1], quantptr[DCTSIZE*1]);
-    tmp5 = DEQUANTIZE(inptr[DCTSIZE*3], quantptr[DCTSIZE*3]);
-    tmp6 = DEQUANTIZE(inptr[DCTSIZE*5], quantptr[DCTSIZE*5]);
-    tmp7 = DEQUANTIZE(inptr[DCTSIZE*7], quantptr[DCTSIZE*7]);
-
-    z13 = tmp6 + tmp5;		/* phase 6 */
-    z10 = tmp6 - tmp5;
-    z11 = tmp4 + tmp7;
-    z12 = tmp4 - tmp7;
-
-    tmp7 = z11 + z13;		/* phase 5 */
-    tmp11 = (z11 - z13) * ((FAST_FLOAT) 1.414213562); /* 2*c4 */
-
-    z5 = (z10 + z12) * ((FAST_FLOAT) 1.847759065); /* 2*c2 */
-    tmp10 = ((FAST_FLOAT) 1.082392200) * z12 - z5; /* 2*(c2-c6) */
-    tmp12 = ((FAST_FLOAT) -2.613125930) * z10 + z5; /* -2*(c2+c6) */
-
-    tmp6 = tmp12 - tmp7;	/* phase 2 */
-    tmp5 = tmp11 - tmp6;
-    tmp4 = tmp10 + tmp5;
-
-    wsptr[DCTSIZE*0] = tmp0 + tmp7;
-    wsptr[DCTSIZE*7] = tmp0 - tmp7;
-    wsptr[DCTSIZE*1] = tmp1 + tmp6;
-    wsptr[DCTSIZE*6] = tmp1 - tmp6;
-    wsptr[DCTSIZE*2] = tmp2 + tmp5;
-    wsptr[DCTSIZE*5] = tmp2 - tmp5;
-    wsptr[DCTSIZE*4] = tmp3 + tmp4;
-    wsptr[DCTSIZE*3] = tmp3 - tmp4;
-
-    inptr++;			/* advance pointers to next column */
-    quantptr++;
-    wsptr++;
-  }
-  
-  /* Pass 2: process rows from work array, store into output array. */
-  /* Note that we must descale the results by a factor of 8 == 2**3. */
-
-  wsptr = workspace;
-  for (ctr = 0; ctr < DCTSIZE; ctr++) {
-    outptr = output_buf[ctr] + output_col;
-    /* Rows of zeroes can be exploited in the same way as we did with columns.
-     * However, the column calculation has created many nonzero AC terms, so
-     * the simplification applies less often (typically 5% to 10% of the time).
-     * And testing floats for zero is relatively expensive, so we don't bother.
-     */
-    
-    /* Even part */
-
-    tmp10 = wsptr[0] + wsptr[4];
-    tmp11 = wsptr[0] - wsptr[4];
-
-    tmp13 = wsptr[2] + wsptr[6];
-    tmp12 = (wsptr[2] - wsptr[6]) * ((FAST_FLOAT) 1.414213562) - tmp13;
-
-    tmp0 = tmp10 + tmp13;
-    tmp3 = tmp10 - tmp13;
-    tmp1 = tmp11 + tmp12;
-    tmp2 = tmp11 - tmp12;
-
-    /* Odd part */
-
-    z13 = wsptr[5] + wsptr[3];
-    z10 = wsptr[5] - wsptr[3];
-    z11 = wsptr[1] + wsptr[7];
-    z12 = wsptr[1] - wsptr[7];
-
-    tmp7 = z11 + z13;
-    tmp11 = (z11 - z13) * ((FAST_FLOAT) 1.414213562);
-
-    z5 = (z10 + z12) * ((FAST_FLOAT) 1.847759065); /* 2*c2 */
-    tmp10 = ((FAST_FLOAT) 1.082392200) * z12 - z5; /* 2*(c2-c6) */
-    tmp12 = ((FAST_FLOAT) -2.613125930) * z10 + z5; /* -2*(c2+c6) */
-
-    tmp6 = tmp12 - tmp7;
-    tmp5 = tmp11 - tmp6;
-    tmp4 = tmp10 + tmp5;
-
-    /* Final output stage: scale down by a factor of 8 and range-limit */
-
-    outptr[0] = range_limit[(int) DESCALE((INT32) (tmp0 + tmp7), 3)
-			    & RANGE_MASK];
-    outptr[7] = range_limit[(int) DESCALE((INT32) (tmp0 - tmp7), 3)
-			    & RANGE_MASK];
-    outptr[1] = range_limit[(int) DESCALE((INT32) (tmp1 + tmp6), 3)
-			    & RANGE_MASK];
-    outptr[6] = range_limit[(int) DESCALE((INT32) (tmp1 - tmp6), 3)
-			    & RANGE_MASK];
-    outptr[2] = range_limit[(int) DESCALE((INT32) (tmp2 + tmp5), 3)
-			    & RANGE_MASK];
-    outptr[5] = range_limit[(int) DESCALE((INT32) (tmp2 - tmp5), 3)
-			    & RANGE_MASK];
-    outptr[4] = range_limit[(int) DESCALE((INT32) (tmp3 + tmp4), 3)
-			    & RANGE_MASK];
-    outptr[3] = range_limit[(int) DESCALE((INT32) (tmp3 - tmp4), 3)
-			    & RANGE_MASK];
-    
-    wsptr += DCTSIZE;		/* advance pointer to next row */
-  }
-}
-
-#endif /* DCT_FLOAT_SUPPORTED */
@@ -1,368 +0,0 @@
-/*
- * jidctfst.c
- *
- * Copyright (C) 1994-1998, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains a fast, not so accurate integer implementation of the
- * inverse DCT (Discrete Cosine Transform).  In the IJG code, this routine
- * must also perform dequantization of the input coefficients.
- *
- * A 2-D IDCT can be done by 1-D IDCT on each column followed by 1-D IDCT
- * on each row (or vice versa, but it's more convenient to emit a row at
- * a time).  Direct algorithms are also available, but they are much more
- * complex and seem not to be any faster when reduced to code.
- *
- * This implementation is based on Arai, Agui, and Nakajima's algorithm for
- * scaled DCT.  Their original paper (Trans. IEICE E-71(11):1095) is in
- * Japanese, but the algorithm is described in the Pennebaker & Mitchell
- * JPEG textbook (see REFERENCES section in file README).  The following code
- * is based directly on figure 4-8 in P&M.
- * While an 8-point DCT cannot be done in less than 11 multiplies, it is
- * possible to arrange the computation so that many of the multiplies are
- * simple scalings of the final outputs.  These multiplies can then be
- * folded into the multiplications or divisions by the JPEG quantization
- * table entries.  The AA&N method leaves only 5 multiplies and 29 adds
- * to be done in the DCT itself.
- * The primary disadvantage of this method is that with fixed-point math,
- * accuracy is lost due to imprecise representation of the scaled
- * quantization values.  The smaller the quantization table entry, the less
- * precise the scaled value, so this implementation does worse with high-
- * quality-setting files than with low-quality ones.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-#include "jdct.h"		/* Private declarations for DCT subsystem */
-
-#ifdef DCT_IFAST_SUPPORTED
-
-
-/*
- * This module is specialized to the case DCTSIZE = 8.
- */
-
-#if DCTSIZE != 8
-  Sorry, this code only copes with 8x8 DCTs. /* deliberate syntax err */
-#endif
-
-
-/* Scaling decisions are generally the same as in the LL&M algorithm;
- * see jidctint.c for more details.  However, we choose to descale
- * (right shift) multiplication products as soon as they are formed,
- * rather than carrying additional fractional bits into subsequent additions.
- * This compromises accuracy slightly, but it lets us save a few shifts.
- * More importantly, 16-bit arithmetic is then adequate (for 8-bit samples)
- * everywhere except in the multiplications proper; this saves a good deal
- * of work on 16-bit-int machines.
- *
- * The dequantized coefficients are not integers because the AA&N scaling
- * factors have been incorporated.  We represent them scaled up by PASS1_BITS,
- * so that the first and second IDCT rounds have the same input scaling.
- * For 8-bit JSAMPLEs, we choose IFAST_SCALE_BITS = PASS1_BITS so as to
- * avoid a descaling shift; this compromises accuracy rather drastically
- * for small quantization table entries, but it saves a lot of shifts.
- * For 12-bit JSAMPLEs, there's no hope of using 16x16 multiplies anyway,
- * so we use a much larger scaling factor to preserve accuracy.
- *
- * A final compromise is to represent the multiplicative constants to only
- * 8 fractional bits, rather than 13.  This saves some shifting work on some
- * machines, and may also reduce the cost of multiplication (since there
- * are fewer one-bits in the constants).
- */
-
-#if BITS_IN_JSAMPLE == 8
-#define CONST_BITS  8
-#define PASS1_BITS  2
-#else
-#define CONST_BITS  8
-#define PASS1_BITS  1		/* lose a little precision to avoid overflow */
-#endif
-
-/* Some C compilers fail to reduce "FIX(constant)" at compile time, thus
- * causing a lot of useless floating-point operations at run time.
- * To get around this we use the following pre-calculated constants.
- * If you change CONST_BITS you may want to add appropriate values.
- * (With a reasonable C compiler, you can just rely on the FIX() macro...)
- */
-
-#if CONST_BITS == 8
-#define FIX_1_082392200  ((INT32)  277)		/* FIX(1.082392200) */
-#define FIX_1_414213562  ((INT32)  362)		/* FIX(1.414213562) */
-#define FIX_1_847759065  ((INT32)  473)		/* FIX(1.847759065) */
-#define FIX_2_613125930  ((INT32)  669)		/* FIX(2.613125930) */
-#else
-#define FIX_1_082392200  FIX(1.082392200)
-#define FIX_1_414213562  FIX(1.414213562)
-#define FIX_1_847759065  FIX(1.847759065)
-#define FIX_2_613125930  FIX(2.613125930)
-#endif
-
-
-/* We can gain a little more speed, with a further compromise in accuracy,
- * by omitting the addition in a descaling shift.  This yields an incorrectly
- * rounded result half the time...
- */
-
-#ifndef USE_ACCURATE_ROUNDING
-#undef DESCALE
-#define DESCALE(x,n)  RIGHT_SHIFT(x, n)
-#endif
-
-
-/* Multiply a DCTELEM variable by an INT32 constant, and immediately
- * descale to yield a DCTELEM result.
- */
-
-#define MULTIPLY(var,const)  ((DCTELEM) DESCALE((var) * (const), CONST_BITS))
-
-
-/* Dequantize a coefficient by multiplying it by the multiplier-table
- * entry; produce a DCTELEM result.  For 8-bit data a 16x16->16
- * multiplication will do.  For 12-bit data, the multiplier table is
- * declared INT32, so a 32-bit multiply will be used.
- */
-
-#if BITS_IN_JSAMPLE == 8
-#define DEQUANTIZE(coef,quantval)  (((IFAST_MULT_TYPE) (coef)) * (quantval))
-#else
-#define DEQUANTIZE(coef,quantval)  \
-	DESCALE((coef)*(quantval), IFAST_SCALE_BITS-PASS1_BITS)
-#endif
-
-
-/* Like DESCALE, but applies to a DCTELEM and produces an int.
- * We assume that int right shift is unsigned if INT32 right shift is.
- */
-
-#ifdef RIGHT_SHIFT_IS_UNSIGNED
-#define ISHIFT_TEMPS	DCTELEM ishift_temp;
-#if BITS_IN_JSAMPLE == 8
-#define DCTELEMBITS  16		/* DCTELEM may be 16 or 32 bits */
-#else
-#define DCTELEMBITS  32		/* DCTELEM must be 32 bits */
-#endif
-#define IRIGHT_SHIFT(x,shft)  \
-    ((ishift_temp = (x)) < 0 ? \
-     (ishift_temp >> (shft)) | ((~((DCTELEM) 0)) << (DCTELEMBITS-(shft))) : \
-     (ishift_temp >> (shft)))
-#else
-#define ISHIFT_TEMPS
-#define IRIGHT_SHIFT(x,shft)	((x) >> (shft))
-#endif
-
-#ifdef USE_ACCURATE_ROUNDING
-#define IDESCALE(x,n)  ((int) IRIGHT_SHIFT((x) + (1 << ((n)-1)), n))
-#else
-#define IDESCALE(x,n)  ((int) IRIGHT_SHIFT(x, n))
-#endif
-
-
-/*
- * Perform dequantization and inverse DCT on one block of coefficients.
- */
-
-GLOBAL(void)
-jpeg_idct_ifast (j_decompress_ptr cinfo, jpeg_component_info * compptr,
-		 JCOEFPTR coef_block,
-		 JSAMPARRAY output_buf, JDIMENSION output_col)
-{
-  DCTELEM tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
-  DCTELEM tmp10, tmp11, tmp12, tmp13;
-  DCTELEM z5, z10, z11, z12, z13;
-  JCOEFPTR inptr;
-  IFAST_MULT_TYPE * quantptr;
-  int * wsptr;
-  JSAMPROW outptr;
-  JSAMPLE *range_limit = IDCT_range_limit(cinfo);
-  int ctr;
-  int workspace[DCTSIZE2];	/* buffers data between passes */
-  SHIFT_TEMPS			/* for DESCALE */
-  ISHIFT_TEMPS			/* for IDESCALE */
-
-  /* Pass 1: process columns from input, store into work array. */
-
-  inptr = coef_block;
-  quantptr = (IFAST_MULT_TYPE *) compptr->dct_table;
-  wsptr = workspace;
-  for (ctr = DCTSIZE; ctr > 0; ctr--) {
-    /* Due to quantization, we will usually find that many of the input
-     * coefficients are zero, especially the AC terms.  We can exploit this
-     * by short-circuiting the IDCT calculation for any column in which all
-     * the AC terms are zero.  In that case each output is equal to the
-     * DC coefficient (with scale factor as needed).
-     * With typical images and quantization tables, half or more of the
-     * column DCT calculations can be simplified this way.
-     */
-    
-    if (inptr[DCTSIZE*1] == 0 && inptr[DCTSIZE*2] == 0 &&
-	inptr[DCTSIZE*3] == 0 && inptr[DCTSIZE*4] == 0 &&
-	inptr[DCTSIZE*5] == 0 && inptr[DCTSIZE*6] == 0 &&
-	inptr[DCTSIZE*7] == 0) {
-      /* AC terms all zero */
-      int dcval = (int) DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]);
-
-      wsptr[DCTSIZE*0] = dcval;
-      wsptr[DCTSIZE*1] = dcval;
-      wsptr[DCTSIZE*2] = dcval;
-      wsptr[DCTSIZE*3] = dcval;
-      wsptr[DCTSIZE*4] = dcval;
-      wsptr[DCTSIZE*5] = dcval;
-      wsptr[DCTSIZE*6] = dcval;
-      wsptr[DCTSIZE*7] = dcval;
-      
-      inptr++;			/* advance pointers to next column */
-      quantptr++;
-      wsptr++;
-      continue;
-    }
-    
-    /* Even part */
-
-    tmp0 = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]);
-    tmp1 = DEQUANTIZE(inptr[DCTSIZE*2], quantptr[DCTSIZE*2]);
-    tmp2 = DEQUANTIZE(inptr[DCTSIZE*4], quantptr[DCTSIZE*4]);
-    tmp3 = DEQUANTIZE(inptr[DCTSIZE*6], quantptr[DCTSIZE*6]);
-
-    tmp10 = tmp0 + tmp2;	/* phase 3 */
-    tmp11 = tmp0 - tmp2;
-
-    tmp13 = tmp1 + tmp3;	/* phases 5-3 */
-    tmp12 = MULTIPLY(tmp1 - tmp3, FIX_1_414213562) - tmp13; /* 2*c4 */
-
-    tmp0 = tmp10 + tmp13;	/* phase 2 */
-    tmp3 = tmp10 - tmp13;
-    tmp1 = tmp11 + tmp12;
-    tmp2 = tmp11 - tmp12;
-    
-    /* Odd part */
-
-    tmp4 = DEQUANTIZE(inptr[DCTSIZE*1], quantptr[DCTSIZE*1]);
-    tmp5 = DEQUANTIZE(inptr[DCTSIZE*3], quantptr[DCTSIZE*3]);
-    tmp6 = DEQUANTIZE(inptr[DCTSIZE*5], quantptr[DCTSIZE*5]);
-    tmp7 = DEQUANTIZE(inptr[DCTSIZE*7], quantptr[DCTSIZE*7]);
-
-    z13 = tmp6 + tmp5;		/* phase 6 */
-    z10 = tmp6 - tmp5;
-    z11 = tmp4 + tmp7;
-    z12 = tmp4 - tmp7;
-
-    tmp7 = z11 + z13;		/* phase 5 */
-    tmp11 = MULTIPLY(z11 - z13, FIX_1_414213562); /* 2*c4 */
-
-    z5 = MULTIPLY(z10 + z12, FIX_1_847759065); /* 2*c2 */
-    tmp10 = MULTIPLY(z12, FIX_1_082392200) - z5; /* 2*(c2-c6) */
-    tmp12 = MULTIPLY(z10, - FIX_2_613125930) + z5; /* -2*(c2+c6) */
-
-    tmp6 = tmp12 - tmp7;	/* phase 2 */
-    tmp5 = tmp11 - tmp6;
-    tmp4 = tmp10 + tmp5;
-
-    wsptr[DCTSIZE*0] = (int) (tmp0 + tmp7);
-    wsptr[DCTSIZE*7] = (int) (tmp0 - tmp7);
-    wsptr[DCTSIZE*1] = (int) (tmp1 + tmp6);
-    wsptr[DCTSIZE*6] = (int) (tmp1 - tmp6);
-    wsptr[DCTSIZE*2] = (int) (tmp2 + tmp5);
-    wsptr[DCTSIZE*5] = (int) (tmp2 - tmp5);
-    wsptr[DCTSIZE*4] = (int) (tmp3 + tmp4);
-    wsptr[DCTSIZE*3] = (int) (tmp3 - tmp4);
-
-    inptr++;			/* advance pointers to next column */
-    quantptr++;
-    wsptr++;
-  }
-  
-  /* Pass 2: process rows from work array, store into output array. */
-  /* Note that we must descale the results by a factor of 8 == 2**3, */
-  /* and also undo the PASS1_BITS scaling. */
-
-  wsptr = workspace;
-  for (ctr = 0; ctr < DCTSIZE; ctr++) {
-    outptr = output_buf[ctr] + output_col;
-    /* Rows of zeroes can be exploited in the same way as we did with columns.
-     * However, the column calculation has created many nonzero AC terms, so
-     * the simplification applies less often (typically 5% to 10% of the time).
-     * On machines with very fast multiplication, it's possible that the
-     * test takes more time than it's worth.  In that case this section
-     * may be commented out.
-     */
-    
-#ifndef NO_ZERO_ROW_TEST
-    if (wsptr[1] == 0 && wsptr[2] == 0 && wsptr[3] == 0 && wsptr[4] == 0 &&
-	wsptr[5] == 0 && wsptr[6] == 0 && wsptr[7] == 0) {
-      /* AC terms all zero */
-      JSAMPLE dcval = range_limit[IDESCALE(wsptr[0], PASS1_BITS+3)
-				  & RANGE_MASK];
-      
-      outptr[0] = dcval;
-      outptr[1] = dcval;
-      outptr[2] = dcval;
-      outptr[3] = dcval;
-      outptr[4] = dcval;
-      outptr[5] = dcval;
-      outptr[6] = dcval;
-      outptr[7] = dcval;
-
-      wsptr += DCTSIZE;		/* advance pointer to next row */
-      continue;
-    }
-#endif
-    
-    /* Even part */
-
-    tmp10 = ((DCTELEM) wsptr[0] + (DCTELEM) wsptr[4]);
-    tmp11 = ((DCTELEM) wsptr[0] - (DCTELEM) wsptr[4]);
-
-    tmp13 = ((DCTELEM) wsptr[2] + (DCTELEM) wsptr[6]);
-    tmp12 = MULTIPLY((DCTELEM) wsptr[2] - (DCTELEM) wsptr[6], FIX_1_414213562)
-	    - tmp13;
-
-    tmp0 = tmp10 + tmp13;
-    tmp3 = tmp10 - tmp13;
-    tmp1 = tmp11 + tmp12;
-    tmp2 = tmp11 - tmp12;
-
-    /* Odd part */
-
-    z13 = (DCTELEM) wsptr[5] + (DCTELEM) wsptr[3];
-    z10 = (DCTELEM) wsptr[5] - (DCTELEM) wsptr[3];
-    z11 = (DCTELEM) wsptr[1] + (DCTELEM) wsptr[7];
-    z12 = (DCTELEM) wsptr[1] - (DCTELEM) wsptr[7];
-
-    tmp7 = z11 + z13;		/* phase 5 */
-    tmp11 = MULTIPLY(z11 - z13, FIX_1_414213562); /* 2*c4 */
-
-    z5 = MULTIPLY(z10 + z12, FIX_1_847759065); /* 2*c2 */
-    tmp10 = MULTIPLY(z12, FIX_1_082392200) - z5; /* 2*(c2-c6) */
-    tmp12 = MULTIPLY(z10, - FIX_2_613125930) + z5; /* -2*(c2+c6) */
-
-    tmp6 = tmp12 - tmp7;	/* phase 2 */
-    tmp5 = tmp11 - tmp6;
-    tmp4 = tmp10 + tmp5;
-
-    /* Final output stage: scale down by a factor of 8 and range-limit */
-
-    outptr[0] = range_limit[IDESCALE(tmp0 + tmp7, PASS1_BITS+3)
-			    & RANGE_MASK];
-    outptr[7] = range_limit[IDESCALE(tmp0 - tmp7, PASS1_BITS+3)
-			    & RANGE_MASK];
-    outptr[1] = range_limit[IDESCALE(tmp1 + tmp6, PASS1_BITS+3)
-			    & RANGE_MASK];
-    outptr[6] = range_limit[IDESCALE(tmp1 - tmp6, PASS1_BITS+3)
-			    & RANGE_MASK];
-    outptr[2] = range_limit[IDESCALE(tmp2 + tmp5, PASS1_BITS+3)
-			    & RANGE_MASK];
-    outptr[5] = range_limit[IDESCALE(tmp2 - tmp5, PASS1_BITS+3)
-			    & RANGE_MASK];
-    outptr[4] = range_limit[IDESCALE(tmp3 + tmp4, PASS1_BITS+3)
-			    & RANGE_MASK];
-    outptr[3] = range_limit[IDESCALE(tmp3 - tmp4, PASS1_BITS+3)
-			    & RANGE_MASK];
-
-    wsptr += DCTSIZE;		/* advance pointer to next row */
-  }
-}
-
-#endif /* DCT_IFAST_SUPPORTED */
@@ -1,389 +0,0 @@
-/*
- * jidctint.c
- *
- * Copyright (C) 1991-1998, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains a slow-but-accurate integer implementation of the
- * inverse DCT (Discrete Cosine Transform).  In the IJG code, this routine
- * must also perform dequantization of the input coefficients.
- *
- * A 2-D IDCT can be done by 1-D IDCT on each column followed by 1-D IDCT
- * on each row (or vice versa, but it's more convenient to emit a row at
- * a time).  Direct algorithms are also available, but they are much more
- * complex and seem not to be any faster when reduced to code.
- *
- * This implementation is based on an algorithm described in
- *   C. Loeffler, A. Ligtenberg and G. Moschytz, "Practical Fast 1-D DCT
- *   Algorithms with 11 Multiplications", Proc. Int'l. Conf. on Acoustics,
- *   Speech, and Signal Processing 1989 (ICASSP '89), pp. 988-991.
- * The primary algorithm described there uses 11 multiplies and 29 adds.
- * We use their alternate method with 12 multiplies and 32 adds.
- * The advantage of this method is that no data path contains more than one
- * multiplication; this allows a very simple and accurate implementation in
- * scaled fixed-point arithmetic, with a minimal number of shifts.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-#include "jdct.h"		/* Private declarations for DCT subsystem */
-
-#ifdef DCT_ISLOW_SUPPORTED
-
-
-/*
- * This module is specialized to the case DCTSIZE = 8.
- */
-
-#if DCTSIZE != 8
-  Sorry, this code only copes with 8x8 DCTs. /* deliberate syntax err */
-#endif
-
-
-/*
- * The poop on this scaling stuff is as follows:
- *
- * Each 1-D IDCT step produces outputs which are a factor of sqrt(N)
- * larger than the true IDCT outputs.  The final outputs are therefore
- * a factor of N larger than desired; since N=8 this can be cured by
- * a simple right shift at the end of the algorithm.  The advantage of
- * this arrangement is that we save two multiplications per 1-D IDCT,
- * because the y0 and y4 inputs need not be divided by sqrt(N).
- *
- * We have to do addition and subtraction of the integer inputs, which
- * is no problem, and multiplication by fractional constants, which is
- * a problem to do in integer arithmetic.  We multiply all the constants
- * by CONST_SCALE and convert them to integer constants (thus retaining
- * CONST_BITS bits of precision in the constants).  After doing a
- * multiplication we have to divide the product by CONST_SCALE, with proper
- * rounding, to produce the correct output.  This division can be done
- * cheaply as a right shift of CONST_BITS bits.  We postpone shifting
- * as long as possible so that partial sums can be added together with
- * full fractional precision.
- *
- * The outputs of the first pass are scaled up by PASS1_BITS bits so that
- * they are represented to better-than-integral precision.  These outputs
- * require BITS_IN_JSAMPLE + PASS1_BITS + 3 bits; this fits in a 16-bit word
- * with the recommended scaling.  (To scale up 12-bit sample data further, an
- * intermediate INT32 array would be needed.)
- *
- * To avoid overflow of the 32-bit intermediate results in pass 2, we must
- * have BITS_IN_JSAMPLE + CONST_BITS + PASS1_BITS <= 26.  Error analysis
- * shows that the values given below are the most effective.
- */
-
-#if BITS_IN_JSAMPLE == 8
-#define CONST_BITS  13
-#define PASS1_BITS  2
-#else
-#define CONST_BITS  13
-#define PASS1_BITS  1		/* lose a little precision to avoid overflow */
-#endif
-
-/* Some C compilers fail to reduce "FIX(constant)" at compile time, thus
- * causing a lot of useless floating-point operations at run time.
- * To get around this we use the following pre-calculated constants.
- * If you change CONST_BITS you may want to add appropriate values.
- * (With a reasonable C compiler, you can just rely on the FIX() macro...)
- */
-
-#if CONST_BITS == 13
-#define FIX_0_298631336  ((INT32)  2446)	/* FIX(0.298631336) */
-#define FIX_0_390180644  ((INT32)  3196)	/* FIX(0.390180644) */
-#define FIX_0_541196100  ((INT32)  4433)	/* FIX(0.541196100) */
-#define FIX_0_765366865  ((INT32)  6270)	/* FIX(0.765366865) */
-#define FIX_0_899976223  ((INT32)  7373)	/* FIX(0.899976223) */
-#define FIX_1_175875602  ((INT32)  9633)	/* FIX(1.175875602) */
-#define FIX_1_501321110  ((INT32)  12299)	/* FIX(1.501321110) */
-#define FIX_1_847759065  ((INT32)  15137)	/* FIX(1.847759065) */
-#define FIX_1_961570560  ((INT32)  16069)	/* FIX(1.961570560) */
-#define FIX_2_053119869  ((INT32)  16819)	/* FIX(2.053119869) */
-#define FIX_2_562915447  ((INT32)  20995)	/* FIX(2.562915447) */
-#define FIX_3_072711026  ((INT32)  25172)	/* FIX(3.072711026) */
-#else
-#define FIX_0_298631336  FIX(0.298631336)
-#define FIX_0_390180644  FIX(0.390180644)
-#define FIX_0_541196100  FIX(0.541196100)
-#define FIX_0_765366865  FIX(0.765366865)
-#define FIX_0_899976223  FIX(0.899976223)
-#define FIX_1_175875602  FIX(1.175875602)
-#define FIX_1_501321110  FIX(1.501321110)
-#define FIX_1_847759065  FIX(1.847759065)
-#define FIX_1_961570560  FIX(1.961570560)
-#define FIX_2_053119869  FIX(2.053119869)
-#define FIX_2_562915447  FIX(2.562915447)
-#define FIX_3_072711026  FIX(3.072711026)
-#endif
-
-
-/* Multiply an INT32 variable by an INT32 constant to yield an INT32 result.
- * For 8-bit samples with the recommended scaling, all the variable
- * and constant values involved are no more than 16 bits wide, so a
- * 16x16->32 bit multiply can be used instead of a full 32x32 multiply.
- * For 12-bit samples, a full 32-bit multiplication will be needed.
- */
-
-#if BITS_IN_JSAMPLE == 8
-#define MULTIPLY(var,const)  MULTIPLY16C16(var,const)
-#else
-#define MULTIPLY(var,const)  ((var) * (const))
-#endif
-
-
-/* Dequantize a coefficient by multiplying it by the multiplier-table
- * entry; produce an int result.  In this module, both inputs and result
- * are 16 bits or less, so either int or short multiply will work.
- */
-
-#define DEQUANTIZE(coef,quantval)  (((ISLOW_MULT_TYPE) (coef)) * (quantval))
-
-
-/*
- * Perform dequantization and inverse DCT on one block of coefficients.
- */
-
-GLOBAL(void)
-jpeg_idct_islow (j_decompress_ptr cinfo, jpeg_component_info * compptr,
-		 JCOEFPTR coef_block,
-		 JSAMPARRAY output_buf, JDIMENSION output_col)
-{
-  INT32 tmp0, tmp1, tmp2, tmp3;
-  INT32 tmp10, tmp11, tmp12, tmp13;
-  INT32 z1, z2, z3, z4, z5;
-  JCOEFPTR inptr;
-  ISLOW_MULT_TYPE * quantptr;
-  int * wsptr;
-  JSAMPROW outptr;
-  JSAMPLE *range_limit = IDCT_range_limit(cinfo);
-  int ctr;
-  int workspace[DCTSIZE2];	/* buffers data between passes */
-  SHIFT_TEMPS
-
-  /* Pass 1: process columns from input, store into work array. */
-  /* Note results are scaled up by sqrt(8) compared to a true IDCT; */
-  /* furthermore, we scale the results by 2**PASS1_BITS. */
-
-  inptr = coef_block;
-  quantptr = (ISLOW_MULT_TYPE *) compptr->dct_table;
-  wsptr = workspace;
-  for (ctr = DCTSIZE; ctr > 0; ctr--) {
-    /* Due to quantization, we will usually find that many of the input
-     * coefficients are zero, especially the AC terms.  We can exploit this
-     * by short-circuiting the IDCT calculation for any column in which all
-     * the AC terms are zero.  In that case each output is equal to the
-     * DC coefficient (with scale factor as needed).
-     * With typical images and quantization tables, half or more of the
-     * column DCT calculations can be simplified this way.
-     */
-    
-    if (inptr[DCTSIZE*1] == 0 && inptr[DCTSIZE*2] == 0 &&
-	inptr[DCTSIZE*3] == 0 && inptr[DCTSIZE*4] == 0 &&
-	inptr[DCTSIZE*5] == 0 && inptr[DCTSIZE*6] == 0 &&
-	inptr[DCTSIZE*7] == 0) {
-      /* AC terms all zero */
-      int dcval = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]) << PASS1_BITS;
-      
-      wsptr[DCTSIZE*0] = dcval;
-      wsptr[DCTSIZE*1] = dcval;
-      wsptr[DCTSIZE*2] = dcval;
-      wsptr[DCTSIZE*3] = dcval;
-      wsptr[DCTSIZE*4] = dcval;
-      wsptr[DCTSIZE*5] = dcval;
-      wsptr[DCTSIZE*6] = dcval;
-      wsptr[DCTSIZE*7] = dcval;
-      
-      inptr++;			/* advance pointers to next column */
-      quantptr++;
-      wsptr++;
-      continue;
-    }
-    
-    /* Even part: reverse the even part of the forward DCT. */
-    /* The rotator is sqrt(2)*c(-6). */
-    
-    z2 = DEQUANTIZE(inptr[DCTSIZE*2], quantptr[DCTSIZE*2]);
-    z3 = DEQUANTIZE(inptr[DCTSIZE*6], quantptr[DCTSIZE*6]);
-    
-    z1 = MULTIPLY(z2 + z3, FIX_0_541196100);
-    tmp2 = z1 + MULTIPLY(z3, - FIX_1_847759065);
-    tmp3 = z1 + MULTIPLY(z2, FIX_0_765366865);
-    
-    z2 = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]);
-    z3 = DEQUANTIZE(inptr[DCTSIZE*4], quantptr[DCTSIZE*4]);
-
-    tmp0 = (z2 + z3) << CONST_BITS;
-    tmp1 = (z2 - z3) << CONST_BITS;
-    
-    tmp10 = tmp0 + tmp3;
-    tmp13 = tmp0 - tmp3;
-    tmp11 = tmp1 + tmp2;
-    tmp12 = tmp1 - tmp2;
-    
-    /* Odd part per figure 8; the matrix is unitary and hence its
-     * transpose is its inverse.  i0..i3 are y7,y5,y3,y1 respectively.
-     */
-    
-    tmp0 = DEQUANTIZE(inptr[DCTSIZE*7], quantptr[DCTSIZE*7]);
-    tmp1 = DEQUANTIZE(inptr[DCTSIZE*5], quantptr[DCTSIZE*5]);
-    tmp2 = DEQUANTIZE(inptr[DCTSIZE*3], quantptr[DCTSIZE*3]);
-    tmp3 = DEQUANTIZE(inptr[DCTSIZE*1], quantptr[DCTSIZE*1]);
-    
-    z1 = tmp0 + tmp3;
-    z2 = tmp1 + tmp2;
-    z3 = tmp0 + tmp2;
-    z4 = tmp1 + tmp3;
-    z5 = MULTIPLY(z3 + z4, FIX_1_175875602); /* sqrt(2) * c3 */
-    
-    tmp0 = MULTIPLY(tmp0, FIX_0_298631336); /* sqrt(2) * (-c1+c3+c5-c7) */
-    tmp1 = MULTIPLY(tmp1, FIX_2_053119869); /* sqrt(2) * ( c1+c3-c5+c7) */
-    tmp2 = MULTIPLY(tmp2, FIX_3_072711026); /* sqrt(2) * ( c1+c3+c5-c7) */
-    tmp3 = MULTIPLY(tmp3, FIX_1_501321110); /* sqrt(2) * ( c1+c3-c5-c7) */
-    z1 = MULTIPLY(z1, - FIX_0_899976223); /* sqrt(2) * (c7-c3) */
-    z2 = MULTIPLY(z2, - FIX_2_562915447); /* sqrt(2) * (-c1-c3) */
-    z3 = MULTIPLY(z3, - FIX_1_961570560); /* sqrt(2) * (-c3-c5) */
-    z4 = MULTIPLY(z4, - FIX_0_390180644); /* sqrt(2) * (c5-c3) */
-    
-    z3 += z5;
-    z4 += z5;
-    
-    tmp0 += z1 + z3;
-    tmp1 += z2 + z4;
-    tmp2 += z2 + z3;
-    tmp3 += z1 + z4;
-    
-    /* Final output stage: inputs are tmp10..tmp13, tmp0..tmp3 */
-    
-    wsptr[DCTSIZE*0] = (int) DESCALE(tmp10 + tmp3, CONST_BITS-PASS1_BITS);
-    wsptr[DCTSIZE*7] = (int) DESCALE(tmp10 - tmp3, CONST_BITS-PASS1_BITS);
-    wsptr[DCTSIZE*1] = (int) DESCALE(tmp11 + tmp2, CONST_BITS-PASS1_BITS);
-    wsptr[DCTSIZE*6] = (int) DESCALE(tmp11 - tmp2, CONST_BITS-PASS1_BITS);
-    wsptr[DCTSIZE*2] = (int) DESCALE(tmp12 + tmp1, CONST_BITS-PASS1_BITS);
-    wsptr[DCTSIZE*5] = (int) DESCALE(tmp12 - tmp1, CONST_BITS-PASS1_BITS);
-    wsptr[DCTSIZE*3] = (int) DESCALE(tmp13 + tmp0, CONST_BITS-PASS1_BITS);
-    wsptr[DCTSIZE*4] = (int) DESCALE(tmp13 - tmp0, CONST_BITS-PASS1_BITS);
-    
-    inptr++;			/* advance pointers to next column */
-    quantptr++;
-    wsptr++;
-  }
-  
-  /* Pass 2: process rows from work array, store into output array. */
-  /* Note that we must descale the results by a factor of 8 == 2**3, */
-  /* and also undo the PASS1_BITS scaling. */
-
-  wsptr = workspace;
-  for (ctr = 0; ctr < DCTSIZE; ctr++) {
-    outptr = output_buf[ctr] + output_col;
-    /* Rows of zeroes can be exploited in the same way as we did with columns.
-     * However, the column calculation has created many nonzero AC terms, so
-     * the simplification applies less often (typically 5% to 10% of the time).
-     * On machines with very fast multiplication, it's possible that the
-     * test takes more time than it's worth.  In that case this section
-     * may be commented out.
-     */
-    
-#ifndef NO_ZERO_ROW_TEST
-    if (wsptr[1] == 0 && wsptr[2] == 0 && wsptr[3] == 0 && wsptr[4] == 0 &&
-	wsptr[5] == 0 && wsptr[6] == 0 && wsptr[7] == 0) {
-      /* AC terms all zero */
-      JSAMPLE dcval = range_limit[(int) DESCALE((INT32) wsptr[0], PASS1_BITS+3)
-				  & RANGE_MASK];
-      
-      outptr[0] = dcval;
-      outptr[1] = dcval;
-      outptr[2] = dcval;
-      outptr[3] = dcval;
-      outptr[4] = dcval;
-      outptr[5] = dcval;
-      outptr[6] = dcval;
-      outptr[7] = dcval;
-
-      wsptr += DCTSIZE;		/* advance pointer to next row */
-      continue;
-    }
-#endif
-    
-    /* Even part: reverse the even part of the forward DCT. */
-    /* The rotator is sqrt(2)*c(-6). */
-    
-    z2 = (INT32) wsptr[2];
-    z3 = (INT32) wsptr[6];
-    
-    z1 = MULTIPLY(z2 + z3, FIX_0_541196100);
-    tmp2 = z1 + MULTIPLY(z3, - FIX_1_847759065);
-    tmp3 = z1 + MULTIPLY(z2, FIX_0_765366865);
-    
-    tmp0 = ((INT32) wsptr[0] + (INT32) wsptr[4]) << CONST_BITS;
-    tmp1 = ((INT32) wsptr[0] - (INT32) wsptr[4]) << CONST_BITS;
-    
-    tmp10 = tmp0 + tmp3;
-    tmp13 = tmp0 - tmp3;
-    tmp11 = tmp1 + tmp2;
-    tmp12 = tmp1 - tmp2;
-    
-    /* Odd part per figure 8; the matrix is unitary and hence its
-     * transpose is its inverse.  i0..i3 are y7,y5,y3,y1 respectively.
-     */
-    
-    tmp0 = (INT32) wsptr[7];
-    tmp1 = (INT32) wsptr[5];
-    tmp2 = (INT32) wsptr[3];
-    tmp3 = (INT32) wsptr[1];
-    
-    z1 = tmp0 + tmp3;
-    z2 = tmp1 + tmp2;
-    z3 = tmp0 + tmp2;
-    z4 = tmp1 + tmp3;
-    z5 = MULTIPLY(z3 + z4, FIX_1_175875602); /* sqrt(2) * c3 */
-    
-    tmp0 = MULTIPLY(tmp0, FIX_0_298631336); /* sqrt(2) * (-c1+c3+c5-c7) */
-    tmp1 = MULTIPLY(tmp1, FIX_2_053119869); /* sqrt(2) * ( c1+c3-c5+c7) */
-    tmp2 = MULTIPLY(tmp2, FIX_3_072711026); /* sqrt(2) * ( c1+c3+c5-c7) */
-    tmp3 = MULTIPLY(tmp3, FIX_1_501321110); /* sqrt(2) * ( c1+c3-c5-c7) */
-    z1 = MULTIPLY(z1, - FIX_0_899976223); /* sqrt(2) * (c7-c3) */
-    z2 = MULTIPLY(z2, - FIX_2_562915447); /* sqrt(2) * (-c1-c3) */
-    z3 = MULTIPLY(z3, - FIX_1_961570560); /* sqrt(2) * (-c3-c5) */
-    z4 = MULTIPLY(z4, - FIX_0_390180644); /* sqrt(2) * (c5-c3) */
-    
-    z3 += z5;
-    z4 += z5;
-    
-    tmp0 += z1 + z3;
-    tmp1 += z2 + z4;
-    tmp2 += z2 + z3;
-    tmp3 += z1 + z4;
-    
-    /* Final output stage: inputs are tmp10..tmp13, tmp0..tmp3 */
-    
-    outptr[0] = range_limit[(int) DESCALE(tmp10 + tmp3,
-					  CONST_BITS+PASS1_BITS+3)
-			    & RANGE_MASK];
-    outptr[7] = range_limit[(int) DESCALE(tmp10 - tmp3,
-					  CONST_BITS+PASS1_BITS+3)
-			    & RANGE_MASK];
-    outptr[1] = range_limit[(int) DESCALE(tmp11 + tmp2,
-					  CONST_BITS+PASS1_BITS+3)
-			    & RANGE_MASK];
-    outptr[6] = range_limit[(int) DESCALE(tmp11 - tmp2,
-					  CONST_BITS+PASS1_BITS+3)
-			    & RANGE_MASK];
-    outptr[2] = range_limit[(int) DESCALE(tmp12 + tmp1,
-					  CONST_BITS+PASS1_BITS+3)
-			    & RANGE_MASK];
-    outptr[5] = range_limit[(int) DESCALE(tmp12 - tmp1,
-					  CONST_BITS+PASS1_BITS+3)
-			    & RANGE_MASK];
-    outptr[3] = range_limit[(int) DESCALE(tmp13 + tmp0,
-					  CONST_BITS+PASS1_BITS+3)
-			    & RANGE_MASK];
-    outptr[4] = range_limit[(int) DESCALE(tmp13 - tmp0,
-					  CONST_BITS+PASS1_BITS+3)
-			    & RANGE_MASK];
-    
-    wsptr += DCTSIZE;		/* advance pointer to next row */
-  }
-}
-
-#endif /* DCT_ISLOW_SUPPORTED */
@@ -1,398 +0,0 @@
-/*
- * jidctred.c
- *
- * Copyright (C) 1994-1998, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains inverse-DCT routines that produce reduced-size output:
- * either 4x4, 2x2, or 1x1 pixels from an 8x8 DCT block.
- *
- * The implementation is based on the Loeffler, Ligtenberg and Moschytz (LL&M)
- * algorithm used in jidctint.c.  We simply replace each 8-to-8 1-D IDCT step
- * with an 8-to-4 step that produces the four averages of two adjacent outputs
- * (or an 8-to-2 step producing two averages of four outputs, for 2x2 output).
- * These steps were derived by computing the corresponding values at the end
- * of the normal LL&M code, then simplifying as much as possible.
- *
- * 1x1 is trivial: just take the DC coefficient divided by 8.
- *
- * See jidctint.c for additional comments.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-#include "jdct.h"		/* Private declarations for DCT subsystem */
-
-#ifdef IDCT_SCALING_SUPPORTED
-
-
-/*
- * This module is specialized to the case DCTSIZE = 8.
- */
-
-#if DCTSIZE != 8
-  Sorry, this code only copes with 8x8 DCTs. /* deliberate syntax err */
-#endif
-
-
-/* Scaling is the same as in jidctint.c. */
-
-#if BITS_IN_JSAMPLE == 8
-#define CONST_BITS  13
-#define PASS1_BITS  2
-#else
-#define CONST_BITS  13
-#define PASS1_BITS  1		/* lose a little precision to avoid overflow */
-#endif
-
-/* Some C compilers fail to reduce "FIX(constant)" at compile time, thus
- * causing a lot of useless floating-point operations at run time.
- * To get around this we use the following pre-calculated constants.
- * If you change CONST_BITS you may want to add appropriate values.
- * (With a reasonable C compiler, you can just rely on the FIX() macro...)
- */
-
-#if CONST_BITS == 13
-#define FIX_0_211164243  ((INT32)  1730)	/* FIX(0.211164243) */
-#define FIX_0_509795579  ((INT32)  4176)	/* FIX(0.509795579) */
-#define FIX_0_601344887  ((INT32)  4926)	/* FIX(0.601344887) */
-#define FIX_0_720959822  ((INT32)  5906)	/* FIX(0.720959822) */
-#define FIX_0_765366865  ((INT32)  6270)	/* FIX(0.765366865) */
-#define FIX_0_850430095  ((INT32)  6967)	/* FIX(0.850430095) */
-#define FIX_0_899976223  ((INT32)  7373)	/* FIX(0.899976223) */
-#define FIX_1_061594337  ((INT32)  8697)	/* FIX(1.061594337) */
-#define FIX_1_272758580  ((INT32)  10426)	/* FIX(1.272758580) */
-#define FIX_1_451774981  ((INT32)  11893)	/* FIX(1.451774981) */
-#define FIX_1_847759065  ((INT32)  15137)	/* FIX(1.847759065) */
-#define FIX_2_172734803  ((INT32)  17799)	/* FIX(2.172734803) */
-#define FIX_2_562915447  ((INT32)  20995)	/* FIX(2.562915447) */
-#define FIX_3_624509785  ((INT32)  29692)	/* FIX(3.624509785) */
-#else
-#define FIX_0_211164243  FIX(0.211164243)
-#define FIX_0_509795579  FIX(0.509795579)
-#define FIX_0_601344887  FIX(0.601344887)
-#define FIX_0_720959822  FIX(0.720959822)
-#define FIX_0_765366865  FIX(0.765366865)
-#define FIX_0_850430095  FIX(0.850430095)
-#define FIX_0_899976223  FIX(0.899976223)
-#define FIX_1_061594337  FIX(1.061594337)
-#define FIX_1_272758580  FIX(1.272758580)
-#define FIX_1_451774981  FIX(1.451774981)
-#define FIX_1_847759065  FIX(1.847759065)
-#define FIX_2_172734803  FIX(2.172734803)
-#define FIX_2_562915447  FIX(2.562915447)
-#define FIX_3_624509785  FIX(3.624509785)
-#endif
-
-
-/* Multiply an INT32 variable by an INT32 constant to yield an INT32 result.
- * For 8-bit samples with the recommended scaling, all the variable
- * and constant values involved are no more than 16 bits wide, so a
- * 16x16->32 bit multiply can be used instead of a full 32x32 multiply.
- * For 12-bit samples, a full 32-bit multiplication will be needed.
- */
-
-#if BITS_IN_JSAMPLE == 8
-#define MULTIPLY(var,const)  MULTIPLY16C16(var,const)
-#else
-#define MULTIPLY(var,const)  ((var) * (const))
-#endif
-
-
-/* Dequantize a coefficient by multiplying it by the multiplier-table
- * entry; produce an int result.  In this module, both inputs and result
- * are 16 bits or less, so either int or short multiply will work.
- */
-
-#define DEQUANTIZE(coef,quantval)  (((ISLOW_MULT_TYPE) (coef)) * (quantval))
-
-
-/*
- * Perform dequantization and inverse DCT on one block of coefficients,
- * producing a reduced-size 4x4 output block.
- */
-
-GLOBAL(void)
-jpeg_idct_4x4 (j_decompress_ptr cinfo, jpeg_component_info * compptr,
-	       JCOEFPTR coef_block,
-	       JSAMPARRAY output_buf, JDIMENSION output_col)
-{
-  INT32 tmp0, tmp2, tmp10, tmp12;
-  INT32 z1, z2, z3, z4;
-  JCOEFPTR inptr;
-  ISLOW_MULT_TYPE * quantptr;
-  int * wsptr;
-  JSAMPROW outptr;
-  JSAMPLE *range_limit = IDCT_range_limit(cinfo);
-  int ctr;
-  int workspace[DCTSIZE*4];	/* buffers data between passes */
-  SHIFT_TEMPS
-
-  /* Pass 1: process columns from input, store into work array. */
-
-  inptr = coef_block;
-  quantptr = (ISLOW_MULT_TYPE *) compptr->dct_table;
-  wsptr = workspace;
-  for (ctr = DCTSIZE; ctr > 0; inptr++, quantptr++, wsptr++, ctr--) {
-    /* Don't bother to process column 4, because second pass won't use it */
-    if (ctr == DCTSIZE-4)
-      continue;
-    if (inptr[DCTSIZE*1] == 0 && inptr[DCTSIZE*2] == 0 &&
-	inptr[DCTSIZE*3] == 0 && inptr[DCTSIZE*5] == 0 &&
-	inptr[DCTSIZE*6] == 0 && inptr[DCTSIZE*7] == 0) {
-      /* AC terms all zero; we need not examine term 4 for 4x4 output */
-      int dcval = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]) << PASS1_BITS;
-      
-      wsptr[DCTSIZE*0] = dcval;
-      wsptr[DCTSIZE*1] = dcval;
-      wsptr[DCTSIZE*2] = dcval;
-      wsptr[DCTSIZE*3] = dcval;
-      
-      continue;
-    }
-    
-    /* Even part */
-    
-    tmp0 = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]);
-    tmp0 <<= (CONST_BITS+1);
-    
-    z2 = DEQUANTIZE(inptr[DCTSIZE*2], quantptr[DCTSIZE*2]);
-    z3 = DEQUANTIZE(inptr[DCTSIZE*6], quantptr[DCTSIZE*6]);
-
-    tmp2 = MULTIPLY(z2, FIX_1_847759065) + MULTIPLY(z3, - FIX_0_765366865);
-    
-    tmp10 = tmp0 + tmp2;
-    tmp12 = tmp0 - tmp2;
-    
-    /* Odd part */
-    
-    z1 = DEQUANTIZE(inptr[DCTSIZE*7], quantptr[DCTSIZE*7]);
-    z2 = DEQUANTIZE(inptr[DCTSIZE*5], quantptr[DCTSIZE*5]);
-    z3 = DEQUANTIZE(inptr[DCTSIZE*3], quantptr[DCTSIZE*3]);
-    z4 = DEQUANTIZE(inptr[DCTSIZE*1], quantptr[DCTSIZE*1]);
-    
-    tmp0 = MULTIPLY(z1, - FIX_0_211164243) /* sqrt(2) * (c3-c1) */
-	 + MULTIPLY(z2, FIX_1_451774981) /* sqrt(2) * (c3+c7) */
-	 + MULTIPLY(z3, - FIX_2_172734803) /* sqrt(2) * (-c1-c5) */
-	 + MULTIPLY(z4, FIX_1_061594337); /* sqrt(2) * (c5+c7) */
-    
-    tmp2 = MULTIPLY(z1, - FIX_0_509795579) /* sqrt(2) * (c7-c5) */
-	 + MULTIPLY(z2, - FIX_0_601344887) /* sqrt(2) * (c5-c1) */
-	 + MULTIPLY(z3, FIX_0_899976223) /* sqrt(2) * (c3-c7) */
-	 + MULTIPLY(z4, FIX_2_562915447); /* sqrt(2) * (c1+c3) */
-
-    /* Final output stage */
-    
-    wsptr[DCTSIZE*0] = (int) DESCALE(tmp10 + tmp2, CONST_BITS-PASS1_BITS+1);
-    wsptr[DCTSIZE*3] = (int) DESCALE(tmp10 - tmp2, CONST_BITS-PASS1_BITS+1);
-    wsptr[DCTSIZE*1] = (int) DESCALE(tmp12 + tmp0, CONST_BITS-PASS1_BITS+1);
-    wsptr[DCTSIZE*2] = (int) DESCALE(tmp12 - tmp0, CONST_BITS-PASS1_BITS+1);
-  }
-  
-  /* Pass 2: process 4 rows from work array, store into output array. */
-
-  wsptr = workspace;
-  for (ctr = 0; ctr < 4; ctr++) {
-    outptr = output_buf[ctr] + output_col;
-    /* It's not clear whether a zero row test is worthwhile here ... */
-
-#ifndef NO_ZERO_ROW_TEST
-    if (wsptr[1] == 0 && wsptr[2] == 0 && wsptr[3] == 0 &&
-	wsptr[5] == 0 && wsptr[6] == 0 && wsptr[7] == 0) {
-      /* AC terms all zero */
-      JSAMPLE dcval = range_limit[(int) DESCALE((INT32) wsptr[0], PASS1_BITS+3)
-				  & RANGE_MASK];
-      
-      outptr[0] = dcval;
-      outptr[1] = dcval;
-      outptr[2] = dcval;
-      outptr[3] = dcval;
-      
-      wsptr += DCTSIZE;		/* advance pointer to next row */
-      continue;
-    }
-#endif
-    
-    /* Even part */
-    
-    tmp0 = ((INT32) wsptr[0]) << (CONST_BITS+1);
-    
-    tmp2 = MULTIPLY((INT32) wsptr[2], FIX_1_847759065)
-	 + MULTIPLY((INT32) wsptr[6], - FIX_0_765366865);
-    
-    tmp10 = tmp0 + tmp2;
-    tmp12 = tmp0 - tmp2;
-    
-    /* Odd part */
-    
-    z1 = (INT32) wsptr[7];
-    z2 = (INT32) wsptr[5];
-    z3 = (INT32) wsptr[3];
-    z4 = (INT32) wsptr[1];
-    
-    tmp0 = MULTIPLY(z1, - FIX_0_211164243) /* sqrt(2) * (c3-c1) */
-	 + MULTIPLY(z2, FIX_1_451774981) /* sqrt(2) * (c3+c7) */
-	 + MULTIPLY(z3, - FIX_2_172734803) /* sqrt(2) * (-c1-c5) */
-	 + MULTIPLY(z4, FIX_1_061594337); /* sqrt(2) * (c5+c7) */
-    
-    tmp2 = MULTIPLY(z1, - FIX_0_509795579) /* sqrt(2) * (c7-c5) */
-	 + MULTIPLY(z2, - FIX_0_601344887) /* sqrt(2) * (c5-c1) */
-	 + MULTIPLY(z3, FIX_0_899976223) /* sqrt(2) * (c3-c7) */
-	 + MULTIPLY(z4, FIX_2_562915447); /* sqrt(2) * (c1+c3) */
-
-    /* Final output stage */
-    
-    outptr[0] = range_limit[(int) DESCALE(tmp10 + tmp2,
-					  CONST_BITS+PASS1_BITS+3+1)
-			    & RANGE_MASK];
-    outptr[3] = range_limit[(int) DESCALE(tmp10 - tmp2,
-					  CONST_BITS+PASS1_BITS+3+1)
-			    & RANGE_MASK];
-    outptr[1] = range_limit[(int) DESCALE(tmp12 + tmp0,
-					  CONST_BITS+PASS1_BITS+3+1)
-			    & RANGE_MASK];
-    outptr[2] = range_limit[(int) DESCALE(tmp12 - tmp0,
-					  CONST_BITS+PASS1_BITS+3+1)
-			    & RANGE_MASK];
-    
-    wsptr += DCTSIZE;		/* advance pointer to next row */
-  }
-}
-
-
-/*
- * Perform dequantization and inverse DCT on one block of coefficients,
- * producing a reduced-size 2x2 output block.
- */
-
-GLOBAL(void)
-jpeg_idct_2x2 (j_decompress_ptr cinfo, jpeg_component_info * compptr,
-	       JCOEFPTR coef_block,
-	       JSAMPARRAY output_buf, JDIMENSION output_col)
-{
-  INT32 tmp0, tmp10, z1;
-  JCOEFPTR inptr;
-  ISLOW_MULT_TYPE * quantptr;
-  int * wsptr;
-  JSAMPROW outptr;
-  JSAMPLE *range_limit = IDCT_range_limit(cinfo);
-  int ctr;
-  int workspace[DCTSIZE*2];	/* buffers data between passes */
-  SHIFT_TEMPS
-
-  /* Pass 1: process columns from input, store into work array. */
-
-  inptr = coef_block;
-  quantptr = (ISLOW_MULT_TYPE *) compptr->dct_table;
-  wsptr = workspace;
-  for (ctr = DCTSIZE; ctr > 0; inptr++, quantptr++, wsptr++, ctr--) {
-    /* Don't bother to process columns 2,4,6 */
-    if (ctr == DCTSIZE-2 || ctr == DCTSIZE-4 || ctr == DCTSIZE-6)
-      continue;
-    if (inptr[DCTSIZE*1] == 0 && inptr[DCTSIZE*3] == 0 &&
-	inptr[DCTSIZE*5] == 0 && inptr[DCTSIZE*7] == 0) {
-      /* AC terms all zero; we need not examine terms 2,4,6 for 2x2 output */
-      int dcval = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]) << PASS1_BITS;
-      
-      wsptr[DCTSIZE*0] = dcval;
-      wsptr[DCTSIZE*1] = dcval;
-      
-      continue;
-    }
-    
-    /* Even part */
-    
-    z1 = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]);
-    tmp10 = z1 << (CONST_BITS+2);
-    
-    /* Odd part */
-
-    z1 = DEQUANTIZE(inptr[DCTSIZE*7], quantptr[DCTSIZE*7]);
-    tmp0 = MULTIPLY(z1, - FIX_0_720959822); /* sqrt(2) * (c7-c5+c3-c1) */
-    z1 = DEQUANTIZE(inptr[DCTSIZE*5], quantptr[DCTSIZE*5]);
-    tmp0 += MULTIPLY(z1, FIX_0_850430095); /* sqrt(2) * (-c1+c3+c5+c7) */
-    z1 = DEQUANTIZE(inptr[DCTSIZE*3], quantptr[DCTSIZE*3]);
-    tmp0 += MULTIPLY(z1, - FIX_1_272758580); /* sqrt(2) * (-c1+c3-c5-c7) */
-    z1 = DEQUANTIZE(inptr[DCTSIZE*1], quantptr[DCTSIZE*1]);
-    tmp0 += MULTIPLY(z1, FIX_3_624509785); /* sqrt(2) * (c1+c3+c5+c7) */
-
-    /* Final output stage */
-    
-    wsptr[DCTSIZE*0] = (int) DESCALE(tmp10 + tmp0, CONST_BITS-PASS1_BITS+2);
-    wsptr[DCTSIZE*1] = (int) DESCALE(tmp10 - tmp0, CONST_BITS-PASS1_BITS+2);
-  }
-  
-  /* Pass 2: process 2 rows from work array, store into output array. */
-
-  wsptr = workspace;
-  for (ctr = 0; ctr < 2; ctr++) {
-    outptr = output_buf[ctr] + output_col;
-    /* It's not clear whether a zero row test is worthwhile here ... */
-
-#ifndef NO_ZERO_ROW_TEST
-    if (wsptr[1] == 0 && wsptr[3] == 0 && wsptr[5] == 0 && wsptr[7] == 0) {
-      /* AC terms all zero */
-      JSAMPLE dcval = range_limit[(int) DESCALE((INT32) wsptr[0], PASS1_BITS+3)
-				  & RANGE_MASK];
-      
-      outptr[0] = dcval;
-      outptr[1] = dcval;
-      
-      wsptr += DCTSIZE;		/* advance pointer to next row */
-      continue;
-    }
-#endif
-    
-    /* Even part */
-    
-    tmp10 = ((INT32) wsptr[0]) << (CONST_BITS+2);
-    
-    /* Odd part */
-
-    tmp0 = MULTIPLY((INT32) wsptr[7], - FIX_0_720959822) /* sqrt(2) * (c7-c5+c3-c1) */
-	 + MULTIPLY((INT32) wsptr[5], FIX_0_850430095) /* sqrt(2) * (-c1+c3+c5+c7) */
-	 + MULTIPLY((INT32) wsptr[3], - FIX_1_272758580) /* sqrt(2) * (-c1+c3-c5-c7) */
-	 + MULTIPLY((INT32) wsptr[1], FIX_3_624509785); /* sqrt(2) * (c1+c3+c5+c7) */
-
-    /* Final output stage */
-    
-    outptr[0] = range_limit[(int) DESCALE(tmp10 + tmp0,
-					  CONST_BITS+PASS1_BITS+3+2)
-			    & RANGE_MASK];
-    outptr[1] = range_limit[(int) DESCALE(tmp10 - tmp0,
-					  CONST_BITS+PASS1_BITS+3+2)
-			    & RANGE_MASK];
-    
-    wsptr += DCTSIZE;		/* advance pointer to next row */
-  }
-}
-
-
-/*
- * Perform dequantization and inverse DCT on one block of coefficients,
- * producing a reduced-size 1x1 output block.
- */
-
-GLOBAL(void)
-jpeg_idct_1x1 (j_decompress_ptr cinfo, jpeg_component_info * compptr,
-	       JCOEFPTR coef_block,
-	       JSAMPARRAY output_buf, JDIMENSION output_col)
-{
-  int dcval;
-  ISLOW_MULT_TYPE * quantptr;
-  JSAMPLE *range_limit = IDCT_range_limit(cinfo);
-  SHIFT_TEMPS
-
-  /* We hardly need an inverse DCT routine for this: just take the
-   * average pixel value, which is one-eighth of the DC coefficient.
-   */
-  quantptr = (ISLOW_MULT_TYPE *) compptr->dct_table;
-  dcval = DEQUANTIZE(coef_block[0], quantptr[0]);
-  dcval = (int) DESCALE((INT32) dcval, 3);
-
-  output_buf[0][output_col] = range_limit[dcval & RANGE_MASK];
-}
-
-#endif /* IDCT_SCALING_SUPPORTED */
@@ -1,91 +0,0 @@
-/*
- * jinclude.h
- *
- * Copyright (C) 1991-1994, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file exists to provide a single place to fix any problems with
- * including the wrong system include files.  (Common problems are taken
- * care of by the standard jconfig symbols, but on really weird systems
- * you may have to edit this file.)
- *
- * NOTE: this file is NOT intended to be included by applications using the
- * JPEG library.  Most applications need only include jpeglib.h.
- */
-
-
-/* Include auto-config file to find out which system include files we need. */
-
-#include "jconfig.h"		/* auto configuration options */
-#define JCONFIG_INCLUDED	/* so that jpeglib.h doesn't do it again */
-
-/*
- * We need the NULL macro and size_t typedef.
- * On an ANSI-conforming system it is sufficient to include <stddef.h>.
- * Otherwise, we get them from <stdlib.h> or <stdio.h>; we may have to
- * pull in <sys/types.h> as well.
- * Note that the core JPEG library does not require <stdio.h>;
- * only the default error handler and data source/destination modules do.
- * But we must pull it in because of the references to FILE in jpeglib.h.
- * You can remove those references if you want to compile without <stdio.h>.
- */
-
-#ifdef HAVE_STDDEF_H
-#include <stddef.h>
-#endif
-
-#ifdef HAVE_STDLIB_H
-#include <stdlib.h>
-#endif
-
-#ifdef NEED_SYS_TYPES_H
-#include <sys/types.h>
-#endif
-
-#include <stdio.h>
-
-/*
- * We need memory copying and zeroing functions, plus strncpy().
- * ANSI and System V implementations declare these in <string.h>.
- * BSD doesn't have the mem() functions, but it does have bcopy()/bzero().
- * Some systems may declare memset and memcpy in <memory.h>.
- *
- * NOTE: we assume the size parameters to these functions are of type size_t.
- * Change the casts in these macros if not!
- */
-
-#ifdef NEED_BSD_STRINGS
-
-#include <strings.h>
-#define MEMZERO(target,size)	bzero((void *)(target), (size_t)(size))
-#define MEMCOPY(dest,src,size)	bcopy((const void *)(src), (void *)(dest), (size_t)(size))
-
-#else /* not BSD, assume ANSI/SysV string lib */
-
-#include <string.h>
-#define MEMZERO(target,size)	memset((void *)(target), 0, (size_t)(size))
-#define MEMCOPY(dest,src,size)	memcpy((void *)(dest), (const void *)(src), (size_t)(size))
-
-#endif
-
-/*
- * In ANSI C, and indeed any rational implementation, size_t is also the
- * type returned by sizeof().  However, it seems there are some irrational
- * implementations out there, in which sizeof() returns an int even though
- * size_t is defined as long or unsigned long.  To ensure consistent results
- * we always use this SIZEOF() macro in place of using sizeof() directly.
- */
-
-#define SIZEOF(object)	((size_t) sizeof(object))
-
-/*
- * The modules that use fread() and fwrite() always invoke them through
- * these macros.  On some systems you may need to twiddle the argument casts.
- * CAUTION: argument order is different from underlying functions!
- */
-
-#define JFREAD(file,buf,sizeofbuf)  \
-  ((size_t) fread((void *) (buf), (size_t) 1, (size_t) (sizeofbuf), (file)))
-#define JFWRITE(file,buf,sizeofbuf)  \
-  ((size_t) fwrite((const void *) (buf), (size_t) 1, (size_t) (sizeofbuf), (file)))
@@ -1,109 +0,0 @@
-/*
- * jmemnobs.c
- *
- * Copyright (C) 1992-1996, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file provides a really simple implementation of the system-
- * dependent portion of the JPEG memory manager.  This implementation
- * assumes that no backing-store files are needed: all required space
- * can be obtained from malloc().
- * This is very portable in the sense that it'll compile on almost anything,
- * but you'd better have lots of main memory (or virtual memory) if you want
- * to process big images.
- * Note that the max_memory_to_use option is ignored by this implementation.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-#include "jmemsys.h"		/* import the system-dependent declarations */
-
-#ifndef HAVE_STDLIB_H		/* <stdlib.h> should declare malloc(),free() */
-extern void * malloc JPP((size_t size));
-extern void free JPP((void *ptr));
-#endif
-
-
-/*
- * Memory allocation and freeing are controlled by the regular library
- * routines malloc() and free().
- */
-
-GLOBAL(void *)
-jpeg_get_small (j_common_ptr cinfo, size_t sizeofobject)
-{
-  return (void *) malloc(sizeofobject);
-}
-
-GLOBAL(void)
-jpeg_free_small (j_common_ptr cinfo, void * object, size_t sizeofobject)
-{
-  free(object);
-}
-
-
-/*
- * "Large" objects are treated the same as "small" ones.
- * NB: although we include FAR keywords in the routine declarations,
- * this file won't actually work in 80x86 small/medium model; at least,
- * you probably won't be able to process useful-size images in only 64KB.
- */
-
-GLOBAL(void FAR *)
-jpeg_get_large (j_common_ptr cinfo, size_t sizeofobject)
-{
-  return (void FAR *) malloc(sizeofobject);
-}
-
-GLOBAL(void)
-jpeg_free_large (j_common_ptr cinfo, void FAR * object, size_t sizeofobject)
-{
-  free(object);
-}
-
-
-/*
- * This routine computes the total memory space available for allocation.
- * Here we always say, "we got all you want bud!"
- */
-
-GLOBAL(long)
-jpeg_mem_available (j_common_ptr cinfo, long min_bytes_needed,
-		    long max_bytes_needed, long already_allocated)
-{
-  return max_bytes_needed;
-}
-
-
-/*
- * Backing store (temporary file) management.
- * Since jpeg_mem_available always promised the moon,
- * this should never be called and we can just error out.
- */
-
-GLOBAL(void)
-jpeg_open_backing_store (j_common_ptr cinfo, backing_store_ptr info,
-			 long total_bytes_needed)
-{
-  ERREXIT(cinfo, JERR_NO_BACKING_STORE);
-}
-
-
-/*
- * These routines take care of any system-dependent initialization and
- * cleanup required.  Here, there isn't any.
- */
-
-GLOBAL(long)
-jpeg_mem_init (j_common_ptr cinfo)
-{
-  return 0;			/* just set max_memory_to_use to 0 */
-}
-
-GLOBAL(void)
-jpeg_mem_term (j_common_ptr cinfo)
-{
-  /* no work */
-}
@@ -1,204 +0,0 @@
-/*
- * jmemsys.h
- *
- * Copyright (C) 1992-1997, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This include file defines the interface between the system-independent
- * and system-dependent portions of the JPEG memory manager.  No other
- * modules need include it.  (The system-independent portion is jmemmgr.c;
- * there are several different versions of the system-dependent portion.)
- *
- * This file works as-is for the system-dependent memory managers supplied
- * in the IJG distribution.  You may need to modify it if you write a
- * custom memory manager.  If system-dependent changes are needed in
- * this file, the best method is to #ifdef them based on a configuration
- * symbol supplied in jconfig.h, as we have done with USE_MSDOS_MEMMGR
- * and USE_MAC_MEMMGR.
- */
-
-
-/* Short forms of external names for systems with brain-damaged linkers. */
-
-#ifdef NEED_SHORT_EXTERNAL_NAMES
-#define jpeg_get_small		jGetSmall
-#define jpeg_free_small		jFreeSmall
-#define jpeg_get_large		jGetLarge
-#define jpeg_free_large		jFreeLarge
-#define jpeg_mem_available	jMemAvail
-#define jpeg_open_backing_store	jOpenBackStore
-#define jpeg_mem_init		jMemInit
-#define jpeg_mem_term		jMemTerm
-#endif /* NEED_SHORT_EXTERNAL_NAMES */
-
-
-/*
- * These two functions are used to allocate and release small chunks of
- * memory.  (Typically the total amount requested through jpeg_get_small is
- * no more than 20K or so; this will be requested in chunks of a few K each.)
- * Behavior should be the same as for the standard library functions malloc
- * and free; in particular, jpeg_get_small must return NULL on failure.
- * On most systems, these ARE malloc and free.  jpeg_free_small is passed the
- * size of the object being freed, just in case it's needed.
- * On an 80x86 machine using small-data memory model, these manage near heap.
- */
-
-EXTERN(void *) jpeg_get_small JPP((j_common_ptr cinfo, size_t sizeofobject));
-EXTERN(void) jpeg_free_small JPP((j_common_ptr cinfo, void * object,
-				  size_t sizeofobject));
-
-/*
- * These two functions are used to allocate and release large chunks of
- * memory (up to the total free space designated by jpeg_mem_available).
- * The interface is the same as above, except that on an 80x86 machine,
- * far pointers are used.  On most other machines these are identical to
- * the jpeg_get/free_small routines; but we keep them separate anyway,
- * in case a different allocation strategy is desirable for large chunks.
- */
-
-EXTERN(void FAR *) jpeg_get_large JPP((j_common_ptr cinfo,
-				       size_t sizeofobject));
-EXTERN(void) jpeg_free_large JPP((j_common_ptr cinfo, void FAR * object,
-				  size_t sizeofobject));
-
-/*
- * The macro MAX_ALLOC_CHUNK designates the maximum number of bytes that may
- * be requested in a single call to jpeg_get_large (and jpeg_get_small for that
- * matter, but that case should never come into play).  This macro is needed
- * to model the 64Kb-segment-size limit of far addressing on 80x86 machines.
- * On those machines, we expect that jconfig.h will provide a proper value.
- * On machines with 32-bit flat address spaces, any large constant may be used.
- *
- * NB: jmemmgr.c expects that MAX_ALLOC_CHUNK will be representable as type
- * size_t and will be a multiple of sizeof(align_type).
- */
-
-#ifndef MAX_ALLOC_CHUNK		/* may be overridden in jconfig.h */
-#define MAX_ALLOC_CHUNK  1000000000L
-#endif
-
-/*
- * This routine computes the total space still available for allocation by
- * jpeg_get_large.  If more space than this is needed, backing store will be
- * used.  NOTE: any memory already allocated must not be counted.
- *
- * There is a minimum space requirement, corresponding to the minimum
- * feasible buffer sizes; jmemmgr.c will request that much space even if
- * jpeg_mem_available returns zero.  The maximum space needed, enough to hold
- * all working storage in memory, is also passed in case it is useful.
- * Finally, the total space already allocated is passed.  If no better
- * method is available, cinfo->mem->max_memory_to_use - already_allocated
- * is often a suitable calculation.
- *
- * It is OK for jpeg_mem_available to underestimate the space available
- * (that'll just lead to more backing-store access than is really necessary).
- * However, an overestimate will lead to failure.  Hence it's wise to subtract
- * a slop factor from the true available space.  5% should be enough.
- *
- * On machines with lots of virtual memory, any large constant may be returned.
- * Conversely, zero may be returned to always use the minimum amount of memory.
- */
-
-EXTERN(long) jpeg_mem_available JPP((j_common_ptr cinfo,
-				     long min_bytes_needed,
-				     long max_bytes_needed,
-				     long already_allocated));
-
-
-/*
- * This structure holds whatever state is needed to access a single
- * backing-store object.  The read/write/close method pointers are called
- * by jmemmgr.c to manipulate the backing-store object; all other fields
- * are private to the system-dependent backing store routines.
- */
-
-#define TEMP_NAME_LENGTH   64	/* max length of a temporary file's name */
-
-
-#ifdef USE_MSDOS_MEMMGR		/* DOS-specific junk */
-
-typedef unsigned short XMSH;	/* type of extended-memory handles */
-typedef unsigned short EMSH;	/* type of expanded-memory handles */
-
-typedef union {
-  short file_handle;		/* DOS file handle if it's a temp file */
-  XMSH xms_handle;		/* handle if it's a chunk of XMS */
-  EMSH ems_handle;		/* handle if it's a chunk of EMS */
-} handle_union;
-
-#endif /* USE_MSDOS_MEMMGR */
-
-#ifdef USE_MAC_MEMMGR		/* Mac-specific junk */
-#include <Files.h>
-#endif /* USE_MAC_MEMMGR */
-
-
-typedef struct backing_store_struct * backing_store_ptr;
-
-typedef struct backing_store_struct {
-  /* Methods for reading/writing/closing this backing-store object */
-  JMETHOD(void, read_backing_store, (j_common_ptr cinfo,
-				     backing_store_ptr info,
-				     void FAR * buffer_address,
-				     long file_offset, long byte_count));
-  JMETHOD(void, write_backing_store, (j_common_ptr cinfo,
-				      backing_store_ptr info,
-				      void FAR * buffer_address,
-				      long file_offset, long byte_count));
-  JMETHOD(void, close_backing_store, (j_common_ptr cinfo,
-				      backing_store_ptr info));
-
-  /* Private fields for system-dependent backing-store management */
-#ifdef USE_MSDOS_MEMMGR
-  /* For the MS-DOS manager (jmemdos.c), we need: */
-  handle_union handle;		/* reference to backing-store storage object */
-  char temp_name[TEMP_NAME_LENGTH]; /* name if it's a file */
-#else
-#ifdef USE_MAC_MEMMGR
-  /* For the Mac manager (jmemmac.c), we need: */
-  short temp_file;		/* file reference number to temp file */
-  FSSpec tempSpec;		/* the FSSpec for the temp file */
-  char temp_name[TEMP_NAME_LENGTH]; /* name if it's a file */
-#else
-#ifdef USE_ANDROID_ASHMEM
-  short temp_file;		/* file reference number to temp file */
-  unsigned char* addr;  /* the memory address mapped to ashmem */
-  long size;            /* the requested ashmem size */
-#else
-  /* For a typical implementation with temp files, we need: */
-  FILE * temp_file;		/* stdio reference to temp file */
-  char temp_name[TEMP_NAME_LENGTH]; /* name of temp file */
-#endif
-#endif
-#endif
-} backing_store_info;
-
-
-/*
- * Initial opening of a backing-store object.  This must fill in the
- * read/write/close pointers in the object.  The read/write routines
- * may take an error exit if the specified maximum file size is exceeded.
- * (If jpeg_mem_available always returns a large value, this routine can
- * just take an error exit.)
- */
-
-EXTERN(void) jpeg_open_backing_store JPP((j_common_ptr cinfo,
-					  backing_store_ptr info,
-					  long total_bytes_needed));
-
-
-/*
- * These routines take care of any system-dependent initialization and
- * cleanup required.  jpeg_mem_init will be called before anything is
- * allocated (and, therefore, nothing in cinfo is of use except the error
- * manager pointer).  It should return a suitable default value for
- * max_memory_to_use; this may subsequently be overridden by the surrounding
- * application.  (Note that max_memory_to_use is only important if
- * jpeg_mem_available chooses to consult it ... no one else will.)
- * jpeg_mem_term may assume that all requested memory has been freed and that
- * all opened backing-store objects have been closed.
- */
-
-EXTERN(long) jpeg_mem_init JPP((j_common_ptr cinfo));
-EXTERN(void) jpeg_mem_term JPP((j_common_ptr cinfo));
@@ -1,395 +0,0 @@
-/*
- * jmorecfg.h
- *
- * Copyright (C) 1991-1997, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains additional configuration options that customize the
- * JPEG software for special applications or support machine-dependent
- * optimizations.  Most users will not need to touch this file.
- */
-
-/*
- * Define ANDROID_RGB to enable specific optimizations for Android
- *   JCS_RGBA_8888 support
- *   JCS_RGB_565 support
- * 
- */
-
-#define ANDROID_RGB
-
-#ifdef ANDROID_RGB
-#define PACK_SHORT_565(r,g,b)  ((((r)<<8)&0xf800)|(((g)<<3)&0x7E0)|((b)>>3))
-#define PACK_TWO_PIXELS(l,r)   ((r<<16) | l)
-#define PACK_NEED_ALIGNMENT(ptr) (((uintptr_t)(ptr))&3)
-#define WRITE_TWO_PIXELS(addr, pixels) do {     \
-         ((INT16*)(addr))[0] = (pixels);        \
-         ((INT16*)(addr))[1] = (pixels)>>16;    \
-    } while(0)
-#define WRITE_TWO_ALIGNED_PIXELS(addr, pixels)  ((*(INT32*)(addr)) = pixels)
-#define DITHER_565_R(r, dither) ((r) + ((dither)&0xFF))
-#define DITHER_565_G(g, dither) ((g) + (((dither)&0xFF)>>1))
-#define DITHER_565_B(b, dither) ((b) + ((dither)&0xFF))
-#endif
-
-/*
- * Define BITS_IN_JSAMPLE as either
- *   8   for 8-bit sample values (the usual setting)
- *   12  for 12-bit sample values
- * Only 8 and 12 are legal data precisions for lossy JPEG according to the
- * JPEG standard, and the IJG code does not support anything else!
- * We do not support run-time selection of data precision, sorry.
- */
-
-#define BITS_IN_JSAMPLE  8	/* use 8 or 12 */
-
-
-/*
- * Maximum number of components (color channels) allowed in JPEG image.
- * To meet the letter of the JPEG spec, set this to 255.  However, darn
- * few applications need more than 4 channels (maybe 5 for CMYK + alpha
- * mask).  We recommend 10 as a reasonable compromise; use 4 if you are
- * really short on memory.  (Each allowed component costs a hundred or so
- * bytes of storage, whether actually used in an image or not.)
- */
-
-#define MAX_COMPONENTS  10	/* maximum number of image components */
-
-
-/*
- * Basic data types.
- * You may need to change these if you have a machine with unusual data
- * type sizes; for example, "char" not 8 bits, "short" not 16 bits,
- * or "long" not 32 bits.  We don't care whether "int" is 16 or 32 bits,
- * but it had better be at least 16.
- */
-
-/* Representation of a single sample (pixel element value).
- * We frequently allocate large arrays of these, so it's important to keep
- * them small.  But if you have memory to burn and access to char or short
- * arrays is very slow on your hardware, you might want to change these.
- */
-
-#if BITS_IN_JSAMPLE == 8
-/* JSAMPLE should be the smallest type that will hold the values 0..255.
- * You can use a signed char by having GETJSAMPLE mask it with 0xFF.
- */
-
-#ifdef HAVE_UNSIGNED_CHAR
-
-typedef unsigned char JSAMPLE;
-#define GETJSAMPLE(value)  ((int) (value))
-
-#else /* not HAVE_UNSIGNED_CHAR */
-
-typedef char JSAMPLE;
-#ifdef CHAR_IS_UNSIGNED
-#define GETJSAMPLE(value)  ((int) (value))
-#else
-#define GETJSAMPLE(value)  ((int) (value) & 0xFF)
-#endif /* CHAR_IS_UNSIGNED */
-
-#endif /* HAVE_UNSIGNED_CHAR */
-
-#define MAXJSAMPLE	255
-#define CENTERJSAMPLE	128
-
-#endif /* BITS_IN_JSAMPLE == 8 */
-
-
-#if BITS_IN_JSAMPLE == 12
-/* JSAMPLE should be the smallest type that will hold the values 0..4095.
- * On nearly all machines "short" will do nicely.
- */
-
-typedef short JSAMPLE;
-#define GETJSAMPLE(value)  ((int) (value))
-
-#define MAXJSAMPLE	4095
-#define CENTERJSAMPLE	2048
-
-#endif /* BITS_IN_JSAMPLE == 12 */
-
-
-/* Representation of a DCT frequency coefficient.
- * This should be a signed value of at least 16 bits; "short" is usually OK.
- * Again, we allocate large arrays of these, but you can change to int
- * if you have memory to burn and "short" is really slow.
- */
-
-typedef short JCOEF;
-
-
-/* Compressed datastreams are represented as arrays of JOCTET.
- * These must be EXACTLY 8 bits wide, at least once they are written to
- * external storage.  Note that when using the stdio data source/destination
- * managers, this is also the data type passed to fread/fwrite.
- */
-
-#ifdef HAVE_UNSIGNED_CHAR
-
-typedef unsigned char JOCTET;
-#define GETJOCTET(value)  (value)
-
-#else /* not HAVE_UNSIGNED_CHAR */
-
-typedef char JOCTET;
-#ifdef CHAR_IS_UNSIGNED
-#define GETJOCTET(value)  (value)
-#else
-#define GETJOCTET(value)  ((value) & 0xFF)
-#endif /* CHAR_IS_UNSIGNED */
-
-#endif /* HAVE_UNSIGNED_CHAR */
-
-
-/* These typedefs are used for various table entries and so forth.
- * They must be at least as wide as specified; but making them too big
- * won't cost a huge amount of memory, so we don't provide special
- * extraction code like we did for JSAMPLE.  (In other words, these
- * typedefs live at a different point on the speed/space tradeoff curve.)
- */
-
-/* UINT8 must hold at least the values 0..255. */
-
-#ifdef HAVE_UNSIGNED_CHAR
-typedef unsigned char UINT8;
-#else /* not HAVE_UNSIGNED_CHAR */
-#ifdef CHAR_IS_UNSIGNED
-typedef char UINT8;
-#else /* not CHAR_IS_UNSIGNED */
-typedef short UINT8;
-#endif /* CHAR_IS_UNSIGNED */
-#endif /* HAVE_UNSIGNED_CHAR */
-
-/* UINT16 must hold at least the values 0..65535. */
-
-#ifdef HAVE_UNSIGNED_SHORT
-typedef unsigned short UINT16;
-#else /* not HAVE_UNSIGNED_SHORT */
-typedef unsigned int UINT16;
-#endif /* HAVE_UNSIGNED_SHORT */
-
-/* INT16 must hold at least the values -32768..32767. */
-
-#ifndef XMD_H			/* X11/xmd.h correctly defines INT16 */
-typedef short INT16;
-#endif
-
-/* INT32 must hold at least signed 32-bit values. */
-
-#ifndef XMD_H			/* X11/xmd.h correctly defines INT32 */
-typedef long INT32;
-#endif
-
-/* Datatype used for image dimensions.  The JPEG standard only supports
- * images up to 64K*64K due to 16-bit fields in SOF markers.  Therefore
- * "unsigned int" is sufficient on all machines.  However, if you need to
- * handle larger images and you don't mind deviating from the spec, you
- * can change this datatype.
- */
-
-typedef unsigned int JDIMENSION;
-
-#define JPEG_MAX_DIMENSION  65500L  /* a tad under 64K to prevent overflows */
-
-
-/* These macros are used in all function definitions and extern declarations.
- * You could modify them if you need to change function linkage conventions;
- * in particular, you'll need to do that to make the library a Windows DLL.
- * Another application is to make all functions global for use with debuggers
- * or code profilers that require it.
- */
-
-/* a function called through method pointers: */
-#define METHODDEF(type)		static type
-/* a function used only in its module: */
-#define LOCAL(type)		static type
-/* a function referenced thru EXTERNs: */
-#define GLOBAL(type)		type
-/* a reference to a GLOBAL function: */
-#define EXTERN(type)		extern type
-
-
-/* This macro is used to declare a "method", that is, a function pointer.
- * We want to supply prototype parameters if the compiler can cope.
- * Note that the arglist parameter must be parenthesized!
- * Again, you can customize this if you need special linkage keywords.
- */
-
-#ifdef HAVE_PROTOTYPES
-#define JMETHOD(type,methodname,arglist)  type (*methodname) arglist
-#else
-#define JMETHOD(type,methodname,arglist)  type (*methodname) ()
-#endif
-
-
-/* Here is the pseudo-keyword for declaring pointers that must be "far"
- * on 80x86 machines.  Most of the specialized coding for 80x86 is handled
- * by just saying "FAR *" where such a pointer is needed.  In a few places
- * explicit coding is needed; see uses of the NEED_FAR_POINTERS symbol.
- */
-
-#ifdef NEED_FAR_POINTERS
-#define FAR  far
-#else
-#define FAR
-#endif
-
-
-/*
- * On a few systems, type boolean and/or its values FALSE, TRUE may appear
- * in standard header files.  Or you may have conflicts with application-
- * specific header files that you want to include together with these files.
- * Defining HAVE_BOOLEAN before including jpeglib.h should make it work.
- */
-
-#ifndef HAVE_BOOLEAN
-typedef int boolean;
-#endif
-#ifndef FALSE			/* in case these macros already exist */
-#define FALSE	0		/* values of boolean */
-#endif
-#ifndef TRUE
-#define TRUE	1
-#endif
-
-
-/*
- * The remaining options affect code selection within the JPEG library,
- * but they don't need to be visible to most applications using the library.
- * To minimize application namespace pollution, the symbols won't be
- * defined unless JPEG_INTERNALS or JPEG_INTERNAL_OPTIONS has been defined.
- */
-
-#ifdef JPEG_INTERNALS
-#define JPEG_INTERNAL_OPTIONS
-#endif
-
-#ifdef JPEG_INTERNAL_OPTIONS
-
-
-/*
- * These defines indicate whether to include various optional functions.
- * Undefining some of these symbols will produce a smaller but less capable
- * library.  Note that you can leave certain source files out of the
- * compilation/linking process if you've #undef'd the corresponding symbols.
- * (You may HAVE to do that if your compiler doesn't like null source files.)
- */
-
-/* Arithmetic coding is unsupported for legal reasons.  Complaints to IBM. */
-
-/* Capability options common to encoder and decoder: */
-
-#define DCT_ISLOW_SUPPORTED	/* slow but accurate integer algorithm */
-#define DCT_IFAST_SUPPORTED	/* faster, less accurate integer method */
-#define DCT_FLOAT_SUPPORTED	/* floating-point: accurate, fast on fast HW */
-
-/* Encoder capability options: */
-
-#undef  C_ARITH_CODING_SUPPORTED    /* Arithmetic coding back end? */
-#define C_MULTISCAN_FILES_SUPPORTED /* Multiple-scan JPEG files? */
-#define C_PROGRESSIVE_SUPPORTED	    /* Progressive JPEG? (Requires MULTISCAN)*/
-#define ENTROPY_OPT_SUPPORTED	    /* Optimization of entropy coding parms? */
-/* Note: if you selected 12-bit data precision, it is dangerous to turn off
- * ENTROPY_OPT_SUPPORTED.  The standard Huffman tables are only good for 8-bit
- * precision, so jchuff.c normally uses entropy optimization to compute
- * usable tables for higher precision.  If you don't want to do optimization,
- * you'll have to supply different default Huffman tables.
- * The exact same statements apply for progressive JPEG: the default tables
- * don't work for progressive mode.  (This may get fixed, however.)
- */
-#define INPUT_SMOOTHING_SUPPORTED   /* Input image smoothing option? */
-
-/* Decoder capability options: */
-
-#undef  D_ARITH_CODING_SUPPORTED    /* Arithmetic coding back end? */
-#define D_MULTISCAN_FILES_SUPPORTED /* Multiple-scan JPEG files? */
-#define D_PROGRESSIVE_SUPPORTED	    /* Progressive JPEG? (Requires MULTISCAN)*/
-#define SAVE_MARKERS_SUPPORTED	    /* jpeg_save_markers() needed? */
-#define BLOCK_SMOOTHING_SUPPORTED   /* Block smoothing? (Progressive only) */
-#define IDCT_SCALING_SUPPORTED	    /* Output rescaling via IDCT? */
-#undef  UPSAMPLE_SCALING_SUPPORTED  /* Output rescaling at upsample stage? */
-#define UPSAMPLE_MERGING_SUPPORTED  /* Fast path for sloppy upsampling? */
-#define QUANT_1PASS_SUPPORTED	    /* 1-pass color quantization? */
-#define QUANT_2PASS_SUPPORTED	    /* 2-pass color quantization? */
-
-/* more capability options later, no doubt */
-
-
-/*
- * Ordering of RGB data in scanlines passed to or from the application.
- * If your application wants to deal with data in the order B,G,R, just
- * change these macros.  You can also deal with formats such as R,G,B,X
- * (one extra byte per pixel) by changing RGB_PIXELSIZE.  Note that changing
- * the offsets will also change the order in which colormap data is organized.
- * RESTRICTIONS:
- * 1. The sample applications cjpeg,djpeg do NOT support modified RGB formats.
- * 2. These macros only affect RGB<=>YCbCr color conversion, so they are not
- *    useful if you are using JPEG color spaces other than YCbCr or grayscale.
- * 3. The color quantizer modules will not behave desirably if RGB_PIXELSIZE
- *    is not 3 (they don't understand about dummy color components!).  So you
- *    can't use color quantization if you change that value.
- */
-
-#define RGB_RED		0	/* Offset of Red in an RGB scanline element */
-#define RGB_GREEN	1	/* Offset of Green */
-#define RGB_BLUE	2	/* Offset of Blue */
-#ifdef ANDROID_RGB
-#define RGB_ALPHA   3   /* Offset of Alpha */
-#endif
-#define RGB_PIXELSIZE   3   /* JSAMPLEs per RGB scanline element */
-
-/* Definitions for speed-related optimizations. */
-
-
-/* If your compiler supports inline functions, define INLINE
- * as the inline keyword; otherwise define it as empty.
- */
-
-#ifndef INLINE
-#ifdef __GNUC__			/* for instance, GNU C knows about inline */
-#define INLINE __inline__
-#endif
-#ifndef INLINE
-#define INLINE			/* default is to define it as empty */
-#endif
-#endif
-
-
-/* On some machines (notably 68000 series) "int" is 32 bits, but multiplying
- * two 16-bit shorts is faster than multiplying two ints.  Define MULTIPLIER
- * as short on such a machine.  MULTIPLIER must be at least 16 bits wide.
- */
-
-#ifndef MULTIPLIER
-#ifdef ANDROID_INTELSSE2_IDCT
-  #define MULTIPLIER short
-#elif ANDROID_MIPS_IDCT
-  #define MULTIPLIER  short
-#elif NV_ARM_NEON
-  #define MULTIPLIER short
-#else
-  #define MULTIPLIER  int		/* type for fastest integer multiply */
-#endif
-#endif
-
-
-/* FAST_FLOAT should be either float or double, whichever is done faster
- * by your compiler.  (Note that this type is only used in the floating point
- * DCT routines, so it only matters if you've defined DCT_FLOAT_SUPPORTED.)
- * Typically, float is faster in ANSI C compilers, while double is faster in
- * pre-ANSI compilers (because they insist on converting to double anyway).
- * The code below therefore chooses float if we have ANSI-style prototypes.
- */
-
-#ifndef FAST_FLOAT
-#ifdef HAVE_PROTOTYPES
-#define FAST_FLOAT  float
-#else
-#define FAST_FLOAT  double
-#endif
-#endif
-
-#endif /* JPEG_INTERNAL_OPTIONS */
@@ -1,432 +0,0 @@
-/*
- * jpegint.h
- *
- * Copyright (C) 1991-1997, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file provides common declarations for the various JPEG modules.
- * These declarations are considered internal to the JPEG library; most
- * applications using the library shouldn't need to include this file.
- */
-
-
-/* Declarations for both compression & decompression */
-
-typedef enum {			/* Operating modes for buffer controllers */
-	JBUF_PASS_THRU,		/* Plain stripwise operation */
-	/* Remaining modes require a full-image buffer to have been created */
-	JBUF_SAVE_SOURCE,	/* Run source subobject only, save output */
-	JBUF_CRANK_DEST,	/* Run dest subobject only, using saved data */
-	JBUF_SAVE_AND_PASS	/* Run both subobjects, save output */
-} J_BUF_MODE;
-
-/* Values of global_state field (jdapi.c has some dependencies on ordering!) */
-#define CSTATE_START	100	/* after create_compress */
-#define CSTATE_SCANNING	101	/* start_compress done, write_scanlines OK */
-#define CSTATE_RAW_OK	102	/* start_compress done, write_raw_data OK */
-#define CSTATE_WRCOEFS	103	/* jpeg_write_coefficients done */
-#define DSTATE_START	200	/* after create_decompress */
-#define DSTATE_INHEADER	201	/* reading header markers, no SOS yet */
-#define DSTATE_READY	202	/* found SOS, ready for start_decompress */
-#define DSTATE_PRELOAD	203	/* reading multiscan file in start_decompress*/
-#define DSTATE_PRESCAN	204	/* performing dummy pass for 2-pass quant */
-#define DSTATE_SCANNING	205	/* start_decompress done, read_scanlines OK */
-#define DSTATE_RAW_OK	206	/* start_decompress done, read_raw_data OK */
-#define DSTATE_BUFIMAGE	207	/* expecting jpeg_start_output */
-#define DSTATE_BUFPOST	208	/* looking for SOS/EOI in jpeg_finish_output */
-#define DSTATE_RDCOEFS	209	/* reading file in jpeg_read_coefficients */
-#define DSTATE_STOPPING	210	/* looking for EOI in jpeg_finish_decompress */
-
-
-/* Declarations for compression modules */
-
-/* Master control module */
-struct jpeg_comp_master {
-  JMETHOD(void, prepare_for_pass, (j_compress_ptr cinfo));
-  JMETHOD(void, pass_startup, (j_compress_ptr cinfo));
-  JMETHOD(void, finish_pass, (j_compress_ptr cinfo));
-
-  /* State variables made visible to other modules */
-  boolean call_pass_startup;	/* True if pass_startup must be called */
-  boolean is_last_pass;		/* True during last pass */
-};
-
-/* Main buffer control (downsampled-data buffer) */
-struct jpeg_c_main_controller {
-  JMETHOD(void, start_pass, (j_compress_ptr cinfo, J_BUF_MODE pass_mode));
-  JMETHOD(void, process_data, (j_compress_ptr cinfo,
-			       JSAMPARRAY input_buf, JDIMENSION *in_row_ctr,
-			       JDIMENSION in_rows_avail));
-};
-
-/* Compression preprocessing (downsampling input buffer control) */
-struct jpeg_c_prep_controller {
-  JMETHOD(void, start_pass, (j_compress_ptr cinfo, J_BUF_MODE pass_mode));
-  JMETHOD(void, pre_process_data, (j_compress_ptr cinfo,
-				   JSAMPARRAY input_buf,
-				   JDIMENSION *in_row_ctr,
-				   JDIMENSION in_rows_avail,
-				   JSAMPIMAGE output_buf,
-				   JDIMENSION *out_row_group_ctr,
-				   JDIMENSION out_row_groups_avail));
-};
-
-/* Coefficient buffer control */
-struct jpeg_c_coef_controller {
-  JMETHOD(void, start_pass, (j_compress_ptr cinfo, J_BUF_MODE pass_mode));
-  JMETHOD(boolean, compress_data, (j_compress_ptr cinfo,
-				   JSAMPIMAGE input_buf));
-};
-
-/* Colorspace conversion */
-struct jpeg_color_converter {
-  JMETHOD(void, start_pass, (j_compress_ptr cinfo));
-  JMETHOD(void, color_convert, (j_compress_ptr cinfo,
-				JSAMPARRAY input_buf, JSAMPIMAGE output_buf,
-				JDIMENSION output_row, int num_rows));
-};
-
-/* Downsampling */
-struct jpeg_downsampler {
-  JMETHOD(void, start_pass, (j_compress_ptr cinfo));
-  JMETHOD(void, downsample, (j_compress_ptr cinfo,
-			     JSAMPIMAGE input_buf, JDIMENSION in_row_index,
-			     JSAMPIMAGE output_buf,
-			     JDIMENSION out_row_group_index));
-
-  boolean need_context_rows;	/* TRUE if need rows above & below */
-};
-
-/* Forward DCT (also controls coefficient quantization) */
-struct jpeg_forward_dct {
-  JMETHOD(void, start_pass, (j_compress_ptr cinfo));
-  /* perhaps this should be an array??? */
-  JMETHOD(void, forward_DCT, (j_compress_ptr cinfo,
-			      jpeg_component_info * compptr,
-			      JSAMPARRAY sample_data, JBLOCKROW coef_blocks,
-			      JDIMENSION start_row, JDIMENSION start_col,
-			      JDIMENSION num_blocks));
-};
-
-/* Entropy encoding */
-struct jpeg_entropy_encoder {
-  JMETHOD(void, start_pass, (j_compress_ptr cinfo, boolean gather_statistics));
-  JMETHOD(boolean, encode_mcu, (j_compress_ptr cinfo, JBLOCKROW *MCU_data));
-  JMETHOD(void, finish_pass, (j_compress_ptr cinfo));
-};
-
-/* Marker writing */
-struct jpeg_marker_writer {
-  JMETHOD(void, write_file_header, (j_compress_ptr cinfo));
-  JMETHOD(void, write_frame_header, (j_compress_ptr cinfo));
-  JMETHOD(void, write_scan_header, (j_compress_ptr cinfo));
-  JMETHOD(void, write_file_trailer, (j_compress_ptr cinfo));
-  JMETHOD(void, write_tables_only, (j_compress_ptr cinfo));
-  /* These routines are exported to allow insertion of extra markers */
-  /* Probably only COM and APPn markers should be written this way */
-  JMETHOD(void, write_marker_header, (j_compress_ptr cinfo, int marker,
-				      unsigned int datalen));
-  JMETHOD(void, write_marker_byte, (j_compress_ptr cinfo, int val));
-};
-
-
-/* Declarations for decompression modules */
-
-/* Master control module */
-struct jpeg_decomp_master {
-  JMETHOD(void, prepare_for_output_pass, (j_decompress_ptr cinfo));
-  JMETHOD(void, finish_output_pass, (j_decompress_ptr cinfo));
-
-  /* State variables made visible to other modules */
-  boolean is_dummy_pass;	/* True during 1st pass for 2-pass quant */
-};
-
-/* Input control module */
-struct jpeg_input_controller {
-  JMETHOD(int, consume_input, (j_decompress_ptr cinfo));
-  JMETHOD(int, consume_input_build_huffman_index, (j_decompress_ptr cinfo,
-                    huffman_index *index, int scan_count));
-  JMETHOD(int, consume_markers, (j_decompress_ptr cinfo,
-                    huffman_index *index, int scan_count));
-  JMETHOD(void, reset_input_controller, (j_decompress_ptr cinfo));
-  JMETHOD(void, start_input_pass, (j_decompress_ptr cinfo));
-  JMETHOD(void, finish_input_pass, (j_decompress_ptr cinfo));
-
-  /* State variables made visible to other modules */
-  boolean has_multiple_scans;	/* True if file has multiple scans */
-  boolean eoi_reached;		/* True when EOI has been consumed */
-};
-
-/* Main buffer control (downsampled-data buffer) */
-struct jpeg_d_main_controller {
-  JMETHOD(void, start_pass, (j_decompress_ptr cinfo, J_BUF_MODE pass_mode));
-  JMETHOD(void, process_data, (j_decompress_ptr cinfo,
-			       JSAMPARRAY output_buf, JDIMENSION *out_row_ctr,
-			       JDIMENSION out_rows_avail));
-};
-
-/* Coefficient buffer control */
-struct jpeg_d_coef_controller {
-  JMETHOD(void, start_input_pass, (j_decompress_ptr cinfo));
-  JMETHOD(int, consume_data, (j_decompress_ptr cinfo));
-  JMETHOD(int, consume_data_build_huffman_index, (j_decompress_ptr cinfo,
-                    huffman_index* index, int scan_count));
-  JMETHOD(void, start_output_pass, (j_decompress_ptr cinfo));
-  JMETHOD(int, decompress_data, (j_decompress_ptr cinfo,
-				 JSAMPIMAGE output_buf));
-  /* Pointer to array of coefficient virtual arrays, or NULL if none */
-  jvirt_barray_ptr *coef_arrays;
-
-  /* column number of the first and last tile, respectively */
-  int column_left_boundary;
-  int column_right_boundary;
-
-  /* column number of the first and last MCU, respectively */
-  int MCU_column_left_boundary;
-  int MCU_column_right_boundary;
-
-  /* the number of MCU columns to skip from the indexed MCU, iM,
-   * to the requested MCU boundary, rM, where iM is the MCU that we sample
-   * into our index and is the nearest one to the left of rM.
-   */
-  int MCU_columns_to_skip;
-};
-
-/* Decompression postprocessing (color quantization buffer control) */
-struct jpeg_d_post_controller {
-  JMETHOD(void, start_pass, (j_decompress_ptr cinfo, J_BUF_MODE pass_mode));
-  JMETHOD(void, post_process_data, (j_decompress_ptr cinfo,
-				    JSAMPIMAGE input_buf,
-				    JDIMENSION *in_row_group_ctr,
-				    JDIMENSION in_row_groups_avail,
-				    JSAMPARRAY output_buf,
-				    JDIMENSION *out_row_ctr,
-				    JDIMENSION out_rows_avail));
-};
-
-/* Marker reading & parsing */
-struct jpeg_marker_reader {
-  JMETHOD(void, reset_marker_reader, (j_decompress_ptr cinfo));
-  /* Read markers until SOS or EOI.
-   * Returns same codes as are defined for jpeg_consume_input:
-   * JPEG_SUSPENDED, JPEG_REACHED_SOS, or JPEG_REACHED_EOI.
-   */
-  JMETHOD(int, read_markers, (j_decompress_ptr cinfo));
-  JMETHOD(void, get_sos_marker_position, (j_decompress_ptr cinfo,
-                    huffman_index *index));
-  /* Read a restart marker --- exported for use by entropy decoder only */
-  jpeg_marker_parser_method read_restart_marker;
-
-  /* State of marker reader --- nominally internal, but applications
-   * supplying COM or APPn handlers might like to know the state.
-   */
-  boolean saw_SOI;		/* found SOI? */
-  boolean saw_SOF;		/* found SOF? */
-  int next_restart_num;		/* next restart number expected (0-7) */
-  int current_sos_marker_position;
-  unsigned int discarded_bytes;	/* # of bytes skipped looking for a marker */
-};
-
-/* Entropy decoding */
-struct jpeg_entropy_decoder {
-  JMETHOD(void, start_pass, (j_decompress_ptr cinfo));
-  JMETHOD(boolean, decode_mcu, (j_decompress_ptr cinfo,
-				JBLOCKROW *MCU_data));
-  JMETHOD(boolean, decode_mcu_discard_coef, (j_decompress_ptr cinfo));
-  JMETHOD(void, configure_huffman_decoder, (j_decompress_ptr cinfo,
-                    huffman_offset_data offset));
-  JMETHOD(void, get_huffman_decoder_configuration, (j_decompress_ptr cinfo,
-                    huffman_offset_data *offset));
-
-  /* This is here to share code between baseline and progressive decoders; */
-  /* other modules probably should not use it */
-  boolean insufficient_data;	/* set TRUE after emitting warning */
-
-  huffman_index *index;
-};
-
-/* Inverse DCT (also performs dequantization) */
-typedef JMETHOD(void, inverse_DCT_method_ptr,
-		(j_decompress_ptr cinfo, jpeg_component_info * compptr,
-		 JCOEFPTR coef_block,
-		 JSAMPARRAY output_buf, JDIMENSION output_col));
-
-struct jpeg_inverse_dct {
-  JMETHOD(void, start_pass, (j_decompress_ptr cinfo));
-  /* It is useful to allow each component to have a separate IDCT method. */
-  inverse_DCT_method_ptr inverse_DCT[MAX_COMPONENTS];
-};
-
-/* Upsampling (note that upsampler must also call color converter) */
-struct jpeg_upsampler {
-  JMETHOD(void, start_pass, (j_decompress_ptr cinfo));
-  JMETHOD(void, upsample, (j_decompress_ptr cinfo,
-			   JSAMPIMAGE input_buf,
-			   JDIMENSION *in_row_group_ctr,
-			   JDIMENSION in_row_groups_avail,
-			   JSAMPARRAY output_buf,
-			   JDIMENSION *out_row_ctr,
-			   JDIMENSION out_rows_avail));
-
-  boolean need_context_rows;	/* TRUE if need rows above & below */
-};
-
-/* Colorspace conversion */
-struct jpeg_color_deconverter {
-  JMETHOD(void, start_pass, (j_decompress_ptr cinfo));
-  JMETHOD(void, color_convert, (j_decompress_ptr cinfo,
-				JSAMPIMAGE input_buf, JDIMENSION input_row,
-				JSAMPARRAY output_buf, int num_rows));
-};
-
-/* Color quantization or color precision reduction */
-struct jpeg_color_quantizer {
-  JMETHOD(void, start_pass, (j_decompress_ptr cinfo, boolean is_pre_scan));
-  JMETHOD(void, color_quantize, (j_decompress_ptr cinfo,
-				 JSAMPARRAY input_buf, JSAMPARRAY output_buf,
-				 int num_rows));
-  JMETHOD(void, finish_pass, (j_decompress_ptr cinfo));
-  JMETHOD(void, new_color_map, (j_decompress_ptr cinfo));
-};
-
-
-/* Miscellaneous useful macros */
-
-#undef MAX
-#define MAX(a,b)	((a) > (b) ? (a) : (b))
-#undef MIN
-#define MIN(a,b)	((a) < (b) ? (a) : (b))
-
-
-/* We assume that right shift corresponds to signed division by 2 with
- * rounding towards minus infinity.  This is correct for typical "arithmetic
- * shift" instructions that shift in copies of the sign bit.  But some
- * C compilers implement >> with an unsigned shift.  For these machines you
- * must define RIGHT_SHIFT_IS_UNSIGNED.
- * RIGHT_SHIFT provides a proper signed right shift of an INT32 quantity.
- * It is only applied with constant shift counts.  SHIFT_TEMPS must be
- * included in the variables of any routine using RIGHT_SHIFT.
- */
-
-#ifdef RIGHT_SHIFT_IS_UNSIGNED
-#define SHIFT_TEMPS	INT32 shift_temp;
-#define RIGHT_SHIFT(x,shft)  \
-	((shift_temp = (x)) < 0 ? \
-	 (shift_temp >> (shft)) | ((~((INT32) 0)) << (32-(shft))) : \
-	 (shift_temp >> (shft)))
-#else
-#define SHIFT_TEMPS
-#define RIGHT_SHIFT(x,shft)	((x) >> (shft))
-#endif
-
-
-/* Short forms of external names for systems with brain-damaged linkers. */
-
-#ifdef NEED_SHORT_EXTERNAL_NAMES
-#define jinit_compress_master	jICompress
-#define jinit_c_master_control	jICMaster
-#define jinit_c_main_controller	jICMainC
-#define jinit_c_prep_controller	jICPrepC
-#define jinit_c_coef_controller	jICCoefC
-#define jinit_color_converter	jICColor
-#define jinit_downsampler	jIDownsampler
-#define jinit_forward_dct	jIFDCT
-#define jinit_huff_encoder	jIHEncoder
-#define jinit_phuff_encoder	jIPHEncoder
-#define jinit_marker_writer	jIMWriter
-#define jinit_master_decompress	jIDMaster
-#define jinit_d_main_controller	jIDMainC
-#define jinit_d_coef_controller	jIDCoefC
-#define jinit_d_post_controller	jIDPostC
-#define jinit_input_controller	jIInCtlr
-#define jinit_marker_reader	jIMReader
-#define jinit_huff_decoder	jIHDecoder
-#define jinit_phuff_decoder	jIPHDecoder
-#define jinit_inverse_dct	jIIDCT
-#define jinit_upsampler		jIUpsampler
-#define jinit_color_deconverter	jIDColor
-#define jinit_1pass_quantizer	jI1Quant
-#define jinit_2pass_quantizer	jI2Quant
-#define jinit_merged_upsampler	jIMUpsampler
-#define jinit_memory_mgr	jIMemMgr
-#define jdiv_round_up		jDivRound
-#define jround_up		jRound
-#define jcopy_sample_rows	jCopySamples
-#define jcopy_block_row		jCopyBlocks
-#define jzero_far		jZeroFar
-#define jpeg_zigzag_order	jZIGTable
-#define jpeg_natural_order	jZAGTable
-#endif /* NEED_SHORT_EXTERNAL_NAMES */
-
-
-/* Compression module initialization routines */
-EXTERN(void) jinit_compress_master JPP((j_compress_ptr cinfo));
-EXTERN(void) jinit_c_master_control JPP((j_compress_ptr cinfo,
-					 boolean transcode_only));
-EXTERN(void) jinit_c_main_controller JPP((j_compress_ptr cinfo,
-					  boolean need_full_buffer));
-EXTERN(void) jinit_c_prep_controller JPP((j_compress_ptr cinfo,
-					  boolean need_full_buffer));
-EXTERN(void) jinit_c_coef_controller JPP((j_compress_ptr cinfo,
-					  boolean need_full_buffer));
-EXTERN(void) jinit_color_converter JPP((j_compress_ptr cinfo));
-EXTERN(void) jinit_downsampler JPP((j_compress_ptr cinfo));
-EXTERN(void) jinit_forward_dct JPP((j_compress_ptr cinfo));
-EXTERN(void) jinit_huff_encoder JPP((j_compress_ptr cinfo));
-EXTERN(void) jinit_phuff_encoder JPP((j_compress_ptr cinfo));
-EXTERN(void) jinit_marker_writer JPP((j_compress_ptr cinfo));
-/* Decompression module initialization routines */
-EXTERN(void) jinit_master_decompress JPP((j_decompress_ptr cinfo));
-EXTERN(void) jinit_d_main_controller JPP((j_decompress_ptr cinfo,
-					  boolean need_full_buffer));
-EXTERN(void) jinit_d_coef_controller JPP((j_decompress_ptr cinfo,
-					  boolean need_full_buffer));
-EXTERN(void) jinit_d_post_controller JPP((j_decompress_ptr cinfo,
-					  boolean need_full_buffer));
-EXTERN(void) jinit_input_controller JPP((j_decompress_ptr cinfo));
-EXTERN(void) jinit_marker_reader JPP((j_decompress_ptr cinfo));
-EXTERN(void) jinit_huff_decoder JPP((j_decompress_ptr cinfo));
-EXTERN(void) jinit_huff_decoder_no_data JPP((j_decompress_ptr cinfo));
-EXTERN(void) jinit_phuff_decoder JPP((j_decompress_ptr cinfo));
-EXTERN(void) jinit_inverse_dct JPP((j_decompress_ptr cinfo));
-EXTERN(void) jinit_upsampler JPP((j_decompress_ptr cinfo));
-EXTERN(void) jinit_color_deconverter JPP((j_decompress_ptr cinfo));
-EXTERN(void) jinit_1pass_quantizer JPP((j_decompress_ptr cinfo));
-EXTERN(void) jinit_2pass_quantizer JPP((j_decompress_ptr cinfo));
-EXTERN(void) jinit_merged_upsampler JPP((j_decompress_ptr cinfo));
-EXTERN(void) jpeg_decompress_per_scan_setup (j_decompress_ptr cinfo);
-/* Memory manager initialization */
-EXTERN(void) jinit_memory_mgr JPP((j_common_ptr cinfo));
-
-/* Utility routines in jutils.c */
-EXTERN(long) jdiv_round_up JPP((long a, long b));
-EXTERN(long) jround_up JPP((long a, long b));
-EXTERN(long) jmin JPP((long a, long b));
-EXTERN(void) jcopy_sample_rows JPP((JSAMPARRAY input_array, int source_row,
-				    JSAMPARRAY output_array, int dest_row,
-				    int num_rows, JDIMENSION num_cols));
-EXTERN(void) jcopy_block_row JPP((JBLOCKROW input_row, JBLOCKROW output_row,
-				  JDIMENSION num_blocks));
-EXTERN(void) jzero_far JPP((void FAR * target, size_t bytestozero));
-
-EXTERN(void) jset_input_stream_position JPP((j_decompress_ptr cinfo,
-                    int offset));
-EXTERN(void) jset_input_stream_position_bit JPP((j_decompress_ptr cinfo,
-                    int byte_offset, int bit_left, INT32 buf));
-
-EXTERN(int) jget_input_stream_position JPP((j_decompress_ptr cinfo));
-/* Constant tables in jutils.c */
-#if 0				/* This table is not actually needed in v6a */
-extern const int jpeg_zigzag_order[]; /* natural coef order to zigzag order */
-#endif
-extern const int jpeg_natural_order[]; /* zigzag coef order to natural order */
-
-/* Suppress undefined-structure complaints if necessary. */
-
-#ifdef INCOMPLETE_TYPES_BROKEN
-#ifndef AM_MEMORY_MANAGER	/* only jmemmgr.c defines these */
-struct jvirt_sarray_control { long dummy; };
-struct jvirt_barray_control { long dummy; };
-#endif
-#endif /* INCOMPLETE_TYPES_BROKEN */
@@ -1,856 +0,0 @@
-/*
- * jquant1.c
- *
- * Copyright (C) 1991-1996, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains 1-pass color quantization (color mapping) routines.
- * These routines provide mapping to a fixed color map using equally spaced
- * color values.  Optional Floyd-Steinberg or ordered dithering is available.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-
-#ifdef QUANT_1PASS_SUPPORTED
-
-
-/*
- * The main purpose of 1-pass quantization is to provide a fast, if not very
- * high quality, colormapped output capability.  A 2-pass quantizer usually
- * gives better visual quality; however, for quantized grayscale output this
- * quantizer is perfectly adequate.  Dithering is highly recommended with this
- * quantizer, though you can turn it off if you really want to.
- *
- * In 1-pass quantization the colormap must be chosen in advance of seeing the
- * image.  We use a map consisting of all combinations of Ncolors[i] color
- * values for the i'th component.  The Ncolors[] values are chosen so that
- * their product, the total number of colors, is no more than that requested.
- * (In most cases, the product will be somewhat less.)
- *
- * Since the colormap is orthogonal, the representative value for each color
- * component can be determined without considering the other components;
- * then these indexes can be combined into a colormap index by a standard
- * N-dimensional-array-subscript calculation.  Most of the arithmetic involved
- * can be precalculated and stored in the lookup table colorindex[].
- * colorindex[i][j] maps pixel value j in component i to the nearest
- * representative value (grid plane) for that component; this index is
- * multiplied by the array stride for component i, so that the
- * index of the colormap entry closest to a given pixel value is just
- *    sum( colorindex[component-number][pixel-component-value] )
- * Aside from being fast, this scheme allows for variable spacing between
- * representative values with no additional lookup cost.
- *
- * If gamma correction has been applied in color conversion, it might be wise
- * to adjust the color grid spacing so that the representative colors are
- * equidistant in linear space.  At this writing, gamma correction is not
- * implemented by jdcolor, so nothing is done here.
- */
-
-
-/* Declarations for ordered dithering.
- *
- * We use a standard 16x16 ordered dither array.  The basic concept of ordered
- * dithering is described in many references, for instance Dale Schumacher's
- * chapter II.2 of Graphics Gems II (James Arvo, ed. Academic Press, 1991).
- * In place of Schumacher's comparisons against a "threshold" value, we add a
- * "dither" value to the input pixel and then round the result to the nearest
- * output value.  The dither value is equivalent to (0.5 - threshold) times
- * the distance between output values.  For ordered dithering, we assume that
- * the output colors are equally spaced; if not, results will probably be
- * worse, since the dither may be too much or too little at a given point.
- *
- * The normal calculation would be to form pixel value + dither, range-limit
- * this to 0..MAXJSAMPLE, and then index into the colorindex table as usual.
- * We can skip the separate range-limiting step by extending the colorindex
- * table in both directions.
- */
-
-#define ODITHER_SIZE  16	/* dimension of dither matrix */
-/* NB: if ODITHER_SIZE is not a power of 2, ODITHER_MASK uses will break */
-#define ODITHER_CELLS (ODITHER_SIZE*ODITHER_SIZE)	/* # cells in matrix */
-#define ODITHER_MASK  (ODITHER_SIZE-1) /* mask for wrapping around counters */
-
-typedef int ODITHER_MATRIX[ODITHER_SIZE][ODITHER_SIZE];
-typedef int (*ODITHER_MATRIX_PTR)[ODITHER_SIZE];
-
-static const UINT8 base_dither_matrix[ODITHER_SIZE][ODITHER_SIZE] = {
-  /* Bayer's order-4 dither array.  Generated by the code given in
-   * Stephen Hawley's article "Ordered Dithering" in Graphics Gems I.
-   * The values in this array must range from 0 to ODITHER_CELLS-1.
-   */
-  {   0,192, 48,240, 12,204, 60,252,  3,195, 51,243, 15,207, 63,255 },
-  { 128, 64,176,112,140, 76,188,124,131, 67,179,115,143, 79,191,127 },
-  {  32,224, 16,208, 44,236, 28,220, 35,227, 19,211, 47,239, 31,223 },
-  { 160, 96,144, 80,172,108,156, 92,163, 99,147, 83,175,111,159, 95 },
-  {   8,200, 56,248,  4,196, 52,244, 11,203, 59,251,  7,199, 55,247 },
-  { 136, 72,184,120,132, 68,180,116,139, 75,187,123,135, 71,183,119 },
-  {  40,232, 24,216, 36,228, 20,212, 43,235, 27,219, 39,231, 23,215 },
-  { 168,104,152, 88,164,100,148, 84,171,107,155, 91,167,103,151, 87 },
-  {   2,194, 50,242, 14,206, 62,254,  1,193, 49,241, 13,205, 61,253 },
-  { 130, 66,178,114,142, 78,190,126,129, 65,177,113,141, 77,189,125 },
-  {  34,226, 18,210, 46,238, 30,222, 33,225, 17,209, 45,237, 29,221 },
-  { 162, 98,146, 82,174,110,158, 94,161, 97,145, 81,173,109,157, 93 },
-  {  10,202, 58,250,  6,198, 54,246,  9,201, 57,249,  5,197, 53,245 },
-  { 138, 74,186,122,134, 70,182,118,137, 73,185,121,133, 69,181,117 },
-  {  42,234, 26,218, 38,230, 22,214, 41,233, 25,217, 37,229, 21,213 },
-  { 170,106,154, 90,166,102,150, 86,169,105,153, 89,165,101,149, 85 }
-};
-
-
-/* Declarations for Floyd-Steinberg dithering.
- *
- * Errors are accumulated into the array fserrors[], at a resolution of
- * 1/16th of a pixel count.  The error at a given pixel is propagated
- * to its not-yet-processed neighbors using the standard F-S fractions,
- *		...	(here)	7/16
- *		3/16	5/16	1/16
- * We work left-to-right on even rows, right-to-left on odd rows.
- *
- * We can get away with a single array (holding one row's worth of errors)
- * by using it to store the current row's errors at pixel columns not yet
- * processed, but the next row's errors at columns already processed.  We
- * need only a few extra variables to hold the errors immediately around the
- * current column.  (If we are lucky, those variables are in registers, but
- * even if not, they're probably cheaper to access than array elements are.)
- *
- * The fserrors[] array is indexed [component#][position].
- * We provide (#columns + 2) entries per component; the extra entry at each
- * end saves us from special-casing the first and last pixels.
- *
- * Note: on a wide image, we might not have enough room in a PC's near data
- * segment to hold the error array; so it is allocated with alloc_large.
- */
-
-#if BITS_IN_JSAMPLE == 8
-typedef INT16 FSERROR;		/* 16 bits should be enough */
-typedef int LOCFSERROR;		/* use 'int' for calculation temps */
-#else
-typedef INT32 FSERROR;		/* may need more than 16 bits */
-typedef INT32 LOCFSERROR;	/* be sure calculation temps are big enough */
-#endif
-
-typedef FSERROR FAR *FSERRPTR;	/* pointer to error array (in FAR storage!) */
-
-
-/* Private subobject */
-
-#define MAX_Q_COMPS 4		/* max components I can handle */
-
-typedef struct {
-  struct jpeg_color_quantizer pub; /* public fields */
-
-  /* Initially allocated colormap is saved here */
-  JSAMPARRAY sv_colormap;	/* The color map as a 2-D pixel array */
-  int sv_actual;		/* number of entries in use */
-
-  JSAMPARRAY colorindex;	/* Precomputed mapping for speed */
-  /* colorindex[i][j] = index of color closest to pixel value j in component i,
-   * premultiplied as described above.  Since colormap indexes must fit into
-   * JSAMPLEs, the entries of this array will too.
-   */
-  boolean is_padded;		/* is the colorindex padded for odither? */
-
-  int Ncolors[MAX_Q_COMPS];	/* # of values alloced to each component */
-
-  /* Variables for ordered dithering */
-  int row_index;		/* cur row's vertical index in dither matrix */
-  ODITHER_MATRIX_PTR odither[MAX_Q_COMPS]; /* one dither array per component */
-
-  /* Variables for Floyd-Steinberg dithering */
-  FSERRPTR fserrors[MAX_Q_COMPS]; /* accumulated errors */
-  boolean on_odd_row;		/* flag to remember which row we are on */
-} my_cquantizer;
-
-typedef my_cquantizer * my_cquantize_ptr;
-
-
-/*
- * Policy-making subroutines for create_colormap and create_colorindex.
- * These routines determine the colormap to be used.  The rest of the module
- * only assumes that the colormap is orthogonal.
- *
- *  * select_ncolors decides how to divvy up the available colors
- *    among the components.
- *  * output_value defines the set of representative values for a component.
- *  * largest_input_value defines the mapping from input values to
- *    representative values for a component.
- * Note that the latter two routines may impose different policies for
- * different components, though this is not currently done.
- */
-
-
-LOCAL(int)
-select_ncolors (j_decompress_ptr cinfo, int Ncolors[])
-/* Determine allocation of desired colors to components, */
-/* and fill in Ncolors[] array to indicate choice. */
-/* Return value is total number of colors (product of Ncolors[] values). */
-{
-  int nc = cinfo->out_color_components; /* number of color components */
-  int max_colors = cinfo->desired_number_of_colors;
-  int total_colors, iroot, i, j;
-  boolean changed;
-  long temp;
-  static const int RGB_order[3] = { RGB_GREEN, RGB_RED, RGB_BLUE };
-
-  /* We can allocate at least the nc'th root of max_colors per component. */
-  /* Compute floor(nc'th root of max_colors). */
-  iroot = 1;
-  do {
-    iroot++;
-    temp = iroot;		/* set temp = iroot ** nc */
-    for (i = 1; i < nc; i++)
-      temp *= iroot;
-  } while (temp <= (long) max_colors); /* repeat till iroot exceeds root */
-  iroot--;			/* now iroot = floor(root) */
-
-  /* Must have at least 2 color values per component */
-  if (iroot < 2)
-    ERREXIT1(cinfo, JERR_QUANT_FEW_COLORS, (int) temp);
-
-  /* Initialize to iroot color values for each component */
-  total_colors = 1;
-  for (i = 0; i < nc; i++) {
-    Ncolors[i] = iroot;
-    total_colors *= iroot;
-  }
-  /* We may be able to increment the count for one or more components without
-   * exceeding max_colors, though we know not all can be incremented.
-   * Sometimes, the first component can be incremented more than once!
-   * (Example: for 16 colors, we start at 2*2*2, go to 3*2*2, then 4*2*2.)
-   * In RGB colorspace, try to increment G first, then R, then B.
-   */
-  do {
-    changed = FALSE;
-    for (i = 0; i < nc; i++) {
-      j = (cinfo->out_color_space == JCS_RGB ? RGB_order[i] : i);
-      /* calculate new total_colors if Ncolors[j] is incremented */
-      temp = total_colors / Ncolors[j];
-      temp *= Ncolors[j]+1;	/* done in long arith to avoid oflo */
-      if (temp > (long) max_colors)
-	break;			/* won't fit, done with this pass */
-      Ncolors[j]++;		/* OK, apply the increment */
-      total_colors = (int) temp;
-      changed = TRUE;
-    }
-  } while (changed);
-
-  return total_colors;
-}
-
-
-LOCAL(int)
-output_value (j_decompress_ptr cinfo, int ci, int j, int maxj)
-/* Return j'th output value, where j will range from 0 to maxj */
-/* The output values must fall in 0..MAXJSAMPLE in increasing order */
-{
-  /* We always provide values 0 and MAXJSAMPLE for each component;
-   * any additional values are equally spaced between these limits.
-   * (Forcing the upper and lower values to the limits ensures that
-   * dithering can't produce a color outside the selected gamut.)
-   */
-  return (int) (((INT32) j * MAXJSAMPLE + maxj/2) / maxj);
-}
-
-
-LOCAL(int)
-largest_input_value (j_decompress_ptr cinfo, int ci, int j, int maxj)
-/* Return largest input value that should map to j'th output value */
-/* Must have largest(j=0) >= 0, and largest(j=maxj) >= MAXJSAMPLE */
-{
-  /* Breakpoints are halfway between values returned by output_value */
-  return (int) (((INT32) (2*j + 1) * MAXJSAMPLE + maxj) / (2*maxj));
-}
-
-
-/*
- * Create the colormap.
- */
-
-LOCAL(void)
-create_colormap (j_decompress_ptr cinfo)
-{
-  my_cquantize_ptr cquantize = (my_cquantize_ptr) cinfo->cquantize;
-  JSAMPARRAY colormap;		/* Created colormap */
-  int total_colors;		/* Number of distinct output colors */
-  int i,j,k, nci, blksize, blkdist, ptr, val;
-
-  /* Select number of colors for each component */
-  total_colors = select_ncolors(cinfo, cquantize->Ncolors);
-
-  /* Report selected color counts */
-  if (cinfo->out_color_components == 3)
-    TRACEMS4(cinfo, 1, JTRC_QUANT_3_NCOLORS,
-	     total_colors, cquantize->Ncolors[0],
-	     cquantize->Ncolors[1], cquantize->Ncolors[2]);
-  else
-    TRACEMS1(cinfo, 1, JTRC_QUANT_NCOLORS, total_colors);
-
-  /* Allocate and fill in the colormap. */
-  /* The colors are ordered in the map in standard row-major order, */
-  /* i.e. rightmost (highest-indexed) color changes most rapidly. */
-
-  colormap = (*cinfo->mem->alloc_sarray)
-    ((j_common_ptr) cinfo, JPOOL_IMAGE,
-     (JDIMENSION) total_colors, (JDIMENSION) cinfo->out_color_components);
-
-  /* blksize is number of adjacent repeated entries for a component */
-  /* blkdist is distance between groups of identical entries for a component */
-  blkdist = total_colors;
-
-  for (i = 0; i < cinfo->out_color_components; i++) {
-    /* fill in colormap entries for i'th color component */
-    nci = cquantize->Ncolors[i]; /* # of distinct values for this color */
-    blksize = blkdist / nci;
-    for (j = 0; j < nci; j++) {
-      /* Compute j'th output value (out of nci) for component */
-      val = output_value(cinfo, i, j, nci-1);
-      /* Fill in all colormap entries that have this value of this component */
-      for (ptr = j * blksize; ptr < total_colors; ptr += blkdist) {
-	/* fill in blksize entries beginning at ptr */
-	for (k = 0; k < blksize; k++)
-	  colormap[i][ptr+k] = (JSAMPLE) val;
-      }
-    }
-    blkdist = blksize;		/* blksize of this color is blkdist of next */
-  }
-
-  /* Save the colormap in private storage,
-   * where it will survive color quantization mode changes.
-   */
-  cquantize->sv_colormap = colormap;
-  cquantize->sv_actual = total_colors;
-}
-
-
-/*
- * Create the color index table.
- */
-
-LOCAL(void)
-create_colorindex (j_decompress_ptr cinfo)
-{
-  my_cquantize_ptr cquantize = (my_cquantize_ptr) cinfo->cquantize;
-  JSAMPROW indexptr;
-  int i,j,k, nci, blksize, val, pad;
-
-  /* For ordered dither, we pad the color index tables by MAXJSAMPLE in
-   * each direction (input index values can be -MAXJSAMPLE .. 2*MAXJSAMPLE).
-   * This is not necessary in the other dithering modes.  However, we
-   * flag whether it was done in case user changes dithering mode.
-   */
-  if (cinfo->dither_mode == JDITHER_ORDERED) {
-    pad = MAXJSAMPLE*2;
-    cquantize->is_padded = TRUE;
-  } else {
-    pad = 0;
-    cquantize->is_padded = FALSE;
-  }
-
-  cquantize->colorindex = (*cinfo->mem->alloc_sarray)
-    ((j_common_ptr) cinfo, JPOOL_IMAGE,
-     (JDIMENSION) (MAXJSAMPLE+1 + pad),
-     (JDIMENSION) cinfo->out_color_components);
-
-  /* blksize is number of adjacent repeated entries for a component */
-  blksize = cquantize->sv_actual;
-
-  for (i = 0; i < cinfo->out_color_components; i++) {
-    /* fill in colorindex entries for i'th color component */
-    nci = cquantize->Ncolors[i]; /* # of distinct values for this color */
-    blksize = blksize / nci;
-
-    /* adjust colorindex pointers to provide padding at negative indexes. */
-    if (pad)
-      cquantize->colorindex[i] += MAXJSAMPLE;
-
-    /* in loop, val = index of current output value, */
-    /* and k = largest j that maps to current val */
-    indexptr = cquantize->colorindex[i];
-    val = 0;
-    k = largest_input_value(cinfo, i, 0, nci-1);
-    for (j = 0; j <= MAXJSAMPLE; j++) {
-      while (j > k)		/* advance val if past boundary */
-	k = largest_input_value(cinfo, i, ++val, nci-1);
-      /* premultiply so that no multiplication needed in main processing */
-      indexptr[j] = (JSAMPLE) (val * blksize);
-    }
-    /* Pad at both ends if necessary */
-    if (pad)
-      for (j = 1; j <= MAXJSAMPLE; j++) {
-	indexptr[-j] = indexptr[0];
-	indexptr[MAXJSAMPLE+j] = indexptr[MAXJSAMPLE];
-      }
-  }
-}
-
-
-/*
- * Create an ordered-dither array for a component having ncolors
- * distinct output values.
- */
-
-LOCAL(ODITHER_MATRIX_PTR)
-make_odither_array (j_decompress_ptr cinfo, int ncolors)
-{
-  ODITHER_MATRIX_PTR odither;
-  int j,k;
-  INT32 num,den;
-
-  odither = (ODITHER_MATRIX_PTR)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				SIZEOF(ODITHER_MATRIX));
-  /* The inter-value distance for this color is MAXJSAMPLE/(ncolors-1).
-   * Hence the dither value for the matrix cell with fill order f
-   * (f=0..N-1) should be (N-1-2*f)/(2*N) * MAXJSAMPLE/(ncolors-1).
-   * On 16-bit-int machine, be careful to avoid overflow.
-   */
-  den = 2 * ODITHER_CELLS * ((INT32) (ncolors - 1));
-  for (j = 0; j < ODITHER_SIZE; j++) {
-    for (k = 0; k < ODITHER_SIZE; k++) {
-      num = ((INT32) (ODITHER_CELLS-1 - 2*((int)base_dither_matrix[j][k])))
-	    * MAXJSAMPLE;
-      /* Ensure round towards zero despite C's lack of consistency
-       * about rounding negative values in integer division...
-       */
-      odither[j][k] = (int) (num<0 ? -((-num)/den) : num/den);
-    }
-  }
-  return odither;
-}
-
-
-/*
- * Create the ordered-dither tables.
- * Components having the same number of representative colors may 
- * share a dither table.
- */
-
-LOCAL(void)
-create_odither_tables (j_decompress_ptr cinfo)
-{
-  my_cquantize_ptr cquantize = (my_cquantize_ptr) cinfo->cquantize;
-  ODITHER_MATRIX_PTR odither;
-  int i, j, nci;
-
-  for (i = 0; i < cinfo->out_color_components; i++) {
-    nci = cquantize->Ncolors[i]; /* # of distinct values for this color */
-    odither = NULL;		/* search for matching prior component */
-    for (j = 0; j < i; j++) {
-      if (nci == cquantize->Ncolors[j]) {
-	odither = cquantize->odither[j];
-	break;
-      }
-    }
-    if (odither == NULL)	/* need a new table? */
-      odither = make_odither_array(cinfo, nci);
-    cquantize->odither[i] = odither;
-  }
-}
-
-
-/*
- * Map some rows of pixels to the output colormapped representation.
- */
-
-METHODDEF(void)
-color_quantize (j_decompress_ptr cinfo, JSAMPARRAY input_buf,
-		JSAMPARRAY output_buf, int num_rows)
-/* General case, no dithering */
-{
-  my_cquantize_ptr cquantize = (my_cquantize_ptr) cinfo->cquantize;
-  JSAMPARRAY colorindex = cquantize->colorindex;
-  register int pixcode, ci;
-  register JSAMPROW ptrin, ptrout;
-  int row;
-  JDIMENSION col;
-  JDIMENSION width = cinfo->output_width;
-  register int nc = cinfo->out_color_components;
-
-  for (row = 0; row < num_rows; row++) {
-    ptrin = input_buf[row];
-    ptrout = output_buf[row];
-    for (col = width; col > 0; col--) {
-      pixcode = 0;
-      for (ci = 0; ci < nc; ci++) {
-	pixcode += GETJSAMPLE(colorindex[ci][GETJSAMPLE(*ptrin++)]);
-      }
-      *ptrout++ = (JSAMPLE) pixcode;
-    }
-  }
-}
-
-
-METHODDEF(void)
-color_quantize3 (j_decompress_ptr cinfo, JSAMPARRAY input_buf,
-		 JSAMPARRAY output_buf, int num_rows)
-/* Fast path for out_color_components==3, no dithering */
-{
-  my_cquantize_ptr cquantize = (my_cquantize_ptr) cinfo->cquantize;
-  register int pixcode;
-  register JSAMPROW ptrin, ptrout;
-  JSAMPROW colorindex0 = cquantize->colorindex[0];
-  JSAMPROW colorindex1 = cquantize->colorindex[1];
-  JSAMPROW colorindex2 = cquantize->colorindex[2];
-  int row;
-  JDIMENSION col;
-  JDIMENSION width = cinfo->output_width;
-
-  for (row = 0; row < num_rows; row++) {
-    ptrin = input_buf[row];
-    ptrout = output_buf[row];
-    for (col = width; col > 0; col--) {
-      pixcode  = GETJSAMPLE(colorindex0[GETJSAMPLE(*ptrin++)]);
-      pixcode += GETJSAMPLE(colorindex1[GETJSAMPLE(*ptrin++)]);
-      pixcode += GETJSAMPLE(colorindex2[GETJSAMPLE(*ptrin++)]);
-      *ptrout++ = (JSAMPLE) pixcode;
-    }
-  }
-}
-
-
-METHODDEF(void)
-quantize_ord_dither (j_decompress_ptr cinfo, JSAMPARRAY input_buf,
-		     JSAMPARRAY output_buf, int num_rows)
-/* General case, with ordered dithering */
-{
-  my_cquantize_ptr cquantize = (my_cquantize_ptr) cinfo->cquantize;
-  register JSAMPROW input_ptr;
-  register JSAMPROW output_ptr;
-  JSAMPROW colorindex_ci;
-  int * dither;			/* points to active row of dither matrix */
-  int row_index, col_index;	/* current indexes into dither matrix */
-  int nc = cinfo->out_color_components;
-  int ci;
-  int row;
-  JDIMENSION col;
-  JDIMENSION width = cinfo->output_width;
-
-  for (row = 0; row < num_rows; row++) {
-    /* Initialize output values to 0 so can process components separately */
-    jzero_far((void FAR *) output_buf[row],
-	      (size_t) (width * SIZEOF(JSAMPLE)));
-    row_index = cquantize->row_index;
-    for (ci = 0; ci < nc; ci++) {
-      input_ptr = input_buf[row] + ci;
-      output_ptr = output_buf[row];
-      colorindex_ci = cquantize->colorindex[ci];
-      dither = cquantize->odither[ci][row_index];
-      col_index = 0;
-
-      for (col = width; col > 0; col--) {
-	/* Form pixel value + dither, range-limit to 0..MAXJSAMPLE,
-	 * select output value, accumulate into output code for this pixel.
-	 * Range-limiting need not be done explicitly, as we have extended
-	 * the colorindex table to produce the right answers for out-of-range
-	 * inputs.  The maximum dither is +- MAXJSAMPLE; this sets the
-	 * required amount of padding.
-	 */
-	*output_ptr += colorindex_ci[GETJSAMPLE(*input_ptr)+dither[col_index]];
-	input_ptr += nc;
-	output_ptr++;
-	col_index = (col_index + 1) & ODITHER_MASK;
-      }
-    }
-    /* Advance row index for next row */
-    row_index = (row_index + 1) & ODITHER_MASK;
-    cquantize->row_index = row_index;
-  }
-}
-
-
-METHODDEF(void)
-quantize3_ord_dither (j_decompress_ptr cinfo, JSAMPARRAY input_buf,
-		      JSAMPARRAY output_buf, int num_rows)
-/* Fast path for out_color_components==3, with ordered dithering */
-{
-  my_cquantize_ptr cquantize = (my_cquantize_ptr) cinfo->cquantize;
-  register int pixcode;
-  register JSAMPROW input_ptr;
-  register JSAMPROW output_ptr;
-  JSAMPROW colorindex0 = cquantize->colorindex[0];
-  JSAMPROW colorindex1 = cquantize->colorindex[1];
-  JSAMPROW colorindex2 = cquantize->colorindex[2];
-  int * dither0;		/* points to active row of dither matrix */
-  int * dither1;
-  int * dither2;
-  int row_index, col_index;	/* current indexes into dither matrix */
-  int row;
-  JDIMENSION col;
-  JDIMENSION width = cinfo->output_width;
-
-  for (row = 0; row < num_rows; row++) {
-    row_index = cquantize->row_index;
-    input_ptr = input_buf[row];
-    output_ptr = output_buf[row];
-    dither0 = cquantize->odither[0][row_index];
-    dither1 = cquantize->odither[1][row_index];
-    dither2 = cquantize->odither[2][row_index];
-    col_index = 0;
-
-    for (col = width; col > 0; col--) {
-      pixcode  = GETJSAMPLE(colorindex0[GETJSAMPLE(*input_ptr++) +
-					dither0[col_index]]);
-      pixcode += GETJSAMPLE(colorindex1[GETJSAMPLE(*input_ptr++) +
-					dither1[col_index]]);
-      pixcode += GETJSAMPLE(colorindex2[GETJSAMPLE(*input_ptr++) +
-					dither2[col_index]]);
-      *output_ptr++ = (JSAMPLE) pixcode;
-      col_index = (col_index + 1) & ODITHER_MASK;
-    }
-    row_index = (row_index + 1) & ODITHER_MASK;
-    cquantize->row_index = row_index;
-  }
-}
-
-
-METHODDEF(void)
-quantize_fs_dither (j_decompress_ptr cinfo, JSAMPARRAY input_buf,
-		    JSAMPARRAY output_buf, int num_rows)
-/* General case, with Floyd-Steinberg dithering */
-{
-  my_cquantize_ptr cquantize = (my_cquantize_ptr) cinfo->cquantize;
-  register LOCFSERROR cur;	/* current error or pixel value */
-  LOCFSERROR belowerr;		/* error for pixel below cur */
-  LOCFSERROR bpreverr;		/* error for below/prev col */
-  LOCFSERROR bnexterr;		/* error for below/next col */
-  LOCFSERROR delta;
-  register FSERRPTR errorptr;	/* => fserrors[] at column before current */
-  register JSAMPROW input_ptr;
-  register JSAMPROW output_ptr;
-  JSAMPROW colorindex_ci;
-  JSAMPROW colormap_ci;
-  int pixcode;
-  int nc = cinfo->out_color_components;
-  int dir;			/* 1 for left-to-right, -1 for right-to-left */
-  int dirnc;			/* dir * nc */
-  int ci;
-  int row;
-  JDIMENSION col;
-  JDIMENSION width = cinfo->output_width;
-  JSAMPLE *range_limit = cinfo->sample_range_limit;
-  SHIFT_TEMPS
-
-  for (row = 0; row < num_rows; row++) {
-    /* Initialize output values to 0 so can process components separately */
-    jzero_far((void FAR *) output_buf[row],
-	      (size_t) (width * SIZEOF(JSAMPLE)));
-    for (ci = 0; ci < nc; ci++) {
-      input_ptr = input_buf[row] + ci;
-      output_ptr = output_buf[row];
-      if (cquantize->on_odd_row) {
-	/* work right to left in this row */
-	input_ptr += (width-1) * nc; /* so point to rightmost pixel */
-	output_ptr += width-1;
-	dir = -1;
-	dirnc = -nc;
-	errorptr = cquantize->fserrors[ci] + (width+1); /* => entry after last column */
-      } else {
-	/* work left to right in this row */
-	dir = 1;
-	dirnc = nc;
-	errorptr = cquantize->fserrors[ci]; /* => entry before first column */
-      }
-      colorindex_ci = cquantize->colorindex[ci];
-      colormap_ci = cquantize->sv_colormap[ci];
-      /* Preset error values: no error propagated to first pixel from left */
-      cur = 0;
-      /* and no error propagated to row below yet */
-      belowerr = bpreverr = 0;
-
-      for (col = width; col > 0; col--) {
-	/* cur holds the error propagated from the previous pixel on the
-	 * current line.  Add the error propagated from the previous line
-	 * to form the complete error correction term for this pixel, and
-	 * round the error term (which is expressed * 16) to an integer.
-	 * RIGHT_SHIFT rounds towards minus infinity, so adding 8 is correct
-	 * for either sign of the error value.
-	 * Note: errorptr points to *previous* column's array entry.
-	 */
-	cur = RIGHT_SHIFT(cur + errorptr[dir] + 8, 4);
-	/* Form pixel value + error, and range-limit to 0..MAXJSAMPLE.
-	 * The maximum error is +- MAXJSAMPLE; this sets the required size
-	 * of the range_limit array.
-	 */
-	cur += GETJSAMPLE(*input_ptr);
-	cur = GETJSAMPLE(range_limit[cur]);
-	/* Select output value, accumulate into output code for this pixel */
-	pixcode = GETJSAMPLE(colorindex_ci[cur]);
-	*output_ptr += (JSAMPLE) pixcode;
-	/* Compute actual representation error at this pixel */
-	/* Note: we can do this even though we don't have the final */
-	/* pixel code, because the colormap is orthogonal. */
-	cur -= GETJSAMPLE(colormap_ci[pixcode]);
-	/* Compute error fractions to be propagated to adjacent pixels.
-	 * Add these into the running sums, and simultaneously shift the
-	 * next-line error sums left by 1 column.
-	 */
-	bnexterr = cur;
-	delta = cur * 2;
-	cur += delta;		/* form error * 3 */
-	errorptr[0] = (FSERROR) (bpreverr + cur);
-	cur += delta;		/* form error * 5 */
-	bpreverr = belowerr + cur;
-	belowerr = bnexterr;
-	cur += delta;		/* form error * 7 */
-	/* At this point cur contains the 7/16 error value to be propagated
-	 * to the next pixel on the current line, and all the errors for the
-	 * next line have been shifted over. We are therefore ready to move on.
-	 */
-	input_ptr += dirnc;	/* advance input ptr to next column */
-	output_ptr += dir;	/* advance output ptr to next column */
-	errorptr += dir;	/* advance errorptr to current column */
-      }
-      /* Post-loop cleanup: we must unload the final error value into the
-       * final fserrors[] entry.  Note we need not unload belowerr because
-       * it is for the dummy column before or after the actual array.
-       */
-      errorptr[0] = (FSERROR) bpreverr; /* unload prev err into array */
-    }
-    cquantize->on_odd_row = (cquantize->on_odd_row ? FALSE : TRUE);
-  }
-}
-
-
-/*
- * Allocate workspace for Floyd-Steinberg errors.
- */
-
-LOCAL(void)
-alloc_fs_workspace (j_decompress_ptr cinfo)
-{
-  my_cquantize_ptr cquantize = (my_cquantize_ptr) cinfo->cquantize;
-  size_t arraysize;
-  int i;
-
-  arraysize = (size_t) ((cinfo->output_width + 2) * SIZEOF(FSERROR));
-  for (i = 0; i < cinfo->out_color_components; i++) {
-    cquantize->fserrors[i] = (FSERRPTR)
-      (*cinfo->mem->alloc_large)((j_common_ptr) cinfo, JPOOL_IMAGE, arraysize);
-  }
-}
-
-
-/*
- * Initialize for one-pass color quantization.
- */
-
-METHODDEF(void)
-start_pass_1_quant (j_decompress_ptr cinfo, boolean is_pre_scan)
-{
-  my_cquantize_ptr cquantize = (my_cquantize_ptr) cinfo->cquantize;
-  size_t arraysize;
-  int i;
-
-  /* Install my colormap. */
-  cinfo->colormap = cquantize->sv_colormap;
-  cinfo->actual_number_of_colors = cquantize->sv_actual;
-
-  /* Initialize for desired dithering mode. */
-  switch (cinfo->dither_mode) {
-  case JDITHER_NONE:
-    if (cinfo->out_color_components == 3)
-      cquantize->pub.color_quantize = color_quantize3;
-    else
-      cquantize->pub.color_quantize = color_quantize;
-    break;
-  case JDITHER_ORDERED:
-    if (cinfo->out_color_components == 3)
-      cquantize->pub.color_quantize = quantize3_ord_dither;
-    else
-      cquantize->pub.color_quantize = quantize_ord_dither;
-    cquantize->row_index = 0;	/* initialize state for ordered dither */
-    /* If user changed to ordered dither from another mode,
-     * we must recreate the color index table with padding.
-     * This will cost extra space, but probably isn't very likely.
-     */
-    if (! cquantize->is_padded)
-      create_colorindex(cinfo);
-    /* Create ordered-dither tables if we didn't already. */
-    if (cquantize->odither[0] == NULL)
-      create_odither_tables(cinfo);
-    break;
-  case JDITHER_FS:
-    cquantize->pub.color_quantize = quantize_fs_dither;
-    cquantize->on_odd_row = FALSE; /* initialize state for F-S dither */
-    /* Allocate Floyd-Steinberg workspace if didn't already. */
-    if (cquantize->fserrors[0] == NULL)
-      alloc_fs_workspace(cinfo);
-    /* Initialize the propagated errors to zero. */
-    arraysize = (size_t) ((cinfo->output_width + 2) * SIZEOF(FSERROR));
-    for (i = 0; i < cinfo->out_color_components; i++)
-      jzero_far((void FAR *) cquantize->fserrors[i], arraysize);
-    break;
-  default:
-    ERREXIT(cinfo, JERR_NOT_COMPILED);
-    break;
-  }
-}
-
-
-/*
- * Finish up at the end of the pass.
- */
-
-METHODDEF(void)
-finish_pass_1_quant (j_decompress_ptr cinfo)
-{
-  /* no work in 1-pass case */
-}
-
-
-/*
- * Switch to a new external colormap between output passes.
- * Shouldn't get to this module!
- */
-
-METHODDEF(void)
-new_color_map_1_quant (j_decompress_ptr cinfo)
-{
-  ERREXIT(cinfo, JERR_MODE_CHANGE);
-}
-
-
-/*
- * Module initialization routine for 1-pass color quantization.
- */
-
-GLOBAL(void)
-jinit_1pass_quantizer (j_decompress_ptr cinfo)
-{
-  my_cquantize_ptr cquantize;
-
-  cquantize = (my_cquantize_ptr)
-    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				SIZEOF(my_cquantizer));
-  cinfo->cquantize = (struct jpeg_color_quantizer *) cquantize;
-  cquantize->pub.start_pass = start_pass_1_quant;
-  cquantize->pub.finish_pass = finish_pass_1_quant;
-  cquantize->pub.new_color_map = new_color_map_1_quant;
-  cquantize->fserrors[0] = NULL; /* Flag FS workspace not allocated */
-  cquantize->odither[0] = NULL;	/* Also flag odither arrays not allocated */
-
-  /* Make sure my internal arrays won't overflow */
-  if (cinfo->out_color_components > MAX_Q_COMPS)
-    ERREXIT1(cinfo, JERR_QUANT_COMPONENTS, MAX_Q_COMPS);
-  /* Make sure colormap indexes can be represented by JSAMPLEs */
-  if (cinfo->desired_number_of_colors > (MAXJSAMPLE+1))
-    ERREXIT1(cinfo, JERR_QUANT_MANY_COLORS, MAXJSAMPLE+1);
-
-  /* Create the colormap and color index table. */
-  create_colormap(cinfo);
-  create_colorindex(cinfo);
-
-  /* Allocate Floyd-Steinberg workspace now if requested.
-   * We do this now since it is FAR storage and may affect the memory
-   * manager's space calculations.  If the user changes to FS dither
-   * mode in a later pass, we will allocate the space then, and will
-   * possibly overrun the max_memory_to_use setting.
-   */
-  if (cinfo->dither_mode == JDITHER_FS)
-    alloc_fs_workspace(cinfo);
-}
-
-#endif /* QUANT_1PASS_SUPPORTED */
@@ -1,185 +0,0 @@
-/*
- * jutils.c
- *
- * Copyright (C) 1991-1996, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains tables and miscellaneous utility routines needed
- * for both compression and decompression.
- * Note we prefix all global names with "j" to minimize conflicts with
- * a surrounding application.
- */
-
-#define JPEG_INTERNALS
-#include "jinclude.h"
-#include "jpeglib.h"
-
-
-/*
- * jpeg_zigzag_order[i] is the zigzag-order position of the i'th element
- * of a DCT block read in natural order (left to right, top to bottom).
- */
-
-#if 0				/* This table is not actually needed in v6a */
-
-const int jpeg_zigzag_order[DCTSIZE2] = {
-   0,  1,  5,  6, 14, 15, 27, 28,
-   2,  4,  7, 13, 16, 26, 29, 42,
-   3,  8, 12, 17, 25, 30, 41, 43,
-   9, 11, 18, 24, 31, 40, 44, 53,
-  10, 19, 23, 32, 39, 45, 52, 54,
-  20, 22, 33, 38, 46, 51, 55, 60,
-  21, 34, 37, 47, 50, 56, 59, 61,
-  35, 36, 48, 49, 57, 58, 62, 63
-};
-
-#endif
-
-/*
- * jpeg_natural_order[i] is the natural-order position of the i'th element
- * of zigzag order.
- *
- * When reading corrupted data, the Huffman decoders could attempt
- * to reference an entry beyond the end of this array (if the decoded
- * zero run length reaches past the end of the block).  To prevent
- * wild stores without adding an inner-loop test, we put some extra
- * "63"s after the real entries.  This will cause the extra coefficient
- * to be stored in location 63 of the block, not somewhere random.
- * The worst case would be a run-length of 15, which means we need 16
- * fake entries.
- */
-
-const int jpeg_natural_order[DCTSIZE2+16] = {
-  0,  1,  8, 16,  9,  2,  3, 10,
- 17, 24, 32, 25, 18, 11,  4,  5,
- 12, 19, 26, 33, 40, 48, 41, 34,
- 27, 20, 13,  6,  7, 14, 21, 28,
- 35, 42, 49, 56, 57, 50, 43, 36,
- 29, 22, 15, 23, 30, 37, 44, 51,
- 58, 59, 52, 45, 38, 31, 39, 46,
- 53, 60, 61, 54, 47, 55, 62, 63,
- 63, 63, 63, 63, 63, 63, 63, 63, /* extra entries for safety in decoder */
- 63, 63, 63, 63, 63, 63, 63, 63
-};
-
-
-/*
- * Arithmetic utilities
- */
-
-GLOBAL(long)
-jdiv_round_up (long a, long b)
-/* Compute a/b rounded up to next integer, ie, ceil(a/b) */
-/* Assumes a >= 0, b > 0 */
-{
-  return (a + b - 1L) / b;
-}
-
-
-GLOBAL(long)
-jround_up (long a, long b)
-/* Compute a rounded up to next multiple of b, ie, ceil(a/b)*b */
-/* Assumes a >= 0, b > 0 */
-{
-  a += b - 1L;
-  return a - (a % b);
-}
-
-GLOBAL(long)
-jmin (long a, long b)
-{
-  return a < b ? a : b;
-}
-
-
-/* On normal machines we can apply MEMCOPY() and MEMZERO() to sample arrays
- * and coefficient-block arrays.  This won't work on 80x86 because the arrays
- * are FAR and we're assuming a small-pointer memory model.  However, some
- * DOS compilers provide far-pointer versions of memcpy() and memset() even
- * in the small-model libraries.  These will be used if USE_FMEM is defined.
- * Otherwise, the routines below do it the hard way.  (The performance cost
- * is not all that great, because these routines aren't very heavily used.)
- */
-
-#ifndef NEED_FAR_POINTERS	/* normal case, same as regular macros */
-#define FMEMCOPY(dest,src,size)	MEMCOPY(dest,src,size)
-#define FMEMZERO(target,size)	MEMZERO(target,size)
-#else				/* 80x86 case, define if we can */
-#ifdef USE_FMEM
-#define FMEMCOPY(dest,src,size)	_fmemcpy((void FAR *)(dest), (const void FAR *)(src), (size_t)(size))
-#define FMEMZERO(target,size)	_fmemset((void FAR *)(target), 0, (size_t)(size))
-#endif
-#endif
-
-
-GLOBAL(void)
-jcopy_sample_rows (JSAMPARRAY input_array, int source_row,
-		   JSAMPARRAY output_array, int dest_row,
-		   int num_rows, JDIMENSION num_cols)
-/* Copy some rows of samples from one place to another.
- * num_rows rows are copied from input_array[source_row++]
- * to output_array[dest_row++]; these areas may overlap for duplication.
- * The source and destination arrays must be at least as wide as num_cols.
- */
-{
-  register JSAMPROW inptr, outptr;
-#ifdef FMEMCOPY
-  register size_t count = (size_t) (num_cols * SIZEOF(JSAMPLE));
-#else
-  register JDIMENSION count;
-#endif
-  register int row;
-
-  input_array += source_row;
-  output_array += dest_row;
-
-  for (row = num_rows; row > 0; row--) {
-    inptr = *input_array++;
-    outptr = *output_array++;
-#ifdef FMEMCOPY
-    FMEMCOPY(outptr, inptr, count);
-#else
-    for (count = num_cols; count > 0; count--)
-      *outptr++ = *inptr++;	/* needn't bother with GETJSAMPLE() here */
-#endif
-  }
-}
-
-
-GLOBAL(void)
-jcopy_block_row (JBLOCKROW input_row, JBLOCKROW output_row,
-		 JDIMENSION num_blocks)
-/* Copy a row of coefficient blocks from one place to another. */
-{
-#ifdef FMEMCOPY
-  FMEMCOPY(output_row, input_row, num_blocks * (DCTSIZE2 * SIZEOF(JCOEF)));
-#else
-  register JCOEFPTR inptr, outptr;
-  register long count;
-
-  inptr = (JCOEFPTR) input_row;
-  outptr = (JCOEFPTR) output_row;
-  for (count = (long) num_blocks * DCTSIZE2; count > 0; count--) {
-    *outptr++ = *inptr++;
-  }
-#endif
-}
-
-
-GLOBAL(void)
-jzero_far (void FAR * target, size_t bytestozero)
-/* Zero out a chunk of FAR memory. */
-/* This might be sample-array data, block-array data, or alloc_large data. */
-{
-#ifdef FMEMZERO
-  FMEMZERO(target, bytestozero);
-#else
-  register char FAR * ptr = (char FAR *) target;
-  register size_t count;
-
-  for (count = bytestozero; count > 0; count--) {
-    *ptr++ = 0;
-  }
-#endif
-}
@@ -1,14 +0,0 @@
-/*
- * jversion.h
- *
- * Copyright (C) 1991-1998, Thomas G. Lane.
- * This file is part of the Independent JPEG Group's software.
- * For conditions of distribution and use, see the accompanying README file.
- *
- * This file contains software version identification.
- */
-
-
-#define JVERSION	"6b  27-Mar-1998"
-
-#define JCOPYRIGHT	"Copyright (C) 1998, Thomas G. Lane"
@@ -1,24 +0,0 @@
-Contributors:
- Charles Munger (clm at google dot com)
- Christian Duvivier (cduvivier at google dot com)
- Djordje Pesut (djordje dot pesut at imgtec dot com)
- James Zern (jzern at google dot com)
- Jan Engelhardt (jengelh at medozas dot de)
- Johann (johann dot koenig at duck dot com)
- Jovan Zelincevic (jovan dot zelincevic at imgtec dot com)
- Jyrki Alakuijala (jyrki at google dot com)
- levytamar82 (tamar dot levy at intel dot com)
- Lou Quillio (louquillio at google dot com)
- Mans Rullgard (mans at mansr dot com)
- Martin Olsson (mnemo at minimum dot se)
- Mikołaj Zalewski (mikolajz at google dot com)
- Noel Chromium (noel at chromium dot org)
- Pascal Massimino (pascal dot massimino at gmail dot com)
- Paweł Hajdan, Jr (phajdan dot jr at chromium dot org)
- Pierre Joye (pierre dot php at gmail dot com)
- Scott LaVarnway (slavarnway at google dot com)
- Scott Talbot (s at chikachow dot org)
- Slobodan Prijic (slobodan dot prijic at imgtec dot com)
- Somnath Banerjee (somnath dot banerjee at gmail dot com)
- Urvang Joshi (urvang at google dot com)
- Vikas Arora (vikasa at google dot com)
@@ -1,30 +0,0 @@
-Copyright (c) 2010, Google Inc. All rights reserved.
-
-Redistribution and use in source and binary forms, with or without
-modification, are permitted provided that the following conditions are
-met:
-
-  * Redistributions of source code must retain the above copyright
-    notice, this list of conditions and the following disclaimer.
-
-  * Redistributions in binary form must reproduce the above copyright
-    notice, this list of conditions and the following disclaimer in
-    the documentation and/or other materials provided with the
-    distribution.
-
-  * Neither the name of Google nor the names of its contributors may
-    be used to endorse or promote products derived from this software
-    without specific prior written permission.
-
-THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
-"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
-LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
-A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
-HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
-SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
-LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
-DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
-THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
-(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
-OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
-
@@ -1,23 +0,0 @@
-Additional IP Rights Grant (Patents)
------------------------------------
-
-"These implementations" means the copyrightable works that implement the WebM
-codecs distributed by Google as part of the WebM Project.
-
-Google hereby grants to you a perpetual, worldwide, non-exclusive, no-charge,
-royalty-free, irrevocable (except as stated in this section) patent license to
-make, have made, use, offer to sell, sell, import, transfer, and otherwise
-run, modify and propagate the contents of these implementations of WebM, where
-such license applies only to those patent claims, both currently owned by
-Google and acquired in the future, licensable by Google that are necessarily
-infringed by these implementations of WebM. This grant does not include claims
-that would be infringed only as a consequence of further modification of these
-implementations. If you or your agent or exclusive licensee institute or order
-or agree to the institution of patent litigation or any other patent
-enforcement activity against any entity (including a cross-claim or
-counterclaim in a lawsuit) alleging that any of these implementations of WebM
-or any code incorporated within any of these implementations of WebM
-constitute direct or contributory patent infringement, or inducement of
-patent infringement, then any patent rights granted to you under this License
-for these implementations of WebM shall terminate as of the date such
-litigation is filed.
@@ -1,165 +0,0 @@
-// Copyright 2011 Google Inc. All Rights Reserved.
-//
-// Use of this source code is governed by a BSD-style license
-// that can be found in the COPYING file in the root of the source
-// tree. An additional intellectual property rights grant can be found
-// in the file PATENTS. All contributing project authors may
-// be found in the AUTHORS file in the root of the source tree.
-// -----------------------------------------------------------------------------
-//
-// Alpha-plane decompression.
-//
-// Author: Skal (pascal.massimino@gmail.com)
-
-#include <stdlib.h>
-#include "./alphai.h"
-#include "./vp8i.h"
-#include "./vp8li.h"
-#include "../utils/quant_levels_dec.h"
-#include "../utils/utils.h"
-#include "../webp/format_constants.h"
-
-//------------------------------------------------------------------------------
-// ALPHDecoder object.
-
-ALPHDecoder* ALPHNew(void) {
-  ALPHDecoder* const dec = (ALPHDecoder*)WebPSafeCalloc(1ULL, sizeof(*dec));
-  return dec;
-}
-
-void ALPHDelete(ALPHDecoder* const dec) {
-  if (dec != NULL) {
-    VP8LDelete(dec->vp8l_dec_);
-    dec->vp8l_dec_ = NULL;
-    WebPSafeFree(dec);
-  }
-}
-
-//------------------------------------------------------------------------------
-// Decoding.
-
-// Initialize alpha decoding by parsing the alpha header and decoding the image
-// header for alpha data stored using lossless compression.
-// Returns false in case of error in alpha header (data too short, invalid
-// compression method or filter, error in lossless header data etc).
-static int ALPHInit(ALPHDecoder* const dec, const uint8_t* data,
-                    size_t data_size, int width, int height, uint8_t* output) {
-  int ok = 0;
-  const uint8_t* const alpha_data = data + ALPHA_HEADER_LEN;
-  const size_t alpha_data_size = data_size - ALPHA_HEADER_LEN;
-  int rsrv;
-
-  assert(width > 0 && height > 0);
-  assert(data != NULL && output != NULL);
-
-  dec->width_ = width;
-  dec->height_ = height;
-
-  if (data_size <= ALPHA_HEADER_LEN) {
-    return 0;
-  }
-
-  dec->method_ = (data[0] >> 0) & 0x03;
-  dec->filter_ = (data[0] >> 2) & 0x03;
-  dec->pre_processing_ = (data[0] >> 4) & 0x03;
-  rsrv = (data[0] >> 6) & 0x03;
-  if (dec->method_ < ALPHA_NO_COMPRESSION ||
-      dec->method_ > ALPHA_LOSSLESS_COMPRESSION ||
-      dec->filter_ >= WEBP_FILTER_LAST ||
-      dec->pre_processing_ > ALPHA_PREPROCESSED_LEVELS ||
-      rsrv != 0) {
-    return 0;
-  }
-
-  if (dec->method_ == ALPHA_NO_COMPRESSION) {
-    const size_t alpha_decoded_size = dec->width_ * dec->height_;
-    ok = (alpha_data_size >= alpha_decoded_size);
-  } else {
-    assert(dec->method_ == ALPHA_LOSSLESS_COMPRESSION);
-    ok = VP8LDecodeAlphaHeader(dec, alpha_data, alpha_data_size, output);
-  }
-  return ok;
-}
-
-// Decodes, unfilters and dequantizes *at least* 'num_rows' rows of alpha
-// starting from row number 'row'. It assumes that rows up to (row - 1) have
-// already been decoded.
-// Returns false in case of bitstream error.
-static int ALPHDecode(VP8Decoder* const dec, int row, int num_rows) {
-  ALPHDecoder* const alph_dec = dec->alph_dec_;
-  const int width = alph_dec->width_;
-  const int height = alph_dec->height_;
-  WebPUnfilterFunc unfilter_func = WebPUnfilters[alph_dec->filter_];
-  uint8_t* const output = dec->alpha_plane_;
-  if (alph_dec->method_ == ALPHA_NO_COMPRESSION) {
-    const size_t offset = row * width;
-    const size_t num_pixels = num_rows * width;
-    assert(dec->alpha_data_size_ >= ALPHA_HEADER_LEN + offset + num_pixels);
-    memcpy(dec->alpha_plane_ + offset,
-           dec->alpha_data_ + ALPHA_HEADER_LEN + offset, num_pixels);
-  } else {  // alph_dec->method_ == ALPHA_LOSSLESS_COMPRESSION
-    assert(alph_dec->vp8l_dec_ != NULL);
-    if (!VP8LDecodeAlphaImageStream(alph_dec, row + num_rows)) {
-      return 0;
-    }
-  }
-
-  if (unfilter_func != NULL) {
-    unfilter_func(width, height, width, row, num_rows, output);
-  }
-
-  if (row + num_rows == dec->pic_hdr_.height_) {
-    dec->is_alpha_decoded_ = 1;
-  }
-  return 1;
-}
-
-//------------------------------------------------------------------------------
-// Main entry point.
-
-const uint8_t* VP8DecompressAlphaRows(VP8Decoder* const dec,
-                                      int row, int num_rows) {
-  const int width = dec->pic_hdr_.width_;
-  const int height = dec->pic_hdr_.height_;
-
-  if (row < 0 || num_rows <= 0 || row + num_rows > height) {
-    return NULL;    // sanity check.
-  }
-
-  if (row == 0) {
-    // Initialize decoding.
-    assert(dec->alpha_plane_ != NULL);
-    dec->alph_dec_ = ALPHNew();
-    if (dec->alph_dec_ == NULL) return NULL;
-    if (!ALPHInit(dec->alph_dec_, dec->alpha_data_, dec->alpha_data_size_,
-                  width, height, dec->alpha_plane_)) {
-      ALPHDelete(dec->alph_dec_);
-      dec->alph_dec_ = NULL;
-      return NULL;
-    }
-    // if we allowed use of alpha dithering, check whether it's needed at all
-    if (dec->alph_dec_->pre_processing_ != ALPHA_PREPROCESSED_LEVELS) {
-      dec->alpha_dithering_ = 0;  // disable dithering
-    } else {
-      num_rows = height;          // decode everything in one pass
-    }
-  }
-
-  if (!dec->is_alpha_decoded_) {
-    int ok = 0;
-    assert(dec->alph_dec_ != NULL);
-    ok = ALPHDecode(dec, row, num_rows);
-    if (ok && dec->alpha_dithering_ > 0) {
-      ok = WebPDequantizeLevels(dec->alpha_plane_, width, height,
-                                dec->alpha_dithering_);
-    }
-    if (!ok || dec->is_alpha_decoded_) {
-      ALPHDelete(dec->alph_dec_);
-      dec->alph_dec_ = NULL;
-    }
-    if (!ok) return NULL;  // Error.
-  }
-
-  // Return a pointer to the current decoded row.
-  return dec->alpha_plane_ + row * width;
-}
@@ -1,55 +0,0 @@
-// Copyright 2013 Google Inc. All Rights Reserved.
-//
-// Use of this source code is governed by a BSD-style license
-// that can be found in the COPYING file in the root of the source
-// tree. An additional intellectual property rights grant can be found
-// in the file PATENTS. All contributing project authors may
-// be found in the AUTHORS file in the root of the source tree.
-// -----------------------------------------------------------------------------
-//
-// Alpha decoder: internal header.
-//
-// Author: Urvang (urvang@google.com)
-
-#ifndef WEBP_DEC_ALPHAI_H_
-#define WEBP_DEC_ALPHAI_H_
-
-#include "./webpi.h"
-#include "../utils/filters.h"
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-struct VP8LDecoder;  // Defined in dec/vp8li.h.
-
-typedef struct ALPHDecoder ALPHDecoder;
-struct ALPHDecoder {
-  int width_;
-  int height_;
-  int method_;
-  WEBP_FILTER_TYPE filter_;
-  int pre_processing_;
-  struct VP8LDecoder* vp8l_dec_;
-  VP8Io io_;
-  int use_8b_decode;  // Although alpha channel requires only 1 byte per
-                      // pixel, sometimes VP8LDecoder may need to allocate
-                      // 4 bytes per pixel internally during decode.
-};
-
-//------------------------------------------------------------------------------
-// internal functions. Not public.
-
-// Allocates a new alpha decoder instance.
-ALPHDecoder* ALPHNew(void);
-
-// Clears and deallocates an alpha decoder instance.
-void ALPHDelete(ALPHDecoder* const dec);
-
-//------------------------------------------------------------------------------
-
-#ifdef __cplusplus
-}    // extern "C"
-#endif
-
-#endif  /* WEBP_DEC_ALPHAI_H_ */
@@ -1,251 +0,0 @@
-// Copyright 2011 Google Inc. All Rights Reserved.
-//
-// Use of this source code is governed by a BSD-style license
-// that can be found in the COPYING file in the root of the source
-// tree. An additional intellectual property rights grant can be found
-// in the file PATENTS. All contributing project authors may
-// be found in the AUTHORS file in the root of the source tree.
-// -----------------------------------------------------------------------------
-//
-// Everything about WebPDecBuffer
-//
-// Author: Skal (pascal.massimino@gmail.com)
-
-#include <stdlib.h>
-
-#include "./vp8i.h"
-#include "./webpi.h"
-#include "../utils/utils.h"
-
-//------------------------------------------------------------------------------
-// WebPDecBuffer
-
-// Number of bytes per pixel for the different color-spaces.
-static const int kModeBpp[MODE_LAST] = {
-  3, 4, 3, 4, 4, 2, 2,
-  4, 4, 4, 2,    // pre-multiplied modes
-  1, 1 };
-
-// Check that webp_csp_mode is within the bounds of WEBP_CSP_MODE.
-// Convert to an integer to handle both the unsigned/signed enum cases
-// without the need for casting to remove type limit warnings.
-static int IsValidColorspace(int webp_csp_mode) {
-  return (webp_csp_mode >= MODE_RGB && webp_csp_mode < MODE_LAST);
-}
-
-static VP8StatusCode CheckDecBuffer(const WebPDecBuffer* const buffer) {
-  int ok = 1;
-  const WEBP_CSP_MODE mode = buffer->colorspace;
-  const int width = buffer->width;
-  const int height = buffer->height;
-  if (!IsValidColorspace(mode)) {
-    ok = 0;
-  } else if (!WebPIsRGBMode(mode)) {   // YUV checks
-    const WebPYUVABuffer* const buf = &buffer->u.YUVA;
-    const int y_stride = abs(buf->y_stride);
-    const int u_stride = abs(buf->u_stride);
-    const int v_stride = abs(buf->v_stride);
-    const int a_stride = abs(buf->a_stride);
-    const uint64_t y_size = (uint64_t)y_stride * height;
-    const uint64_t u_size = (uint64_t)u_stride * ((height + 1) / 2);
-    const uint64_t v_size = (uint64_t)v_stride * ((height + 1) / 2);
-    const uint64_t a_size = (uint64_t)a_stride * height;
-    ok &= (y_size <= buf->y_size);
-    ok &= (u_size <= buf->u_size);
-    ok &= (v_size <= buf->v_size);
-    ok &= (y_stride >= width);
-    ok &= (u_stride >= (width + 1) / 2);
-    ok &= (v_stride >= (width + 1) / 2);
-    ok &= (buf->y != NULL);
-    ok &= (buf->u != NULL);
-    ok &= (buf->v != NULL);
-    if (mode == MODE_YUVA) {
-      ok &= (a_stride >= width);
-      ok &= (a_size <= buf->a_size);
-      ok &= (buf->a != NULL);
-    }
-  } else {    // RGB checks
-    const WebPRGBABuffer* const buf = &buffer->u.RGBA;
-    const int stride = abs(buf->stride);
-    const uint64_t size = (uint64_t)stride * height;
-    ok &= (size <= buf->size);
-    ok &= (stride >= width * kModeBpp[mode]);
-    ok &= (buf->rgba != NULL);
-  }
-  return ok ? VP8_STATUS_OK : VP8_STATUS_INVALID_PARAM;
-}
-
-static VP8StatusCode AllocateBuffer(WebPDecBuffer* const buffer) {
-  const int w = buffer->width;
-  const int h = buffer->height;
-  const WEBP_CSP_MODE mode = buffer->colorspace;
-
-  if (w <= 0 || h <= 0 || !IsValidColorspace(mode)) {
-    return VP8_STATUS_INVALID_PARAM;
-  }
-
-  if (!buffer->is_external_memory && buffer->private_memory == NULL) {
-    uint8_t* output;
-    int uv_stride = 0, a_stride = 0;
-    uint64_t uv_size = 0, a_size = 0, total_size;
-    // We need memory and it hasn't been allocated yet.
-    // => initialize output buffer, now that dimensions are known.
-    const int stride = w * kModeBpp[mode];
-    const uint64_t size = (uint64_t)stride * h;
-
-    if (!WebPIsRGBMode(mode)) {
-      uv_stride = (w + 1) / 2;
-      uv_size = (uint64_t)uv_stride * ((h + 1) / 2);
-      if (mode == MODE_YUVA) {
-        a_stride = w;
-        a_size = (uint64_t)a_stride * h;
-      }
-    }
-    total_size = size + 2 * uv_size + a_size;
-
-    // Security/sanity checks
-    output = (uint8_t*)WebPSafeMalloc(total_size, sizeof(*output));
-    if (output == NULL) {
-      return VP8_STATUS_OUT_OF_MEMORY;
-    }
-    buffer->private_memory = output;
-
-    if (!WebPIsRGBMode(mode)) {   // YUVA initialization
-      WebPYUVABuffer* const buf = &buffer->u.YUVA;
-      buf->y = output;
-      buf->y_stride = stride;
-      buf->y_size = (size_t)size;
-      buf->u = output + size;
-      buf->u_stride = uv_stride;
-      buf->u_size = (size_t)uv_size;
-      buf->v = output + size + uv_size;
-      buf->v_stride = uv_stride;
-      buf->v_size = (size_t)uv_size;
-      if (mode == MODE_YUVA) {
-        buf->a = output + size + 2 * uv_size;
-      }
-      buf->a_size = (size_t)a_size;
-      buf->a_stride = a_stride;
-    } else {  // RGBA initialization
-      WebPRGBABuffer* const buf = &buffer->u.RGBA;
-      buf->rgba = output;
-      buf->stride = stride;
-      buf->size = (size_t)size;
-    }
-  }
-  return CheckDecBuffer(buffer);
-}
-
-VP8StatusCode WebPFlipBuffer(WebPDecBuffer* const buffer) {
-  if (buffer == NULL) {
-    return VP8_STATUS_INVALID_PARAM;
-  }
-  if (WebPIsRGBMode(buffer->colorspace)) {
-    WebPRGBABuffer* const buf = &buffer->u.RGBA;
-    buf->rgba += (buffer->height - 1) * buf->stride;
-    buf->stride = -buf->stride;
-  } else {
-    WebPYUVABuffer* const buf = &buffer->u.YUVA;
-    const int H = buffer->height;
-    buf->y += (H - 1) * buf->y_stride;
-    buf->y_stride = -buf->y_stride;
-    buf->u += ((H - 1) >> 1) * buf->u_stride;
-    buf->u_stride = -buf->u_stride;
-    buf->v += ((H - 1) >> 1) * buf->v_stride;
-    buf->v_stride = -buf->v_stride;
-    if (buf->a != NULL) {
-      buf->a += (H - 1) * buf->a_stride;
-      buf->a_stride = -buf->a_stride;
-    }
-  }
-  return VP8_STATUS_OK;
-}
-
-VP8StatusCode WebPAllocateDecBuffer(int w, int h,
-                                    const WebPDecoderOptions* const options,
-                                    WebPDecBuffer* const out) {
-  VP8StatusCode status;
-  if (out == NULL || w <= 0 || h <= 0) {
-    return VP8_STATUS_INVALID_PARAM;
-  }
-  if (options != NULL) {    // First, apply options if there is any.
-    if (options->use_cropping) {
-      const int cw = options->crop_width;
-      const int ch = options->crop_height;
-      const int x = options->crop_left & ~1;
-      const int y = options->crop_top & ~1;
-      if (x < 0 || y < 0 || cw <= 0 || ch <= 0 || x + cw > w || y + ch > h) {
-        return VP8_STATUS_INVALID_PARAM;   // out of frame boundary.
-      }
-      w = cw;
-      h = ch;
-    }
-    if (options->use_scaling) {
-      if (options->scaled_width <= 0 || options->scaled_height <= 0) {
-        return VP8_STATUS_INVALID_PARAM;
-      }
-      w = options->scaled_width;
-      h = options->scaled_height;
-    }
-  }
-  out->width = w;
-  out->height = h;
-
-  // Then, allocate buffer for real.
-  status = AllocateBuffer(out);
-  if (status != VP8_STATUS_OK) return status;
-
-#if WEBP_DECODER_ABI_VERSION > 0x0203
-  // Use the stride trick if vertical flip is needed.
-  if (options != NULL && options->flip) {
-    status = WebPFlipBuffer(out);
-  }
-#endif
-  return status;
-}
-
-//------------------------------------------------------------------------------
-// constructors / destructors
-
-int WebPInitDecBufferInternal(WebPDecBuffer* buffer, int version) {
-  if (WEBP_ABI_IS_INCOMPATIBLE(version, WEBP_DECODER_ABI_VERSION)) {
-    return 0;  // version mismatch
-  }
-  if (buffer == NULL) return 0;
-  memset(buffer, 0, sizeof(*buffer));
-  return 1;
-}
-
-void WebPFreeDecBuffer(WebPDecBuffer* buffer) {
-  if (buffer != NULL) {
-    if (!buffer->is_external_memory) {
-      WebPSafeFree(buffer->private_memory);
-    }
-    buffer->private_memory = NULL;
-  }
-}
-
-void WebPCopyDecBuffer(const WebPDecBuffer* const src,
-                       WebPDecBuffer* const dst) {
-  if (src != NULL && dst != NULL) {
-    *dst = *src;
-    if (src->private_memory != NULL) {
-      dst->is_external_memory = 1;   // dst buffer doesn't own the memory.
-      dst->private_memory = NULL;
-    }
-  }
-}
-
-// Copy and transfer ownership from src to dst (beware of parameter order!)
-void WebPGrabDecBuffer(WebPDecBuffer* const src, WebPDecBuffer* const dst) {
-  if (src != NULL && dst != NULL) {
-    *dst = *src;
-    if (src->private_memory != NULL) {
-      src->is_external_memory = 1;   // src relinquishes ownership
-      src->private_memory = NULL;
-    }
-  }
-}
-
-//------------------------------------------------------------------------------
-
@@ -1,185 +0,0 @@
-// Copyright 2010 Google Inc. All Rights Reserved.
-//
-// Use of this source code is governed by a BSD-style license
-// that can be found in the COPYING file in the root of the source
-// tree. An additional intellectual property rights grant can be found
-// in the file PATENTS. All contributing project authors may
-// be found in the AUTHORS file in the root of the source tree.
-// -----------------------------------------------------------------------------
-//
-//  Low-level API for VP8 decoder
-//
-// Author: Skal (pascal.massimino@gmail.com)
-
-#ifndef WEBP_WEBP_DECODE_VP8_H_
-#define WEBP_WEBP_DECODE_VP8_H_
-
-#include "../webp/decode.h"
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-//------------------------------------------------------------------------------
-// Lower-level API
-//
-// These functions provide fine-grained control of the decoding process.
-// The call flow should resemble:
-//
-//   VP8Io io;
-//   VP8InitIo(&io);
-//   io.data = data;
-//   io.data_size = size;
-//   /* customize io's functions (setup()/put()/teardown()) if needed. */
-//
-//   VP8Decoder* dec = VP8New();
-//   bool ok = VP8Decode(dec);
-//   if (!ok) printf("Error: %s\n", VP8StatusMessage(dec));
-//   VP8Delete(dec);
-//   return ok;
-
-// Input / Output
-typedef struct VP8Io VP8Io;
-typedef int (*VP8IoPutHook)(const VP8Io* io);
-typedef int (*VP8IoSetupHook)(VP8Io* io);
-typedef void (*VP8IoTeardownHook)(const VP8Io* io);
-
-struct VP8Io {
-  // set by VP8GetHeaders()
-  int width, height;         // picture dimensions, in pixels (invariable).
-                             // These are the original, uncropped dimensions.
-                             // The actual area passed to put() is stored
-                             // in mb_w / mb_h fields.
-
-  // set before calling put()
-  int mb_y;                  // position of the current rows (in pixels)
-  int mb_w;                  // number of columns in the sample
-  int mb_h;                  // number of rows in the sample
-  const uint8_t* y, *u, *v;  // rows to copy (in yuv420 format)
-  int y_stride;              // row stride for luma
-  int uv_stride;             // row stride for chroma
-
-  void* opaque;              // user data
-
-  // called when fresh samples are available. Currently, samples are in
-  // YUV420 format, and can be up to width x 24 in size (depending on the
-  // in-loop filtering level, e.g.). Should return false in case of error
-  // or abort request. The actual size of the area to update is mb_w x mb_h
-  // in size, taking cropping into account.
-  VP8IoPutHook put;
-
-  // called just before starting to decode the blocks.
-  // Must return false in case of setup error, true otherwise. If false is
-  // returned, teardown() will NOT be called. But if the setup succeeded
-  // and true is returned, then teardown() will always be called afterward.
-  VP8IoSetupHook setup;
-
-  // Called just after block decoding is finished (or when an error occurred
-  // during put()). Is NOT called if setup() failed.
-  VP8IoTeardownHook teardown;
-
-  // this is a recommendation for the user-side yuv->rgb converter. This flag
-  // is set when calling setup() hook and can be overwritten by it. It then
-  // can be taken into consideration during the put() method.
-  int fancy_upsampling;
-
-  // Input buffer.
-  size_t data_size;
-  const uint8_t* data;
-
-  // If true, in-loop filtering will not be performed even if present in the
-  // bitstream. Switching off filtering may speed up decoding at the expense
-  // of more visible blocking. Note that output will also be non-compliant
-  // with the VP8 specifications.
-  int bypass_filtering;
-
-  // Cropping parameters.
-  int use_cropping;
-  int crop_left, crop_right, crop_top, crop_bottom;
-
-  // Scaling parameters.
-  int use_scaling;
-  int scaled_width, scaled_height;
-
-  // If non NULL, pointer to the alpha data (if present) corresponding to the
-  // start of the current row (That is: it is pre-offset by mb_y and takes
-  // cropping into account).
-  const uint8_t* a;
-};
-
-// Internal, version-checked, entry point
-int VP8InitIoInternal(VP8Io* const, int);
-
-// Set the custom IO function pointers and user-data. The setter for IO hooks
-// should be called before initiating incremental decoding. Returns true if
-// WebPIDecoder object is successfully modified, false otherwise.
-int WebPISetIOHooks(WebPIDecoder* const idec,
-                    VP8IoPutHook put,
-                    VP8IoSetupHook setup,
-                    VP8IoTeardownHook teardown,
-                    void* user_data);
-
-// Main decoding object. This is an opaque structure.
-typedef struct VP8Decoder VP8Decoder;
-
-// Create a new decoder object.
-VP8Decoder* VP8New(void);
-
-// Must be called to make sure 'io' is initialized properly.
-// Returns false in case of version mismatch. Upon such failure, no other
-// decoding function should be called (VP8Decode, VP8GetHeaders, ...)
-static WEBP_INLINE int VP8InitIo(VP8Io* const io) {
-  return VP8InitIoInternal(io, WEBP_DECODER_ABI_VERSION);
-}
-
-// Decode the VP8 frame header. Returns true if ok.
-// Note: 'io->data' must be pointing to the start of the VP8 frame header.
-int VP8GetHeaders(VP8Decoder* const dec, VP8Io* const io);
-
-// Decode a picture. Will call VP8GetHeaders() if it wasn't done already.
-// Returns false in case of error.
-int VP8Decode(VP8Decoder* const dec, VP8Io* const io);
-
-// Return current status of the decoder:
-VP8StatusCode VP8Status(VP8Decoder* const dec);
-
-// return readable string corresponding to the last status.
-const char* VP8StatusMessage(VP8Decoder* const dec);
-
-// Resets the decoder in its initial state, reclaiming memory.
-// Not a mandatory call between calls to VP8Decode().
-void VP8Clear(VP8Decoder* const dec);
-
-// Destroy the decoder object.
-void VP8Delete(VP8Decoder* const dec);
-
-//------------------------------------------------------------------------------
-// Miscellaneous VP8/VP8L bitstream probing functions.
-
-// Returns true if the next 3 bytes in data contain the VP8 signature.
-WEBP_EXTERN(int) VP8CheckSignature(const uint8_t* const data, size_t data_size);
-
-// Validates the VP8 data-header and retrieves basic header information viz
-// width and height. Returns 0 in case of formatting error. *width/*height
-// can be passed NULL.
-WEBP_EXTERN(int) VP8GetInfo(
-    const uint8_t* data,
-    size_t data_size,    // data available so far
-    size_t chunk_size,   // total data size expected in the chunk
-    int* const width, int* const height);
-
-// Returns true if the next byte(s) in data is a VP8L signature.
-WEBP_EXTERN(int) VP8LCheckSignature(const uint8_t* const data, size_t size);
-
-// Validates the VP8L data-header and retrieves basic header information viz
-// width, height and alpha. Returns 0 in case of formatting error.
-// width/height/has_alpha can be passed NULL.
-WEBP_EXTERN(int) VP8LGetInfo(
-    const uint8_t* data, size_t data_size,  // data available so far
-    int* const width, int* const height, int* const has_alpha);
-
-#ifdef __cplusplus
-}    // extern "C"
-#endif
-
-#endif  /* WEBP_WEBP_DECODE_VP8_H_ */
@@ -1,828 +0,0 @@
-// Copyright 2010 Google Inc. All Rights Reserved.
-//
-// Use of this source code is governed by a BSD-style license
-// that can be found in the COPYING file in the root of the source
-// tree. An additional intellectual property rights grant can be found
-// in the file PATENTS. All contributing project authors may
-// be found in the AUTHORS file in the root of the source tree.
-// -----------------------------------------------------------------------------
-//
-// Frame-reconstruction function. Memory allocation.
-//
-// Author: Skal (pascal.massimino@gmail.com)
-
-#include <stdlib.h>
-#include "./vp8i.h"
-#include "../utils/utils.h"
-
-#define ALIGN_MASK (32 - 1)
-
-static void ReconstructRow(const VP8Decoder* const dec,
-                           const VP8ThreadContext* ctx);  // TODO(skal): remove
-
-//------------------------------------------------------------------------------
-// Filtering
-
-// kFilterExtraRows[] = How many extra lines are needed on the MB boundary
-// for caching, given a filtering level.
-// Simple filter:  up to 2 luma samples are read and 1 is written.
-// Complex filter: up to 4 luma samples are read and 3 are written. Same for
-//                 U/V, so it's 8 samples total (because of the 2x upsampling).
-static const uint8_t kFilterExtraRows[3] = { 0, 2, 8 };
-
-static void DoFilter(const VP8Decoder* const dec, int mb_x, int mb_y) {
-  const VP8ThreadContext* const ctx = &dec->thread_ctx_;
-  const int cache_id = ctx->id_;
-  const int y_bps = dec->cache_y_stride_;
-  const VP8FInfo* const f_info = ctx->f_info_ + mb_x;
-  uint8_t* const y_dst = dec->cache_y_ + cache_id * 16 * y_bps + mb_x * 16;
-  const int ilevel = f_info->f_ilevel_;
-  const int limit = f_info->f_limit_;
-  if (limit == 0) {
-    return;
-  }
-  assert(limit >= 3);
-  if (dec->filter_type_ == 1) {   // simple
-    if (mb_x > 0) {
-      VP8SimpleHFilter16(y_dst, y_bps, limit + 4);
-    }
-    if (f_info->f_inner_) {
-      VP8SimpleHFilter16i(y_dst, y_bps, limit);
-    }
-    if (mb_y > 0) {
-      VP8SimpleVFilter16(y_dst, y_bps, limit + 4);
-    }
-    if (f_info->f_inner_) {
-      VP8SimpleVFilter16i(y_dst, y_bps, limit);
-    }
-  } else {    // complex
-    const int uv_bps = dec->cache_uv_stride_;
-    uint8_t* const u_dst = dec->cache_u_ + cache_id * 8 * uv_bps + mb_x * 8;
-    uint8_t* const v_dst = dec->cache_v_ + cache_id * 8 * uv_bps + mb_x * 8;
-    const int hev_thresh = f_info->hev_thresh_;
-    if (mb_x > 0) {
-      VP8HFilter16(y_dst, y_bps, limit + 4, ilevel, hev_thresh);
-      VP8HFilter8(u_dst, v_dst, uv_bps, limit + 4, ilevel, hev_thresh);
-    }
-    if (f_info->f_inner_) {
-      VP8HFilter16i(y_dst, y_bps, limit, ilevel, hev_thresh);
-      VP8HFilter8i(u_dst, v_dst, uv_bps, limit, ilevel, hev_thresh);
-    }
-    if (mb_y > 0) {
-      VP8VFilter16(y_dst, y_bps, limit + 4, ilevel, hev_thresh);
-      VP8VFilter8(u_dst, v_dst, uv_bps, limit + 4, ilevel, hev_thresh);
-    }
-    if (f_info->f_inner_) {
-      VP8VFilter16i(y_dst, y_bps, limit, ilevel, hev_thresh);
-      VP8VFilter8i(u_dst, v_dst, uv_bps, limit, ilevel, hev_thresh);
-    }
-  }
-}
-
-// Filter the decoded macroblock row (if needed)
-static void FilterRow(const VP8Decoder* const dec) {
-  int mb_x;
-  const int mb_y = dec->thread_ctx_.mb_y_;
-  assert(dec->thread_ctx_.filter_row_);
-  for (mb_x = dec->tl_mb_x_; mb_x < dec->br_mb_x_; ++mb_x) {
-    DoFilter(dec, mb_x, mb_y);
-  }
-}
-
-//------------------------------------------------------------------------------
-// Precompute the filtering strength for each segment and each i4x4/i16x16 mode.
-
-static void PrecomputeFilterStrengths(VP8Decoder* const dec) {
-  if (dec->filter_type_ > 0) {
-    int s;
-    const VP8FilterHeader* const hdr = &dec->filter_hdr_;
-    for (s = 0; s < NUM_MB_SEGMENTS; ++s) {
-      int i4x4;
-      // First, compute the initial level
-      int base_level;
-      if (dec->segment_hdr_.use_segment_) {
-        base_level = dec->segment_hdr_.filter_strength_[s];
-        if (!dec->segment_hdr_.absolute_delta_) {
-          base_level += hdr->level_;
-        }
-      } else {
-        base_level = hdr->level_;
-      }
-      for (i4x4 = 0; i4x4 <= 1; ++i4x4) {
-        VP8FInfo* const info = &dec->fstrengths_[s][i4x4];
-        int level = base_level;
-        if (hdr->use_lf_delta_) {
-          // TODO(skal): only CURRENT is handled for now.
-          level += hdr->ref_lf_delta_[0];
-          if (i4x4) {
-            level += hdr->mode_lf_delta_[0];
-          }
-        }
-        level = (level < 0) ? 0 : (level > 63) ? 63 : level;
-        if (level > 0) {
-          int ilevel = level;
-          if (hdr->sharpness_ > 0) {
-            if (hdr->sharpness_ > 4) {
-              ilevel >>= 2;
-            } else {
-              ilevel >>= 1;
-            }
-            if (ilevel > 9 - hdr->sharpness_) {
-              ilevel = 9 - hdr->sharpness_;
-            }
-          }
-          if (ilevel < 1) ilevel = 1;
-          info->f_ilevel_ = ilevel;
-          info->f_limit_ = 2 * level + ilevel;
-          info->hev_thresh_ = (level >= 40) ? 2 : (level >= 15) ? 1 : 0;
-        } else {
-          info->f_limit_ = 0;  // no filtering
-        }
-        info->f_inner_ = i4x4;
-      }
-    }
-  }
-}
-
-//------------------------------------------------------------------------------
-// Dithering
-
-#define DITHER_AMP_TAB_SIZE 12
-static const int kQuantToDitherAmp[DITHER_AMP_TAB_SIZE] = {
-  // roughly, it's dqm->uv_mat_[1]
-  8, 7, 6, 4, 4, 2, 2, 2, 1, 1, 1, 1
-};
-
-void VP8InitDithering(const WebPDecoderOptions* const options,
-                      VP8Decoder* const dec) {
-  assert(dec != NULL);
-  if (options != NULL) {
-    const int d = options->dithering_strength;
-    const int max_amp = (1 << VP8_RANDOM_DITHER_FIX) - 1;
-    const int f = (d < 0) ? 0 : (d > 100) ? max_amp : (d * max_amp / 100);
-    if (f > 0) {
-      int s;
-      int all_amp = 0;
-      for (s = 0; s < NUM_MB_SEGMENTS; ++s) {
-        VP8QuantMatrix* const dqm = &dec->dqm_[s];
-        if (dqm->uv_quant_ < DITHER_AMP_TAB_SIZE) {
-          // TODO(skal): should we specially dither more for uv_quant_ < 0?
-          const int idx = (dqm->uv_quant_ < 0) ? 0 : dqm->uv_quant_;
-          dqm->dither_ = (f * kQuantToDitherAmp[idx]) >> 3;
-        }
-        all_amp |= dqm->dither_;
-      }
-      if (all_amp != 0) {
-        VP8InitRandom(&dec->dithering_rg_, 1.0f);
-        dec->dither_ = 1;
-      }
-    }
-#if WEBP_DECODER_ABI_VERSION > 0x0204
-    // potentially allow alpha dithering
-    dec->alpha_dithering_ = options->alpha_dithering_strength;
-    if (dec->alpha_dithering_ > 100) {
-      dec->alpha_dithering_ = 100;
-    } else if (dec->alpha_dithering_ < 0) {
-      dec->alpha_dithering_ = 0;
-    }
-#endif
-  }
-}
-
-// minimal amp that will provide a non-zero dithering effect
-#define MIN_DITHER_AMP 4
-#define DITHER_DESCALE 4
-#define DITHER_DESCALE_ROUNDER (1 << (DITHER_DESCALE - 1))
-#define DITHER_AMP_BITS 8
-#define DITHER_AMP_CENTER (1 << DITHER_AMP_BITS)
-
-static void Dither8x8(VP8Random* const rg, uint8_t* dst, int bps, int amp) {
-  int i, j;
-  for (j = 0; j < 8; ++j) {
-    for (i = 0; i < 8; ++i) {
-      // TODO: could be made faster with SSE2
-      const int bits =
-          VP8RandomBits2(rg, DITHER_AMP_BITS + 1, amp) - DITHER_AMP_CENTER;
-      // Convert to range: [-2,2] for dither=50, [-4,4] for dither=100
-      const int delta = (bits + DITHER_DESCALE_ROUNDER) >> DITHER_DESCALE;
-      const int v = (int)dst[i] + delta;
-      dst[i] = (v < 0) ? 0 : (v > 255) ? 255u : (uint8_t)v;
-    }
-    dst += bps;
-  }
-}
-
-static void DitherRow(VP8Decoder* const dec) {
-  int mb_x;
-  assert(dec->dither_);
-  for (mb_x = dec->tl_mb_x_; mb_x < dec->br_mb_x_; ++mb_x) {
-    const VP8ThreadContext* const ctx = &dec->thread_ctx_;
-    const VP8MBData* const data = ctx->mb_data_ + mb_x;
-    const int cache_id = ctx->id_;
-    const int uv_bps = dec->cache_uv_stride_;
-    if (data->dither_ >= MIN_DITHER_AMP) {
-      uint8_t* const u_dst = dec->cache_u_ + cache_id * 8 * uv_bps + mb_x * 8;
-      uint8_t* const v_dst = dec->cache_v_ + cache_id * 8 * uv_bps + mb_x * 8;
-      Dither8x8(&dec->dithering_rg_, u_dst, uv_bps, data->dither_);
-      Dither8x8(&dec->dithering_rg_, v_dst, uv_bps, data->dither_);
-    }
-  }
-}
-
-//------------------------------------------------------------------------------
-// This function is called after a row of macroblocks is finished decoding.
-// It also takes into account the following restrictions:
-//  * In case of in-loop filtering, we must hold off sending some of the bottom
-//    pixels as they are yet unfiltered. They will be when the next macroblock
-//    row is decoded. Meanwhile, we must preserve them by rotating them in the
-//    cache area. This doesn't hold for the very bottom row of the uncropped
-//    picture of course.
-//  * we must clip the remaining pixels against the cropping area. The VP8Io
-//    struct must have the following fields set correctly before calling put():
-
-#define MACROBLOCK_VPOS(mb_y)  ((mb_y) * 16)    // vertical position of a MB
-
-// Finalize and transmit a complete row. Return false in case of user-abort.
-static int FinishRow(VP8Decoder* const dec, VP8Io* const io) {
-  int ok = 1;
-  const VP8ThreadContext* const ctx = &dec->thread_ctx_;
-  const int cache_id = ctx->id_;
-  const int extra_y_rows = kFilterExtraRows[dec->filter_type_];
-  const int ysize = extra_y_rows * dec->cache_y_stride_;
-  const int uvsize = (extra_y_rows / 2) * dec->cache_uv_stride_;
-  const int y_offset = cache_id * 16 * dec->cache_y_stride_;
-  const int uv_offset = cache_id * 8 * dec->cache_uv_stride_;
-  uint8_t* const ydst = dec->cache_y_ - ysize + y_offset;
-  uint8_t* const udst = dec->cache_u_ - uvsize + uv_offset;
-  uint8_t* const vdst = dec->cache_v_ - uvsize + uv_offset;
-  const int mb_y = ctx->mb_y_;
-  const int is_first_row = (mb_y == 0);
-  const int is_last_row = (mb_y >= dec->br_mb_y_ - 1);
-
-  if (dec->mt_method_ == 2) {
-    ReconstructRow(dec, ctx);
-  }
-
-  if (ctx->filter_row_) {
-    FilterRow(dec);
-  }
-
-  if (dec->dither_) {
-    DitherRow(dec);
-  }
-
-  if (io->put != NULL) {
-    int y_start = MACROBLOCK_VPOS(mb_y);
-    int y_end = MACROBLOCK_VPOS(mb_y + 1);
-    if (!is_first_row) {
-      y_start -= extra_y_rows;
-      io->y = ydst;
-      io->u = udst;
-      io->v = vdst;
-    } else {
-      io->y = dec->cache_y_ + y_offset;
-      io->u = dec->cache_u_ + uv_offset;
-      io->v = dec->cache_v_ + uv_offset;
-    }
-
-    if (!is_last_row) {
-      y_end -= extra_y_rows;
-    }
-    if (y_end > io->crop_bottom) {
-      y_end = io->crop_bottom;    // make sure we don't overflow on last row.
-    }
-    io->a = NULL;
-    if (dec->alpha_data_ != NULL && y_start < y_end) {
-      // TODO(skal): testing presence of alpha with dec->alpha_data_ is not a
-      // good idea.
-      io->a = VP8DecompressAlphaRows(dec, y_start, y_end - y_start);
-      if (io->a == NULL) {
-        return VP8SetError(dec, VP8_STATUS_BITSTREAM_ERROR,
-                           "Could not decode alpha data.");
-      }
-    }
-    if (y_start < io->crop_top) {
-      const int delta_y = io->crop_top - y_start;
-      y_start = io->crop_top;
-      assert(!(delta_y & 1));
-      io->y += dec->cache_y_stride_ * delta_y;
-      io->u += dec->cache_uv_stride_ * (delta_y >> 1);
-      io->v += dec->cache_uv_stride_ * (delta_y >> 1);
-      if (io->a != NULL) {
-        io->a += io->width * delta_y;
-      }
-    }
-    if (y_start < y_end) {
-      io->y += io->crop_left;
-      io->u += io->crop_left >> 1;
-      io->v += io->crop_left >> 1;
-      if (io->a != NULL) {
-        io->a += io->crop_left;
-      }
-      io->mb_y = y_start - io->crop_top;
-      io->mb_w = io->crop_right - io->crop_left;
-      io->mb_h = y_end - y_start;
-      ok = io->put(io);
-    }
-  }
-  // rotate top samples if needed
-  if (cache_id + 1 == dec->num_caches_) {
-    if (!is_last_row) {
-      memcpy(dec->cache_y_ - ysize, ydst + 16 * dec->cache_y_stride_, ysize);
-      memcpy(dec->cache_u_ - uvsize, udst + 8 * dec->cache_uv_stride_, uvsize);
-      memcpy(dec->cache_v_ - uvsize, vdst + 8 * dec->cache_uv_stride_, uvsize);
-    }
-  }
-
-  return ok;
-}
-
-#undef MACROBLOCK_VPOS
-
-//------------------------------------------------------------------------------
-
-int VP8ProcessRow(VP8Decoder* const dec, VP8Io* const io) {
-  int ok = 1;
-  VP8ThreadContext* const ctx = &dec->thread_ctx_;
-  const int filter_row =
-      (dec->filter_type_ > 0) &&
-      (dec->mb_y_ >= dec->tl_mb_y_) && (dec->mb_y_ <= dec->br_mb_y_);
-  if (dec->mt_method_ == 0) {
-    // ctx->id_ and ctx->f_info_ are already set
-    ctx->mb_y_ = dec->mb_y_;
-    ctx->filter_row_ = filter_row;
-    ReconstructRow(dec, ctx);
-    ok = FinishRow(dec, io);
-  } else {
-    WebPWorker* const worker = &dec->worker_;
-    // Finish previous job *before* updating context
-    ok &= WebPGetWorkerInterface()->Sync(worker);
-    assert(worker->status_ == OK);
-    if (ok) {   // spawn a new deblocking/output job
-      ctx->io_ = *io;
-      ctx->id_ = dec->cache_id_;
-      ctx->mb_y_ = dec->mb_y_;
-      ctx->filter_row_ = filter_row;
-      if (dec->mt_method_ == 2) {  // swap macroblock data
-        VP8MBData* const tmp = ctx->mb_data_;
-        ctx->mb_data_ = dec->mb_data_;
-        dec->mb_data_ = tmp;
-      } else {
-        // perform reconstruction directly in main thread
-        ReconstructRow(dec, ctx);
-      }
-      if (filter_row) {            // swap filter info
-        VP8FInfo* const tmp = ctx->f_info_;
-        ctx->f_info_ = dec->f_info_;
-        dec->f_info_ = tmp;
-      }
-      // (reconstruct)+filter in parallel
-      WebPGetWorkerInterface()->Launch(worker);
-      if (++dec->cache_id_ == dec->num_caches_) {
-        dec->cache_id_ = 0;
-      }
-    }
-  }
-  return ok;
-}
-
-//------------------------------------------------------------------------------
-// Finish setting up the decoding parameter once user's setup() is called.
-
-VP8StatusCode VP8EnterCritical(VP8Decoder* const dec, VP8Io* const io) {
-  // Call setup() first. This may trigger additional decoding features on 'io'.
-  // Note: Afterward, we must call teardown() no matter what.
-  if (io->setup != NULL && !io->setup(io)) {
-    VP8SetError(dec, VP8_STATUS_USER_ABORT, "Frame setup failed");
-    return dec->status_;
-  }
-
-  // Disable filtering per user request
-  if (io->bypass_filtering) {
-    dec->filter_type_ = 0;
-  }
-  // TODO(skal): filter type / strength / sharpness forcing
-
-  // Define the area where we can skip in-loop filtering, in case of cropping.
-  //
-  // 'Simple' filter reads two luma samples outside of the macroblock
-  // and filters one. It doesn't filter the chroma samples. Hence, we can
-  // avoid doing the in-loop filtering before crop_top/crop_left position.
-  // For the 'Complex' filter, 3 samples are read and up to 3 are filtered.
-  // Means: there's a dependency chain that goes all the way up to the
-  // top-left corner of the picture (MB #0). We must filter all the previous
-  // macroblocks.
-  // TODO(skal): add an 'approximate_decoding' option, that won't produce
-  // a 1:1 bit-exactness for complex filtering?
-  {
-    const int extra_pixels = kFilterExtraRows[dec->filter_type_];
-    if (dec->filter_type_ == 2) {
-      // For complex filter, we need to preserve the dependency chain.
-      dec->tl_mb_x_ = 0;
-      dec->tl_mb_y_ = 0;
-    } else {
-      // For simple filter, we can filter only the cropped region.
-      // We include 'extra_pixels' on the other side of the boundary, since
-      // vertical or horizontal filtering of the previous macroblock can
-      // modify some abutting pixels.
-      dec->tl_mb_x_ = (io->crop_left - extra_pixels) >> 4;
-      dec->tl_mb_y_ = (io->crop_top - extra_pixels) >> 4;
-      if (dec->tl_mb_x_ < 0) dec->tl_mb_x_ = 0;
-      if (dec->tl_mb_y_ < 0) dec->tl_mb_y_ = 0;
-    }
-    // We need some 'extra' pixels on the right/bottom.
-    dec->br_mb_y_ = (io->crop_bottom + 15 + extra_pixels) >> 4;
-    dec->br_mb_x_ = (io->crop_right + 15 + extra_pixels) >> 4;
-    if (dec->br_mb_x_ > dec->mb_w_) {
-      dec->br_mb_x_ = dec->mb_w_;
-    }
-    if (dec->br_mb_y_ > dec->mb_h_) {
-      dec->br_mb_y_ = dec->mb_h_;
-    }
-  }
-  PrecomputeFilterStrengths(dec);
-  return VP8_STATUS_OK;
-}
-
-int VP8ExitCritical(VP8Decoder* const dec, VP8Io* const io) {
-  int ok = 1;
-  if (dec->mt_method_ > 0) {
-    ok = WebPGetWorkerInterface()->Sync(&dec->worker_);
-  }
-
-  if (io->teardown != NULL) {
-    io->teardown(io);
-  }
-  return ok;
-}
-
-//------------------------------------------------------------------------------
-// For multi-threaded decoding we need to use 3 rows of 16 pixels as delay line.
-//
-// Reason is: the deblocking filter cannot deblock the bottom horizontal edges
-// immediately, and needs to wait for first few rows of the next macroblock to
-// be decoded. Hence, deblocking is lagging behind by 4 or 8 pixels (depending
-// on strength).
-// With two threads, the vertical positions of the rows being decoded are:
-// Decode:  [ 0..15][16..31][32..47][48..63][64..79][...
-// Deblock:         [ 0..11][12..27][28..43][44..59][...
-// If we use two threads and two caches of 16 pixels, the sequence would be:
-// Decode:  [ 0..15][16..31][ 0..15!!][16..31][ 0..15][...
-// Deblock:         [ 0..11][12..27!!][-4..11][12..27][...
-// The problem occurs during row [12..15!!] that both the decoding and
-// deblocking threads are writing simultaneously.
-// With 3 cache lines, one get a safe write pattern:
-// Decode:  [ 0..15][16..31][32..47][ 0..15][16..31][32..47][0..
-// Deblock:         [ 0..11][12..27][28..43][-4..11][12..27][28...
-// Note that multi-threaded output _without_ deblocking can make use of two
-// cache lines of 16 pixels only, since there's no lagging behind. The decoding
-// and output process have non-concurrent writing:
-// Decode:  [ 0..15][16..31][ 0..15][16..31][...
-// io->put:         [ 0..15][16..31][ 0..15][...
-
-#define MT_CACHE_LINES 3
-#define ST_CACHE_LINES 1   // 1 cache row only for single-threaded case
-
-// Initialize multi/single-thread worker
-static int InitThreadContext(VP8Decoder* const dec) {
-  dec->cache_id_ = 0;
-  if (dec->mt_method_ > 0) {
-    WebPWorker* const worker = &dec->worker_;
-    if (!WebPGetWorkerInterface()->Reset(worker)) {
-      return VP8SetError(dec, VP8_STATUS_OUT_OF_MEMORY,
-                         "thread initialization failed.");
-    }
-    worker->data1 = dec;
-    worker->data2 = (void*)&dec->thread_ctx_.io_;
-    worker->hook = (WebPWorkerHook)FinishRow;
-    dec->num_caches_ =
-      (dec->filter_type_ > 0) ? MT_CACHE_LINES : MT_CACHE_LINES - 1;
-  } else {
-    dec->num_caches_ = ST_CACHE_LINES;
-  }
-  return 1;
-}
-
-int VP8GetThreadMethod(const WebPDecoderOptions* const options,
-                       const WebPHeaderStructure* const headers,
-                       int width, int height) {
-  if (options == NULL || options->use_threads == 0) {
-    return 0;
-  }
-  (void)headers;
-  (void)width;
-  (void)height;
-  assert(headers == NULL || !headers->is_lossless);
-#if defined(WEBP_USE_THREAD)
-  if (width < MIN_WIDTH_FOR_THREADS) return 0;
-  // TODO(skal): tune the heuristic further
-#if 0
-  if (height < 2 * width) return 2;
-#endif
-  return 2;
-#else   // !WEBP_USE_THREAD
-  return 0;
-#endif
-}
-
-#undef MT_CACHE_LINES
-#undef ST_CACHE_LINES
-
-//------------------------------------------------------------------------------
-// Memory setup
-
-static int AllocateMemory(VP8Decoder* const dec) {
-  const int num_caches = dec->num_caches_;
-  const int mb_w = dec->mb_w_;
-  // Note: we use 'size_t' when there's no overflow risk, uint64_t otherwise.
-  const size_t intra_pred_mode_size = 4 * mb_w * sizeof(uint8_t);
-  const size_t top_size = sizeof(VP8TopSamples) * mb_w;
-  const size_t mb_info_size = (mb_w + 1) * sizeof(VP8MB);
-  const size_t f_info_size =
-      (dec->filter_type_ > 0) ?
-          mb_w * (dec->mt_method_ > 0 ? 2 : 1) * sizeof(VP8FInfo)
-        : 0;
-  const size_t yuv_size = YUV_SIZE * sizeof(*dec->yuv_b_);
-  const size_t mb_data_size =
-      (dec->mt_method_ == 2 ? 2 : 1) * mb_w * sizeof(*dec->mb_data_);
-  const size_t cache_height = (16 * num_caches
-                            + kFilterExtraRows[dec->filter_type_]) * 3 / 2;
-  const size_t cache_size = top_size * cache_height;
-  // alpha_size is the only one that scales as width x height.
-  const uint64_t alpha_size = (dec->alpha_data_ != NULL) ?
-      (uint64_t)dec->pic_hdr_.width_ * dec->pic_hdr_.height_ : 0ULL;
-  const uint64_t needed = (uint64_t)intra_pred_mode_size
-                        + top_size + mb_info_size + f_info_size
-                        + yuv_size + mb_data_size
-                        + cache_size + alpha_size + ALIGN_MASK;
-  uint8_t* mem;
-
-  if (needed != (size_t)needed) return 0;  // check for overflow
-  if (needed > dec->mem_size_) {
-    WebPSafeFree(dec->mem_);
-    dec->mem_size_ = 0;
-    dec->mem_ = WebPSafeMalloc(needed, sizeof(uint8_t));
-    if (dec->mem_ == NULL) {
-      return VP8SetError(dec, VP8_STATUS_OUT_OF_MEMORY,
-                         "no memory during frame initialization.");
-    }
-    // down-cast is ok, thanks to WebPSafeAlloc() above.
-    dec->mem_size_ = (size_t)needed;
-  }
-
-  mem = (uint8_t*)dec->mem_;
-  dec->intra_t_ = (uint8_t*)mem;
-  mem += intra_pred_mode_size;
-
-  dec->yuv_t_ = (VP8TopSamples*)mem;
-  mem += top_size;
-
-  dec->mb_info_ = ((VP8MB*)mem) + 1;
-  mem += mb_info_size;
-
-  dec->f_info_ = f_info_size ? (VP8FInfo*)mem : NULL;
-  mem += f_info_size;
-  dec->thread_ctx_.id_ = 0;
-  dec->thread_ctx_.f_info_ = dec->f_info_;
-  if (dec->mt_method_ > 0) {
-    // secondary cache line. The deblocking process need to make use of the
-    // filtering strength from previous macroblock row, while the new ones
-    // are being decoded in parallel. We'll just swap the pointers.
-    dec->thread_ctx_.f_info_ += mb_w;
-  }
-
-  mem = (uint8_t*)((uintptr_t)(mem + ALIGN_MASK) & ~ALIGN_MASK);
-  assert((yuv_size & ALIGN_MASK) == 0);
-  dec->yuv_b_ = (uint8_t*)mem;
-  mem += yuv_size;
-
-  dec->mb_data_ = (VP8MBData*)mem;
-  dec->thread_ctx_.mb_data_ = (VP8MBData*)mem;
-  if (dec->mt_method_ == 2) {
-    dec->thread_ctx_.mb_data_ += mb_w;
-  }
-  mem += mb_data_size;
-
-  dec->cache_y_stride_ = 16 * mb_w;
-  dec->cache_uv_stride_ = 8 * mb_w;
-  {
-    const int extra_rows = kFilterExtraRows[dec->filter_type_];
-    const int extra_y = extra_rows * dec->cache_y_stride_;
-    const int extra_uv = (extra_rows / 2) * dec->cache_uv_stride_;
-    dec->cache_y_ = ((uint8_t*)mem) + extra_y;
-    dec->cache_u_ = dec->cache_y_
-                  + 16 * num_caches * dec->cache_y_stride_ + extra_uv;
-    dec->cache_v_ = dec->cache_u_
-                  + 8 * num_caches * dec->cache_uv_stride_ + extra_uv;
-    dec->cache_id_ = 0;
-  }
-  mem += cache_size;
-
-  // alpha plane
-  dec->alpha_plane_ = alpha_size ? (uint8_t*)mem : NULL;
-  mem += alpha_size;
-  assert(mem <= (uint8_t*)dec->mem_ + dec->mem_size_);
-
-  // note: left/top-info is initialized once for all.
-  memset(dec->mb_info_ - 1, 0, mb_info_size);
-  VP8InitScanline(dec);   // initialize left too.
-
-  // initialize top
-  memset(dec->intra_t_, B_DC_PRED, intra_pred_mode_size);
-
-  return 1;
-}
-
-static void InitIo(VP8Decoder* const dec, VP8Io* io) {
-  // prepare 'io'
-  io->mb_y = 0;
-  io->y = dec->cache_y_;
-  io->u = dec->cache_u_;
-  io->v = dec->cache_v_;
-  io->y_stride = dec->cache_y_stride_;
-  io->uv_stride = dec->cache_uv_stride_;
-  io->a = NULL;
-}
-
-int VP8InitFrame(VP8Decoder* const dec, VP8Io* io) {
-  if (!InitThreadContext(dec)) return 0;  // call first. Sets dec->num_caches_.
-  if (!AllocateMemory(dec)) return 0;
-  InitIo(dec, io);
-  VP8DspInit();  // Init critical function pointers and look-up tables.
-  return 1;
-}
-
-//------------------------------------------------------------------------------
-// Main reconstruction function.
-
-static const int kScan[16] = {
-  0 +  0 * BPS,  4 +  0 * BPS, 8 +  0 * BPS, 12 +  0 * BPS,
-  0 +  4 * BPS,  4 +  4 * BPS, 8 +  4 * BPS, 12 +  4 * BPS,
-  0 +  8 * BPS,  4 +  8 * BPS, 8 +  8 * BPS, 12 +  8 * BPS,
-  0 + 12 * BPS,  4 + 12 * BPS, 8 + 12 * BPS, 12 + 12 * BPS
-};
-
-static int CheckMode(int mb_x, int mb_y, int mode) {
-  if (mode == B_DC_PRED) {
-    if (mb_x == 0) {
-      return (mb_y == 0) ? B_DC_PRED_NOTOPLEFT : B_DC_PRED_NOLEFT;
-    } else {
-      return (mb_y == 0) ? B_DC_PRED_NOTOP : B_DC_PRED;
-    }
-  }
-  return mode;
-}
-
-static void Copy32b(uint8_t* dst, uint8_t* src) {
-  memcpy(dst, src, 4);
-}
-
-static WEBP_INLINE void DoTransform(uint32_t bits, const int16_t* const src,
-                                    uint8_t* const dst) {
-  switch (bits >> 30) {
-    case 3:
-      VP8Transform(src, dst, 0);
-      break;
-    case 2:
-      VP8TransformAC3(src, dst);
-      break;
-    case 1:
-      VP8TransformDC(src, dst);
-      break;
-    default:
-      break;
-  }
-}
-
-static void DoUVTransform(uint32_t bits, const int16_t* const src,
-                          uint8_t* const dst) {
-  if (bits & 0xff) {    // any non-zero coeff at all?
-    if (bits & 0xaa) {  // any non-zero AC coefficient?
-      VP8TransformUV(src, dst);   // note we don't use the AC3 variant for U/V
-    } else {
-      VP8TransformDCUV(src, dst);
-    }
-  }
-}
-
-static void ReconstructRow(const VP8Decoder* const dec,
-                           const VP8ThreadContext* ctx) {
-  int j;
-  int mb_x;
-  const int mb_y = ctx->mb_y_;
-  const int cache_id = ctx->id_;
-  uint8_t* const y_dst = dec->yuv_b_ + Y_OFF;
-  uint8_t* const u_dst = dec->yuv_b_ + U_OFF;
-  uint8_t* const v_dst = dec->yuv_b_ + V_OFF;
-  for (mb_x = 0; mb_x < dec->mb_w_; ++mb_x) {
-    const VP8MBData* const block = ctx->mb_data_ + mb_x;
-
-    // Rotate in the left samples from previously decoded block. We move four
-    // pixels at a time for alignment reason, and because of in-loop filter.
-    if (mb_x > 0) {
-      for (j = -1; j < 16; ++j) {
-        Copy32b(&y_dst[j * BPS - 4], &y_dst[j * BPS + 12]);
-      }
-      for (j = -1; j < 8; ++j) {
-        Copy32b(&u_dst[j * BPS - 4], &u_dst[j * BPS + 4]);
-        Copy32b(&v_dst[j * BPS - 4], &v_dst[j * BPS + 4]);
-      }
-    } else {
-      for (j = 0; j < 16; ++j) {
-        y_dst[j * BPS - 1] = 129;
-      }
-      for (j = 0; j < 8; ++j) {
-        u_dst[j * BPS - 1] = 129;
-        v_dst[j * BPS - 1] = 129;
-      }
-      // Init top-left sample on left column too
-      if (mb_y > 0) {
-        y_dst[-1 - BPS] = u_dst[-1 - BPS] = v_dst[-1 - BPS] = 129;
-      }
-    }
-    {
-      // bring top samples into the cache
-      VP8TopSamples* const top_yuv = dec->yuv_t_ + mb_x;
-      const int16_t* const coeffs = block->coeffs_;
-      uint32_t bits = block->non_zero_y_;
-      int n;
-
-      if (mb_y > 0) {
-        memcpy(y_dst - BPS, top_yuv[0].y, 16);
-        memcpy(u_dst - BPS, top_yuv[0].u, 8);
-        memcpy(v_dst - BPS, top_yuv[0].v, 8);
-      } else if (mb_x == 0) {
-        // we only need to do this init once at block (0,0).
-        // Afterward, it remains valid for the whole topmost row.
-        memset(y_dst - BPS - 1, 127, 16 + 4 + 1);
-        memset(u_dst - BPS - 1, 127, 8 + 1);
-        memset(v_dst - BPS - 1, 127, 8 + 1);
-      }
-
-      // predict and add residuals
-      if (block->is_i4x4_) {   // 4x4
-        uint32_t* const top_right = (uint32_t*)(y_dst - BPS + 16);
-
-        if (mb_y > 0) {
-          if (mb_x >= dec->mb_w_ - 1) {    // on rightmost border
-            memset(top_right, top_yuv[0].y[15], sizeof(*top_right));
-          } else {
-            memcpy(top_right, top_yuv[1].y, sizeof(*top_right));
-          }
-        }
-        // replicate the top-right pixels below
-        top_right[BPS] = top_right[2 * BPS] = top_right[3 * BPS] = top_right[0];
-
-        // predict and add residuals for all 4x4 blocks in turn.
-        for (n = 0; n < 16; ++n, bits <<= 2) {
-          uint8_t* const dst = y_dst + kScan[n];
-          VP8PredLuma4[block->imodes_[n]](dst);
-          DoTransform(bits, coeffs + n * 16, dst);
-        }
-      } else {    // 16x16
-        const int pred_func = CheckMode(mb_x, mb_y,
-                                        block->imodes_[0]);
-        VP8PredLuma16[pred_func](y_dst);
-        if (bits != 0) {
-          for (n = 0; n < 16; ++n, bits <<= 2) {
-            DoTransform(bits, coeffs + n * 16, y_dst + kScan[n]);
-          }
-        }
-      }
-      {
-        // Chroma
-        const uint32_t bits_uv = block->non_zero_uv_;
-        const int pred_func = CheckMode(mb_x, mb_y, block->uvmode_);
-        VP8PredChroma8[pred_func](u_dst);
-        VP8PredChroma8[pred_func](v_dst);
-        DoUVTransform(bits_uv >> 0, coeffs + 16 * 16, u_dst);
-        DoUVTransform(bits_uv >> 8, coeffs + 20 * 16, v_dst);
-      }
-
-      // stash away top samples for next block
-      if (mb_y < dec->mb_h_ - 1) {
-        memcpy(top_yuv[0].y, y_dst + 15 * BPS, 16);
-        memcpy(top_yuv[0].u, u_dst +  7 * BPS,  8);
-        memcpy(top_yuv[0].v, v_dst +  7 * BPS,  8);
-      }
-    }
-    // Transfer reconstructed samples from yuv_b_ cache to final destination.
-    {
-      const int y_offset = cache_id * 16 * dec->cache_y_stride_;
-      const int uv_offset = cache_id * 8 * dec->cache_uv_stride_;
-      uint8_t* const y_out = dec->cache_y_ + mb_x * 16 + y_offset;
-      uint8_t* const u_out = dec->cache_u_ + mb_x * 8 + uv_offset;
-      uint8_t* const v_out = dec->cache_v_ + mb_x * 8 + uv_offset;
-      for (j = 0; j < 16; ++j) {
-        memcpy(y_out + j * dec->cache_y_stride_, y_dst + j * BPS, 16);
-      }
-      for (j = 0; j < 8; ++j) {
-        memcpy(u_out + j * dec->cache_uv_stride_, u_dst + j * BPS, 8);
-        memcpy(v_out + j * dec->cache_uv_stride_, v_dst + j * BPS, 8);
-      }
-    }
-  }
-}
-
-//------------------------------------------------------------------------------
-
@@ -1,857 +0,0 @@
-// Copyright 2011 Google Inc. All Rights Reserved.
-//
-// Use of this source code is governed by a BSD-style license
-// that can be found in the COPYING file in the root of the source
-// tree. An additional intellectual property rights grant can be found
-// in the file PATENTS. All contributing project authors may
-// be found in the AUTHORS file in the root of the source tree.
-// -----------------------------------------------------------------------------
-//
-// Incremental decoding
-//
-// Author: somnath@google.com (Somnath Banerjee)
-
-#include <assert.h>
-#include <string.h>
-#include <stdlib.h>
-
-#include "./alphai.h"
-#include "./webpi.h"
-#include "./vp8i.h"
-#include "../utils/utils.h"
-
-// In append mode, buffer allocations increase as multiples of this value.
-// Needs to be a power of 2.
-#define CHUNK_SIZE 4096
-#define MAX_MB_SIZE 4096
-
-//------------------------------------------------------------------------------
-// Data structures for memory and states
-
-// Decoding states. State normally flows as:
-// WEBP_HEADER->VP8_HEADER->VP8_PARTS0->VP8_DATA->DONE for a lossy image, and
-// WEBP_HEADER->VP8L_HEADER->VP8L_DATA->DONE for a lossless image.
-// If there is any error the decoder goes into state ERROR.
-typedef enum {
-  STATE_WEBP_HEADER,  // All the data before that of the VP8/VP8L chunk.
-  STATE_VP8_HEADER,   // The VP8 Frame header (within the VP8 chunk).
-  STATE_VP8_PARTS0,
-  STATE_VP8_DATA,
-  STATE_VP8L_HEADER,
-  STATE_VP8L_DATA,
-  STATE_DONE,
-  STATE_ERROR
-} DecState;
-
-// Operating state for the MemBuffer
-typedef enum {
-  MEM_MODE_NONE = 0,
-  MEM_MODE_APPEND,
-  MEM_MODE_MAP
-} MemBufferMode;
-
-// storage for partition #0 and partial data (in a rolling fashion)
-typedef struct {
-  MemBufferMode mode_;  // Operation mode
-  size_t start_;        // start location of the data to be decoded
-  size_t end_;          // end location
-  size_t buf_size_;     // size of the allocated buffer
-  uint8_t* buf_;        // We don't own this buffer in case WebPIUpdate()
-
-  size_t part0_size_;         // size of partition #0
-  const uint8_t* part0_buf_;  // buffer to store partition #0
-} MemBuffer;
-
-struct WebPIDecoder {
-  DecState state_;         // current decoding state
-  WebPDecParams params_;   // Params to store output info
-  int is_lossless_;        // for down-casting 'dec_'.
-  void* dec_;              // either a VP8Decoder or a VP8LDecoder instance
-  VP8Io io_;
-
-  MemBuffer mem_;          // input memory buffer.
-  WebPDecBuffer output_;   // output buffer (when no external one is supplied)
-  size_t chunk_size_;      // Compressed VP8/VP8L size extracted from Header.
-
-  int last_mb_y_;          // last row reached for intra-mode decoding
-};
-
-// MB context to restore in case VP8DecodeMB() fails
-typedef struct {
-  VP8MB left_;
-  VP8MB info_;
-  VP8BitReader token_br_;
-} MBContext;
-
-//------------------------------------------------------------------------------
-// MemBuffer: incoming data handling
-
-static WEBP_INLINE size_t MemDataSize(const MemBuffer* mem) {
-  return (mem->end_ - mem->start_);
-}
-
-// Check if we need to preserve the compressed alpha data, as it may not have
-// been decoded yet.
-static int NeedCompressedAlpha(const WebPIDecoder* const idec) {
-  if (idec->state_ == STATE_WEBP_HEADER) {
-    // We haven't parsed the headers yet, so we don't know whether the image is
-    // lossy or lossless. This also means that we haven't parsed the ALPH chunk.
-    return 0;
-  }
-  if (idec->is_lossless_) {
-    return 0;  // ALPH chunk is not present for lossless images.
-  } else {
-    const VP8Decoder* const dec = (VP8Decoder*)idec->dec_;
-    assert(dec != NULL);  // Must be true as idec->state_ != STATE_WEBP_HEADER.
-    return (dec->alpha_data_ != NULL) && !dec->is_alpha_decoded_;
-  }
-}
-
-static void DoRemap(WebPIDecoder* const idec, ptrdiff_t offset) {
-  MemBuffer* const mem = &idec->mem_;
-  const uint8_t* const new_base = mem->buf_ + mem->start_;
-  // note: for VP8, setting up idec->io_ is only really needed at the beginning
-  // of the decoding, till partition #0 is complete.
-  idec->io_.data = new_base;
-  idec->io_.data_size = MemDataSize(mem);
-
-  if (idec->dec_ != NULL) {
-    if (!idec->is_lossless_) {
-      VP8Decoder* const dec = (VP8Decoder*)idec->dec_;
-      const int last_part = dec->num_parts_ - 1;
-      if (offset != 0) {
-        int p;
-        for (p = 0; p <= last_part; ++p) {
-          VP8RemapBitReader(dec->parts_ + p, offset);
-        }
-        // Remap partition #0 data pointer to new offset, but only in MAP
-        // mode (in APPEND mode, partition #0 is copied into a fixed memory).
-        if (mem->mode_ == MEM_MODE_MAP) {
-          VP8RemapBitReader(&dec->br_, offset);
-        }
-      }
-      assert(last_part >= 0);
-      dec->parts_[last_part].buf_end_ = mem->buf_ + mem->end_;
-      if (NeedCompressedAlpha(idec)) {
-        ALPHDecoder* const alph_dec = dec->alph_dec_;
-        dec->alpha_data_ += offset;
-        if (alph_dec != NULL) {
-          if (alph_dec->method_ == ALPHA_LOSSLESS_COMPRESSION) {
-            VP8LDecoder* const alph_vp8l_dec = alph_dec->vp8l_dec_;
-            assert(alph_vp8l_dec != NULL);
-            assert(dec->alpha_data_size_ >= ALPHA_HEADER_LEN);
-            VP8LBitReaderSetBuffer(&alph_vp8l_dec->br_,
-                                   dec->alpha_data_ + ALPHA_HEADER_LEN,
-                                   dec->alpha_data_size_ - ALPHA_HEADER_LEN);
-          } else {  // alph_dec->method_ == ALPHA_NO_COMPRESSION
-            // Nothing special to do in this case.
-          }
-        }
-      }
-    } else {    // Resize lossless bitreader
-      VP8LDecoder* const dec = (VP8LDecoder*)idec->dec_;
-      VP8LBitReaderSetBuffer(&dec->br_, new_base, MemDataSize(mem));
-    }
-  }
-}
-
-// Appends data to the end of MemBuffer->buf_. It expands the allocated memory
-// size if required and also updates VP8BitReader's if new memory is allocated.
-static int AppendToMemBuffer(WebPIDecoder* const idec,
-                             const uint8_t* const data, size_t data_size) {
-  VP8Decoder* const dec = (VP8Decoder*)idec->dec_;
-  MemBuffer* const mem = &idec->mem_;
-  const int need_compressed_alpha = NeedCompressedAlpha(idec);
-  const uint8_t* const old_start = mem->buf_ + mem->start_;
-  const uint8_t* const old_base =
-      need_compressed_alpha ? dec->alpha_data_ : old_start;
-  assert(mem->mode_ == MEM_MODE_APPEND);
-  if (data_size > MAX_CHUNK_PAYLOAD) {
-    // security safeguard: trying to allocate more than what the format
-    // allows for a chunk should be considered a smoke smell.
-    return 0;
-  }
-
-  if (mem->end_ + data_size > mem->buf_size_) {  // Need some free memory
-    const size_t new_mem_start = old_start - old_base;
-    const size_t current_size = MemDataSize(mem) + new_mem_start;
-    const uint64_t new_size = (uint64_t)current_size + data_size;
-    const uint64_t extra_size = (new_size + CHUNK_SIZE - 1) & ~(CHUNK_SIZE - 1);
-    uint8_t* const new_buf =
-        (uint8_t*)WebPSafeMalloc(extra_size, sizeof(*new_buf));
-    if (new_buf == NULL) return 0;
-    memcpy(new_buf, old_base, current_size);
-    WebPSafeFree(mem->buf_);
-    mem->buf_ = new_buf;
-    mem->buf_size_ = (size_t)extra_size;
-    mem->start_ = new_mem_start;
-    mem->end_ = current_size;
-  }
-
-  memcpy(mem->buf_ + mem->end_, data, data_size);
-  mem->end_ += data_size;
-  assert(mem->end_ <= mem->buf_size_);
-
-  DoRemap(idec, mem->buf_ + mem->start_ - old_start);
-  return 1;
-}
-
-static int RemapMemBuffer(WebPIDecoder* const idec,
-                          const uint8_t* const data, size_t data_size) {
-  MemBuffer* const mem = &idec->mem_;
-  const uint8_t* const old_buf = mem->buf_;
-  const uint8_t* const old_start = old_buf + mem->start_;
-  assert(mem->mode_ == MEM_MODE_MAP);
-
-  if (data_size < mem->buf_size_) return 0;  // can't remap to a shorter buffer!
-
-  mem->buf_ = (uint8_t*)data;
-  mem->end_ = mem->buf_size_ = data_size;
-
-  DoRemap(idec, mem->buf_ + mem->start_ - old_start);
-  return 1;
-}
-
-static void InitMemBuffer(MemBuffer* const mem) {
-  mem->mode_       = MEM_MODE_NONE;
-  mem->buf_        = NULL;
-  mem->buf_size_   = 0;
-  mem->part0_buf_  = NULL;
-  mem->part0_size_ = 0;
-}
-
-static void ClearMemBuffer(MemBuffer* const mem) {
-  assert(mem);
-  if (mem->mode_ == MEM_MODE_APPEND) {
-    WebPSafeFree(mem->buf_);
-    WebPSafeFree((void*)mem->part0_buf_);
-  }
-}
-
-static int CheckMemBufferMode(MemBuffer* const mem, MemBufferMode expected) {
-  if (mem->mode_ == MEM_MODE_NONE) {
-    mem->mode_ = expected;    // switch to the expected mode
-  } else if (mem->mode_ != expected) {
-    return 0;         // we mixed the modes => error
-  }
-  assert(mem->mode_ == expected);   // mode is ok
-  return 1;
-}
-
-// To be called last.
-static VP8StatusCode FinishDecoding(WebPIDecoder* const idec) {
-#if WEBP_DECODER_ABI_VERSION > 0x0203
-  const WebPDecoderOptions* const options = idec->params_.options;
-  WebPDecBuffer* const output = idec->params_.output;
-
-  idec->state_ = STATE_DONE;
-  if (options != NULL && options->flip) {
-    return WebPFlipBuffer(output);
-  }
-#endif
-  idec->state_ = STATE_DONE;
-  return VP8_STATUS_OK;
-}
-
-//------------------------------------------------------------------------------
-// Macroblock-decoding contexts
-
-static void SaveContext(const VP8Decoder* dec, const VP8BitReader* token_br,
-                        MBContext* const context) {
-  context->left_ = dec->mb_info_[-1];
-  context->info_ = dec->mb_info_[dec->mb_x_];
-  context->token_br_ = *token_br;
-}
-
-static void RestoreContext(const MBContext* context, VP8Decoder* const dec,
-                           VP8BitReader* const token_br) {
-  dec->mb_info_[-1] = context->left_;
-  dec->mb_info_[dec->mb_x_] = context->info_;
-  *token_br = context->token_br_;
-}
-
-//------------------------------------------------------------------------------
-
-static VP8StatusCode IDecError(WebPIDecoder* const idec, VP8StatusCode error) {
-  if (idec->state_ == STATE_VP8_DATA) {
-    VP8Io* const io = &idec->io_;
-    if (io->teardown != NULL) {
-      io->teardown(io);
-    }
-  }
-  idec->state_ = STATE_ERROR;
-  return error;
-}
-
-static void ChangeState(WebPIDecoder* const idec, DecState new_state,
-                        size_t consumed_bytes) {
-  MemBuffer* const mem = &idec->mem_;
-  idec->state_ = new_state;
-  mem->start_ += consumed_bytes;
-  assert(mem->start_ <= mem->end_);
-  idec->io_.data = mem->buf_ + mem->start_;
-  idec->io_.data_size = MemDataSize(mem);
-}
-
-// Headers
-static VP8StatusCode DecodeWebPHeaders(WebPIDecoder* const idec) {
-  MemBuffer* const mem = &idec->mem_;
-  const uint8_t* data = mem->buf_ + mem->start_;
-  size_t curr_size = MemDataSize(mem);
-  VP8StatusCode status;
-  WebPHeaderStructure headers;
-
-  headers.data = data;
-  headers.data_size = curr_size;
-  headers.have_all_data = 0;
-  status = WebPParseHeaders(&headers);
-  if (status == VP8_STATUS_NOT_ENOUGH_DATA) {
-    return VP8_STATUS_SUSPENDED;  // We haven't found a VP8 chunk yet.
-  } else if (status != VP8_STATUS_OK) {
-    return IDecError(idec, status);
-  }
-
-  idec->chunk_size_ = headers.compressed_size;
-  idec->is_lossless_ = headers.is_lossless;
-  if (!idec->is_lossless_) {
-    VP8Decoder* const dec = VP8New();
-    if (dec == NULL) {
-      return VP8_STATUS_OUT_OF_MEMORY;
-    }
-    idec->dec_ = dec;
-    dec->alpha_data_ = headers.alpha_data;
-    dec->alpha_data_size_ = headers.alpha_data_size;
-    ChangeState(idec, STATE_VP8_HEADER, headers.offset);
-  } else {
-    VP8LDecoder* const dec = VP8LNew();
-    if (dec == NULL) {
-      return VP8_STATUS_OUT_OF_MEMORY;
-    }
-    idec->dec_ = dec;
-    ChangeState(idec, STATE_VP8L_HEADER, headers.offset);
-  }
-  return VP8_STATUS_OK;
-}
-
-static VP8StatusCode DecodeVP8FrameHeader(WebPIDecoder* const idec) {
-  const uint8_t* data = idec->mem_.buf_ + idec->mem_.start_;
-  const size_t curr_size = MemDataSize(&idec->mem_);
-  int width, height;
-  uint32_t bits;
-
-  if (curr_size < VP8_FRAME_HEADER_SIZE) {
-    // Not enough data bytes to extract VP8 Frame Header.
-    return VP8_STATUS_SUSPENDED;
-  }
-  if (!VP8GetInfo(data, curr_size, idec->chunk_size_, &width, &height)) {
-    return IDecError(idec, VP8_STATUS_BITSTREAM_ERROR);
-  }
-
-  bits = data[0] | (data[1] << 8) | (data[2] << 16);
-  idec->mem_.part0_size_ = (bits >> 5) + VP8_FRAME_HEADER_SIZE;
-
-  idec->io_.data = data;
-  idec->io_.data_size = curr_size;
-  idec->state_ = STATE_VP8_PARTS0;
-  return VP8_STATUS_OK;
-}
-
-// Partition #0
-static int CopyParts0Data(WebPIDecoder* const idec) {
-  VP8Decoder* const dec = (VP8Decoder*)idec->dec_;
-  VP8BitReader* const br = &dec->br_;
-  const size_t psize = br->buf_end_ - br->buf_;
-  MemBuffer* const mem = &idec->mem_;
-  assert(!idec->is_lossless_);
-  assert(mem->part0_buf_ == NULL);
-  assert(psize > 0);
-  assert(psize <= mem->part0_size_);  // Format limit: no need for runtime check
-  if (mem->mode_ == MEM_MODE_APPEND) {
-    // We copy and grab ownership of the partition #0 data.
-    uint8_t* const part0_buf = (uint8_t*)WebPSafeMalloc(1ULL, psize);
-    if (part0_buf == NULL) {
-      return 0;
-    }
-    memcpy(part0_buf, br->buf_, psize);
-    mem->part0_buf_ = part0_buf;
-    br->buf_ = part0_buf;
-    br->buf_end_ = part0_buf + psize;
-  } else {
-    // Else: just keep pointers to the partition #0's data in dec_->br_.
-  }
-  mem->start_ += psize;
-  return 1;
-}
-
-static VP8StatusCode DecodePartition0(WebPIDecoder* const idec) {
-  VP8Decoder* const dec = (VP8Decoder*)idec->dec_;
-  VP8Io* const io = &idec->io_;
-  const WebPDecParams* const params = &idec->params_;
-  WebPDecBuffer* const output = params->output;
-
-  // Wait till we have enough data for the whole partition #0
-  if (MemDataSize(&idec->mem_) < idec->mem_.part0_size_) {
-    return VP8_STATUS_SUSPENDED;
-  }
-
-  if (!VP8GetHeaders(dec, io)) {
-    const VP8StatusCode status = dec->status_;
-    if (status == VP8_STATUS_SUSPENDED ||
-        status == VP8_STATUS_NOT_ENOUGH_DATA) {
-      // treating NOT_ENOUGH_DATA as SUSPENDED state
-      return VP8_STATUS_SUSPENDED;
-    }
-    return IDecError(idec, status);
-  }
-
-  // Allocate/Verify output buffer now
-  dec->status_ = WebPAllocateDecBuffer(io->width, io->height, params->options,
-                                       output);
-  if (dec->status_ != VP8_STATUS_OK) {
-    return IDecError(idec, dec->status_);
-  }
-  // This change must be done before calling VP8InitFrame()
-  dec->mt_method_ = VP8GetThreadMethod(params->options, NULL,
-                                       io->width, io->height);
-  VP8InitDithering(params->options, dec);
-  if (!CopyParts0Data(idec)) {
-    return IDecError(idec, VP8_STATUS_OUT_OF_MEMORY);
-  }
-
-  // Finish setting up the decoding parameters. Will call io->setup().
-  if (VP8EnterCritical(dec, io) != VP8_STATUS_OK) {
-    return IDecError(idec, dec->status_);
-  }
-
-  // Note: past this point, teardown() must always be called
-  // in case of error.
-  idec->state_ = STATE_VP8_DATA;
-  // Allocate memory and prepare everything.
-  if (!VP8InitFrame(dec, io)) {
-    return IDecError(idec, dec->status_);
-  }
-  return VP8_STATUS_OK;
-}
-
-// Remaining partitions
-static VP8StatusCode DecodeRemaining(WebPIDecoder* const idec) {
-  VP8Decoder* const dec = (VP8Decoder*)idec->dec_;
-  VP8Io* const io = &idec->io_;
-
-  assert(dec->ready_);
-  for (; dec->mb_y_ < dec->mb_h_; ++dec->mb_y_) {
-    if (idec->last_mb_y_ != dec->mb_y_) {
-      if (!VP8ParseIntraModeRow(&dec->br_, dec)) {
-        // note: normally, error shouldn't occur since we already have the whole
-        // partition0 available here in DecodeRemaining(). Reaching EOF while
-        // reading intra modes really means a BITSTREAM_ERROR.
-        return IDecError(idec, VP8_STATUS_BITSTREAM_ERROR);
-      }
-      idec->last_mb_y_ = dec->mb_y_;
-    }
-    for (; dec->mb_x_ < dec->mb_w_; ++dec->mb_x_) {
-      VP8BitReader* const token_br =
-          &dec->parts_[dec->mb_y_ & (dec->num_parts_ - 1)];
-      MBContext context;
-      SaveContext(dec, token_br, &context);
-      if (!VP8DecodeMB(dec, token_br)) {
-        // We shouldn't fail when MAX_MB data was available
-        if (dec->num_parts_ == 1 && MemDataSize(&idec->mem_) > MAX_MB_SIZE) {
-          return IDecError(idec, VP8_STATUS_BITSTREAM_ERROR);
-        }
-        RestoreContext(&context, dec, token_br);
-        return VP8_STATUS_SUSPENDED;
-      }
-      // Release buffer only if there is only one partition
-      if (dec->num_parts_ == 1) {
-        idec->mem_.start_ = token_br->buf_ - idec->mem_.buf_;
-        assert(idec->mem_.start_ <= idec->mem_.end_);
-      }
-    }
-    VP8InitScanline(dec);   // Prepare for next scanline
-
-    // Reconstruct, filter and emit the row.
-    if (!VP8ProcessRow(dec, io)) {
-      return IDecError(idec, VP8_STATUS_USER_ABORT);
-    }
-  }
-  // Synchronize the thread and check for errors.
-  if (!VP8ExitCritical(dec, io)) {
-    return IDecError(idec, VP8_STATUS_USER_ABORT);
-  }
-  dec->ready_ = 0;
-  return FinishDecoding(idec);
-}
-
-static VP8StatusCode ErrorStatusLossless(WebPIDecoder* const idec,
-                                         VP8StatusCode status) {
-  if (status == VP8_STATUS_SUSPENDED || status == VP8_STATUS_NOT_ENOUGH_DATA) {
-    return VP8_STATUS_SUSPENDED;
-  }
-  return IDecError(idec, status);
-}
-
-static VP8StatusCode DecodeVP8LHeader(WebPIDecoder* const idec) {
-  VP8Io* const io = &idec->io_;
-  VP8LDecoder* const dec = (VP8LDecoder*)idec->dec_;
-  const WebPDecParams* const params = &idec->params_;
-  WebPDecBuffer* const output = params->output;
-  size_t curr_size = MemDataSize(&idec->mem_);
-  assert(idec->is_lossless_);
-
-  // Wait until there's enough data for decoding header.
-  if (curr_size < (idec->chunk_size_ >> 3)) {
-    return VP8_STATUS_SUSPENDED;
-  }
-  if (!VP8LDecodeHeader(dec, io)) {
-    return ErrorStatusLossless(idec, dec->status_);
-  }
-  // Allocate/verify output buffer now.
-  dec->status_ = WebPAllocateDecBuffer(io->width, io->height, params->options,
-                                       output);
-  if (dec->status_ != VP8_STATUS_OK) {
-    return IDecError(idec, dec->status_);
-  }
-
-  idec->state_ = STATE_VP8L_DATA;
-  return VP8_STATUS_OK;
-}
-
-static VP8StatusCode DecodeVP8LData(WebPIDecoder* const idec) {
-  VP8LDecoder* const dec = (VP8LDecoder*)idec->dec_;
-  const size_t curr_size = MemDataSize(&idec->mem_);
-  assert(idec->is_lossless_);
-
-  // At present Lossless decoder can't decode image incrementally. So wait till
-  // all the image data is aggregated before image can be decoded.
-  if (curr_size < idec->chunk_size_) {
-    return VP8_STATUS_SUSPENDED;
-  }
-
-  if (!VP8LDecodeImage(dec)) {
-    // The decoding is called after all the data-bytes are aggregated. Change
-    // the error to VP8_BITSTREAM_ERROR in case lossless decoder fails to decode
-    // all the pixels (VP8_STATUS_SUSPENDED).
-    if (dec->status_ == VP8_STATUS_SUSPENDED) {
-      dec->status_ = VP8_STATUS_BITSTREAM_ERROR;
-    }
-    return ErrorStatusLossless(idec, dec->status_);
-  }
-
-  return FinishDecoding(idec);
-}
-
-  // Main decoding loop
-static VP8StatusCode IDecode(WebPIDecoder* idec) {
-  VP8StatusCode status = VP8_STATUS_SUSPENDED;
-
-  if (idec->state_ == STATE_WEBP_HEADER) {
-    status = DecodeWebPHeaders(idec);
-  } else {
-    if (idec->dec_ == NULL) {
-      return VP8_STATUS_SUSPENDED;    // can't continue if we have no decoder.
-    }
-  }
-  if (idec->state_ == STATE_VP8_HEADER) {
-    status = DecodeVP8FrameHeader(idec);
-  }
-  if (idec->state_ == STATE_VP8_PARTS0) {
-    status = DecodePartition0(idec);
-  }
-  if (idec->state_ == STATE_VP8_DATA) {
-    status = DecodeRemaining(idec);
-  }
-  if (idec->state_ == STATE_VP8L_HEADER) {
-    status = DecodeVP8LHeader(idec);
-  }
-  if (idec->state_ == STATE_VP8L_DATA) {
-    status = DecodeVP8LData(idec);
-  }
-  return status;
-}
-
-//------------------------------------------------------------------------------
-// Public functions
-
-WebPIDecoder* WebPINewDecoder(WebPDecBuffer* output_buffer) {
-  WebPIDecoder* idec = (WebPIDecoder*)WebPSafeCalloc(1ULL, sizeof(*idec));
-  if (idec == NULL) {
-    return NULL;
-  }
-
-  idec->state_ = STATE_WEBP_HEADER;
-  idec->chunk_size_ = 0;
-
-  idec->last_mb_y_ = -1;
-
-  InitMemBuffer(&idec->mem_);
-  WebPInitDecBuffer(&idec->output_);
-  VP8InitIo(&idec->io_);
-
-  WebPResetDecParams(&idec->params_);
-  idec->params_.output = (output_buffer != NULL) ? output_buffer
-                                                 : &idec->output_;
-  WebPInitCustomIo(&idec->params_, &idec->io_);  // Plug the I/O functions.
-
-  return idec;
-}
-
-WebPIDecoder* WebPIDecode(const uint8_t* data, size_t data_size,
-                          WebPDecoderConfig* config) {
-  WebPIDecoder* idec;
-
-  // Parse the bitstream's features, if requested:
-  if (data != NULL && data_size > 0 && config != NULL) {
-    if (WebPGetFeatures(data, data_size, &config->input) != VP8_STATUS_OK) {
-      return NULL;
-    }
-  }
-  // Create an instance of the incremental decoder
-  idec = WebPINewDecoder(config ? &config->output : NULL);
-  if (idec == NULL) {
-    return NULL;
-  }
-  // Finish initialization
-  if (config != NULL) {
-    idec->params_.options = &config->options;
-  }
-  return idec;
-}
-
-void WebPIDelete(WebPIDecoder* idec) {
-  if (idec == NULL) return;
-  if (idec->dec_ != NULL) {
-    if (!idec->is_lossless_) {
-      if (idec->state_ == STATE_VP8_DATA) {
-        // Synchronize the thread, clean-up and check for errors.
-        VP8ExitCritical((VP8Decoder*)idec->dec_, &idec->io_);
-      }
-      VP8Delete((VP8Decoder*)idec->dec_);
-    } else {
-      VP8LDelete((VP8LDecoder*)idec->dec_);
-    }
-  }
-  ClearMemBuffer(&idec->mem_);
-  WebPFreeDecBuffer(&idec->output_);
-  WebPSafeFree(idec);
-}
-
-//------------------------------------------------------------------------------
-// Wrapper toward WebPINewDecoder
-
-WebPIDecoder* WebPINewRGB(WEBP_CSP_MODE mode, uint8_t* output_buffer,
-                          size_t output_buffer_size, int output_stride) {
-  const int is_external_memory = (output_buffer != NULL);
-  WebPIDecoder* idec;
-
-  if (mode >= MODE_YUV) return NULL;
-  if (!is_external_memory) {    // Overwrite parameters to sane values.
-    output_buffer_size = 0;
-    output_stride = 0;
-  } else {  // A buffer was passed. Validate the other params.
-    if (output_stride == 0 || output_buffer_size == 0) {
-      return NULL;   // invalid parameter.
-    }
-  }
-  idec = WebPINewDecoder(NULL);
-  if (idec == NULL) return NULL;
-  idec->output_.colorspace = mode;
-  idec->output_.is_external_memory = is_external_memory;
-  idec->output_.u.RGBA.rgba = output_buffer;
-  idec->output_.u.RGBA.stride = output_stride;
-  idec->output_.u.RGBA.size = output_buffer_size;
-  return idec;
-}
-
-WebPIDecoder* WebPINewYUVA(uint8_t* luma, size_t luma_size, int luma_stride,
-                           uint8_t* u, size_t u_size, int u_stride,
-                           uint8_t* v, size_t v_size, int v_stride,
-                           uint8_t* a, size_t a_size, int a_stride) {
-  const int is_external_memory = (luma != NULL);
-  WebPIDecoder* idec;
-  WEBP_CSP_MODE colorspace;
-
-  if (!is_external_memory) {    // Overwrite parameters to sane values.
-    luma_size = u_size = v_size = a_size = 0;
-    luma_stride = u_stride = v_stride = a_stride = 0;
-    u = v = a = NULL;
-    colorspace = MODE_YUVA;
-  } else {  // A luma buffer was passed. Validate the other parameters.
-    if (u == NULL || v == NULL) return NULL;
-    if (luma_size == 0 || u_size == 0 || v_size == 0) return NULL;
-    if (luma_stride == 0 || u_stride == 0 || v_stride == 0) return NULL;
-    if (a != NULL) {
-      if (a_size == 0 || a_stride == 0) return NULL;
-    }
-    colorspace = (a == NULL) ? MODE_YUV : MODE_YUVA;
-  }
-
-  idec = WebPINewDecoder(NULL);
-  if (idec == NULL) return NULL;
-
-  idec->output_.colorspace = colorspace;
-  idec->output_.is_external_memory = is_external_memory;
-  idec->output_.u.YUVA.y = luma;
-  idec->output_.u.YUVA.y_stride = luma_stride;
-  idec->output_.u.YUVA.y_size = luma_size;
-  idec->output_.u.YUVA.u = u;
-  idec->output_.u.YUVA.u_stride = u_stride;
-  idec->output_.u.YUVA.u_size = u_size;
-  idec->output_.u.YUVA.v = v;
-  idec->output_.u.YUVA.v_stride = v_stride;
-  idec->output_.u.YUVA.v_size = v_size;
-  idec->output_.u.YUVA.a = a;
-  idec->output_.u.YUVA.a_stride = a_stride;
-  idec->output_.u.YUVA.a_size = a_size;
-  return idec;
-}
-
-WebPIDecoder* WebPINewYUV(uint8_t* luma, size_t luma_size, int luma_stride,
-                          uint8_t* u, size_t u_size, int u_stride,
-                          uint8_t* v, size_t v_size, int v_stride) {
-  return WebPINewYUVA(luma, luma_size, luma_stride,
-                      u, u_size, u_stride,
-                      v, v_size, v_stride,
-                      NULL, 0, 0);
-}
-
-//------------------------------------------------------------------------------
-
-static VP8StatusCode IDecCheckStatus(const WebPIDecoder* const idec) {
-  assert(idec);
-  if (idec->state_ == STATE_ERROR) {
-    return VP8_STATUS_BITSTREAM_ERROR;
-  }
-  if (idec->state_ == STATE_DONE) {
-    return VP8_STATUS_OK;
-  }
-  return VP8_STATUS_SUSPENDED;
-}
-
-VP8StatusCode WebPIAppend(WebPIDecoder* idec,
-                          const uint8_t* data, size_t data_size) {
-  VP8StatusCode status;
-  if (idec == NULL || data == NULL) {
-    return VP8_STATUS_INVALID_PARAM;
-  }
-  status = IDecCheckStatus(idec);
-  if (status != VP8_STATUS_SUSPENDED) {
-    return status;
-  }
-  // Check mixed calls between RemapMemBuffer and AppendToMemBuffer.
-  if (!CheckMemBufferMode(&idec->mem_, MEM_MODE_APPEND)) {
-    return VP8_STATUS_INVALID_PARAM;
-  }
-  // Append data to memory buffer
-  if (!AppendToMemBuffer(idec, data, data_size)) {
-    return VP8_STATUS_OUT_OF_MEMORY;
-  }
-  return IDecode(idec);
-}
-
-VP8StatusCode WebPIUpdate(WebPIDecoder* idec,
-                          const uint8_t* data, size_t data_size) {
-  VP8StatusCode status;
-  if (idec == NULL || data == NULL) {
-    return VP8_STATUS_INVALID_PARAM;
-  }
-  status = IDecCheckStatus(idec);
-  if (status != VP8_STATUS_SUSPENDED) {
-    return status;
-  }
-  // Check mixed calls between RemapMemBuffer and AppendToMemBuffer.
-  if (!CheckMemBufferMode(&idec->mem_, MEM_MODE_MAP)) {
-    return VP8_STATUS_INVALID_PARAM;
-  }
-  // Make the memory buffer point to the new buffer
-  if (!RemapMemBuffer(idec, data, data_size)) {
-    return VP8_STATUS_INVALID_PARAM;
-  }
-  return IDecode(idec);
-}
-
-//------------------------------------------------------------------------------
-
-static const WebPDecBuffer* GetOutputBuffer(const WebPIDecoder* const idec) {
-  if (idec == NULL || idec->dec_ == NULL) {
-    return NULL;
-  }
-  if (idec->state_ <= STATE_VP8_PARTS0) {
-    return NULL;
-  }
-  return idec->params_.output;
-}
-
-const WebPDecBuffer* WebPIDecodedArea(const WebPIDecoder* idec,
-                                      int* left, int* top,
-                                      int* width, int* height) {
-  const WebPDecBuffer* const src = GetOutputBuffer(idec);
-  if (left != NULL) *left = 0;
-  if (top != NULL) *top = 0;
-  // TODO(skal): later include handling of rotations.
-  if (src) {
-    if (width != NULL) *width = src->width;
-    if (height != NULL) *height = idec->params_.last_y;
-  } else {
-    if (width != NULL) *width = 0;
-    if (height != NULL) *height = 0;
-  }
-  return src;
-}
-
-uint8_t* WebPIDecGetRGB(const WebPIDecoder* idec, int* last_y,
-                        int* width, int* height, int* stride) {
-  const WebPDecBuffer* const src = GetOutputBuffer(idec);
-  if (src == NULL) return NULL;
-  if (src->colorspace >= MODE_YUV) {
-    return NULL;
-  }
-
-  if (last_y != NULL) *last_y = idec->params_.last_y;
-  if (width != NULL) *width = src->width;
-  if (height != NULL) *height = src->height;
-  if (stride != NULL) *stride = src->u.RGBA.stride;
-
-  return src->u.RGBA.rgba;
-}
-
-uint8_t* WebPIDecGetYUVA(const WebPIDecoder* idec, int* last_y,
-                         uint8_t** u, uint8_t** v, uint8_t** a,
-                         int* width, int* height,
-                         int* stride, int* uv_stride, int* a_stride) {
-  const WebPDecBuffer* const src = GetOutputBuffer(idec);
-  if (src == NULL) return NULL;
-  if (src->colorspace < MODE_YUV) {
-    return NULL;
-  }
-
-  if (last_y != NULL) *last_y = idec->params_.last_y;
-  if (u != NULL) *u = src->u.YUVA.u;
-  if (v != NULL) *v = src->u.YUVA.v;
-  if (a != NULL) *a = src->u.YUVA.a;
-  if (width != NULL) *width = src->width;
-  if (height != NULL) *height = src->height;
-  if (stride != NULL) *stride = src->u.YUVA.y_stride;
-  if (uv_stride != NULL) *uv_stride = src->u.YUVA.u_stride;
-  if (a_stride != NULL) *a_stride = src->u.YUVA.a_stride;
-
-  return src->u.YUVA.y;
-}
-
-int WebPISetIOHooks(WebPIDecoder* const idec,
-                    VP8IoPutHook put,
-                    VP8IoSetupHook setup,
-                    VP8IoTeardownHook teardown,
-                    void* user_data) {
-  if (idec == NULL || idec->state_ > STATE_WEBP_HEADER) {
-    return 0;
-  }
-
-  idec->io_.put = put;
-  idec->io_.setup = setup;
-  idec->io_.teardown = teardown;
-  idec->io_.opaque = user_data;
-
-  return 1;
-}
-
@@ -1,648 +0,0 @@
-// Copyright 2011 Google Inc. All Rights Reserved.
-//
-// Use of this source code is governed by a BSD-style license
-// that can be found in the COPYING file in the root of the source
-// tree. An additional intellectual property rights grant can be found
-// in the file PATENTS. All contributing project authors may
-// be found in the AUTHORS file in the root of the source tree.
-// -----------------------------------------------------------------------------
-//
-// functions for sample output.
-//
-// Author: Skal (pascal.massimino@gmail.com)
-
-#include <assert.h>
-#include <stdlib.h>
-#include "../dec/vp8i.h"
-#include "./webpi.h"
-#include "../dsp/dsp.h"
-#include "../dsp/yuv.h"
-#include "../utils/utils.h"
-
-//------------------------------------------------------------------------------
-// Main YUV<->RGB conversion functions
-
-static int EmitYUV(const VP8Io* const io, WebPDecParams* const p) {
-  WebPDecBuffer* output = p->output;
-  const WebPYUVABuffer* const buf = &output->u.YUVA;
-  uint8_t* const y_dst = buf->y + io->mb_y * buf->y_stride;
-  uint8_t* const u_dst = buf->u + (io->mb_y >> 1) * buf->u_stride;
-  uint8_t* const v_dst = buf->v + (io->mb_y >> 1) * buf->v_stride;
-  const int mb_w = io->mb_w;
-  const int mb_h = io->mb_h;
-  const int uv_w = (mb_w + 1) / 2;
-  const int uv_h = (mb_h + 1) / 2;
-  int j;
-  for (j = 0; j < mb_h; ++j) {
-    memcpy(y_dst + j * buf->y_stride, io->y + j * io->y_stride, mb_w);
-  }
-  for (j = 0; j < uv_h; ++j) {
-    memcpy(u_dst + j * buf->u_stride, io->u + j * io->uv_stride, uv_w);
-    memcpy(v_dst + j * buf->v_stride, io->v + j * io->uv_stride, uv_w);
-  }
-  return io->mb_h;
-}
-
-// Point-sampling U/V sampler.
-static int EmitSampledRGB(const VP8Io* const io, WebPDecParams* const p) {
-  WebPDecBuffer* const output = p->output;
-  WebPRGBABuffer* const buf = &output->u.RGBA;
-  uint8_t* const dst = buf->rgba + io->mb_y * buf->stride;
-  WebPSamplerProcessPlane(io->y, io->y_stride,
-                          io->u, io->v, io->uv_stride,
-                          dst, buf->stride, io->mb_w, io->mb_h,
-                          WebPSamplers[output->colorspace]);
-  return io->mb_h;
-}
-
-//------------------------------------------------------------------------------
-// YUV444 -> RGB conversion
-
-#if 0   // TODO(skal): this is for future rescaling.
-static int EmitRGB(const VP8Io* const io, WebPDecParams* const p) {
-  WebPDecBuffer* output = p->output;
-  const WebPRGBABuffer* const buf = &output->u.RGBA;
-  uint8_t* dst = buf->rgba + io->mb_y * buf->stride;
-  const uint8_t* y_src = io->y;
-  const uint8_t* u_src = io->u;
-  const uint8_t* v_src = io->v;
-  const WebPYUV444Converter convert = WebPYUV444Converters[output->colorspace];
-  const int mb_w = io->mb_w;
-  const int last = io->mb_h;
-  int j;
-  for (j = 0; j < last; ++j) {
-    convert(y_src, u_src, v_src, dst, mb_w);
-    y_src += io->y_stride;
-    u_src += io->uv_stride;
-    v_src += io->uv_stride;
-    dst += buf->stride;
-  }
-  return io->mb_h;
-}
-#endif
-
-//------------------------------------------------------------------------------
-// Fancy upsampling
-
-#ifdef FANCY_UPSAMPLING
-static int EmitFancyRGB(const VP8Io* const io, WebPDecParams* const p) {
-  int num_lines_out = io->mb_h;   // a priori guess
-  const WebPRGBABuffer* const buf = &p->output->u.RGBA;
-  uint8_t* dst = buf->rgba + io->mb_y * buf->stride;
-  WebPUpsampleLinePairFunc upsample = WebPUpsamplers[p->output->colorspace];
-  const uint8_t* cur_y = io->y;
-  const uint8_t* cur_u = io->u;
-  const uint8_t* cur_v = io->v;
-  const uint8_t* top_u = p->tmp_u;
-  const uint8_t* top_v = p->tmp_v;
-  int y = io->mb_y;
-  const int y_end = io->mb_y + io->mb_h;
-  const int mb_w = io->mb_w;
-  const int uv_w = (mb_w + 1) / 2;
-
-  if (y == 0) {
-    // First line is special cased. We mirror the u/v samples at boundary.
-    upsample(cur_y, NULL, cur_u, cur_v, cur_u, cur_v, dst, NULL, mb_w);
-  } else {
-    // We can finish the left-over line from previous call.
-    upsample(p->tmp_y, cur_y, top_u, top_v, cur_u, cur_v,
-             dst - buf->stride, dst, mb_w);
-    ++num_lines_out;
-  }
-  // Loop over each output pairs of row.
-  for (; y + 2 < y_end; y += 2) {
-    top_u = cur_u;
-    top_v = cur_v;
-    cur_u += io->uv_stride;
-    cur_v += io->uv_stride;
-    dst += 2 * buf->stride;
-    cur_y += 2 * io->y_stride;
-    upsample(cur_y - io->y_stride, cur_y,
-             top_u, top_v, cur_u, cur_v,
-             dst - buf->stride, dst, mb_w);
-  }
-  // move to last row
-  cur_y += io->y_stride;
-  if (io->crop_top + y_end < io->crop_bottom) {
-    // Save the unfinished samples for next call (as we're not done yet).
-    memcpy(p->tmp_y, cur_y, mb_w * sizeof(*p->tmp_y));
-    memcpy(p->tmp_u, cur_u, uv_w * sizeof(*p->tmp_u));
-    memcpy(p->tmp_v, cur_v, uv_w * sizeof(*p->tmp_v));
-    // The fancy upsampler leaves a row unfinished behind
-    // (except for the very last row)
-    num_lines_out--;
-  } else {
-    // Process the very last row of even-sized picture
-    if (!(y_end & 1)) {
-      upsample(cur_y, NULL, cur_u, cur_v, cur_u, cur_v,
-               dst + buf->stride, NULL, mb_w);
-    }
-  }
-  return num_lines_out;
-}
-
-#endif    /* FANCY_UPSAMPLING */
-
-//------------------------------------------------------------------------------
-
-static int EmitAlphaYUV(const VP8Io* const io, WebPDecParams* const p) {
-  const uint8_t* alpha = io->a;
-  const WebPYUVABuffer* const buf = &p->output->u.YUVA;
-  const int mb_w = io->mb_w;
-  const int mb_h = io->mb_h;
-  uint8_t* dst = buf->a + io->mb_y * buf->a_stride;
-  int j;
-
-  if (alpha != NULL) {
-    for (j = 0; j < mb_h; ++j) {
-      memcpy(dst, alpha, mb_w * sizeof(*dst));
-      alpha += io->width;
-      dst += buf->a_stride;
-    }
-  } else if (buf->a != NULL) {
-    // the user requested alpha, but there is none, set it to opaque.
-    for (j = 0; j < mb_h; ++j) {
-      memset(dst, 0xff, mb_w * sizeof(*dst));
-      dst += buf->a_stride;
-    }
-  }
-  return 0;
-}
-
-static int GetAlphaSourceRow(const VP8Io* const io,
-                             const uint8_t** alpha, int* const num_rows) {
-  int start_y = io->mb_y;
-  *num_rows = io->mb_h;
-
-  // Compensate for the 1-line delay of the fancy upscaler.
-  // This is similar to EmitFancyRGB().
-  if (io->fancy_upsampling) {
-    if (start_y == 0) {
-      // We don't process the last row yet. It'll be done during the next call.
-      --*num_rows;
-    } else {
-      --start_y;
-      // Fortunately, *alpha data is persistent, so we can go back
-      // one row and finish alpha blending, now that the fancy upscaler
-      // completed the YUV->RGB interpolation.
-      *alpha -= io->width;
-    }
-    if (io->crop_top + io->mb_y + io->mb_h == io->crop_bottom) {
-      // If it's the very last call, we process all the remaining rows!
-      *num_rows = io->crop_bottom - io->crop_top - start_y;
-    }
-  }
-  return start_y;
-}
-
-static int EmitAlphaRGB(const VP8Io* const io, WebPDecParams* const p) {
-  const uint8_t* alpha = io->a;
-  if (alpha != NULL) {
-    const int mb_w = io->mb_w;
-    const WEBP_CSP_MODE colorspace = p->output->colorspace;
-    const int alpha_first =
-        (colorspace == MODE_ARGB || colorspace == MODE_Argb);
-    const WebPRGBABuffer* const buf = &p->output->u.RGBA;
-    int num_rows;
-    const int start_y = GetAlphaSourceRow(io, &alpha, &num_rows);
-    uint8_t* const base_rgba = buf->rgba + start_y * buf->stride;
-    uint8_t* dst = base_rgba + (alpha_first ? 0 : 3);
-    uint32_t alpha_mask = 0xff;
-    int i, j;
-
-    for (j = 0; j < num_rows; ++j) {
-      for (i = 0; i < mb_w; ++i) {
-        const uint32_t alpha_value = alpha[i];
-        dst[4 * i] = alpha_value;
-        alpha_mask &= alpha_value;
-      }
-      alpha += io->width;
-      dst += buf->stride;
-    }
-    // alpha_mask is < 0xff if there's non-trivial alpha to premultiply with.
-    if (alpha_mask != 0xff && WebPIsPremultipliedMode(colorspace)) {
-      WebPApplyAlphaMultiply(base_rgba, alpha_first,
-                             mb_w, num_rows, buf->stride);
-    }
-  }
-  return 0;
-}
-
-static int EmitAlphaRGBA4444(const VP8Io* const io, WebPDecParams* const p) {
-  const uint8_t* alpha = io->a;
-  if (alpha != NULL) {
-    const int mb_w = io->mb_w;
-    const WEBP_CSP_MODE colorspace = p->output->colorspace;
-    const WebPRGBABuffer* const buf = &p->output->u.RGBA;
-    int num_rows;
-    const int start_y = GetAlphaSourceRow(io, &alpha, &num_rows);
-    uint8_t* const base_rgba = buf->rgba + start_y * buf->stride;
-#ifdef WEBP_SWAP_16BIT_CSP
-    uint8_t* alpha_dst = base_rgba;
-#else
-    uint8_t* alpha_dst = base_rgba + 1;
-#endif
-    uint32_t alpha_mask = 0x0f;
-    int i, j;
-
-    for (j = 0; j < num_rows; ++j) {
-      for (i = 0; i < mb_w; ++i) {
-        // Fill in the alpha value (converted to 4 bits).
-        const uint32_t alpha_value = alpha[i] >> 4;
-        alpha_dst[2 * i] = (alpha_dst[2 * i] & 0xf0) | alpha_value;
-        alpha_mask &= alpha_value;
-      }
-      alpha += io->width;
-      alpha_dst += buf->stride;
-    }
-    if (alpha_mask != 0x0f && WebPIsPremultipliedMode(colorspace)) {
-      WebPApplyAlphaMultiply4444(base_rgba, mb_w, num_rows, buf->stride);
-    }
-  }
-  return 0;
-}
-
-//------------------------------------------------------------------------------
-// YUV rescaling (no final RGB conversion needed)
-
-static int Rescale(const uint8_t* src, int src_stride,
-                   int new_lines, WebPRescaler* const wrk) {
-  int num_lines_out = 0;
-  while (new_lines > 0) {    // import new contributions of source rows.
-    const int lines_in = WebPRescalerImport(wrk, new_lines, src, src_stride);
-    src += lines_in * src_stride;
-    new_lines -= lines_in;
-    num_lines_out += WebPRescalerExport(wrk);    // emit output row(s)
-  }
-  return num_lines_out;
-}
-
-static int EmitRescaledYUV(const VP8Io* const io, WebPDecParams* const p) {
-  const int mb_h = io->mb_h;
-  const int uv_mb_h = (mb_h + 1) >> 1;
-  WebPRescaler* const scaler = &p->scaler_y;
-  int num_lines_out = 0;
-  if (WebPIsAlphaMode(p->output->colorspace) && io->a != NULL) {
-    // Before rescaling, we premultiply the luma directly into the io->y
-    // internal buffer. This is OK since these samples are not used for
-    // intra-prediction (the top samples are saved in cache_y_/u_/v_).
-    // But we need to cast the const away, though.
-    WebPMultRows((uint8_t*)io->y, io->y_stride,
-                 io->a, io->width, io->mb_w, mb_h, 0);
-  }
-  num_lines_out = Rescale(io->y, io->y_stride, mb_h, scaler);
-  Rescale(io->u, io->uv_stride, uv_mb_h, &p->scaler_u);
-  Rescale(io->v, io->uv_stride, uv_mb_h, &p->scaler_v);
-  return num_lines_out;
-}
-
-static int EmitRescaledAlphaYUV(const VP8Io* const io, WebPDecParams* const p) {
-  if (io->a != NULL) {
-    const WebPYUVABuffer* const buf = &p->output->u.YUVA;
-    uint8_t* dst_y = buf->y + p->last_y * buf->y_stride;
-    const uint8_t* src_a = buf->a + p->last_y * buf->a_stride;
-    const int num_lines_out = Rescale(io->a, io->width, io->mb_h, &p->scaler_a);
-    if (num_lines_out > 0) {   // unmultiply the Y
-      WebPMultRows(dst_y, buf->y_stride, src_a, buf->a_stride,
-                   p->scaler_a.dst_width, num_lines_out, 1);
-    }
-  }
-  return 0;
-}
-
-static int InitYUVRescaler(const VP8Io* const io, WebPDecParams* const p) {
-  const int has_alpha = WebPIsAlphaMode(p->output->colorspace);
-  const WebPYUVABuffer* const buf = &p->output->u.YUVA;
-  const int out_width  = io->scaled_width;
-  const int out_height = io->scaled_height;
-  const int uv_out_width  = (out_width + 1) >> 1;
-  const int uv_out_height = (out_height + 1) >> 1;
-  const int uv_in_width  = (io->mb_w + 1) >> 1;
-  const int uv_in_height = (io->mb_h + 1) >> 1;
-  const size_t work_size = 2 * out_width;   // scratch memory for luma rescaler
-  const size_t uv_work_size = 2 * uv_out_width;  // and for each u/v ones
-  size_t tmp_size;
-  int32_t* work;
-
-  tmp_size = (work_size + 2 * uv_work_size) * sizeof(*work);
-  if (has_alpha) {
-    tmp_size += work_size * sizeof(*work);
-  }
-  p->memory = WebPSafeCalloc(1ULL, tmp_size);
-  if (p->memory == NULL) {
-    return 0;   // memory error
-  }
-  work = (int32_t*)p->memory;
-  WebPRescalerInit(&p->scaler_y, io->mb_w, io->mb_h,
-                   buf->y, out_width, out_height, buf->y_stride, 1,
-                   io->mb_w, out_width, io->mb_h, out_height,
-                   work);
-  WebPRescalerInit(&p->scaler_u, uv_in_width, uv_in_height,
-                   buf->u, uv_out_width, uv_out_height, buf->u_stride, 1,
-                   uv_in_width, uv_out_width,
-                   uv_in_height, uv_out_height,
-                   work + work_size);
-  WebPRescalerInit(&p->scaler_v, uv_in_width, uv_in_height,
-                   buf->v, uv_out_width, uv_out_height, buf->v_stride, 1,
-                   uv_in_width, uv_out_width,
-                   uv_in_height, uv_out_height,
-                   work + work_size + uv_work_size);
-  p->emit = EmitRescaledYUV;
-
-  if (has_alpha) {
-    WebPRescalerInit(&p->scaler_a, io->mb_w, io->mb_h,
-                     buf->a, out_width, out_height, buf->a_stride, 1,
-                     io->mb_w, out_width, io->mb_h, out_height,
-                     work + work_size + 2 * uv_work_size);
-    p->emit_alpha = EmitRescaledAlphaYUV;
-    WebPInitAlphaProcessing();
-  }
-  return 1;
-}
-
-//------------------------------------------------------------------------------
-// RGBA rescaling
-
-static int ExportRGB(WebPDecParams* const p, int y_pos) {
-  const WebPYUV444Converter convert =
-      WebPYUV444Converters[p->output->colorspace];
-  const WebPRGBABuffer* const buf = &p->output->u.RGBA;
-  uint8_t* dst = buf->rgba + (p->last_y + y_pos) * buf->stride;
-  int num_lines_out = 0;
-  // For RGB rescaling, because of the YUV420, current scan position
-  // U/V can be +1/-1 line from the Y one.  Hence the double test.
-  while (WebPRescalerHasPendingOutput(&p->scaler_y) &&
-         WebPRescalerHasPendingOutput(&p->scaler_u)) {
-    assert(p->last_y + y_pos + num_lines_out < p->output->height);
-    assert(p->scaler_u.y_accum == p->scaler_v.y_accum);
-    WebPRescalerExportRow(&p->scaler_y, 0);
-    WebPRescalerExportRow(&p->scaler_u, 0);
-    WebPRescalerExportRow(&p->scaler_v, 0);
-    convert(p->scaler_y.dst, p->scaler_u.dst, p->scaler_v.dst,
-            dst, p->scaler_y.dst_width);
-    dst += buf->stride;
-    ++num_lines_out;
-  }
-  return num_lines_out;
-}
-
-static int EmitRescaledRGB(const VP8Io* const io, WebPDecParams* const p) {
-  const int mb_h = io->mb_h;
-  const int uv_mb_h = (mb_h + 1) >> 1;
-  int j = 0, uv_j = 0;
-  int num_lines_out = 0;
-  while (j < mb_h) {
-    const int y_lines_in =
-        WebPRescalerImport(&p->scaler_y, mb_h - j,
-                           io->y + j * io->y_stride, io->y_stride);
-    const int u_lines_in =
-        WebPRescalerImport(&p->scaler_u, uv_mb_h - uv_j,
-                           io->u + uv_j * io->uv_stride, io->uv_stride);
-    const int v_lines_in =
-        WebPRescalerImport(&p->scaler_v, uv_mb_h - uv_j,
-                           io->v + uv_j * io->uv_stride, io->uv_stride);
-    (void)v_lines_in;   // remove a gcc warning
-    assert(u_lines_in == v_lines_in);
-    j += y_lines_in;
-    uv_j += u_lines_in;
-    num_lines_out += ExportRGB(p, num_lines_out);
-  }
-  return num_lines_out;
-}
-
-static int ExportAlpha(WebPDecParams* const p, int y_pos) {
-  const WebPRGBABuffer* const buf = &p->output->u.RGBA;
-  uint8_t* const base_rgba = buf->rgba + (p->last_y + y_pos) * buf->stride;
-  const WEBP_CSP_MODE colorspace = p->output->colorspace;
-  const int alpha_first =
-      (colorspace == MODE_ARGB || colorspace == MODE_Argb);
-  uint8_t* dst = base_rgba + (alpha_first ? 0 : 3);
-  int num_lines_out = 0;
-  const int is_premult_alpha = WebPIsPremultipliedMode(colorspace);
-  uint32_t alpha_mask = 0xff;
-  const int width = p->scaler_a.dst_width;
-
-  while (WebPRescalerHasPendingOutput(&p->scaler_a)) {
-    int i;
-    assert(p->last_y + y_pos + num_lines_out < p->output->height);
-    WebPRescalerExportRow(&p->scaler_a, 0);
-    for (i = 0; i < width; ++i) {
-      const uint32_t alpha_value = p->scaler_a.dst[i];
-      dst[4 * i] = alpha_value;
-      alpha_mask &= alpha_value;
-    }
-    dst += buf->stride;
-    ++num_lines_out;
-  }
-  if (is_premult_alpha && alpha_mask != 0xff) {
-    WebPApplyAlphaMultiply(base_rgba, alpha_first,
-                           width, num_lines_out, buf->stride);
-  }
-  return num_lines_out;
-}
-
-static int ExportAlphaRGBA4444(WebPDecParams* const p, int y_pos) {
-  const WebPRGBABuffer* const buf = &p->output->u.RGBA;
-  uint8_t* const base_rgba = buf->rgba + (p->last_y + y_pos) * buf->stride;
-#ifdef WEBP_SWAP_16BIT_CSP
-  uint8_t* alpha_dst = base_rgba;
-#else
-  uint8_t* alpha_dst = base_rgba + 1;
-#endif
-  int num_lines_out = 0;
-  const WEBP_CSP_MODE colorspace = p->output->colorspace;
-  const int width = p->scaler_a.dst_width;
-  const int is_premult_alpha = WebPIsPremultipliedMode(colorspace);
-  uint32_t alpha_mask = 0x0f;
-
-  while (WebPRescalerHasPendingOutput(&p->scaler_a)) {
-    int i;
-    assert(p->last_y + y_pos + num_lines_out < p->output->height);
-    WebPRescalerExportRow(&p->scaler_a, 0);
-    for (i = 0; i < width; ++i) {
-      // Fill in the alpha value (converted to 4 bits).
-      const uint32_t alpha_value = p->scaler_a.dst[i] >> 4;
-      alpha_dst[2 * i] = (alpha_dst[2 * i] & 0xf0) | alpha_value;
-      alpha_mask &= alpha_value;
-    }
-    alpha_dst += buf->stride;
-    ++num_lines_out;
-  }
-  if (is_premult_alpha && alpha_mask != 0x0f) {
-    WebPApplyAlphaMultiply4444(base_rgba, width, num_lines_out, buf->stride);
-  }
-  return num_lines_out;
-}
-
-static int EmitRescaledAlphaRGB(const VP8Io* const io, WebPDecParams* const p) {
-  if (io->a != NULL) {
-    WebPRescaler* const scaler = &p->scaler_a;
-    int j = 0;
-    int pos = 0;
-    while (j < io->mb_h) {
-      j += WebPRescalerImport(scaler, io->mb_h - j,
-                              io->a + j * io->width, io->width);
-      pos += p->emit_alpha_row(p, pos);
-    }
-  }
-  return 0;
-}
-
-static int InitRGBRescaler(const VP8Io* const io, WebPDecParams* const p) {
-  const int has_alpha = WebPIsAlphaMode(p->output->colorspace);
-  const int out_width  = io->scaled_width;
-  const int out_height = io->scaled_height;
-  const int uv_in_width  = (io->mb_w + 1) >> 1;
-  const int uv_in_height = (io->mb_h + 1) >> 1;
-  const size_t work_size = 2 * out_width;   // scratch memory for one rescaler
-  int32_t* work;  // rescalers work area
-  uint8_t* tmp;   // tmp storage for scaled YUV444 samples before RGB conversion
-  size_t tmp_size1, tmp_size2, total_size;
-
-  tmp_size1 = 3 * work_size;
-  tmp_size2 = 3 * out_width;
-  if (has_alpha) {
-    tmp_size1 += work_size;
-    tmp_size2 += out_width;
-  }
-  total_size = tmp_size1 * sizeof(*work) + tmp_size2 * sizeof(*tmp);
-  p->memory = WebPSafeCalloc(1ULL, total_size);
-  if (p->memory == NULL) {
-    return 0;   // memory error
-  }
-  work = (int32_t*)p->memory;
-  tmp = (uint8_t*)(work + tmp_size1);
-  WebPRescalerInit(&p->scaler_y, io->mb_w, io->mb_h,
-                   tmp + 0 * out_width, out_width, out_height, 0, 1,
-                   io->mb_w, out_width, io->mb_h, out_height,
-                   work + 0 * work_size);
-  WebPRescalerInit(&p->scaler_u, uv_in_width, uv_in_height,
-                   tmp + 1 * out_width, out_width, out_height, 0, 1,
-                   io->mb_w, 2 * out_width, io->mb_h, 2 * out_height,
-                   work + 1 * work_size);
-  WebPRescalerInit(&p->scaler_v, uv_in_width, uv_in_height,
-                   tmp + 2 * out_width, out_width, out_height, 0, 1,
-                   io->mb_w, 2 * out_width, io->mb_h, 2 * out_height,
-                   work + 2 * work_size);
-  p->emit = EmitRescaledRGB;
-
-  if (has_alpha) {
-    WebPRescalerInit(&p->scaler_a, io->mb_w, io->mb_h,
-                     tmp + 3 * out_width, out_width, out_height, 0, 1,
-                     io->mb_w, out_width, io->mb_h, out_height,
-                     work + 3 * work_size);
-    p->emit_alpha = EmitRescaledAlphaRGB;
-    if (p->output->colorspace == MODE_RGBA_4444 ||
-        p->output->colorspace == MODE_rgbA_4444) {
-      p->emit_alpha_row = ExportAlphaRGBA4444;
-    } else {
-      p->emit_alpha_row = ExportAlpha;
-    }
-    WebPInitAlphaProcessing();
-  }
-  return 1;
-}
-
-//------------------------------------------------------------------------------
-// Default custom functions
-
-static int CustomSetup(VP8Io* io) {
-  WebPDecParams* const p = (WebPDecParams*)io->opaque;
-  const WEBP_CSP_MODE colorspace = p->output->colorspace;
-  const int is_rgb = WebPIsRGBMode(colorspace);
-  const int is_alpha = WebPIsAlphaMode(colorspace);
-
-  p->memory = NULL;
-  p->emit = NULL;
-  p->emit_alpha = NULL;
-  p->emit_alpha_row = NULL;
-  if (!WebPIoInitFromOptions(p->options, io, is_alpha ? MODE_YUV : MODE_YUVA)) {
-    return 0;
-  }
-  if (is_alpha && WebPIsPremultipliedMode(colorspace)) {
-    WebPInitUpsamplers();
-  }
-  if (io->use_scaling) {
-    const int ok = is_rgb ? InitRGBRescaler(io, p) : InitYUVRescaler(io, p);
-    if (!ok) {
-      return 0;    // memory error
-    }
-  } else {
-    if (is_rgb) {
-      p->emit = EmitSampledRGB;   // default
-      if (io->fancy_upsampling) {
-#ifdef FANCY_UPSAMPLING
-        const int uv_width = (io->mb_w + 1) >> 1;
-        p->memory = WebPSafeMalloc(1ULL, (size_t)(io->mb_w + 2 * uv_width));
-        if (p->memory == NULL) {
-          return 0;   // memory error.
-        }
-        p->tmp_y = (uint8_t*)p->memory;
-        p->tmp_u = p->tmp_y + io->mb_w;
-        p->tmp_v = p->tmp_u + uv_width;
-        p->emit = EmitFancyRGB;
-        WebPInitUpsamplers();
-#endif
-      } else {
-        WebPInitSamplers();
-      }
-    } else {
-      p->emit = EmitYUV;
-    }
-    if (is_alpha) {  // need transparency output
-      p->emit_alpha =
-          (colorspace == MODE_RGBA_4444 || colorspace == MODE_rgbA_4444) ?
-              EmitAlphaRGBA4444
-          : is_rgb ? EmitAlphaRGB
-          : EmitAlphaYUV;
-      if (is_rgb) {
-        WebPInitAlphaProcessing();
-      }
-    }
-  }
-
-  if (is_rgb) {
-    VP8YUVInit();
-  }
-  return 1;
-}
-
-//------------------------------------------------------------------------------
-
-static int CustomPut(const VP8Io* io) {
-  WebPDecParams* const p = (WebPDecParams*)io->opaque;
-  const int mb_w = io->mb_w;
-  const int mb_h = io->mb_h;
-  int num_lines_out;
-  assert(!(io->mb_y & 1));
-
-  if (mb_w <= 0 || mb_h <= 0) {
-    return 0;
-  }
-  num_lines_out = p->emit(io, p);
-  if (p->emit_alpha != NULL) {
-    p->emit_alpha(io, p);
-  }
-  p->last_y += num_lines_out;
-  return 1;
-}
-
-//------------------------------------------------------------------------------
-
-static void CustomTeardown(const VP8Io* io) {
-  WebPDecParams* const p = (WebPDecParams*)io->opaque;
-  WebPSafeFree(p->memory);
-  p->memory = NULL;
-}
-
-//------------------------------------------------------------------------------
-// Main entry point
-
-void WebPInitCustomIo(WebPDecParams* const params, VP8Io* const io) {
-  io->put      = CustomPut;
-  io->setup    = CustomSetup;
-  io->teardown = CustomTeardown;
-  io->opaque   = params;
-}
-
-//------------------------------------------------------------------------------
@@ -1,110 +0,0 @@
-// Copyright 2010 Google Inc. All Rights Reserved.
-//
-// Use of this source code is governed by a BSD-style license
-// that can be found in the COPYING file in the root of the source
-// tree. An additional intellectual property rights grant can be found
-// in the file PATENTS. All contributing project authors may
-// be found in the AUTHORS file in the root of the source tree.
-// -----------------------------------------------------------------------------
-//
-// Quantizer initialization
-//
-// Author: Skal (pascal.massimino@gmail.com)
-
-#include "./vp8i.h"
-
-static WEBP_INLINE int clip(int v, int M) {
-  return v < 0 ? 0 : v > M ? M : v;
-}
-
-// Paragraph 14.1
-static const uint8_t kDcTable[128] = {
-  4,     5,   6,   7,   8,   9,  10,  10,
-  11,   12,  13,  14,  15,  16,  17,  17,
-  18,   19,  20,  20,  21,  21,  22,  22,
-  23,   23,  24,  25,  25,  26,  27,  28,
-  29,   30,  31,  32,  33,  34,  35,  36,
-  37,   37,  38,  39,  40,  41,  42,  43,
-  44,   45,  46,  46,  47,  48,  49,  50,
-  51,   52,  53,  54,  55,  56,  57,  58,
-  59,   60,  61,  62,  63,  64,  65,  66,
-  67,   68,  69,  70,  71,  72,  73,  74,
-  75,   76,  76,  77,  78,  79,  80,  81,
-  82,   83,  84,  85,  86,  87,  88,  89,
-  91,   93,  95,  96,  98, 100, 101, 102,
-  104, 106, 108, 110, 112, 114, 116, 118,
-  122, 124, 126, 128, 130, 132, 134, 136,
-  138, 140, 143, 145, 148, 151, 154, 157
-};
-
-static const uint16_t kAcTable[128] = {
-  4,     5,   6,   7,   8,   9,  10,  11,
-  12,   13,  14,  15,  16,  17,  18,  19,
-  20,   21,  22,  23,  24,  25,  26,  27,
-  28,   29,  30,  31,  32,  33,  34,  35,
-  36,   37,  38,  39,  40,  41,  42,  43,
-  44,   45,  46,  47,  48,  49,  50,  51,
-  52,   53,  54,  55,  56,  57,  58,  60,
-  62,   64,  66,  68,  70,  72,  74,  76,
-  78,   80,  82,  84,  86,  88,  90,  92,
-  94,   96,  98, 100, 102, 104, 106, 108,
-  110, 112, 114, 116, 119, 122, 125, 128,
-  131, 134, 137, 140, 143, 146, 149, 152,
-  155, 158, 161, 164, 167, 170, 173, 177,
-  181, 185, 189, 193, 197, 201, 205, 209,
-  213, 217, 221, 225, 229, 234, 239, 245,
-  249, 254, 259, 264, 269, 274, 279, 284
-};
-
-//------------------------------------------------------------------------------
-// Paragraph 9.6
-
-void VP8ParseQuant(VP8Decoder* const dec) {
-  VP8BitReader* const br = &dec->br_;
-  const int base_q0 = VP8GetValue(br, 7);
-  const int dqy1_dc = VP8Get(br) ? VP8GetSignedValue(br, 4) : 0;
-  const int dqy2_dc = VP8Get(br) ? VP8GetSignedValue(br, 4) : 0;
-  const int dqy2_ac = VP8Get(br) ? VP8GetSignedValue(br, 4) : 0;
-  const int dquv_dc = VP8Get(br) ? VP8GetSignedValue(br, 4) : 0;
-  const int dquv_ac = VP8Get(br) ? VP8GetSignedValue(br, 4) : 0;
-
-  const VP8SegmentHeader* const hdr = &dec->segment_hdr_;
-  int i;
-
-  for (i = 0; i < NUM_MB_SEGMENTS; ++i) {
-    int q;
-    if (hdr->use_segment_) {
-      q = hdr->quantizer_[i];
-      if (!hdr->absolute_delta_) {
-        q += base_q0;
-      }
-    } else {
-      if (i > 0) {
-        dec->dqm_[i] = dec->dqm_[0];
-        continue;
-      } else {
-        q = base_q0;
-      }
-    }
-    {
-      VP8QuantMatrix* const m = &dec->dqm_[i];
-      m->y1_mat_[0] = kDcTable[clip(q + dqy1_dc, 127)];
-      m->y1_mat_[1] = kAcTable[clip(q + 0,       127)];
-
-      m->y2_mat_[0] = kDcTable[clip(q + dqy2_dc, 127)] * 2;
-      // For all x in [0..284], x*155/100 is bitwise equal to (x*101581) >> 16.
-      // The smallest precision for that is '(x*6349) >> 12' but 16 is a good
-      // word size.
-      m->y2_mat_[1] = (kAcTable[clip(q + dqy2_ac, 127)] * 101581) >> 16;
-      if (m->y2_mat_[1] < 8) m->y2_mat_[1] = 8;
-
-      m->uv_mat_[0] = kDcTable[clip(q + dquv_dc, 117)];
-      m->uv_mat_[1] = kAcTable[clip(q + dquv_ac, 127)];
-
-      m->uv_quant_ = q + dquv_ac;   // for dithering strength evaluation
-    }
-  }
-}
-
-//------------------------------------------------------------------------------
-
@@ -1,516 +0,0 @@
-// Copyright 2010 Google Inc. All Rights Reserved.
-//
-// Use of this source code is governed by a BSD-style license
-// that can be found in the COPYING file in the root of the source
-// tree. An additional intellectual property rights grant can be found
-// in the file PATENTS. All contributing project authors may
-// be found in the AUTHORS file in the root of the source tree.
-// -----------------------------------------------------------------------------
-//
-// Coding trees and probas
-//
-// Author: Skal (pascal.massimino@gmail.com)
-
-#include "./vp8i.h"
-#include "../utils/bit_reader_inl.h"
-
-#define USE_GENERIC_TREE
-
-#ifdef USE_GENERIC_TREE
-static const int8_t kYModesIntra4[18] = {
-  -B_DC_PRED, 1,
-    -B_TM_PRED, 2,
-      -B_VE_PRED, 3,
-        4, 6,
-          -B_HE_PRED, 5,
-            -B_RD_PRED, -B_VR_PRED,
-        -B_LD_PRED, 7,
-          -B_VL_PRED, 8,
-            -B_HD_PRED, -B_HU_PRED
-};
-#endif
-
-//------------------------------------------------------------------------------
-// Default probabilities
-
-// Paragraph 13.5
-static const uint8_t
-  CoeffsProba0[NUM_TYPES][NUM_BANDS][NUM_CTX][NUM_PROBAS] = {
-  { { { 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 },
-      { 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 },
-      { 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 }
-    },
-    { { 253, 136, 254, 255, 228, 219, 128, 128, 128, 128, 128 },
-      { 189, 129, 242, 255, 227, 213, 255, 219, 128, 128, 128 },
-      { 106, 126, 227, 252, 214, 209, 255, 255, 128, 128, 128 }
-    },
-    { { 1, 98, 248, 255, 236, 226, 255, 255, 128, 128, 128 },
-      { 181, 133, 238, 254, 221, 234, 255, 154, 128, 128, 128 },
-      { 78, 134, 202, 247, 198, 180, 255, 219, 128, 128, 128 },
-    },
-    { { 1, 185, 249, 255, 243, 255, 128, 128, 128, 128, 128 },
-      { 184, 150, 247, 255, 236, 224, 128, 128, 128, 128, 128 },
-      { 77, 110, 216, 255, 236, 230, 128, 128, 128, 128, 128 },
-    },
-    { { 1, 101, 251, 255, 241, 255, 128, 128, 128, 128, 128 },
-      { 170, 139, 241, 252, 236, 209, 255, 255, 128, 128, 128 },
-      { 37, 116, 196, 243, 228, 255, 255, 255, 128, 128, 128 }
-    },
-    { { 1, 204, 254, 255, 245, 255, 128, 128, 128, 128, 128 },
-      { 207, 160, 250, 255, 238, 128, 128, 128, 128, 128, 128 },
-      { 102, 103, 231, 255, 211, 171, 128, 128, 128, 128, 128 }
-    },
-    { { 1, 152, 252, 255, 240, 255, 128, 128, 128, 128, 128 },
-      { 177, 135, 243, 255, 234, 225, 128, 128, 128, 128, 128 },
-      { 80, 129, 211, 255, 194, 224, 128, 128, 128, 128, 128 }
-    },
-    { { 1, 1, 255, 128, 128, 128, 128, 128, 128, 128, 128 },
-      { 246, 1, 255, 128, 128, 128, 128, 128, 128, 128, 128 },
-      { 255, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 }
-    }
-  },
-  { { { 198, 35, 237, 223, 193, 187, 162, 160, 145, 155, 62 },
-      { 131, 45, 198, 221, 172, 176, 220, 157, 252, 221, 1 },
-      { 68, 47, 146, 208, 149, 167, 221, 162, 255, 223, 128 }
-    },
-    { { 1, 149, 241, 255, 221, 224, 255, 255, 128, 128, 128 },
-      { 184, 141, 234, 253, 222, 220, 255, 199, 128, 128, 128 },
-      { 81, 99, 181, 242, 176, 190, 249, 202, 255, 255, 128 }
-    },
-    { { 1, 129, 232, 253, 214, 197, 242, 196, 255, 255, 128 },
-      { 99, 121, 210, 250, 201, 198, 255, 202, 128, 128, 128 },
-      { 23, 91, 163, 242, 170, 187, 247, 210, 255, 255, 128 }
-    },
-    { { 1, 200, 246, 255, 234, 255, 128, 128, 128, 128, 128 },
-      { 109, 178, 241, 255, 231, 245, 255, 255, 128, 128, 128 },
-      { 44, 130, 201, 253, 205, 192, 255, 255, 128, 128, 128 }
-    },
-    { { 1, 132, 239, 251, 219, 209, 255, 165, 128, 128, 128 },
-      { 94, 136, 225, 251, 218, 190, 255, 255, 128, 128, 128 },
-      { 22, 100, 174, 245, 186, 161, 255, 199, 128, 128, 128 }
-    },
-    { { 1, 182, 249, 255, 232, 235, 128, 128, 128, 128, 128 },
-      { 124, 143, 241, 255, 227, 234, 128, 128, 128, 128, 128 },
-      { 35, 77, 181, 251, 193, 211, 255, 205, 128, 128, 128 }
-    },
-    { { 1, 157, 247, 255, 236, 231, 255, 255, 128, 128, 128 },
-      { 121, 141, 235, 255, 225, 227, 255, 255, 128, 128, 128 },
-      { 45, 99, 188, 251, 195, 217, 255, 224, 128, 128, 128 }
-    },
-    { { 1, 1, 251, 255, 213, 255, 128, 128, 128, 128, 128 },
-      { 203, 1, 248, 255, 255, 128, 128, 128, 128, 128, 128 },
-      { 137, 1, 177, 255, 224, 255, 128, 128, 128, 128, 128 }
-    }
-  },
-  { { { 253, 9, 248, 251, 207, 208, 255, 192, 128, 128, 128 },
-      { 175, 13, 224, 243, 193, 185, 249, 198, 255, 255, 128 },
-      { 73, 17, 171, 221, 161, 179, 236, 167, 255, 234, 128 }
-    },
-    { { 1, 95, 247, 253, 212, 183, 255, 255, 128, 128, 128 },
-      { 239, 90, 244, 250, 211, 209, 255, 255, 128, 128, 128 },
-      { 155, 77, 195, 248, 188, 195, 255, 255, 128, 128, 128 }
-    },
-    { { 1, 24, 239, 251, 218, 219, 255, 205, 128, 128, 128 },
-      { 201, 51, 219, 255, 196, 186, 128, 128, 128, 128, 128 },
-      { 69, 46, 190, 239, 201, 218, 255, 228, 128, 128, 128 }
-    },
-    { { 1, 191, 251, 255, 255, 128, 128, 128, 128, 128, 128 },
-      { 223, 165, 249, 255, 213, 255, 128, 128, 128, 128, 128 },
-      { 141, 124, 248, 255, 255, 128, 128, 128, 128, 128, 128 }
-    },
-    { { 1, 16, 248, 255, 255, 128, 128, 128, 128, 128, 128 },
-      { 190, 36, 230, 255, 236, 255, 128, 128, 128, 128, 128 },
-      { 149, 1, 255, 128, 128, 128, 128, 128, 128, 128, 128 }
-    },
-    { { 1, 226, 255, 128, 128, 128, 128, 128, 128, 128, 128 },
-      { 247, 192, 255, 128, 128, 128, 128, 128, 128, 128, 128 },
-      { 240, 128, 255, 128, 128, 128, 128, 128, 128, 128, 128 }
-    },
-    { { 1, 134, 252, 255, 255, 128, 128, 128, 128, 128, 128 },
-      { 213, 62, 250, 255, 255, 128, 128, 128, 128, 128, 128 },
-      { 55, 93, 255, 128, 128, 128, 128, 128, 128, 128, 128 }
-    },
-    { { 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 },
-      { 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 },
-      { 128, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 }
-    }
-  },
-  { { { 202, 24, 213, 235, 186, 191, 220, 160, 240, 175, 255 },
-      { 126, 38, 182, 232, 169, 184, 228, 174, 255, 187, 128 },
-      { 61, 46, 138, 219, 151, 178, 240, 170, 255, 216, 128 }
-    },
-    { { 1, 112, 230, 250, 199, 191, 247, 159, 255, 255, 128 },
-      { 166, 109, 228, 252, 211, 215, 255, 174, 128, 128, 128 },
-      { 39, 77, 162, 232, 172, 180, 245, 178, 255, 255, 128 }
-    },
-    { { 1, 52, 220, 246, 198, 199, 249, 220, 255, 255, 128 },
-      { 124, 74, 191, 243, 183, 193, 250, 221, 255, 255, 128 },
-      { 24, 71, 130, 219, 154, 170, 243, 182, 255, 255, 128 }
-    },
-    { { 1, 182, 225, 249, 219, 240, 255, 224, 128, 128, 128 },
-      { 149, 150, 226, 252, 216, 205, 255, 171, 128, 128, 128 },
-      { 28, 108, 170, 242, 183, 194, 254, 223, 255, 255, 128 }
-    },
-    { { 1, 81, 230, 252, 204, 203, 255, 192, 128, 128, 128 },
-      { 123, 102, 209, 247, 188, 196, 255, 233, 128, 128, 128 },
-      { 20, 95, 153, 243, 164, 173, 255, 203, 128, 128, 128 }
-    },
-    { { 1, 222, 248, 255, 216, 213, 128, 128, 128, 128, 128 },
-      { 168, 175, 246, 252, 235, 205, 255, 255, 128, 128, 128 },
-      { 47, 116, 215, 255, 211, 212, 255, 255, 128, 128, 128 }
-    },
-    { { 1, 121, 236, 253, 212, 214, 255, 255, 128, 128, 128 },
-      { 141, 84, 213, 252, 201, 202, 255, 219, 128, 128, 128 },
-      { 42, 80, 160, 240, 162, 185, 255, 205, 128, 128, 128 }
-    },
-    { { 1, 1, 255, 128, 128, 128, 128, 128, 128, 128, 128 },
-      { 244, 1, 255, 128, 128, 128, 128, 128, 128, 128, 128 },
-      { 238, 1, 255, 128, 128, 128, 128, 128, 128, 128, 128 }
-    }
-  }
-};
-
-// Paragraph 11.5
-static const uint8_t kBModesProba[NUM_BMODES][NUM_BMODES][NUM_BMODES - 1] = {
-  { { 231, 120, 48, 89, 115, 113, 120, 152, 112 },
-    { 152, 179, 64, 126, 170, 118, 46, 70, 95 },
-    { 175, 69, 143, 80, 85, 82, 72, 155, 103 },
-    { 56, 58, 10, 171, 218, 189, 17, 13, 152 },
-    { 114, 26, 17, 163, 44, 195, 21, 10, 173 },
-    { 121, 24, 80, 195, 26, 62, 44, 64, 85 },
-    { 144, 71, 10, 38, 171, 213, 144, 34, 26 },
-    { 170, 46, 55, 19, 136, 160, 33, 206, 71 },
-    { 63, 20, 8, 114, 114, 208, 12, 9, 226 },
-    { 81, 40, 11, 96, 182, 84, 29, 16, 36 } },
-  { { 134, 183, 89, 137, 98, 101, 106, 165, 148 },
-    { 72, 187, 100, 130, 157, 111, 32, 75, 80 },
-    { 66, 102, 167, 99, 74, 62, 40, 234, 128 },
-    { 41, 53, 9, 178, 241, 141, 26, 8, 107 },
-    { 74, 43, 26, 146, 73, 166, 49, 23, 157 },
-    { 65, 38, 105, 160, 51, 52, 31, 115, 128 },
-    { 104, 79, 12, 27, 217, 255, 87, 17, 7 },
-    { 87, 68, 71, 44, 114, 51, 15, 186, 23 },
-    { 47, 41, 14, 110, 182, 183, 21, 17, 194 },
-    { 66, 45, 25, 102, 197, 189, 23, 18, 22 } },
-  { { 88, 88, 147, 150, 42, 46, 45, 196, 205 },
-    { 43, 97, 183, 117, 85, 38, 35, 179, 61 },
-    { 39, 53, 200, 87, 26, 21, 43, 232, 171 },
-    { 56, 34, 51, 104, 114, 102, 29, 93, 77 },
-    { 39, 28, 85, 171, 58, 165, 90, 98, 64 },
-    { 34, 22, 116, 206, 23, 34, 43, 166, 73 },
-    { 107, 54, 32, 26, 51, 1, 81, 43, 31 },
-    { 68, 25, 106, 22, 64, 171, 36, 225, 114 },
-    { 34, 19, 21, 102, 132, 188, 16, 76, 124 },
-    { 62, 18, 78, 95, 85, 57, 50, 48, 51 } },
-  { { 193, 101, 35, 159, 215, 111, 89, 46, 111 },
-    { 60, 148, 31, 172, 219, 228, 21, 18, 111 },
-    { 112, 113, 77, 85, 179, 255, 38, 120, 114 },
-    { 40, 42, 1, 196, 245, 209, 10, 25, 109 },
-    { 88, 43, 29, 140, 166, 213, 37, 43, 154 },
-    { 61, 63, 30, 155, 67, 45, 68, 1, 209 },
-    { 100, 80, 8, 43, 154, 1, 51, 26, 71 },
-    { 142, 78, 78, 16, 255, 128, 34, 197, 171 },
-    { 41, 40, 5, 102, 211, 183, 4, 1, 221 },
-    { 51, 50, 17, 168, 209, 192, 23, 25, 82 } },
-  { { 138, 31, 36, 171, 27, 166, 38, 44, 229 },
-    { 67, 87, 58, 169, 82, 115, 26, 59, 179 },
-    { 63, 59, 90, 180, 59, 166, 93, 73, 154 },
-    { 40, 40, 21, 116, 143, 209, 34, 39, 175 },
-    { 47, 15, 16, 183, 34, 223, 49, 45, 183 },
-    { 46, 17, 33, 183, 6, 98, 15, 32, 183 },
-    { 57, 46, 22, 24, 128, 1, 54, 17, 37 },
-    { 65, 32, 73, 115, 28, 128, 23, 128, 205 },
-    { 40, 3, 9, 115, 51, 192, 18, 6, 223 },
-    { 87, 37, 9, 115, 59, 77, 64, 21, 47 } },
-  { { 104, 55, 44, 218, 9, 54, 53, 130, 226 },
-    { 64, 90, 70, 205, 40, 41, 23, 26, 57 },
-    { 54, 57, 112, 184, 5, 41, 38, 166, 213 },
-    { 30, 34, 26, 133, 152, 116, 10, 32, 134 },
-    { 39, 19, 53, 221, 26, 114, 32, 73, 255 },
-    { 31, 9, 65, 234, 2, 15, 1, 118, 73 },
-    { 75, 32, 12, 51, 192, 255, 160, 43, 51 },
-    { 88, 31, 35, 67, 102, 85, 55, 186, 85 },
-    { 56, 21, 23, 111, 59, 205, 45, 37, 192 },
-    { 55, 38, 70, 124, 73, 102, 1, 34, 98 } },
-  { { 125, 98, 42, 88, 104, 85, 117, 175, 82 },
-    { 95, 84, 53, 89, 128, 100, 113, 101, 45 },
-    { 75, 79, 123, 47, 51, 128, 81, 171, 1 },
-    { 57, 17, 5, 71, 102, 57, 53, 41, 49 },
-    { 38, 33, 13, 121, 57, 73, 26, 1, 85 },
-    { 41, 10, 67, 138, 77, 110, 90, 47, 114 },
-    { 115, 21, 2, 10, 102, 255, 166, 23, 6 },
-    { 101, 29, 16, 10, 85, 128, 101, 196, 26 },
-    { 57, 18, 10, 102, 102, 213, 34, 20, 43 },
-    { 117, 20, 15, 36, 163, 128, 68, 1, 26 } },
-  { { 102, 61, 71, 37, 34, 53, 31, 243, 192 },
-    { 69, 60, 71, 38, 73, 119, 28, 222, 37 },
-    { 68, 45, 128, 34, 1, 47, 11, 245, 171 },
-    { 62, 17, 19, 70, 146, 85, 55, 62, 70 },
-    { 37, 43, 37, 154, 100, 163, 85, 160, 1 },
-    { 63, 9, 92, 136, 28, 64, 32, 201, 85 },
-    { 75, 15, 9, 9, 64, 255, 184, 119, 16 },
-    { 86, 6, 28, 5, 64, 255, 25, 248, 1 },
-    { 56, 8, 17, 132, 137, 255, 55, 116, 128 },
-    { 58, 15, 20, 82, 135, 57, 26, 121, 40 } },
-  { { 164, 50, 31, 137, 154, 133, 25, 35, 218 },
-    { 51, 103, 44, 131, 131, 123, 31, 6, 158 },
-    { 86, 40, 64, 135, 148, 224, 45, 183, 128 },
-    { 22, 26, 17, 131, 240, 154, 14, 1, 209 },
-    { 45, 16, 21, 91, 64, 222, 7, 1, 197 },
-    { 56, 21, 39, 155, 60, 138, 23, 102, 213 },
-    { 83, 12, 13, 54, 192, 255, 68, 47, 28 },
-    { 85, 26, 85, 85, 128, 128, 32, 146, 171 },
-    { 18, 11, 7, 63, 144, 171, 4, 4, 246 },
-    { 35, 27, 10, 146, 174, 171, 12, 26, 128 } },
-  { { 190, 80, 35, 99, 180, 80, 126, 54, 45 },
-    { 85, 126, 47, 87, 176, 51, 41, 20, 32 },
-    { 101, 75, 128, 139, 118, 146, 116, 128, 85 },
-    { 56, 41, 15, 176, 236, 85, 37, 9, 62 },
-    { 71, 30, 17, 119, 118, 255, 17, 18, 138 },
-    { 101, 38, 60, 138, 55, 70, 43, 26, 142 },
-    { 146, 36, 19, 30, 171, 255, 97, 27, 20 },
-    { 138, 45, 61, 62, 219, 1, 81, 188, 64 },
-    { 32, 41, 20, 117, 151, 142, 20, 21, 163 },
-    { 112, 19, 12, 61, 195, 128, 48, 4, 24 } }
-};
-
-void VP8ResetProba(VP8Proba* const proba) {
-  memset(proba->segments_, 255u, sizeof(proba->segments_));
-  // proba->bands_[][] is initialized later
-}
-
-static void ParseIntraMode(VP8BitReader* const br,
-                           VP8Decoder* const dec, int mb_x) {
-  uint8_t* const top = dec->intra_t_ + 4 * mb_x;
-  uint8_t* const left = dec->intra_l_;
-  VP8MBData* const block = dec->mb_data_ + mb_x;
-
-  // Note: we don't save segment map (yet), as we don't expect
-  // to decode more than 1 keyframe.
-  if (dec->segment_hdr_.update_map_) {
-    // Hardcoded tree parsing
-    block->segment_ = !VP8GetBit(br, dec->proba_.segments_[0])
-                    ? VP8GetBit(br, dec->proba_.segments_[1])
-                    : 2 + VP8GetBit(br, dec->proba_.segments_[2]);
-  } else {
-    block->segment_ = 0;  // default for intra
-  }
-  if (dec->use_skip_proba_) block->skip_ = VP8GetBit(br, dec->skip_p_);
-
-  block->is_i4x4_ = !VP8GetBit(br, 145);   // decide for B_PRED first
-  if (!block->is_i4x4_) {
-    // Hardcoded 16x16 intra-mode decision tree.
-    const int ymode =
-        VP8GetBit(br, 156) ? (VP8GetBit(br, 128) ? TM_PRED : H_PRED)
-                           : (VP8GetBit(br, 163) ? V_PRED : DC_PRED);
-    block->imodes_[0] = ymode;
-    memset(top, ymode, 4 * sizeof(*top));
-    memset(left, ymode, 4 * sizeof(*left));
-  } else {
-    uint8_t* modes = block->imodes_;
-    int y;
-    for (y = 0; y < 4; ++y) {
-      int ymode = left[y];
-      int x;
-      for (x = 0; x < 4; ++x) {
-        const uint8_t* const prob = kBModesProba[top[x]][ymode];
-#ifdef USE_GENERIC_TREE
-        // Generic tree-parsing
-        int i = kYModesIntra4[VP8GetBit(br, prob[0])];
-        while (i > 0) {
-          i = kYModesIntra4[2 * i + VP8GetBit(br, prob[i])];
-        }
-        ymode = -i;
-#else
-        // Hardcoded tree parsing
-        ymode = !VP8GetBit(br, prob[0]) ? B_DC_PRED :
-                  !VP8GetBit(br, prob[1]) ? B_TM_PRED :
-                    !VP8GetBit(br, prob[2]) ? B_VE_PRED :
-                      !VP8GetBit(br, prob[3]) ?
-                        (!VP8GetBit(br, prob[4]) ? B_HE_PRED :
-                          (!VP8GetBit(br, prob[5]) ? B_RD_PRED : B_VR_PRED)) :
-                        (!VP8GetBit(br, prob[6]) ? B_LD_PRED :
-                          (!VP8GetBit(br, prob[7]) ? B_VL_PRED :
-                            (!VP8GetBit(br, prob[8]) ? B_HD_PRED : B_HU_PRED)));
-#endif    // USE_GENERIC_TREE
-        top[x] = ymode;
-      }
-      memcpy(modes, top, 4 * sizeof(*top));
-      modes += 4;
-      left[y] = ymode;
-    }
-  }
-  // Hardcoded UVMode decision tree
-  block->uvmode_ = !VP8GetBit(br, 142) ? DC_PRED
-                 : !VP8GetBit(br, 114) ? V_PRED
-                 : VP8GetBit(br, 183) ? TM_PRED : H_PRED;
-}
-
-int VP8ParseIntraModeRow(VP8BitReader* const br, VP8Decoder* const dec) {
-  int mb_x;
-  for (mb_x = 0; mb_x < dec->mb_w_; ++mb_x) {
-    ParseIntraMode(br, dec, mb_x);
-  }
-  return !dec->br_.eof_;
-}
-
-//------------------------------------------------------------------------------
-// Paragraph 13
-
-static const uint8_t
-    CoeffsUpdateProba[NUM_TYPES][NUM_BANDS][NUM_CTX][NUM_PROBAS] = {
-  { { { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
-    },
-    { { 176, 246, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 223, 241, 252, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 249, 253, 253, 255, 255, 255, 255, 255, 255, 255, 255 }
-    },
-    { { 255, 244, 252, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 234, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 253, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
-    },
-    { { 255, 246, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 239, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 254, 255, 254, 255, 255, 255, 255, 255, 255, 255, 255 }
-    },
-    { { 255, 248, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 251, 255, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
-    },
-    { { 255, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 251, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 254, 255, 254, 255, 255, 255, 255, 255, 255, 255, 255 }
-    },
-    { { 255, 254, 253, 255, 254, 255, 255, 255, 255, 255, 255 },
-      { 250, 255, 254, 255, 254, 255, 255, 255, 255, 255, 255 },
-      { 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
-    },
-    { { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
-    }
-  },
-  { { { 217, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 225, 252, 241, 253, 255, 255, 254, 255, 255, 255, 255 },
-      { 234, 250, 241, 250, 253, 255, 253, 254, 255, 255, 255 }
-    },
-    { { 255, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 223, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 238, 253, 254, 254, 255, 255, 255, 255, 255, 255, 255 }
-    },
-    { { 255, 248, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 249, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
-    },
-    { { 255, 253, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 247, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
-    },
-    { { 255, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 252, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
-    },
-    { { 255, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 253, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
-    },
-    { { 255, 254, 253, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 250, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
-    },
-    { { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
-    }
-  },
-  { { { 186, 251, 250, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 234, 251, 244, 254, 255, 255, 255, 255, 255, 255, 255 },
-      { 251, 251, 243, 253, 254, 255, 254, 255, 255, 255, 255 }
-    },
-    { { 255, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 236, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 251, 253, 253, 254, 254, 255, 255, 255, 255, 255, 255 }
-    },
-    { { 255, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 254, 254, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
-    },
-    { { 255, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 254, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
-    },
-    { { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
-    },
-    { { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
-    },
-    { { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
-    },
-    { { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
-    }
-  },
-  { { { 248, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 250, 254, 252, 254, 255, 255, 255, 255, 255, 255, 255 },
-      { 248, 254, 249, 253, 255, 255, 255, 255, 255, 255, 255 }
-    },
-    { { 255, 253, 253, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 246, 253, 253, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 252, 254, 251, 254, 254, 255, 255, 255, 255, 255, 255 }
-    },
-    { { 255, 254, 252, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 248, 254, 253, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 253, 255, 254, 254, 255, 255, 255, 255, 255, 255, 255 }
-    },
-    { { 255, 251, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 245, 251, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 253, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255 }
-    },
-    { { 255, 251, 253, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 252, 253, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 255, 254, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
-    },
-    { { 255, 252, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 249, 255, 254, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 255, 255, 254, 255, 255, 255, 255, 255, 255, 255, 255 }
-    },
-    { { 255, 255, 253, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 250, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
-    },
-    { { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 254, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 },
-      { 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255 }
-    }
-  }
-};
-
-// Paragraph 9.9
-void VP8ParseProba(VP8BitReader* const br, VP8Decoder* const dec) {
-  VP8Proba* const proba = &dec->proba_;
-  int t, b, c, p;
-  for (t = 0; t < NUM_TYPES; ++t) {
-    for (b = 0; b < NUM_BANDS; ++b) {
-      for (c = 0; c < NUM_CTX; ++c) {
-        for (p = 0; p < NUM_PROBAS; ++p) {
-          const int v = VP8GetBit(br, CoeffsUpdateProba[t][b][c][p]) ?
-                        VP8GetValue(br, 8) : CoeffsProba0[t][b][c][p];
-          proba->bands_[t][b].probas_[c][p] = v;
-        }
-      }
-    }
-  }
-  dec->use_skip_proba_ = VP8Get(br);
-  if (dec->use_skip_proba_) {
-    dec->skip_p_ = VP8GetValue(br, 8);
-  }
-}
-
@@ -1,668 +0,0 @@
-// Copyright 2010 Google Inc. All Rights Reserved.
-//
-// Use of this source code is governed by a BSD-style license
-// that can be found in the COPYING file in the root of the source
-// tree. An additional intellectual property rights grant can be found
-// in the file PATENTS. All contributing project authors may
-// be found in the AUTHORS file in the root of the source tree.
-// -----------------------------------------------------------------------------
-//
-// main entry for the decoder
-//
-// Author: Skal (pascal.massimino@gmail.com)
-
-#include <stdlib.h>
-
-#include "./alphai.h"
-#include "./vp8i.h"
-#include "./vp8li.h"
-#include "./webpi.h"
-#include "../utils/bit_reader_inl.h"
-#include "../utils/utils.h"
-
-//------------------------------------------------------------------------------
-
-int WebPGetDecoderVersion(void) {
-  return (DEC_MAJ_VERSION << 16) | (DEC_MIN_VERSION << 8) | DEC_REV_VERSION;
-}
-
-//------------------------------------------------------------------------------
-// VP8Decoder
-
-static void SetOk(VP8Decoder* const dec) {
-  dec->status_ = VP8_STATUS_OK;
-  dec->error_msg_ = "OK";
-}
-
-int VP8InitIoInternal(VP8Io* const io, int version) {
-  if (WEBP_ABI_IS_INCOMPATIBLE(version, WEBP_DECODER_ABI_VERSION)) {
-    return 0;  // mismatch error
-  }
-  if (io != NULL) {
-    memset(io, 0, sizeof(*io));
-  }
-  return 1;
-}
-
-VP8Decoder* VP8New(void) {
-  VP8Decoder* const dec = (VP8Decoder*)WebPSafeCalloc(1ULL, sizeof(*dec));
-  if (dec != NULL) {
-    SetOk(dec);
-    WebPGetWorkerInterface()->Init(&dec->worker_);
-    dec->ready_ = 0;
-    dec->num_parts_ = 1;
-  }
-  return dec;
-}
-
-VP8StatusCode VP8Status(VP8Decoder* const dec) {
-  if (!dec) return VP8_STATUS_INVALID_PARAM;
-  return dec->status_;
-}
-
-const char* VP8StatusMessage(VP8Decoder* const dec) {
-  if (dec == NULL) return "no object";
-  if (!dec->error_msg_) return "OK";
-  return dec->error_msg_;
-}
-
-void VP8Delete(VP8Decoder* const dec) {
-  if (dec != NULL) {
-    VP8Clear(dec);
-    WebPSafeFree(dec);
-  }
-}
-
-int VP8SetError(VP8Decoder* const dec,
-                VP8StatusCode error, const char* const msg) {
-  // TODO This check would be unnecessary if alpha decompression was separated
-  // from VP8ProcessRow/FinishRow. This avoids setting 'dec->status_' to
-  // something other than VP8_STATUS_BITSTREAM_ERROR on alpha decompression
-  // failure.
-  if (dec->status_ == VP8_STATUS_OK) {
-    dec->status_ = error;
-    dec->error_msg_ = msg;
-    dec->ready_ = 0;
-  }
-  return 0;
-}
-
-//------------------------------------------------------------------------------
-
-int VP8CheckSignature(const uint8_t* const data, size_t data_size) {
-  return (data_size >= 3 &&
-          data[0] == 0x9d && data[1] == 0x01 && data[2] == 0x2a);
-}
-
-int VP8GetInfo(const uint8_t* data, size_t data_size, size_t chunk_size,
-               int* const width, int* const height) {
-  if (data == NULL || data_size < VP8_FRAME_HEADER_SIZE) {
-    return 0;         // not enough data
-  }
-  // check signature
-  if (!VP8CheckSignature(data + 3, data_size - 3)) {
-    return 0;         // Wrong signature.
-  } else {
-    const uint32_t bits = data[0] | (data[1] << 8) | (data[2] << 16);
-    const int key_frame = !(bits & 1);
-    const int w = ((data[7] << 8) | data[6]) & 0x3fff;
-    const int h = ((data[9] << 8) | data[8]) & 0x3fff;
-
-    if (!key_frame) {   // Not a keyframe.
-      return 0;
-    }
-
-    if (((bits >> 1) & 7) > 3) {
-      return 0;         // unknown profile
-    }
-    if (!((bits >> 4) & 1)) {
-      return 0;         // first frame is invisible!
-    }
-    if (((bits >> 5)) >= chunk_size) {  // partition_length
-      return 0;         // inconsistent size information.
-    }
-    if (w == 0 || h == 0) {
-      return 0;         // We don't support both width and height to be zero.
-    }
-
-    if (width) {
-      *width = w;
-    }
-    if (height) {
-      *height = h;
-    }
-
-    return 1;
-  }
-}
-
-//------------------------------------------------------------------------------
-// Header parsing
-
-static void ResetSegmentHeader(VP8SegmentHeader* const hdr) {
-  assert(hdr != NULL);
-  hdr->use_segment_ = 0;
-  hdr->update_map_ = 0;
-  hdr->absolute_delta_ = 1;
-  memset(hdr->quantizer_, 0, sizeof(hdr->quantizer_));
-  memset(hdr->filter_strength_, 0, sizeof(hdr->filter_strength_));
-}
-
-// Paragraph 9.3
-static int ParseSegmentHeader(VP8BitReader* br,
-                              VP8SegmentHeader* hdr, VP8Proba* proba) {
-  assert(br != NULL);
-  assert(hdr != NULL);
-  hdr->use_segment_ = VP8Get(br);
-  if (hdr->use_segment_) {
-    hdr->update_map_ = VP8Get(br);
-    if (VP8Get(br)) {   // update data
-      int s;
-      hdr->absolute_delta_ = VP8Get(br);
-      for (s = 0; s < NUM_MB_SEGMENTS; ++s) {
-        hdr->quantizer_[s] = VP8Get(br) ? VP8GetSignedValue(br, 7) : 0;
-      }
-      for (s = 0; s < NUM_MB_SEGMENTS; ++s) {
-        hdr->filter_strength_[s] = VP8Get(br) ? VP8GetSignedValue(br, 6) : 0;
-      }
-    }
-    if (hdr->update_map_) {
-      int s;
-      for (s = 0; s < MB_FEATURE_TREE_PROBS; ++s) {
-        proba->segments_[s] = VP8Get(br) ? VP8GetValue(br, 8) : 255u;
-      }
-    }
-  } else {
-    hdr->update_map_ = 0;
-  }
-  return !br->eof_;
-}
-
-// Paragraph 9.5
-// This function returns VP8_STATUS_SUSPENDED if we don't have all the
-// necessary data in 'buf'.
-// This case is not necessarily an error (for incremental decoding).
-// Still, no bitreader is ever initialized to make it possible to read
-// unavailable memory.
-// If we don't even have the partitions' sizes, than VP8_STATUS_NOT_ENOUGH_DATA
-// is returned, and this is an unrecoverable error.
-// If the partitions were positioned ok, VP8_STATUS_OK is returned.
-static VP8StatusCode ParsePartitions(VP8Decoder* const dec,
-                                     const uint8_t* buf, size_t size) {
-  VP8BitReader* const br = &dec->br_;
-  const uint8_t* sz = buf;
-  const uint8_t* buf_end = buf + size;
-  const uint8_t* part_start;
-  int last_part;
-  int p;
-
-  dec->num_parts_ = 1 << VP8GetValue(br, 2);
-  last_part = dec->num_parts_ - 1;
-  part_start = buf + last_part * 3;
-  if (buf_end < part_start) {
-    // we can't even read the sizes with sz[]! That's a failure.
-    return VP8_STATUS_NOT_ENOUGH_DATA;
-  }
-  for (p = 0; p < last_part; ++p) {
-    const uint32_t psize = sz[0] | (sz[1] << 8) | (sz[2] << 16);
-    const uint8_t* part_end = part_start + psize;
-    if (part_end > buf_end) part_end = buf_end;
-    VP8InitBitReader(dec->parts_ + p, part_start, part_end);
-    part_start = part_end;
-    sz += 3;
-  }
-  VP8InitBitReader(dec->parts_ + last_part, part_start, buf_end);
-  return (part_start < buf_end) ? VP8_STATUS_OK :
-           VP8_STATUS_SUSPENDED;   // Init is ok, but there's not enough data
-}
-
-// Paragraph 9.4
-static int ParseFilterHeader(VP8BitReader* br, VP8Decoder* const dec) {
-  VP8FilterHeader* const hdr = &dec->filter_hdr_;
-  hdr->simple_    = VP8Get(br);
-  hdr->level_     = VP8GetValue(br, 6);
-  hdr->sharpness_ = VP8GetValue(br, 3);
-  hdr->use_lf_delta_ = VP8Get(br);
-  if (hdr->use_lf_delta_) {
-    if (VP8Get(br)) {   // update lf-delta?
-      int i;
-      for (i = 0; i < NUM_REF_LF_DELTAS; ++i) {
-        if (VP8Get(br)) {
-          hdr->ref_lf_delta_[i] = VP8GetSignedValue(br, 6);
-        }
-      }
-      for (i = 0; i < NUM_MODE_LF_DELTAS; ++i) {
-        if (VP8Get(br)) {
-          hdr->mode_lf_delta_[i] = VP8GetSignedValue(br, 6);
-        }
-      }
-    }
-  }
-  dec->filter_type_ = (hdr->level_ == 0) ? 0 : hdr->simple_ ? 1 : 2;
-  return !br->eof_;
-}
-
-// Topmost call
-int VP8GetHeaders(VP8Decoder* const dec, VP8Io* const io) {
-  const uint8_t* buf;
-  size_t buf_size;
-  VP8FrameHeader* frm_hdr;
-  VP8PictureHeader* pic_hdr;
-  VP8BitReader* br;
-  VP8StatusCode status;
-
-  if (dec == NULL) {
-    return 0;
-  }
-  SetOk(dec);
-  if (io == NULL) {
-    return VP8SetError(dec, VP8_STATUS_INVALID_PARAM,
-                       "null VP8Io passed to VP8GetHeaders()");
-  }
-  buf = io->data;
-  buf_size = io->data_size;
-  if (buf_size < 4) {
-    return VP8SetError(dec, VP8_STATUS_NOT_ENOUGH_DATA,
-                       "Truncated header.");
-  }
-
-  // Paragraph 9.1
-  {
-    const uint32_t bits = buf[0] | (buf[1] << 8) | (buf[2] << 16);
-    frm_hdr = &dec->frm_hdr_;
-    frm_hdr->key_frame_ = !(bits & 1);
-    frm_hdr->profile_ = (bits >> 1) & 7;
-    frm_hdr->show_ = (bits >> 4) & 1;
-    frm_hdr->partition_length_ = (bits >> 5);
-    if (frm_hdr->profile_ > 3)
-      return VP8SetError(dec, VP8_STATUS_BITSTREAM_ERROR,
-                         "Incorrect keyframe parameters.");
-    if (!frm_hdr->show_)
-      return VP8SetError(dec, VP8_STATUS_UNSUPPORTED_FEATURE,
-                         "Frame not displayable.");
-    buf += 3;
-    buf_size -= 3;
-  }
-
-  pic_hdr = &dec->pic_hdr_;
-  if (frm_hdr->key_frame_) {
-    // Paragraph 9.2
-    if (buf_size < 7) {
-      return VP8SetError(dec, VP8_STATUS_NOT_ENOUGH_DATA,
-                         "cannot parse picture header");
-    }
-    if (!VP8CheckSignature(buf, buf_size)) {
-      return VP8SetError(dec, VP8_STATUS_BITSTREAM_ERROR,
-                         "Bad code word");
-    }
-    pic_hdr->width_ = ((buf[4] << 8) | buf[3]) & 0x3fff;
-    pic_hdr->xscale_ = buf[4] >> 6;   // ratio: 1, 5/4 5/3 or 2
-    pic_hdr->height_ = ((buf[6] << 8) | buf[5]) & 0x3fff;
-    pic_hdr->yscale_ = buf[6] >> 6;
-    buf += 7;
-    buf_size -= 7;
-
-    dec->mb_w_ = (pic_hdr->width_ + 15) >> 4;
-    dec->mb_h_ = (pic_hdr->height_ + 15) >> 4;
-    // Setup default output area (can be later modified during io->setup())
-    io->width = pic_hdr->width_;
-    io->height = pic_hdr->height_;
-    io->use_scaling  = 0;
-    io->use_cropping = 0;
-    io->crop_top  = 0;
-    io->crop_left = 0;
-    io->crop_right  = io->width;
-    io->crop_bottom = io->height;
-    io->mb_w = io->width;   // sanity check
-    io->mb_h = io->height;  // ditto
-
-    VP8ResetProba(&dec->proba_);
-    ResetSegmentHeader(&dec->segment_hdr_);
-  }
-
-  // Check if we have all the partition #0 available, and initialize dec->br_
-  // to read this partition (and this partition only).
-  if (frm_hdr->partition_length_ > buf_size) {
-    return VP8SetError(dec, VP8_STATUS_NOT_ENOUGH_DATA,
-                       "bad partition length");
-  }
-
-  br = &dec->br_;
-  VP8InitBitReader(br, buf, buf + frm_hdr->partition_length_);
-  buf += frm_hdr->partition_length_;
-  buf_size -= frm_hdr->partition_length_;
-
-  if (frm_hdr->key_frame_) {
-    pic_hdr->colorspace_ = VP8Get(br);
-    pic_hdr->clamp_type_ = VP8Get(br);
-  }
-  if (!ParseSegmentHeader(br, &dec->segment_hdr_, &dec->proba_)) {
-    return VP8SetError(dec, VP8_STATUS_BITSTREAM_ERROR,
-                       "cannot parse segment header");
-  }
-  // Filter specs
-  if (!ParseFilterHeader(br, dec)) {
-    return VP8SetError(dec, VP8_STATUS_BITSTREAM_ERROR,
-                       "cannot parse filter header");
-  }
-  status = ParsePartitions(dec, buf, buf_size);
-  if (status != VP8_STATUS_OK) {
-    return VP8SetError(dec, status, "cannot parse partitions");
-  }
-
-  // quantizer change
-  VP8ParseQuant(dec);
-
-  // Frame buffer marking
-  if (!frm_hdr->key_frame_) {
-    return VP8SetError(dec, VP8_STATUS_UNSUPPORTED_FEATURE,
-                       "Not a key frame.");
-  }
-
-  VP8Get(br);   // ignore the value of update_proba_
-
-  VP8ParseProba(br, dec);
-
-  // sanitized state
-  dec->ready_ = 1;
-  return 1;
-}
-
-//------------------------------------------------------------------------------
-// Residual decoding (Paragraph 13.2 / 13.3)
-
-static const int kBands[16 + 1] = {
-  0, 1, 2, 3, 6, 4, 5, 6, 6, 6, 6, 6, 6, 6, 6, 7,
-  0  // extra entry as sentinel
-};
-
-static const uint8_t kCat3[] = { 173, 148, 140, 0 };
-static const uint8_t kCat4[] = { 176, 155, 140, 135, 0 };
-static const uint8_t kCat5[] = { 180, 157, 141, 134, 130, 0 };
-static const uint8_t kCat6[] =
-  { 254, 254, 243, 230, 196, 177, 153, 140, 133, 130, 129, 0 };
-static const uint8_t* const kCat3456[] = { kCat3, kCat4, kCat5, kCat6 };
-static const uint8_t kZigzag[16] = {
-  0, 1, 4, 8,  5, 2, 3, 6,  9, 12, 13, 10,  7, 11, 14, 15
-};
-
-// See section 13-2: http://tools.ietf.org/html/rfc6386#section-13.2
-static int GetLargeValue(VP8BitReader* const br, const uint8_t* const p) {
-  int v;
-  if (!VP8GetBit(br, p[3])) {
-    if (!VP8GetBit(br, p[4])) {
-      v = 2;
-    } else {
-      v = 3 + VP8GetBit(br, p[5]);
-    }
-  } else {
-    if (!VP8GetBit(br, p[6])) {
-      if (!VP8GetBit(br, p[7])) {
-        v = 5 + VP8GetBit(br, 159);
-      } else {
-        v = 7 + 2 * VP8GetBit(br, 165);
-        v += VP8GetBit(br, 145);
-      }
-    } else {
-      const uint8_t* tab;
-      const int bit1 = VP8GetBit(br, p[8]);
-      const int bit0 = VP8GetBit(br, p[9 + bit1]);
-      const int cat = 2 * bit1 + bit0;
-      v = 0;
-      for (tab = kCat3456[cat]; *tab; ++tab) {
-        v += v + VP8GetBit(br, *tab);
-      }
-      v += 3 + (8 << cat);
-    }
-  }
-  return v;
-}
-
-// Returns the position of the last non-zero coeff plus one
-static int GetCoeffs(VP8BitReader* const br, const VP8BandProbas* const prob,
-                     int ctx, const quant_t dq, int n, int16_t* out) {
-  // n is either 0 or 1 here. kBands[n] is not necessary for extracting '*p'.
-  const uint8_t* p = prob[n].probas_[ctx];
-  for (; n < 16; ++n) {
-    if (!VP8GetBit(br, p[0])) {
-      return n;  // previous coeff was last non-zero coeff
-    }
-    while (!VP8GetBit(br, p[1])) {       // sequence of zero coeffs
-      p = prob[kBands[++n]].probas_[0];
-      if (n == 16) return 16;
-    }
-    {        // non zero coeff
-      const VP8ProbaArray* const p_ctx = &prob[kBands[n + 1]].probas_[0];
-      int v;
-      if (!VP8GetBit(br, p[2])) {
-        v = 1;
-        p = p_ctx[1];
-      } else {
-        v = GetLargeValue(br, p);
-        p = p_ctx[2];
-      }
-      out[kZigzag[n]] = VP8GetSigned(br, v) * dq[n > 0];
-    }
-  }
-  return 16;
-}
-
-static WEBP_INLINE uint32_t NzCodeBits(uint32_t nz_coeffs, int nz, int dc_nz) {
-  nz_coeffs <<= 2;
-  nz_coeffs |= (nz > 3) ? 3 : (nz > 1) ? 2 : dc_nz;
-  return nz_coeffs;
-}
-
-static int ParseResiduals(VP8Decoder* const dec,
-                          VP8MB* const mb, VP8BitReader* const token_br) {
-  VP8BandProbas (* const bands)[NUM_BANDS] = dec->proba_.bands_;
-  const VP8BandProbas* ac_proba;
-  VP8MBData* const block = dec->mb_data_ + dec->mb_x_;
-  const VP8QuantMatrix* const q = &dec->dqm_[block->segment_];
-  int16_t* dst = block->coeffs_;
-  VP8MB* const left_mb = dec->mb_info_ - 1;
-  uint8_t tnz, lnz;
-  uint32_t non_zero_y = 0;
-  uint32_t non_zero_uv = 0;
-  int x, y, ch;
-  uint32_t out_t_nz, out_l_nz;
-  int first;
-
-  memset(dst, 0, 384 * sizeof(*dst));
-  if (!block->is_i4x4_) {    // parse DC
-    int16_t dc[16] = { 0 };
-    const int ctx = mb->nz_dc_ + left_mb->nz_dc_;
-    const int nz = GetCoeffs(token_br, bands[1], ctx, q->y2_mat_, 0, dc);
-    mb->nz_dc_ = left_mb->nz_dc_ = (nz > 0);
-    if (nz > 1) {   // more than just the DC -> perform the full transform
-      VP8TransformWHT(dc, dst);
-    } else {        // only DC is non-zero -> inlined simplified transform
-      int i;
-      const int dc0 = (dc[0] + 3) >> 3;
-      for (i = 0; i < 16 * 16; i += 16) dst[i] = dc0;
-    }
-    first = 1;
-    ac_proba = bands[0];
-  } else {
-    first = 0;
-    ac_proba = bands[3];
-  }
-
-  tnz = mb->nz_ & 0x0f;
-  lnz = left_mb->nz_ & 0x0f;
-  for (y = 0; y < 4; ++y) {
-    int l = lnz & 1;
-    uint32_t nz_coeffs = 0;
-    for (x = 0; x < 4; ++x) {
-      const int ctx = l + (tnz & 1);
-      const int nz = GetCoeffs(token_br, ac_proba, ctx, q->y1_mat_, first, dst);
-      l = (nz > first);
-      tnz = (tnz >> 1) | (l << 7);
-      nz_coeffs = NzCodeBits(nz_coeffs, nz, dst[0] != 0);
-      dst += 16;
-    }
-    tnz >>= 4;
-    lnz = (lnz >> 1) | (l << 7);
-    non_zero_y = (non_zero_y << 8) | nz_coeffs;
-  }
-  out_t_nz = tnz;
-  out_l_nz = lnz >> 4;
-
-  for (ch = 0; ch < 4; ch += 2) {
-    uint32_t nz_coeffs = 0;
-    tnz = mb->nz_ >> (4 + ch);
-    lnz = left_mb->nz_ >> (4 + ch);
-    for (y = 0; y < 2; ++y) {
-      int l = lnz & 1;
-      for (x = 0; x < 2; ++x) {
-        const int ctx = l + (tnz & 1);
-        const int nz = GetCoeffs(token_br, bands[2], ctx, q->uv_mat_, 0, dst);
-        l = (nz > 0);
-        tnz = (tnz >> 1) | (l << 3);
-        nz_coeffs = NzCodeBits(nz_coeffs, nz, dst[0] != 0);
-        dst += 16;
-      }
-      tnz >>= 2;
-      lnz = (lnz >> 1) | (l << 5);
-    }
-    // Note: we don't really need the per-4x4 details for U/V blocks.
-    non_zero_uv |= nz_coeffs << (4 * ch);
-    out_t_nz |= (tnz << 4) << ch;
-    out_l_nz |= (lnz & 0xf0) << ch;
-  }
-  mb->nz_ = out_t_nz;
-  left_mb->nz_ = out_l_nz;
-
-  block->non_zero_y_ = non_zero_y;
-  block->non_zero_uv_ = non_zero_uv;
-
-  // We look at the mode-code of each block and check if some blocks have less
-  // than three non-zero coeffs (code < 2). This is to avoid dithering flat and
-  // empty blocks.
-  block->dither_ = (non_zero_uv & 0xaaaa) ? 0 : q->dither_;
-
-  return !(non_zero_y | non_zero_uv);  // will be used for further optimization
-}
-
-//------------------------------------------------------------------------------
-// Main loop
-
-int VP8DecodeMB(VP8Decoder* const dec, VP8BitReader* const token_br) {
-  VP8MB* const left = dec->mb_info_ - 1;
-  VP8MB* const mb = dec->mb_info_ + dec->mb_x_;
-  VP8MBData* const block = dec->mb_data_ + dec->mb_x_;
-  int skip = dec->use_skip_proba_ ? block->skip_ : 0;
-
-  if (!skip) {
-    skip = ParseResiduals(dec, mb, token_br);
-  } else {
-    left->nz_ = mb->nz_ = 0;
-    if (!block->is_i4x4_) {
-      left->nz_dc_ = mb->nz_dc_ = 0;
-    }
-    block->non_zero_y_ = 0;
-    block->non_zero_uv_ = 0;
-  }
-
-  if (dec->filter_type_ > 0) {  // store filter info
-    VP8FInfo* const finfo = dec->f_info_ + dec->mb_x_;
-    *finfo = dec->fstrengths_[block->segment_][block->is_i4x4_];
-    finfo->f_inner_ |= !skip;
-  }
-
-  return !token_br->eof_;
-}
-
-void VP8InitScanline(VP8Decoder* const dec) {
-  VP8MB* const left = dec->mb_info_ - 1;
-  left->nz_ = 0;
-  left->nz_dc_ = 0;
-  memset(dec->intra_l_, B_DC_PRED, sizeof(dec->intra_l_));
-  dec->mb_x_ = 0;
-}
-
-static int ParseFrame(VP8Decoder* const dec, VP8Io* io) {
-  for (dec->mb_y_ = 0; dec->mb_y_ < dec->br_mb_y_; ++dec->mb_y_) {
-    // Parse bitstream for this row.
-    VP8BitReader* const token_br =
-        &dec->parts_[dec->mb_y_ & (dec->num_parts_ - 1)];
-    if (!VP8ParseIntraModeRow(&dec->br_, dec)) {
-      return VP8SetError(dec, VP8_STATUS_NOT_ENOUGH_DATA,
-                         "Premature end-of-partition0 encountered.");
-    }
-    for (; dec->mb_x_ < dec->mb_w_; ++dec->mb_x_) {
-      if (!VP8DecodeMB(dec, token_br)) {
-        return VP8SetError(dec, VP8_STATUS_NOT_ENOUGH_DATA,
-                           "Premature end-of-file encountered.");
-      }
-    }
-    VP8InitScanline(dec);   // Prepare for next scanline
-
-    // Reconstruct, filter and emit the row.
-    if (!VP8ProcessRow(dec, io)) {
-      return VP8SetError(dec, VP8_STATUS_USER_ABORT, "Output aborted.");
-    }
-  }
-  if (dec->mt_method_ > 0) {
-    if (!WebPGetWorkerInterface()->Sync(&dec->worker_)) return 0;
-  }
-
-  return 1;
-}
-
-// Main entry point
-int VP8Decode(VP8Decoder* const dec, VP8Io* const io) {
-  int ok = 0;
-  if (dec == NULL) {
-    return 0;
-  }
-  if (io == NULL) {
-    return VP8SetError(dec, VP8_STATUS_INVALID_PARAM,
-                       "NULL VP8Io parameter in VP8Decode().");
-  }
-
-  if (!dec->ready_) {
-    if (!VP8GetHeaders(dec, io)) {
-      return 0;
-    }
-  }
-  assert(dec->ready_);
-
-  // Finish setting up the decoding parameter. Will call io->setup().
-  ok = (VP8EnterCritical(dec, io) == VP8_STATUS_OK);
-  if (ok) {   // good to go.
-    // Will allocate memory and prepare everything.
-    if (ok) ok = VP8InitFrame(dec, io);
-
-    // Main decoding loop
-    if (ok) ok = ParseFrame(dec, io);
-
-    // Exit.
-    ok &= VP8ExitCritical(dec, io);
-  }
-
-  if (!ok) {
-    VP8Clear(dec);
-    return 0;
-  }
-
-  dec->ready_ = 0;
-  return ok;
-}
-
-void VP8Clear(VP8Decoder* const dec) {
-  if (dec == NULL) {
-    return;
-  }
-  WebPGetWorkerInterface()->End(&dec->worker_);
-  ALPHDelete(dec->alph_dec_);
-  dec->alph_dec_ = NULL;
-  WebPSafeFree(dec->mem_);
-  dec->mem_ = NULL;
-  dec->mem_size_ = 0;
-  memset(&dec->br_, 0, sizeof(dec->br_));
-  dec->ready_ = 0;
-}
-
-//------------------------------------------------------------------------------
-
@@ -1,354 +0,0 @@
-// Copyright 2010 Google Inc. All Rights Reserved.
-//
-// Use of this source code is governed by a BSD-style license
-// that can be found in the COPYING file in the root of the source
-// tree. An additional intellectual property rights grant can be found
-// in the file PATENTS. All contributing project authors may
-// be found in the AUTHORS file in the root of the source tree.
-// -----------------------------------------------------------------------------
-//
-// VP8 decoder: internal header.
-//
-// Author: Skal (pascal.massimino@gmail.com)
-
-#ifndef WEBP_DEC_VP8I_H_
-#define WEBP_DEC_VP8I_H_
-
-#include <string.h>     // for memcpy()
-#include "./vp8li.h"
-#include "../utils/bit_reader.h"
-#include "../utils/random.h"
-#include "../utils/thread.h"
-#include "../dsp/dsp.h"
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-//------------------------------------------------------------------------------
-// Various defines and enums
-
-// version numbers
-#define DEC_MAJ_VERSION 0
-#define DEC_MIN_VERSION 4
-#define DEC_REV_VERSION 2
-
-// intra prediction modes
-enum { B_DC_PRED = 0,   // 4x4 modes
-       B_TM_PRED,
-       B_VE_PRED,
-       B_HE_PRED,
-       B_RD_PRED,
-       B_VR_PRED,
-       B_LD_PRED,
-       B_VL_PRED,
-       B_HD_PRED,
-       B_HU_PRED,
-       NUM_BMODES = B_HU_PRED + 1 - B_DC_PRED,  // = 10
-
-       // Luma16 or UV modes
-       DC_PRED = B_DC_PRED, V_PRED = B_VE_PRED,
-       H_PRED = B_HE_PRED, TM_PRED = B_TM_PRED,
-       B_PRED = NUM_BMODES,   // refined I4x4 mode
-
-       // special modes
-       B_DC_PRED_NOTOP = 4,
-       B_DC_PRED_NOLEFT = 5,
-       B_DC_PRED_NOTOPLEFT = 6,
-       NUM_B_DC_MODES = 7 };
-
-enum { MB_FEATURE_TREE_PROBS = 3,
-       NUM_MB_SEGMENTS = 4,
-       NUM_REF_LF_DELTAS = 4,
-       NUM_MODE_LF_DELTAS = 4,    // I4x4, ZERO, *, SPLIT
-       MAX_NUM_PARTITIONS = 8,
-       // Probabilities
-       NUM_TYPES = 4,
-       NUM_BANDS = 8,
-       NUM_CTX = 3,
-       NUM_PROBAS = 11,
-       NUM_MV_PROBAS = 19 };
-
-// YUV-cache parameters.
-// Constraints are: We need to store one 16x16 block of luma samples (y),
-// and two 8x8 chroma blocks (u/v). These are better be 16-bytes aligned,
-// in order to be SIMD-friendly. We also need to store the top, left and
-// top-left samples (from previously decoded blocks), along with four
-// extra top-right samples for luma (intra4x4 prediction only).
-// One possible layout is, using 32 * (17 + 9) bytes:
-//
-//   .+------   <- only 1 pixel high
-//   .|yyyyt.
-//   .|yyyyt.
-//   .|yyyyt.
-//   .|yyyy..
-//   .+--.+--   <- only 1 pixel high
-//   .|uu.|vv
-//   .|uu.|vv
-//
-// Every character is a 4x4 block, with legend:
-//  '.' = unused
-//  'y' = y-samples   'u' = u-samples     'v' = u-samples
-//  '|' = left sample,   '-' = top sample,    '+' = top-left sample
-//  't' = extra top-right sample for 4x4 modes
-// With this layout, BPS (=Bytes Per Scan-line) is one cacheline size.
-#define BPS       32    // this is the common stride used by yuv[]
-#define YUV_SIZE (BPS * 17 + BPS * 9)
-#define Y_SIZE   (BPS * 17)
-#define Y_OFF    (BPS * 1 + 8)
-#define U_OFF    (Y_OFF + BPS * 16 + BPS)
-#define V_OFF    (U_OFF + 16)
-
-// minimal width under which lossy multi-threading is always disabled
-#define MIN_WIDTH_FOR_THREADS 512
-
-//------------------------------------------------------------------------------
-// Headers
-
-typedef struct {
-  uint8_t key_frame_;
-  uint8_t profile_;
-  uint8_t show_;
-  uint32_t partition_length_;
-} VP8FrameHeader;
-
-typedef struct {
-  uint16_t width_;
-  uint16_t height_;
-  uint8_t xscale_;
-  uint8_t yscale_;
-  uint8_t colorspace_;   // 0 = YCbCr
-  uint8_t clamp_type_;
-} VP8PictureHeader;
-
-// segment features
-typedef struct {
-  int use_segment_;
-  int update_map_;        // whether to update the segment map or not
-  int absolute_delta_;    // absolute or delta values for quantizer and filter
-  int8_t quantizer_[NUM_MB_SEGMENTS];        // quantization changes
-  int8_t filter_strength_[NUM_MB_SEGMENTS];  // filter strength for segments
-} VP8SegmentHeader;
-
-
-// probas associated to one of the contexts
-typedef uint8_t VP8ProbaArray[NUM_PROBAS];
-
-typedef struct {   // all the probas associated to one band
-  VP8ProbaArray probas_[NUM_CTX];
-} VP8BandProbas;
-
-// Struct collecting all frame-persistent probabilities.
-typedef struct {
-  uint8_t segments_[MB_FEATURE_TREE_PROBS];
-  // Type: 0:Intra16-AC  1:Intra16-DC   2:Chroma   3:Intra4
-  VP8BandProbas bands_[NUM_TYPES][NUM_BANDS];
-} VP8Proba;
-
-// Filter parameters
-typedef struct {
-  int simple_;                  // 0=complex, 1=simple
-  int level_;                   // [0..63]
-  int sharpness_;               // [0..7]
-  int use_lf_delta_;
-  int ref_lf_delta_[NUM_REF_LF_DELTAS];
-  int mode_lf_delta_[NUM_MODE_LF_DELTAS];
-} VP8FilterHeader;
-
-//------------------------------------------------------------------------------
-// Informations about the macroblocks.
-
-typedef struct {  // filter specs
-  uint8_t f_limit_;      // filter limit in [3..189], or 0 if no filtering
-  uint8_t f_ilevel_;     // inner limit in [1..63]
-  uint8_t f_inner_;      // do inner filtering?
-  uint8_t hev_thresh_;   // high edge variance threshold in [0..2]
-} VP8FInfo;
-
-typedef struct {  // Top/Left Contexts used for syntax-parsing
-  uint8_t nz_;        // non-zero AC/DC coeffs (4bit for luma + 4bit for chroma)
-  uint8_t nz_dc_;     // non-zero DC coeff (1bit)
-} VP8MB;
-
-// Dequantization matrices
-typedef int quant_t[2];      // [DC / AC].  Can be 'uint16_t[2]' too (~slower).
-typedef struct {
-  quant_t y1_mat_, y2_mat_, uv_mat_;
-
-  int uv_quant_;   // U/V quantizer value
-  int dither_;     // dithering amplitude (0 = off, max=255)
-} VP8QuantMatrix;
-
-// Data needed to reconstruct a macroblock
-typedef struct {
-  int16_t coeffs_[384];   // 384 coeffs = (16+4+4) * 4*4
-  uint8_t is_i4x4_;       // true if intra4x4
-  uint8_t imodes_[16];    // one 16x16 mode (#0) or sixteen 4x4 modes
-  uint8_t uvmode_;        // chroma prediction mode
-  // bit-wise info about the content of each sub-4x4 blocks (in decoding order).
-  // Each of the 4x4 blocks for y/u/v is associated with a 2b code according to:
-  //   code=0 -> no coefficient
-  //   code=1 -> only DC
-  //   code=2 -> first three coefficients are non-zero
-  //   code=3 -> more than three coefficients are non-zero
-  // This allows to call specialized transform functions.
-  uint32_t non_zero_y_;
-  uint32_t non_zero_uv_;
-  uint8_t dither_;      // local dithering strength (deduced from non_zero_*)
-  uint8_t skip_;
-  uint8_t segment_;
-} VP8MBData;
-
-// Persistent information needed by the parallel processing
-typedef struct {
-  int id_;              // cache row to process (in [0..2])
-  int mb_y_;            // macroblock position of the row
-  int filter_row_;      // true if row-filtering is needed
-  VP8FInfo* f_info_;    // filter strengths (swapped with dec->f_info_)
-  VP8MBData* mb_data_;  // reconstruction data (swapped with dec->mb_data_)
-  VP8Io io_;            // copy of the VP8Io to pass to put()
-} VP8ThreadContext;
-
-// Saved top samples, per macroblock. Fits into a cache-line.
-typedef struct {
-  uint8_t y[16], u[8], v[8];
-} VP8TopSamples;
-
-//------------------------------------------------------------------------------
-// VP8Decoder: the main opaque structure handed over to user
-
-struct VP8Decoder {
-  VP8StatusCode status_;
-  int ready_;     // true if ready to decode a picture with VP8Decode()
-  const char* error_msg_;  // set when status_ is not OK.
-
-  // Main data source
-  VP8BitReader br_;
-
-  // headers
-  VP8FrameHeader   frm_hdr_;
-  VP8PictureHeader pic_hdr_;
-  VP8FilterHeader  filter_hdr_;
-  VP8SegmentHeader segment_hdr_;
-
-  // Worker
-  WebPWorker worker_;
-  int mt_method_;      // multi-thread method: 0=off, 1=[parse+recon][filter]
-                       // 2=[parse][recon+filter]
-  int cache_id_;       // current cache row
-  int num_caches_;     // number of cached rows of 16 pixels (1, 2 or 3)
-  VP8ThreadContext thread_ctx_;  // Thread context
-
-  // dimension, in macroblock units.
-  int mb_w_, mb_h_;
-
-  // Macroblock to process/filter, depending on cropping and filter_type.
-  int tl_mb_x_, tl_mb_y_;  // top-left MB that must be in-loop filtered
-  int br_mb_x_, br_mb_y_;  // last bottom-right MB that must be decoded
-
-  // number of partitions.
-  int num_parts_;
-  // per-partition boolean decoders.
-  VP8BitReader parts_[MAX_NUM_PARTITIONS];
-
-  // Dithering strength, deduced from decoding options
-  int dither_;                // whether to use dithering or not
-  VP8Random dithering_rg_;    // random generator for dithering
-
-  // dequantization (one set of DC/AC dequant factor per segment)
-  VP8QuantMatrix dqm_[NUM_MB_SEGMENTS];
-
-  // probabilities
-  VP8Proba proba_;
-  int use_skip_proba_;
-  uint8_t skip_p_;
-
-  // Boundary data cache and persistent buffers.
-  uint8_t* intra_t_;      // top intra modes values: 4 * mb_w_
-  uint8_t  intra_l_[4];   // left intra modes values
-
-  VP8TopSamples* yuv_t_;  // top y/u/v samples
-
-  VP8MB* mb_info_;        // contextual macroblock info (mb_w_ + 1)
-  VP8FInfo* f_info_;      // filter strength info
-  uint8_t* yuv_b_;        // main block for Y/U/V (size = YUV_SIZE)
-
-  uint8_t* cache_y_;      // macroblock row for storing unfiltered samples
-  uint8_t* cache_u_;
-  uint8_t* cache_v_;
-  int cache_y_stride_;
-  int cache_uv_stride_;
-
-  // main memory chunk for the above data. Persistent.
-  void* mem_;
-  size_t mem_size_;
-
-  // Per macroblock non-persistent infos.
-  int mb_x_, mb_y_;       // current position, in macroblock units
-  VP8MBData* mb_data_;    // parsed reconstruction data
-
-  // Filtering side-info
-  int filter_type_;                          // 0=off, 1=simple, 2=complex
-  VP8FInfo fstrengths_[NUM_MB_SEGMENTS][2];  // precalculated per-segment/type
-
-  // Alpha
-  struct ALPHDecoder* alph_dec_;  // alpha-plane decoder object
-  const uint8_t* alpha_data_;     // compressed alpha data (if present)
-  size_t alpha_data_size_;
-  int is_alpha_decoded_;  // true if alpha_data_ is decoded in alpha_plane_
-  uint8_t* alpha_plane_;  // output. Persistent, contains the whole data.
-  int alpha_dithering_;   // derived from decoding options (0=off, 100=full).
-};
-
-//------------------------------------------------------------------------------
-// internal functions. Not public.
-
-// in vp8.c
-int VP8SetError(VP8Decoder* const dec,
-                VP8StatusCode error, const char* const msg);
-
-// in tree.c
-void VP8ResetProba(VP8Proba* const proba);
-void VP8ParseProba(VP8BitReader* const br, VP8Decoder* const dec);
-// parses one row of intra mode data in partition 0, returns !eof
-int VP8ParseIntraModeRow(VP8BitReader* const br, VP8Decoder* const dec);
-
-// in quant.c
-void VP8ParseQuant(VP8Decoder* const dec);
-
-// in frame.c
-int VP8InitFrame(VP8Decoder* const dec, VP8Io* io);
-// Call io->setup() and finish setting up scan parameters.
-// After this call returns, one must always call VP8ExitCritical() with the
-// same parameters. Both functions should be used in pair. Returns VP8_STATUS_OK
-// if ok, otherwise sets and returns the error status on *dec.
-VP8StatusCode VP8EnterCritical(VP8Decoder* const dec, VP8Io* const io);
-// Must always be called in pair with VP8EnterCritical().
-// Returns false in case of error.
-int VP8ExitCritical(VP8Decoder* const dec, VP8Io* const io);
-// Return the multi-threading method to use (0=off), depending
-// on options and bitstream size. Only for lossy decoding.
-int VP8GetThreadMethod(const WebPDecoderOptions* const options,
-                       const WebPHeaderStructure* const headers,
-                       int width, int height);
-// Initialize dithering post-process if needed.
-void VP8InitDithering(const WebPDecoderOptions* const options,
-                      VP8Decoder* const dec);
-// Process the last decoded row (filtering + output).
-int VP8ProcessRow(VP8Decoder* const dec, VP8Io* const io);
-// To be called at the start of a new scanline, to initialize predictors.
-void VP8InitScanline(VP8Decoder* const dec);
-// Decode one macroblock. Returns false if there is not enough data.
-int VP8DecodeMB(VP8Decoder* const dec, VP8BitReader* const token_br);
-
-// in alpha.c
-const uint8_t* VP8DecompressAlphaRows(VP8Decoder* const dec,
-                                      int row, int num_rows);
-
-//------------------------------------------------------------------------------
-
-#ifdef __cplusplus
-}    // extern "C"
-#endif
-
-#endif  /* WEBP_DEC_VP8I_H_ */
@@ -1,132 +0,0 @@
-// Copyright 2012 Google Inc. All Rights Reserved.
-//
-// Use of this source code is governed by a BSD-style license
-// that can be found in the COPYING file in the root of the source
-// tree. An additional intellectual property rights grant can be found
-// in the file PATENTS. All contributing project authors may
-// be found in the AUTHORS file in the root of the source tree.
-// -----------------------------------------------------------------------------
-//
-// Lossless decoder: internal header.
-//
-// Author: Skal (pascal.massimino@gmail.com)
-//         Vikas Arora(vikaas.arora@gmail.com)
-
-#ifndef WEBP_DEC_VP8LI_H_
-#define WEBP_DEC_VP8LI_H_
-
-#include <string.h>     // for memcpy()
-#include "./webpi.h"
-#include "../utils/bit_reader.h"
-#include "../utils/color_cache.h"
-#include "../utils/huffman.h"
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-typedef enum {
-  READ_DATA = 0,
-  READ_HDR = 1,
-  READ_DIM = 2
-} VP8LDecodeState;
-
-typedef struct VP8LTransform VP8LTransform;
-struct VP8LTransform {
-  VP8LImageTransformType type_;   // transform type.
-  int                    bits_;   // subsampling bits defining transform window.
-  int                    xsize_;  // transform window X index.
-  int                    ysize_;  // transform window Y index.
-  uint32_t              *data_;   // transform data.
-};
-
-typedef struct {
-  int             color_cache_size_;
-  VP8LColorCache  color_cache_;
-
-  int             huffman_mask_;
-  int             huffman_subsample_bits_;
-  int             huffman_xsize_;
-  uint32_t       *huffman_image_;
-  int             num_htree_groups_;
-  HTreeGroup     *htree_groups_;
-} VP8LMetadata;
-
-typedef struct VP8LDecoder VP8LDecoder;
-struct VP8LDecoder {
-  VP8StatusCode    status_;
-  VP8LDecodeState  action_;
-  VP8LDecodeState  state_;
-  VP8Io           *io_;
-
-  const WebPDecBuffer *output_;    // shortcut to io->opaque->output
-
-  uint32_t        *pixels_;        // Internal data: either uint8_t* for alpha
-                                   // or uint32_t* for BGRA.
-  uint32_t        *argb_cache_;    // Scratch buffer for temporary BGRA storage.
-
-  VP8LBitReader    br_;
-
-  int              width_;
-  int              height_;
-  int              last_row_;      // last input row decoded so far.
-  int              last_pixel_;    // last pixel decoded so far. However, it may
-                                   // not be transformed, scaled and
-                                   // color-converted yet.
-  int              last_out_row_;  // last row output so far.
-
-  VP8LMetadata     hdr_;
-
-  int              next_transform_;
-  VP8LTransform    transforms_[NUM_TRANSFORMS];
-  // or'd bitset storing the transforms types.
-  uint32_t         transforms_seen_;
-
-  uint8_t         *rescaler_memory;  // Working memory for rescaling work.
-  WebPRescaler    *rescaler;         // Common rescaler for all channels.
-};
-
-//------------------------------------------------------------------------------
-// internal functions. Not public.
-
-struct ALPHDecoder;  // Defined in dec/alphai.h.
-
-// in vp8l.c
-
-// Decodes image header for alpha data stored using lossless compression.
-// Returns false in case of error.
-int VP8LDecodeAlphaHeader(struct ALPHDecoder* const alph_dec,
-                          const uint8_t* const data, size_t data_size,
-                          uint8_t* const output);
-
-// Decodes *at least* 'last_row' rows of alpha. If some of the initial rows are
-// already decoded in previous call(s), it will resume decoding from where it
-// was paused.
-// Returns false in case of bitstream error.
-int VP8LDecodeAlphaImageStream(struct ALPHDecoder* const alph_dec,
-                               int last_row);
-
-// Allocates and initialize a new lossless decoder instance.
-VP8LDecoder* VP8LNew(void);
-
-// Decodes the image header. Returns false in case of error.
-int VP8LDecodeHeader(VP8LDecoder* const dec, VP8Io* const io);
-
-// Decodes an image. It's required to decode the lossless header before calling
-// this function. Returns false in case of error, with updated dec->status_.
-int VP8LDecodeImage(VP8LDecoder* const dec);
-
-// Resets the decoder in its initial state, reclaiming memory.
-// Preserves the dec->status_ value.
-void VP8LClear(VP8LDecoder* const dec);
-
-// Clears and deallocate a lossless decoder instance.
-void VP8LDelete(VP8LDecoder* const dec);
-
-//------------------------------------------------------------------------------
-
-#ifdef __cplusplus
-}    // extern "C"
-#endif
-
-#endif  /* WEBP_DEC_VP8LI_H_ */
@@ -1,836 +0,0 @@
-// Copyright 2010 Google Inc. All Rights Reserved.
-//
-// Use of this source code is governed by a BSD-style license
-// that can be found in the COPYING file in the root of the source
-// tree. An additional intellectual property rights grant can be found
-// in the file PATENTS. All contributing project authors may
-// be found in the AUTHORS file in the root of the source tree.
-// -----------------------------------------------------------------------------
-//
-// Main decoding functions for WEBP images.
-//
-// Author: Skal (pascal.massimino@gmail.com)
-
-#include <stdlib.h>
-
-#include "./vp8i.h"
-#include "./vp8li.h"
-#include "./webpi.h"
-#include "../webp/mux_types.h"  // ALPHA_FLAG
-
-//------------------------------------------------------------------------------
-// RIFF layout is:
-//   Offset  tag
-//   0...3   "RIFF" 4-byte tag
-//   4...7   size of image data (including metadata) starting at offset 8
-//   8...11  "WEBP"   our form-type signature
-// The RIFF container (12 bytes) is followed by appropriate chunks:
-//   12..15  "VP8 ": 4-bytes tags, signaling the use of VP8 video format
-//   16..19  size of the raw VP8 image data, starting at offset 20
-//   20....  the VP8 bytes
-// Or,
-//   12..15  "VP8L": 4-bytes tags, signaling the use of VP8L lossless format
-//   16..19  size of the raw VP8L image data, starting at offset 20
-//   20....  the VP8L bytes
-// Or,
-//   12..15  "VP8X": 4-bytes tags, describing the extended-VP8 chunk.
-//   16..19  size of the VP8X chunk starting at offset 20.
-//   20..23  VP8X flags bit-map corresponding to the chunk-types present.
-//   24..26  Width of the Canvas Image.
-//   27..29  Height of the Canvas Image.
-// There can be extra chunks after the "VP8X" chunk (ICCP, FRGM, ANMF, VP8,
-// VP8L, XMP, EXIF  ...)
-// All sizes are in little-endian order.
-// Note: chunk data size must be padded to multiple of 2 when written.
-
-static WEBP_INLINE uint32_t get_le24(const uint8_t* const data) {
-  return data[0] | (data[1] << 8) | (data[2] << 16);
-}
-
-static WEBP_INLINE uint32_t get_le32(const uint8_t* const data) {
-  return (uint32_t)get_le24(data) | (data[3] << 24);
-}
-
-// Validates the RIFF container (if detected) and skips over it.
-// If a RIFF container is detected, returns:
-//     VP8_STATUS_BITSTREAM_ERROR for invalid header,
-//     VP8_STATUS_NOT_ENOUGH_DATA for truncated data if have_all_data is true,
-// and VP8_STATUS_OK otherwise.
-// In case there are not enough bytes (partial RIFF container), return 0 for
-// *riff_size. Else return the RIFF size extracted from the header.
-static VP8StatusCode ParseRIFF(const uint8_t** const data,
-                               size_t* const data_size, int have_all_data,
-                               size_t* const riff_size) {
-  assert(data != NULL);
-  assert(data_size != NULL);
-  assert(riff_size != NULL);
-
-  *riff_size = 0;  // Default: no RIFF present.
-  if (*data_size >= RIFF_HEADER_SIZE && !memcmp(*data, "RIFF", TAG_SIZE)) {
-    if (memcmp(*data + 8, "WEBP", TAG_SIZE)) {
-      return VP8_STATUS_BITSTREAM_ERROR;  // Wrong image file signature.
-    } else {
-      const uint32_t size = get_le32(*data + TAG_SIZE);
-      // Check that we have at least one chunk (i.e "WEBP" + "VP8?nnnn").
-      if (size < TAG_SIZE + CHUNK_HEADER_SIZE) {
-        return VP8_STATUS_BITSTREAM_ERROR;
-      }
-      if (size > MAX_CHUNK_PAYLOAD) {
-        return VP8_STATUS_BITSTREAM_ERROR;
-      }
-      if (have_all_data && (size > *data_size - CHUNK_HEADER_SIZE)) {
-        return VP8_STATUS_NOT_ENOUGH_DATA;  // Truncated bitstream.
-      }
-      // We have a RIFF container. Skip it.
-      *riff_size = size;
-      *data += RIFF_HEADER_SIZE;
-      *data_size -= RIFF_HEADER_SIZE;
-    }
-  }
-  return VP8_STATUS_OK;
-}
-
-// Validates the VP8X header and skips over it.
-// Returns VP8_STATUS_BITSTREAM_ERROR for invalid VP8X header,
-//         VP8_STATUS_NOT_ENOUGH_DATA in case of insufficient data, and
-//         VP8_STATUS_OK otherwise.
-// If a VP8X chunk is found, found_vp8x is set to true and *width_ptr,
-// *height_ptr and *flags_ptr are set to the corresponding values extracted
-// from the VP8X chunk.
-static VP8StatusCode ParseVP8X(const uint8_t** const data,
-                               size_t* const data_size,
-                               int* const found_vp8x,
-                               int* const width_ptr, int* const height_ptr,
-                               uint32_t* const flags_ptr) {
-  const uint32_t vp8x_size = CHUNK_HEADER_SIZE + VP8X_CHUNK_SIZE;
-  assert(data != NULL);
-  assert(data_size != NULL);
-  assert(found_vp8x != NULL);
-
-  *found_vp8x = 0;
-
-  if (*data_size < CHUNK_HEADER_SIZE) {
-    return VP8_STATUS_NOT_ENOUGH_DATA;  // Insufficient data.
-  }
-
-  if (!memcmp(*data, "VP8X", TAG_SIZE)) {
-    int width, height;
-    uint32_t flags;
-    const uint32_t chunk_size = get_le32(*data + TAG_SIZE);
-    if (chunk_size != VP8X_CHUNK_SIZE) {
-      return VP8_STATUS_BITSTREAM_ERROR;  // Wrong chunk size.
-    }
-
-    // Verify if enough data is available to validate the VP8X chunk.
-    if (*data_size < vp8x_size) {
-      return VP8_STATUS_NOT_ENOUGH_DATA;  // Insufficient data.
-    }
-    flags = get_le32(*data + 8);
-    width = 1 + get_le24(*data + 12);
-    height = 1 + get_le24(*data + 15);
-    if (width * (uint64_t)height >= MAX_IMAGE_AREA) {
-      return VP8_STATUS_BITSTREAM_ERROR;  // image is too large
-    }
-
-    if (flags_ptr != NULL) *flags_ptr = flags;
-    if (width_ptr != NULL) *width_ptr = width;
-    if (height_ptr != NULL) *height_ptr = height;
-    // Skip over VP8X header bytes.
-    *data += vp8x_size;
-    *data_size -= vp8x_size;
-    *found_vp8x = 1;
-  }
-  return VP8_STATUS_OK;
-}
-
-// Skips to the next VP8/VP8L chunk header in the data given the size of the
-// RIFF chunk 'riff_size'.
-// Returns VP8_STATUS_BITSTREAM_ERROR if any invalid chunk size is encountered,
-//         VP8_STATUS_NOT_ENOUGH_DATA in case of insufficient data, and
-//         VP8_STATUS_OK otherwise.
-// If an alpha chunk is found, *alpha_data and *alpha_size are set
-// appropriately.
-static VP8StatusCode ParseOptionalChunks(const uint8_t** const data,
-                                         size_t* const data_size,
-                                         size_t const riff_size,
-                                         const uint8_t** const alpha_data,
-                                         size_t* const alpha_size) {
-  const uint8_t* buf;
-  size_t buf_size;
-  uint32_t total_size = TAG_SIZE +           // "WEBP".
-                        CHUNK_HEADER_SIZE +  // "VP8Xnnnn".
-                        VP8X_CHUNK_SIZE;     // data.
-  assert(data != NULL);
-  assert(data_size != NULL);
-  buf = *data;
-  buf_size = *data_size;
-
-  assert(alpha_data != NULL);
-  assert(alpha_size != NULL);
-  *alpha_data = NULL;
-  *alpha_size = 0;
-
-  while (1) {
-    uint32_t chunk_size;
-    uint32_t disk_chunk_size;   // chunk_size with padding
-
-    *data = buf;
-    *data_size = buf_size;
-
-    if (buf_size < CHUNK_HEADER_SIZE) {  // Insufficient data.
-      return VP8_STATUS_NOT_ENOUGH_DATA;
-    }
-
-    chunk_size = get_le32(buf + TAG_SIZE);
-    if (chunk_size > MAX_CHUNK_PAYLOAD) {
-      return VP8_STATUS_BITSTREAM_ERROR;          // Not a valid chunk size.
-    }
-    // For odd-sized chunk-payload, there's one byte padding at the end.
-    disk_chunk_size = (CHUNK_HEADER_SIZE + chunk_size + 1) & ~1;
-    total_size += disk_chunk_size;
-
-    // Check that total bytes skipped so far does not exceed riff_size.
-    if (riff_size > 0 && (total_size > riff_size)) {
-      return VP8_STATUS_BITSTREAM_ERROR;          // Not a valid chunk size.
-    }
-
-    // Start of a (possibly incomplete) VP8/VP8L chunk implies that we have
-    // parsed all the optional chunks.
-    // Note: This check must occur before the check 'buf_size < disk_chunk_size'
-    // below to allow incomplete VP8/VP8L chunks.
-    if (!memcmp(buf, "VP8 ", TAG_SIZE) ||
-        !memcmp(buf, "VP8L", TAG_SIZE)) {
-      return VP8_STATUS_OK;
-    }
-
-    if (buf_size < disk_chunk_size) {             // Insufficient data.
-      return VP8_STATUS_NOT_ENOUGH_DATA;
-    }
-
-    if (!memcmp(buf, "ALPH", TAG_SIZE)) {         // A valid ALPH header.
-      *alpha_data = buf + CHUNK_HEADER_SIZE;
-      *alpha_size = chunk_size;
-    }
-
-    // We have a full and valid chunk; skip it.
-    buf += disk_chunk_size;
-    buf_size -= disk_chunk_size;
-  }
-}
-
-// Validates the VP8/VP8L Header ("VP8 nnnn" or "VP8L nnnn") and skips over it.
-// Returns VP8_STATUS_BITSTREAM_ERROR for invalid (chunk larger than
-//         riff_size) VP8/VP8L header,
-//         VP8_STATUS_NOT_ENOUGH_DATA in case of insufficient data, and
-//         VP8_STATUS_OK otherwise.
-// If a VP8/VP8L chunk is found, *chunk_size is set to the total number of bytes
-// extracted from the VP8/VP8L chunk header.
-// The flag '*is_lossless' is set to 1 in case of VP8L chunk / raw VP8L data.
-static VP8StatusCode ParseVP8Header(const uint8_t** const data_ptr,
-                                    size_t* const data_size, int have_all_data,
-                                    size_t riff_size, size_t* const chunk_size,
-                                    int* const is_lossless) {
-  const uint8_t* const data = *data_ptr;
-  const int is_vp8 = !memcmp(data, "VP8 ", TAG_SIZE);
-  const int is_vp8l = !memcmp(data, "VP8L", TAG_SIZE);
-  const uint32_t minimal_size =
-      TAG_SIZE + CHUNK_HEADER_SIZE;  // "WEBP" + "VP8 nnnn" OR
-                                     // "WEBP" + "VP8Lnnnn"
-  assert(data != NULL);
-  assert(data_size != NULL);
-  assert(chunk_size != NULL);
-  assert(is_lossless != NULL);
-
-  if (*data_size < CHUNK_HEADER_SIZE) {
-    return VP8_STATUS_NOT_ENOUGH_DATA;  // Insufficient data.
-  }
-
-  if (is_vp8 || is_vp8l) {
-    // Bitstream contains VP8/VP8L header.
-    const uint32_t size = get_le32(data + TAG_SIZE);
-    if ((riff_size >= minimal_size) && (size > riff_size - minimal_size)) {
-      return VP8_STATUS_BITSTREAM_ERROR;  // Inconsistent size information.
-    }
-    if (have_all_data && (size > *data_size - CHUNK_HEADER_SIZE)) {
-      return VP8_STATUS_NOT_ENOUGH_DATA;  // Truncated bitstream.
-    }
-    // Skip over CHUNK_HEADER_SIZE bytes from VP8/VP8L Header.
-    *chunk_size = size;
-    *data_ptr += CHUNK_HEADER_SIZE;
-    *data_size -= CHUNK_HEADER_SIZE;
-    *is_lossless = is_vp8l;
-  } else {
-    // Raw VP8/VP8L bitstream (no header).
-    *is_lossless = VP8LCheckSignature(data, *data_size);
-    *chunk_size = *data_size;
-  }
-
-  return VP8_STATUS_OK;
-}
-
-//------------------------------------------------------------------------------
-
-// Fetch '*width', '*height', '*has_alpha' and fill out 'headers' based on
-// 'data'. All the output parameters may be NULL. If 'headers' is NULL only the
-// minimal amount will be read to fetch the remaining parameters.
-// If 'headers' is non-NULL this function will attempt to locate both alpha
-// data (with or without a VP8X chunk) and the bitstream chunk (VP8/VP8L).
-// Note: The following chunk sequences (before the raw VP8/VP8L data) are
-// considered valid by this function:
-// RIFF + VP8(L)
-// RIFF + VP8X + (optional chunks) + VP8(L)
-// ALPH + VP8 <-- Not a valid WebP format: only allowed for internal purpose.
-// VP8(L)     <-- Not a valid WebP format: only allowed for internal purpose.
-static VP8StatusCode ParseHeadersInternal(const uint8_t* data,
-                                          size_t data_size,
-                                          int* const width,
-                                          int* const height,
-                                          int* const has_alpha,
-                                          int* const has_animation,
-                                          int* const format,
-                                          WebPHeaderStructure* const headers) {
-  int canvas_width = 0;
-  int canvas_height = 0;
-  int image_width = 0;
-  int image_height = 0;
-  int found_riff = 0;
-  int found_vp8x = 0;
-  int animation_present = 0;
-  int fragments_present = 0;
-  const int have_all_data = (headers != NULL) ? headers->have_all_data : 0;
-
-  VP8StatusCode status;
-  WebPHeaderStructure hdrs;
-
-  if (data == NULL || data_size < RIFF_HEADER_SIZE) {
-    return VP8_STATUS_NOT_ENOUGH_DATA;
-  }
-  memset(&hdrs, 0, sizeof(hdrs));
-  hdrs.data = data;
-  hdrs.data_size = data_size;
-
-  // Skip over RIFF header.
-  status = ParseRIFF(&data, &data_size, have_all_data, &hdrs.riff_size);
-  if (status != VP8_STATUS_OK) {
-    return status;   // Wrong RIFF header / insufficient data.
-  }
-  found_riff = (hdrs.riff_size > 0);
-
-  // Skip over VP8X.
-  {
-    uint32_t flags = 0;
-    status = ParseVP8X(&data, &data_size, &found_vp8x,
-                       &canvas_width, &canvas_height, &flags);
-    if (status != VP8_STATUS_OK) {
-      return status;  // Wrong VP8X / insufficient data.
-    }
-    animation_present = !!(flags & ANIMATION_FLAG);
-    fragments_present = !!(flags & FRAGMENTS_FLAG);
-    if (!found_riff && found_vp8x) {
-      // Note: This restriction may be removed in the future, if it becomes
-      // necessary to send VP8X chunk to the decoder.
-      return VP8_STATUS_BITSTREAM_ERROR;
-    }
-    if (has_alpha != NULL) *has_alpha = !!(flags & ALPHA_FLAG);
-    if (has_animation != NULL) *has_animation = animation_present;
-    if (format != NULL) *format = 0;   // default = undefined
-
-    image_width = canvas_width;
-    image_height = canvas_height;
-    if (found_vp8x && (animation_present || fragments_present) &&
-        headers == NULL) {
-      status = VP8_STATUS_OK;
-      goto ReturnWidthHeight;  // Just return features from VP8X header.
-    }
-  }
-
-  if (data_size < TAG_SIZE) {
-    status = VP8_STATUS_NOT_ENOUGH_DATA;
-    goto ReturnWidthHeight;
-  }
-
-  // Skip over optional chunks if data started with "RIFF + VP8X" or "ALPH".
-  if ((found_riff && found_vp8x) ||
-      (!found_riff && !found_vp8x && !memcmp(data, "ALPH", TAG_SIZE))) {
-    status = ParseOptionalChunks(&data, &data_size, hdrs.riff_size,
-                                 &hdrs.alpha_data, &hdrs.alpha_data_size);
-    if (status != VP8_STATUS_OK) {
-      goto ReturnWidthHeight;  // Invalid chunk size / insufficient data.
-    }
-  }
-
-  // Skip over VP8/VP8L header.
-  status = ParseVP8Header(&data, &data_size, have_all_data, hdrs.riff_size,
-                          &hdrs.compressed_size, &hdrs.is_lossless);
-  if (status != VP8_STATUS_OK) {
-    goto ReturnWidthHeight;  // Wrong VP8/VP8L chunk-header / insufficient data.
-  }
-  if (hdrs.compressed_size > MAX_CHUNK_PAYLOAD) {
-    return VP8_STATUS_BITSTREAM_ERROR;
-  }
-
-  if (format != NULL && !(animation_present || fragments_present)) {
-    *format = hdrs.is_lossless ? 2 : 1;
-  }
-
-  if (!hdrs.is_lossless) {
-    if (data_size < VP8_FRAME_HEADER_SIZE) {
-      status = VP8_STATUS_NOT_ENOUGH_DATA;
-      goto ReturnWidthHeight;
-    }
-    // Validates raw VP8 data.
-    if (!VP8GetInfo(data, data_size, (uint32_t)hdrs.compressed_size,
-                    &image_width, &image_height)) {
-      return VP8_STATUS_BITSTREAM_ERROR;
-    }
-  } else {
-    if (data_size < VP8L_FRAME_HEADER_SIZE) {
-      status = VP8_STATUS_NOT_ENOUGH_DATA;
-      goto ReturnWidthHeight;
-    }
-    // Validates raw VP8L data.
-    if (!VP8LGetInfo(data, data_size, &image_width, &image_height, has_alpha)) {
-      return VP8_STATUS_BITSTREAM_ERROR;
-    }
-  }
-  // Validates image size coherency.
-  if (found_vp8x) {
-    if (canvas_width != image_width || canvas_height != image_height) {
-      return VP8_STATUS_BITSTREAM_ERROR;
-    }
-  }
-  if (headers != NULL) {
-    *headers = hdrs;
-    headers->offset = data - headers->data;
-    assert((uint64_t)(data - headers->data) < MAX_CHUNK_PAYLOAD);
-    assert(headers->offset == headers->data_size - data_size);
-  }
- ReturnWidthHeight:
-  if (status == VP8_STATUS_OK ||
-      (status == VP8_STATUS_NOT_ENOUGH_DATA && found_vp8x && headers == NULL)) {
-    if (has_alpha != NULL) {
-      // If the data did not contain a VP8X/VP8L chunk the only definitive way
-      // to set this is by looking for alpha data (from an ALPH chunk).
-      *has_alpha |= (hdrs.alpha_data != NULL);
-    }
-    if (width != NULL) *width = image_width;
-    if (height != NULL) *height = image_height;
-    return VP8_STATUS_OK;
-  } else {
-    return status;
-  }
-}
-
-VP8StatusCode WebPParseHeaders(WebPHeaderStructure* const headers) {
-  VP8StatusCode status;
-  int has_animation = 0;
-  assert(headers != NULL);
-  // fill out headers, ignore width/height/has_alpha.
-  status = ParseHeadersInternal(headers->data, headers->data_size,
-                                NULL, NULL, NULL, &has_animation,
-                                NULL, headers);
-  if (status == VP8_STATUS_OK || status == VP8_STATUS_NOT_ENOUGH_DATA) {
-    // TODO(jzern): full support of animation frames will require API additions.
-    if (has_animation) {
-      status = VP8_STATUS_UNSUPPORTED_FEATURE;
-    }
-  }
-  return status;
-}
-
-//------------------------------------------------------------------------------
-// WebPDecParams
-
-void WebPResetDecParams(WebPDecParams* const params) {
-  if (params != NULL) {
-    memset(params, 0, sizeof(*params));
-  }
-}
-
-//------------------------------------------------------------------------------
-// "Into" decoding variants
-
-// Main flow
-static VP8StatusCode DecodeInto(const uint8_t* const data, size_t data_size,
-                                WebPDecParams* const params) {
-  VP8StatusCode status;
-  VP8Io io;
-  WebPHeaderStructure headers;
-
-  headers.data = data;
-  headers.data_size = data_size;
-  headers.have_all_data = 1;
-  status = WebPParseHeaders(&headers);   // Process Pre-VP8 chunks.
-  if (status != VP8_STATUS_OK) {
-    return status;
-  }
-
-  assert(params != NULL);
-  VP8InitIo(&io);
-  io.data = headers.data + headers.offset;
-  io.data_size = headers.data_size - headers.offset;
-  WebPInitCustomIo(params, &io);  // Plug the I/O functions.
-
-  if (!headers.is_lossless) {
-    VP8Decoder* const dec = VP8New();
-    if (dec == NULL) {
-      return VP8_STATUS_OUT_OF_MEMORY;
-    }
-    dec->alpha_data_ = headers.alpha_data;
-    dec->alpha_data_size_ = headers.alpha_data_size;
-
-    // Decode bitstream header, update io->width/io->height.
-    if (!VP8GetHeaders(dec, &io)) {
-      status = dec->status_;   // An error occurred. Grab error status.
-    } else {
-      // Allocate/check output buffers.
-      status = WebPAllocateDecBuffer(io.width, io.height, params->options,
-                                     params->output);
-      if (status == VP8_STATUS_OK) {  // Decode
-        // This change must be done before calling VP8Decode()
-        dec->mt_method_ = VP8GetThreadMethod(params->options, &headers,
-                                             io.width, io.height);
-        VP8InitDithering(params->options, dec);
-        if (!VP8Decode(dec, &io)) {
-          status = dec->status_;
-        }
-      }
-    }
-    VP8Delete(dec);
-  } else {
-    VP8LDecoder* const dec = VP8LNew();
-    if (dec == NULL) {
-      return VP8_STATUS_OUT_OF_MEMORY;
-    }
-    if (!VP8LDecodeHeader(dec, &io)) {
-      status = dec->status_;   // An error occurred. Grab error status.
-    } else {
-      // Allocate/check output buffers.
-      status = WebPAllocateDecBuffer(io.width, io.height, params->options,
-                                     params->output);
-      if (status == VP8_STATUS_OK) {  // Decode
-        if (!VP8LDecodeImage(dec)) {
-          status = dec->status_;
-        }
-      }
-    }
-    VP8LDelete(dec);
-  }
-
-  if (status != VP8_STATUS_OK) {
-    WebPFreeDecBuffer(params->output);
-  }
-
-#if WEBP_DECODER_ABI_VERSION > 0x0203
-  if (params->options != NULL && params->options->flip) {
-    status = WebPFlipBuffer(params->output);
-  }
-#endif
-  return status;
-}
-
-// Helpers
-static uint8_t* DecodeIntoRGBABuffer(WEBP_CSP_MODE colorspace,
-                                     const uint8_t* const data,
-                                     size_t data_size,
-                                     uint8_t* const rgba,
-                                     int stride, size_t size) {
-  WebPDecParams params;
-  WebPDecBuffer buf;
-  if (rgba == NULL) {
-    return NULL;
-  }
-  WebPInitDecBuffer(&buf);
-  WebPResetDecParams(&params);
-  params.output = &buf;
-  buf.colorspace    = colorspace;
-  buf.u.RGBA.rgba   = rgba;
-  buf.u.RGBA.stride = stride;
-  buf.u.RGBA.size   = size;
-  buf.is_external_memory = 1;
-  if (DecodeInto(data, data_size, &params) != VP8_STATUS_OK) {
-    return NULL;
-  }
-  return rgba;
-}
-
-uint8_t* WebPDecodeRGBInto(const uint8_t* data, size_t data_size,
-                           uint8_t* output, size_t size, int stride) {
-  return DecodeIntoRGBABuffer(MODE_RGB, data, data_size, output, stride, size);
-}
-
-uint8_t* WebPDecodeRGBAInto(const uint8_t* data, size_t data_size,
-                            uint8_t* output, size_t size, int stride) {
-  return DecodeIntoRGBABuffer(MODE_rgbA, data, data_size, output, stride, size);
-}
-
-uint8_t* WebPDecodeARGBInto(const uint8_t* data, size_t data_size,
-                            uint8_t* output, size_t size, int stride) {
-  return DecodeIntoRGBABuffer(MODE_ARGB, data, data_size, output, stride, size);
-}
-
-uint8_t* WebPDecodeBGRInto(const uint8_t* data, size_t data_size,
-                           uint8_t* output, size_t size, int stride) {
-  return DecodeIntoRGBABuffer(MODE_BGR, data, data_size, output, stride, size);
-}
-
-uint8_t* WebPDecodeBGRAInto(const uint8_t* data, size_t data_size,
-                            uint8_t* output, size_t size, int stride) {
-  return DecodeIntoRGBABuffer(MODE_BGRA, data, data_size, output, stride, size);
-}
-
-uint8_t* WebPDecodeYUVInto(const uint8_t* data, size_t data_size,
-                           uint8_t* luma, size_t luma_size, int luma_stride,
-                           uint8_t* u, size_t u_size, int u_stride,
-                           uint8_t* v, size_t v_size, int v_stride) {
-  WebPDecParams params;
-  WebPDecBuffer output;
-  if (luma == NULL) return NULL;
-  WebPInitDecBuffer(&output);
-  WebPResetDecParams(&params);
-  params.output = &output;
-  output.colorspace      = MODE_YUV;
-  output.u.YUVA.y        = luma;
-  output.u.YUVA.y_stride = luma_stride;
-  output.u.YUVA.y_size   = luma_size;
-  output.u.YUVA.u        = u;
-  output.u.YUVA.u_stride = u_stride;
-  output.u.YUVA.u_size   = u_size;
-  output.u.YUVA.v        = v;
-  output.u.YUVA.v_stride = v_stride;
-  output.u.YUVA.v_size   = v_size;
-  output.is_external_memory = 1;
-  if (DecodeInto(data, data_size, &params) != VP8_STATUS_OK) {
-    return NULL;
-  }
-  return luma;
-}
-
-//------------------------------------------------------------------------------
-
-static uint8_t* Decode(WEBP_CSP_MODE mode, const uint8_t* const data,
-                       size_t data_size, int* const width, int* const height,
-                       WebPDecBuffer* const keep_info) {
-  WebPDecParams params;
-  WebPDecBuffer output;
-
-  WebPInitDecBuffer(&output);
-  WebPResetDecParams(&params);
-  params.output = &output;
-  output.colorspace = mode;
-
-  // Retrieve (and report back) the required dimensions from bitstream.
-  if (!WebPGetInfo(data, data_size, &output.width, &output.height)) {
-    return NULL;
-  }
-  if (width != NULL) *width = output.width;
-  if (height != NULL) *height = output.height;
-
-  // Decode
-  if (DecodeInto(data, data_size, &params) != VP8_STATUS_OK) {
-    return NULL;
-  }
-  if (keep_info != NULL) {    // keep track of the side-info
-    WebPCopyDecBuffer(&output, keep_info);
-  }
-  // return decoded samples (don't clear 'output'!)
-  return WebPIsRGBMode(mode) ? output.u.RGBA.rgba : output.u.YUVA.y;
-}
-
-uint8_t* WebPDecodeRGB(const uint8_t* data, size_t data_size,
-                       int* width, int* height) {
-  return Decode(MODE_RGB, data, data_size, width, height, NULL);
-}
-
-uint8_t* WebPDecodeRGBA(const uint8_t* data, size_t data_size,
-                        int* width, int* height) {
-  return Decode(MODE_RGBA, data, data_size, width, height, NULL);
-}
-
-uint8_t* WebPDecodeARGB(const uint8_t* data, size_t data_size,
-                        int* width, int* height) {
-  return Decode(MODE_ARGB, data, data_size, width, height, NULL);
-}
-
-uint8_t* WebPDecodeBGR(const uint8_t* data, size_t data_size,
-                       int* width, int* height) {
-  return Decode(MODE_BGR, data, data_size, width, height, NULL);
-}
-
-uint8_t* WebPDecodeBGRA(const uint8_t* data, size_t data_size,
-                        int* width, int* height) {
-  return Decode(MODE_BGRA, data, data_size, width, height, NULL);
-}
-
-uint8_t* WebPDecodeYUV(const uint8_t* data, size_t data_size,
-                       int* width, int* height, uint8_t** u, uint8_t** v,
-                       int* stride, int* uv_stride) {
-  WebPDecBuffer output;   // only to preserve the side-infos
-  uint8_t* const out = Decode(MODE_YUV, data, data_size,
-                              width, height, &output);
-
-  if (out != NULL) {
-    const WebPYUVABuffer* const buf = &output.u.YUVA;
-    *u = buf->u;
-    *v = buf->v;
-    *stride = buf->y_stride;
-    *uv_stride = buf->u_stride;
-    assert(buf->u_stride == buf->v_stride);
-  }
-  return out;
-}
-
-static void DefaultFeatures(WebPBitstreamFeatures* const features) {
-  assert(features != NULL);
-  memset(features, 0, sizeof(*features));
-}
-
-static VP8StatusCode GetFeatures(const uint8_t* const data, size_t data_size,
-                                 WebPBitstreamFeatures* const features) {
-  if (features == NULL || data == NULL) {
-    return VP8_STATUS_INVALID_PARAM;
-  }
-  DefaultFeatures(features);
-
-  // Only parse enough of the data to retrieve the features.
-  return ParseHeadersInternal(data, data_size,
-                              &features->width, &features->height,
-                              &features->has_alpha, &features->has_animation,
-                              &features->format, NULL);
-}
-
-//------------------------------------------------------------------------------
-// WebPGetInfo()
-
-int WebPGetInfo(const uint8_t* data, size_t data_size,
-                int* width, int* height) {
-  WebPBitstreamFeatures features;
-
-  if (GetFeatures(data, data_size, &features) != VP8_STATUS_OK) {
-    return 0;
-  }
-
-  if (width != NULL) {
-    *width  = features.width;
-  }
-  if (height != NULL) {
-    *height = features.height;
-  }
-
-  return 1;
-}
-
-//------------------------------------------------------------------------------
-// Advance decoding API
-
-int WebPInitDecoderConfigInternal(WebPDecoderConfig* config,
-                                  int version) {
-  if (WEBP_ABI_IS_INCOMPATIBLE(version, WEBP_DECODER_ABI_VERSION)) {
-    return 0;   // version mismatch
-  }
-  if (config == NULL) {
-    return 0;
-  }
-  memset(config, 0, sizeof(*config));
-  DefaultFeatures(&config->input);
-  WebPInitDecBuffer(&config->output);
-  return 1;
-}
-
-VP8StatusCode WebPGetFeaturesInternal(const uint8_t* data, size_t data_size,
-                                      WebPBitstreamFeatures* features,
-                                      int version) {
-  if (WEBP_ABI_IS_INCOMPATIBLE(version, WEBP_DECODER_ABI_VERSION)) {
-    return VP8_STATUS_INVALID_PARAM;   // version mismatch
-  }
-  if (features == NULL) {
-    return VP8_STATUS_INVALID_PARAM;
-  }
-  return GetFeatures(data, data_size, features);
-}
-
-VP8StatusCode WebPDecode(const uint8_t* data, size_t data_size,
-                         WebPDecoderConfig* config) {
-  WebPDecParams params;
-  VP8StatusCode status;
-
-  if (config == NULL) {
-    return VP8_STATUS_INVALID_PARAM;
-  }
-
-  status = GetFeatures(data, data_size, &config->input);
-  if (status != VP8_STATUS_OK) {
-    if (status == VP8_STATUS_NOT_ENOUGH_DATA) {
-      return VP8_STATUS_BITSTREAM_ERROR;  // Not-enough-data treated as error.
-    }
-    return status;
-  }
-
-  WebPResetDecParams(&params);
-  params.output = &config->output;
-  params.options = &config->options;
-  status = DecodeInto(data, data_size, &params);
-
-  return status;
-}
-
-//------------------------------------------------------------------------------
-// Cropping and rescaling.
-
-int WebPIoInitFromOptions(const WebPDecoderOptions* const options,
-                          VP8Io* const io, WEBP_CSP_MODE src_colorspace) {
-  const int W = io->width;
-  const int H = io->height;
-  int x = 0, y = 0, w = W, h = H;
-
-  // Cropping
-  io->use_cropping = (options != NULL) && (options->use_cropping > 0);
-  if (io->use_cropping) {
-    w = options->crop_width;
-    h = options->crop_height;
-    x = options->crop_left;
-    y = options->crop_top;
-    if (!WebPIsRGBMode(src_colorspace)) {   // only snap for YUV420
-      x &= ~1;
-      y &= ~1;
-    }
-    if (x < 0 || y < 0 || w <= 0 || h <= 0 || x + w > W || y + h > H) {
-      return 0;  // out of frame boundary error
-    }
-  }
-  io->crop_left   = x;
-  io->crop_top    = y;
-  io->crop_right  = x + w;
-  io->crop_bottom = y + h;
-  io->mb_w = w;
-  io->mb_h = h;
-
-  // Scaling
-  io->use_scaling = (options != NULL) && (options->use_scaling > 0);
-  if (io->use_scaling) {
-    if (options->scaled_width <= 0 || options->scaled_height <= 0) {
-      return 0;
-    }
-    io->scaled_width = options->scaled_width;
-    io->scaled_height = options->scaled_height;
-  }
-
-  // Filter
-  io->bypass_filtering = options && options->bypass_filtering;
-
-  // Fancy upsampler
-#ifdef FANCY_UPSAMPLING
-  io->fancy_upsampling = (options == NULL) || (!options->no_fancy_upsampling);
-#endif
-
-  if (io->use_scaling) {
-    // disable filter (only for large downscaling ratio).
-    io->bypass_filtering = (io->scaled_width < W * 3 / 4) &&
-                           (io->scaled_height < H * 3 / 4);
-    io->fancy_upsampling = 0;
-  }
-  return 1;
-}
-
-//------------------------------------------------------------------------------
-
@@ -1,120 +0,0 @@
-// Copyright 2011 Google Inc. All Rights Reserved.
-//
-// Use of this source code is governed by a BSD-style license
-// that can be found in the COPYING file in the root of the source
-// tree. An additional intellectual property rights grant can be found
-// in the file PATENTS. All contributing project authors may
-// be found in the AUTHORS file in the root of the source tree.
-// -----------------------------------------------------------------------------
-//
-// Internal header: WebP decoding parameters and custom IO on buffer
-//
-// Author: somnath@google.com (Somnath Banerjee)
-
-#ifndef WEBP_DEC_WEBPI_H_
-#define WEBP_DEC_WEBPI_H_
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-#include "../utils/rescaler.h"
-#include "./decode_vp8.h"
-
-//------------------------------------------------------------------------------
-// WebPDecParams: Decoding output parameters. Transient internal object.
-
-typedef struct WebPDecParams WebPDecParams;
-typedef int (*OutputFunc)(const VP8Io* const io, WebPDecParams* const p);
-typedef int (*OutputRowFunc)(WebPDecParams* const p, int y_pos);
-
-struct WebPDecParams {
-  WebPDecBuffer* output;             // output buffer.
-  uint8_t* tmp_y, *tmp_u, *tmp_v;    // cache for the fancy upsampler
-                                     // or used for tmp rescaling
-
-  int last_y;                 // coordinate of the line that was last output
-  const WebPDecoderOptions* options;  // if not NULL, use alt decoding features
-  // rescalers
-  WebPRescaler scaler_y, scaler_u, scaler_v, scaler_a;
-  void* memory;                  // overall scratch memory for the output work.
-
-  OutputFunc emit;               // output RGB or YUV samples
-  OutputFunc emit_alpha;         // output alpha channel
-  OutputRowFunc emit_alpha_row;  // output one line of rescaled alpha values
-};
-
-// Should be called first, before any use of the WebPDecParams object.
-void WebPResetDecParams(WebPDecParams* const params);
-
-//------------------------------------------------------------------------------
-// Header parsing helpers
-
-// Structure storing a description of the RIFF headers.
-typedef struct {
-  const uint8_t* data;         // input buffer
-  size_t data_size;            // input buffer size
-  int have_all_data;           // true if all data is known to be available
-  size_t offset;               // offset to main data chunk (VP8 or VP8L)
-  const uint8_t* alpha_data;   // points to alpha chunk (if present)
-  size_t alpha_data_size;      // alpha chunk size
-  size_t compressed_size;      // VP8/VP8L compressed data size
-  size_t riff_size;            // size of the riff payload (or 0 if absent)
-  int is_lossless;             // true if a VP8L chunk is present
-} WebPHeaderStructure;
-
-// Skips over all valid chunks prior to the first VP8/VP8L frame header.
-// Returns: VP8_STATUS_OK, VP8_STATUS_BITSTREAM_ERROR (invalid header/chunk),
-// VP8_STATUS_NOT_ENOUGH_DATA (partial input) or VP8_STATUS_UNSUPPORTED_FEATURE
-// in the case of non-decodable features (animation for instance).
-// In 'headers', compressed_size, offset, alpha_data, alpha_size, and lossless
-// fields are updated appropriately upon success.
-VP8StatusCode WebPParseHeaders(WebPHeaderStructure* const headers);
-
-//------------------------------------------------------------------------------
-// Misc utils
-
-// Initializes VP8Io with custom setup, io and teardown functions. The default
-// hooks will use the supplied 'params' as io->opaque handle.
-void WebPInitCustomIo(WebPDecParams* const params, VP8Io* const io);
-
-// Setup crop_xxx fields, mb_w and mb_h in io. 'src_colorspace' refers
-// to the *compressed* format, not the output one.
-int WebPIoInitFromOptions(const WebPDecoderOptions* const options,
-                          VP8Io* const io, WEBP_CSP_MODE src_colorspace);
-
-//------------------------------------------------------------------------------
-// Internal functions regarding WebPDecBuffer memory (in buffer.c).
-// Don't really need to be externally visible for now.
-
-// Prepare 'buffer' with the requested initial dimensions width/height.
-// If no external storage is supplied, initializes buffer by allocating output
-// memory and setting up the stride information. Validate the parameters. Return
-// an error code in case of problem (no memory, or invalid stride / size /
-// dimension / etc.). If *options is not NULL, also verify that the options'
-// parameters are valid and apply them to the width/height dimensions of the
-// output buffer. This takes cropping / scaling / rotation into account.
-// Also incorporates the options->flip flag to flip the buffer parameters if
-// needed.
-VP8StatusCode WebPAllocateDecBuffer(int width, int height,
-                                    const WebPDecoderOptions* const options,
-                                    WebPDecBuffer* const buffer);
-
-// Flip buffer vertically by negating the various strides.
-VP8StatusCode WebPFlipBuffer(WebPDecBuffer* const buffer);
-
-// Copy 'src' into 'dst' buffer, making sure 'dst' is not marked as owner of the
-// memory (still held by 'src').
-void WebPCopyDecBuffer(const WebPDecBuffer* const src,
-                       WebPDecBuffer* const dst);
-
-// Copy and transfer ownership from src to dst (beware of parameter order!)
-void WebPGrabDecBuffer(WebPDecBuffer* const src, WebPDecBuffer* const dst);
-
-//------------------------------------------------------------------------------
-
-#ifdef __cplusplus
-}    // extern "C"
-#endif
-
-#endif  /* WEBP_DEC_WEBPI_H_ */
@@ -1,329 +0,0 @@
-// Copyright 2013 Google Inc. All Rights Reserved.
-//
-// Use of this source code is governed by a BSD-style license
-// that can be found in the COPYING file in the root of the source
-// tree. An additional intellectual property rights grant can be found
-// in the file PATENTS. All contributing project authors may
-// be found in the AUTHORS file in the root of the source tree.
-// -----------------------------------------------------------------------------
-//
-// Utilities for processing transparent channel.
-//
-// Author: Skal (pascal.massimino@gmail.com)
-
-#include <assert.h>
-#include "./dsp.h"
-
-// Tables can be faster on some platform but incur some extra binary size (~2k).
-// #define USE_TABLES_FOR_ALPHA_MULT
-
-// -----------------------------------------------------------------------------
-
-#define MFIX 24    // 24bit fixed-point arithmetic
-#define HALF ((1u << MFIX) >> 1)
-#define KINV_255 ((1u << MFIX) / 255u)
-
-static uint32_t Mult(uint8_t x, uint32_t mult) {
-  const uint32_t v = (x * mult + HALF) >> MFIX;
-  assert(v <= 255);  // <- 24bit precision is enough to ensure that.
-  return v;
-}
-
-#ifdef USE_TABLES_FOR_ALPHA_MULT
-
-static const uint32_t kMultTables[2][256] = {
-  {    // (255u << MFIX) / alpha
-    0x00000000, 0xff000000, 0x7f800000, 0x55000000, 0x3fc00000, 0x33000000,
-    0x2a800000, 0x246db6db, 0x1fe00000, 0x1c555555, 0x19800000, 0x172e8ba2,
-    0x15400000, 0x139d89d8, 0x1236db6d, 0x11000000, 0x0ff00000, 0x0f000000,
-    0x0e2aaaaa, 0x0d6bca1a, 0x0cc00000, 0x0c249249, 0x0b9745d1, 0x0b1642c8,
-    0x0aa00000, 0x0a333333, 0x09cec4ec, 0x0971c71c, 0x091b6db6, 0x08cb08d3,
-    0x08800000, 0x0839ce73, 0x07f80000, 0x07ba2e8b, 0x07800000, 0x07492492,
-    0x07155555, 0x06e45306, 0x06b5e50d, 0x0689d89d, 0x06600000, 0x063831f3,
-    0x06124924, 0x05ee23b8, 0x05cba2e8, 0x05aaaaaa, 0x058b2164, 0x056cefa8,
-    0x05500000, 0x05343eb1, 0x05199999, 0x05000000, 0x04e76276, 0x04cfb2b7,
-    0x04b8e38e, 0x04a2e8ba, 0x048db6db, 0x0479435e, 0x04658469, 0x045270d0,
-    0x04400000, 0x042e29f7, 0x041ce739, 0x040c30c3, 0x03fc0000, 0x03ec4ec4,
-    0x03dd1745, 0x03ce540f, 0x03c00000, 0x03b21642, 0x03a49249, 0x03976fc6,
-    0x038aaaaa, 0x037e3f1f, 0x03722983, 0x03666666, 0x035af286, 0x034fcace,
-    0x0344ec4e, 0x033a5440, 0x03300000, 0x0325ed09, 0x031c18f9, 0x0312818a,
-    0x03092492, 0x03000000, 0x02f711dc, 0x02ee5846, 0x02e5d174, 0x02dd7baf,
-    0x02d55555, 0x02cd5cd5, 0x02c590b2, 0x02bdef7b, 0x02b677d4, 0x02af286b,
-    0x02a80000, 0x02a0fd5c, 0x029a1f58, 0x029364d9, 0x028ccccc, 0x0286562d,
-    0x02800000, 0x0279c952, 0x0273b13b, 0x026db6db, 0x0267d95b, 0x026217ec,
-    0x025c71c7, 0x0256e62a, 0x0251745d, 0x024c1bac, 0x0246db6d, 0x0241b2f9,
-    0x023ca1af, 0x0237a6f4, 0x0232c234, 0x022df2df, 0x02293868, 0x02249249,
-    0x02200000, 0x021b810e, 0x021714fb, 0x0212bb51, 0x020e739c, 0x020a3d70,
-    0x02061861, 0x02020408, 0x01fe0000, 0x01fa0be8, 0x01f62762, 0x01f25213,
-    0x01ee8ba2, 0x01ead3ba, 0x01e72a07, 0x01e38e38, 0x01e00000, 0x01dc7f10,
-    0x01d90b21, 0x01d5a3e9, 0x01d24924, 0x01cefa8d, 0x01cbb7e3, 0x01c880e5,
-    0x01c55555, 0x01c234f7, 0x01bf1f8f, 0x01bc14e5, 0x01b914c1, 0x01b61eed,
-    0x01b33333, 0x01b05160, 0x01ad7943, 0x01aaaaaa, 0x01a7e567, 0x01a5294a,
-    0x01a27627, 0x019fcbd2, 0x019d2a20, 0x019a90e7, 0x01980000, 0x01957741,
-    0x0192f684, 0x01907da4, 0x018e0c7c, 0x018ba2e8, 0x018940c5, 0x0186e5f0,
-    0x01849249, 0x018245ae, 0x01800000, 0x017dc11f, 0x017b88ee, 0x0179574e,
-    0x01772c23, 0x01750750, 0x0172e8ba, 0x0170d045, 0x016ebdd7, 0x016cb157,
-    0x016aaaaa, 0x0168a9b9, 0x0166ae6a, 0x0164b8a7, 0x0162c859, 0x0160dd67,
-    0x015ef7bd, 0x015d1745, 0x015b3bea, 0x01596596, 0x01579435, 0x0155c7b4,
-    0x01540000, 0x01523d03, 0x01507eae, 0x014ec4ec, 0x014d0fac, 0x014b5edc,
-    0x0149b26c, 0x01480a4a, 0x01466666, 0x0144c6af, 0x01432b16, 0x0141938b,
-    0x01400000, 0x013e7063, 0x013ce4a9, 0x013b5cc0, 0x0139d89d, 0x01385830,
-    0x0136db6d, 0x01356246, 0x0133ecad, 0x01327a97, 0x01310bf6, 0x012fa0be,
-    0x012e38e3, 0x012cd459, 0x012b7315, 0x012a150a, 0x0128ba2e, 0x01276276,
-    0x01260dd6, 0x0124bc44, 0x01236db6, 0x01222222, 0x0120d97c, 0x011f93bc,
-    0x011e50d7, 0x011d10c4, 0x011bd37a, 0x011a98ef, 0x0119611a, 0x01182bf2,
-    0x0116f96f, 0x0115c988, 0x01149c34, 0x0113716a, 0x01124924, 0x01112358,
-    0x01100000, 0x010edf12, 0x010dc087, 0x010ca458, 0x010b8a7d, 0x010a72f0,
-    0x01095da8, 0x01084a9f, 0x010739ce, 0x01062b2e, 0x01051eb8, 0x01041465,
-    0x01030c30, 0x01020612, 0x01010204, 0x01000000 },
-  {   // alpha * KINV_255
-    0x00000000, 0x00010101, 0x00020202, 0x00030303, 0x00040404, 0x00050505,
-    0x00060606, 0x00070707, 0x00080808, 0x00090909, 0x000a0a0a, 0x000b0b0b,
-    0x000c0c0c, 0x000d0d0d, 0x000e0e0e, 0x000f0f0f, 0x00101010, 0x00111111,
-    0x00121212, 0x00131313, 0x00141414, 0x00151515, 0x00161616, 0x00171717,
-    0x00181818, 0x00191919, 0x001a1a1a, 0x001b1b1b, 0x001c1c1c, 0x001d1d1d,
-    0x001e1e1e, 0x001f1f1f, 0x00202020, 0x00212121, 0x00222222, 0x00232323,
-    0x00242424, 0x00252525, 0x00262626, 0x00272727, 0x00282828, 0x00292929,
-    0x002a2a2a, 0x002b2b2b, 0x002c2c2c, 0x002d2d2d, 0x002e2e2e, 0x002f2f2f,
-    0x00303030, 0x00313131, 0x00323232, 0x00333333, 0x00343434, 0x00353535,
-    0x00363636, 0x00373737, 0x00383838, 0x00393939, 0x003a3a3a, 0x003b3b3b,
-    0x003c3c3c, 0x003d3d3d, 0x003e3e3e, 0x003f3f3f, 0x00404040, 0x00414141,
-    0x00424242, 0x00434343, 0x00444444, 0x00454545, 0x00464646, 0x00474747,
-    0x00484848, 0x00494949, 0x004a4a4a, 0x004b4b4b, 0x004c4c4c, 0x004d4d4d,
-    0x004e4e4e, 0x004f4f4f, 0x00505050, 0x00515151, 0x00525252, 0x00535353,
-    0x00545454, 0x00555555, 0x00565656, 0x00575757, 0x00585858, 0x00595959,
-    0x005a5a5a, 0x005b5b5b, 0x005c5c5c, 0x005d5d5d, 0x005e5e5e, 0x005f5f5f,
-    0x00606060, 0x00616161, 0x00626262, 0x00636363, 0x00646464, 0x00656565,
-    0x00666666, 0x00676767, 0x00686868, 0x00696969, 0x006a6a6a, 0x006b6b6b,
-    0x006c6c6c, 0x006d6d6d, 0x006e6e6e, 0x006f6f6f, 0x00707070, 0x00717171,
-    0x00727272, 0x00737373, 0x00747474, 0x00757575, 0x00767676, 0x00777777,
-    0x00787878, 0x00797979, 0x007a7a7a, 0x007b7b7b, 0x007c7c7c, 0x007d7d7d,
-    0x007e7e7e, 0x007f7f7f, 0x00808080, 0x00818181, 0x00828282, 0x00838383,
-    0x00848484, 0x00858585, 0x00868686, 0x00878787, 0x00888888, 0x00898989,
-    0x008a8a8a, 0x008b8b8b, 0x008c8c8c, 0x008d8d8d, 0x008e8e8e, 0x008f8f8f,
-    0x00909090, 0x00919191, 0x00929292, 0x00939393, 0x00949494, 0x00959595,
-    0x00969696, 0x00979797, 0x00989898, 0x00999999, 0x009a9a9a, 0x009b9b9b,
-    0x009c9c9c, 0x009d9d9d, 0x009e9e9e, 0x009f9f9f, 0x00a0a0a0, 0x00a1a1a1,
-    0x00a2a2a2, 0x00a3a3a3, 0x00a4a4a4, 0x00a5a5a5, 0x00a6a6a6, 0x00a7a7a7,
-    0x00a8a8a8, 0x00a9a9a9, 0x00aaaaaa, 0x00ababab, 0x00acacac, 0x00adadad,
-    0x00aeaeae, 0x00afafaf, 0x00b0b0b0, 0x00b1b1b1, 0x00b2b2b2, 0x00b3b3b3,
-    0x00b4b4b4, 0x00b5b5b5, 0x00b6b6b6, 0x00b7b7b7, 0x00b8b8b8, 0x00b9b9b9,
-    0x00bababa, 0x00bbbbbb, 0x00bcbcbc, 0x00bdbdbd, 0x00bebebe, 0x00bfbfbf,
-    0x00c0c0c0, 0x00c1c1c1, 0x00c2c2c2, 0x00c3c3c3, 0x00c4c4c4, 0x00c5c5c5,
-    0x00c6c6c6, 0x00c7c7c7, 0x00c8c8c8, 0x00c9c9c9, 0x00cacaca, 0x00cbcbcb,
-    0x00cccccc, 0x00cdcdcd, 0x00cecece, 0x00cfcfcf, 0x00d0d0d0, 0x00d1d1d1,
-    0x00d2d2d2, 0x00d3d3d3, 0x00d4d4d4, 0x00d5d5d5, 0x00d6d6d6, 0x00d7d7d7,
-    0x00d8d8d8, 0x00d9d9d9, 0x00dadada, 0x00dbdbdb, 0x00dcdcdc, 0x00dddddd,
-    0x00dedede, 0x00dfdfdf, 0x00e0e0e0, 0x00e1e1e1, 0x00e2e2e2, 0x00e3e3e3,
-    0x00e4e4e4, 0x00e5e5e5, 0x00e6e6e6, 0x00e7e7e7, 0x00e8e8e8, 0x00e9e9e9,
-    0x00eaeaea, 0x00ebebeb, 0x00ececec, 0x00ededed, 0x00eeeeee, 0x00efefef,
-    0x00f0f0f0, 0x00f1f1f1, 0x00f2f2f2, 0x00f3f3f3, 0x00f4f4f4, 0x00f5f5f5,
-    0x00f6f6f6, 0x00f7f7f7, 0x00f8f8f8, 0x00f9f9f9, 0x00fafafa, 0x00fbfbfb,
-    0x00fcfcfc, 0x00fdfdfd, 0x00fefefe, 0x00ffffff }
-};
-
-static WEBP_INLINE uint32_t GetScale(uint32_t a, int inverse) {
-  return kMultTables[!inverse][a];
-}
-
-#else
-
-static WEBP_INLINE uint32_t GetScale(uint32_t a, int inverse) {
-  return inverse ? (255u << MFIX) / a : a * KINV_255;
-}
-
-#endif    // USE_TABLES_FOR_ALPHA_MULT
-
-static void MultARGBRow(uint32_t* const ptr, int width, int inverse) {
-  int x;
-  for (x = 0; x < width; ++x) {
-    const uint32_t argb = ptr[x];
-    if (argb < 0xff000000u) {      // alpha < 255
-      if (argb <= 0x00ffffffu) {   // alpha == 0
-        ptr[x] = 0;
-      } else {
-        const uint32_t alpha = (argb >> 24) & 0xff;
-        const uint32_t scale = GetScale(alpha, inverse);
-        uint32_t out = argb & 0xff000000u;
-        out |= Mult(argb >>  0, scale) <<  0;
-        out |= Mult(argb >>  8, scale) <<  8;
-        out |= Mult(argb >> 16, scale) << 16;
-        ptr[x] = out;
-      }
-    }
-  }
-}
-
-static void MultRow(uint8_t* const ptr, const uint8_t* const alpha,
-                    int width, int inverse) {
-  int x;
-  for (x = 0; x < width; ++x) {
-    const uint32_t a = alpha[x];
-    if (a != 255) {
-      if (a == 0) {
-        ptr[x] = 0;
-      } else {
-        const uint32_t scale = GetScale(a, inverse);
-        ptr[x] = Mult(ptr[x], scale);
-      }
-    }
-  }
-}
-
-#undef KINV_255
-#undef HALF
-#undef MFIX
-
-void (*WebPMultARGBRow)(uint32_t* const ptr, int width, int inverse);
-void (*WebPMultRow)(uint8_t* const ptr, const uint8_t* const alpha,
-                    int width, int inverse);
-
-//------------------------------------------------------------------------------
-// Generic per-plane calls
-
-void WebPMultARGBRows(uint8_t* ptr, int stride, int width, int num_rows,
-                      int inverse) {
-  int n;
-  for (n = 0; n < num_rows; ++n) {
-    WebPMultARGBRow((uint32_t*)ptr, width, inverse);
-    ptr += stride;
-  }
-}
-
-void WebPMultRows(uint8_t* ptr, int stride,
-                  const uint8_t* alpha, int alpha_stride,
-                  int width, int num_rows, int inverse) {
-  int n;
-  for (n = 0; n < num_rows; ++n) {
-    WebPMultRow(ptr, alpha, width, inverse);
-    ptr += stride;
-    alpha += alpha_stride;
-  }
-}
-
-//------------------------------------------------------------------------------
-// Premultiplied modes
-
-// non dithered-modes
-
-// (x * a * 32897) >> 23 is bit-wise equivalent to (int)(x * a / 255.)
-// for all 8bit x or a. For bit-wise equivalence to (int)(x * a / 255. + .5),
-// one can use instead: (x * a * 65793 + (1 << 23)) >> 24
-#if 1     // (int)(x * a / 255.)
-#define MULTIPLIER(a)   ((a) * 32897U)
-#define PREMULTIPLY(x, m) (((x) * (m)) >> 23)
-#else     // (int)(x * a / 255. + .5)
-#define MULTIPLIER(a) ((a) * 65793U)
-#define PREMULTIPLY(x, m) (((x) * (m) + (1U << 23)) >> 24)
-#endif
-
-static void ApplyAlphaMultiply(uint8_t* rgba, int alpha_first,
-                               int w, int h, int stride) {
-  while (h-- > 0) {
-    uint8_t* const rgb = rgba + (alpha_first ? 1 : 0);
-    const uint8_t* const alpha = rgba + (alpha_first ? 0 : 3);
-    int i;
-    for (i = 0; i < w; ++i) {
-      const uint32_t a = alpha[4 * i];
-      if (a != 0xff) {
-        const uint32_t mult = MULTIPLIER(a);
-        rgb[4 * i + 0] = PREMULTIPLY(rgb[4 * i + 0], mult);
-        rgb[4 * i + 1] = PREMULTIPLY(rgb[4 * i + 1], mult);
-        rgb[4 * i + 2] = PREMULTIPLY(rgb[4 * i + 2], mult);
-      }
-    }
-    rgba += stride;
-  }
-}
-#undef MULTIPLIER
-#undef PREMULTIPLY
-
-// rgbA4444
-
-#define MULTIPLIER(a)  ((a) * 0x1111)    // 0x1111 ~= (1 << 16) / 15
-
-static WEBP_INLINE uint8_t dither_hi(uint8_t x) {
-  return (x & 0xf0) | (x >> 4);
-}
-
-static WEBP_INLINE uint8_t dither_lo(uint8_t x) {
-  return (x & 0x0f) | (x << 4);
-}
-
-static WEBP_INLINE uint8_t multiply(uint8_t x, uint32_t m) {
-  return (x * m) >> 16;
-}
-
-static WEBP_INLINE void ApplyAlphaMultiply4444(uint8_t* rgba4444,
-                                               int w, int h, int stride,
-                                               int rg_byte_pos /* 0 or 1 */) {
-  while (h-- > 0) {
-    int i;
-    for (i = 0; i < w; ++i) {
-      const uint32_t rg = rgba4444[2 * i + rg_byte_pos];
-      const uint32_t ba = rgba4444[2 * i + (rg_byte_pos ^ 1)];
-      const uint8_t a = ba & 0x0f;
-      const uint32_t mult = MULTIPLIER(a);
-      const uint8_t r = multiply(dither_hi(rg), mult);
-      const uint8_t g = multiply(dither_lo(rg), mult);
-      const uint8_t b = multiply(dither_hi(ba), mult);
-      rgba4444[2 * i + rg_byte_pos] = (r & 0xf0) | ((g >> 4) & 0x0f);
-      rgba4444[2 * i + (rg_byte_pos ^ 1)] = (b & 0xf0) | a;
-    }
-    rgba4444 += stride;
-  }
-}
-#undef MULTIPLIER
-
-static void ApplyAlphaMultiply_16b(uint8_t* rgba4444,
-                                   int w, int h, int stride) {
-#ifdef WEBP_SWAP_16BIT_CSP
-  ApplyAlphaMultiply4444(rgba4444, w, h, stride, 1);
-#else
-  ApplyAlphaMultiply4444(rgba4444, w, h, stride, 0);
-#endif
-}
-
-static int ExtractAlpha(const uint8_t* argb, int argb_stride,
-                        int width, int height,
-                        uint8_t* alpha, int alpha_stride) {
-  uint8_t alpha_mask = 0xff;
-  int i, j;
-
-  for (j = 0; j < height; ++j) {
-    for (i = 0; i < width; ++i) {
-      const uint8_t alpha_value = argb[4 * i];
-      alpha[i] = alpha_value;
-      alpha_mask &= alpha_value;
-    }
-    argb += argb_stride;
-    alpha += alpha_stride;
-  }
-  return (alpha_mask == 0xff);
-}
-
-void (*WebPApplyAlphaMultiply)(uint8_t*, int, int, int, int);
-void (*WebPApplyAlphaMultiply4444)(uint8_t*, int, int, int);
-int (*WebPExtractAlpha)(const uint8_t*, int, int, int, uint8_t*, int);
-
-//------------------------------------------------------------------------------
-// Init function
-
-extern void WebPInitAlphaProcessingSSE2(void);
-
-void WebPInitAlphaProcessing(void) {
-  WebPMultARGBRow = MultARGBRow;
-  WebPMultRow = MultRow;
-  WebPApplyAlphaMultiply = ApplyAlphaMultiply;
-  WebPApplyAlphaMultiply4444 = ApplyAlphaMultiply_16b;
-  WebPExtractAlpha = ExtractAlpha;
-
-  // If defined, use CPUInfo() to overwrite some pointers with faster versions.
-  if (VP8GetCPUInfo != NULL) {
-#if defined(WEBP_USE_SSE2)
-    if (VP8GetCPUInfo(kSSE2)) {
-      WebPInitAlphaProcessingSSE2();
-    }
-#endif
-  }
-}
@@ -1,77 +0,0 @@
-// Copyright 2014 Google Inc. All Rights Reserved.
-//
-// Use of this source code is governed by a BSD-style license
-// that can be found in the COPYING file in the root of the source
-// tree. An additional intellectual property rights grant can be found
-// in the file PATENTS. All contributing project authors may
-// be found in the AUTHORS file in the root of the source tree.
-// -----------------------------------------------------------------------------
-//
-// Utilities for processing transparent channel.
-//
-// Author: Skal (pascal.massimino@gmail.com)
-
-#include "./dsp.h"
-
-#if defined(WEBP_USE_SSE2)
-#include <emmintrin.h>
-
-//------------------------------------------------------------------------------
-
-static int ExtractAlpha(const uint8_t* argb, int argb_stride,
-                        int width, int height,
-                        uint8_t* alpha, int alpha_stride) {
-  // alpha_and stores an 'and' operation of all the alpha[] values. The final
-  // value is not 0xff if any of the alpha[] is not equal to 0xff.
-  uint32_t alpha_and = 0xff;
-  int i, j;
-  const __m128i a_mask = _mm_set1_epi32(0xffu);  // to preserve alpha
-  const __m128i all_0xff = _mm_set_epi32(0, 0, ~0u, ~0u);
-  __m128i all_alphas = all_0xff;
-
-  // We must be able to access 3 extra bytes after the last written byte
-  // 'src[4 * width - 4]', because we don't know if alpha is the first or the
-  // last byte of the quadruplet.
-  const int limit = (width - 1) & ~7;
-
-  for (j = 0; j < height; ++j) {
-    const __m128i* src = (const __m128i*)argb;
-    for (i = 0; i < limit; i += 8) {
-      // load 32 argb bytes
-      const __m128i a0 = _mm_loadu_si128(src + 0);
-      const __m128i a1 = _mm_loadu_si128(src + 1);
-      const __m128i b0 = _mm_and_si128(a0, a_mask);
-      const __m128i b1 = _mm_and_si128(a1, a_mask);
-      const __m128i c0 = _mm_packs_epi32(b0, b1);
-      const __m128i d0 = _mm_packus_epi16(c0, c0);
-      // store
-      _mm_storel_epi64((__m128i*)&alpha[i], d0);
-      // accumulate eight alpha 'and' in parallel
-      all_alphas = _mm_and_si128(all_alphas, d0);
-      src += 2;
-    }
-    for (; i < width; ++i) {
-      const uint32_t alpha_value = argb[4 * i];
-      alpha[i] = alpha_value;
-      alpha_and &= alpha_value;
-    }
-    argb += argb_stride;
-    alpha += alpha_stride;
-  }
-  // Combine the eight alpha 'and' into a 8-bit mask.
-  alpha_and &= _mm_movemask_epi8(_mm_cmpeq_epi8(all_alphas, all_0xff));
-  return (alpha_and == 0xff);
-}
-
-#endif   // WEBP_USE_SSE2
-
-//------------------------------------------------------------------------------
-// Init function
-
-extern void WebPInitAlphaProcessingSSE2(void);
-
-void WebPInitAlphaProcessingSSE2(void) {
-#if defined(WEBP_USE_SSE2)
-  WebPExtractAlpha = ExtractAlpha;
-#endif
-}
@@ -1,130 +0,0 @@
-// Copyright 2011 Google Inc. All Rights Reserved.
-//
-// Use of this source code is governed by a BSD-style license
-// that can be found in the COPYING file in the root of the source
-// tree. An additional intellectual property rights grant can be found
-// in the file PATENTS. All contributing project authors may
-// be found in the AUTHORS file in the root of the source tree.
-// -----------------------------------------------------------------------------
-//
-// CPU detection
-//
-// Author: Christian Duvivier (cduvivier@google.com)
-
-#include "./dsp.h"
-
-#if defined(__ANDROID__)
-#include <cpu-features.h>
-#endif
-
-//------------------------------------------------------------------------------
-// SSE2 detection.
-//
-
-// apple/darwin gcc-4.0.1 defines __PIC__, but not __pic__ with -fPIC.
-#if (defined(__pic__) || defined(__PIC__)) && defined(__i386__)
-static WEBP_INLINE void GetCPUInfo(int cpu_info[4], int info_type) {
-  __asm__ volatile (
-    "mov %%ebx, %%edi\n"
-    "cpuid\n"
-    "xchg %%edi, %%ebx\n"
-    : "=a"(cpu_info[0]), "=D"(cpu_info[1]), "=c"(cpu_info[2]), "=d"(cpu_info[3])
-    : "a"(info_type));
-}
-#elif defined(__i386__) || defined(__x86_64__)
-static WEBP_INLINE void GetCPUInfo(int cpu_info[4], int info_type) {
-  __asm__ volatile (
-    "cpuid\n"
-    : "=a"(cpu_info[0]), "=b"(cpu_info[1]), "=c"(cpu_info[2]), "=d"(cpu_info[3])
-    : "a"(info_type));
-}
-#elif defined(_MSC_FULL_VER) && _MSC_FULL_VER >= 150030729  // >= VS2008 SP1
-#define GetCPUInfo(info, type) __cpuidex(info, type, 0)  // set ecx=0
-#elif defined(WEBP_MSC_SSE2)
-#define GetCPUInfo __cpuid
-#endif
-
-// NaCl has no support for xgetbv or the raw opcode.
-#if !defined(__native_client__) && (defined(__i386__) || defined(__x86_64__))
-static WEBP_INLINE uint64_t xgetbv(void) {
-  const uint32_t ecx = 0;
-  uint32_t eax, edx;
-  // Use the raw opcode for xgetbv for compatibility with older toolchains.
-  __asm__ volatile (
-    ".byte 0x0f, 0x01, 0xd0\n"
-    : "=a"(eax), "=d"(edx) : "c" (ecx));
-  return ((uint64_t)edx << 32) | eax;
-}
-#elif defined(_MSC_FULL_VER) && _MSC_FULL_VER >= 160040219  // >= VS2010 SP1
-#define xgetbv() _xgetbv(0)
-#elif defined(_MSC_VER) && defined(_M_IX86)
-static WEBP_INLINE uint64_t xgetbv(void) {
-  uint32_t eax_, edx_;
-  __asm {
-    xor ecx, ecx  // ecx = 0
-    // Use the raw opcode for xgetbv for compatibility with older toolchains.
-    __asm _emit 0x0f __asm _emit 0x01 __asm _emit 0xd0
-    mov eax_, eax
-    mov edx_, edx
-  }
-  return ((uint64_t)edx_ << 32) | eax_;
-}
-#else
-#define xgetbv() 0U  // no AVX for older x64 or unrecognized toolchains.
-#endif
-
-#if defined(__i386__) || defined(__x86_64__) || defined(WEBP_MSC_SSE2)
-static int x86CPUInfo(CPUFeature feature) {
-  int cpu_info[4];
-  GetCPUInfo(cpu_info, 1);
-  if (feature == kSSE2) {
-    return 0 != (cpu_info[3] & 0x04000000);
-  }
-  if (feature == kSSE3) {
-    return 0 != (cpu_info[2] & 0x00000001);
-  }
-  if (feature == kAVX) {
-    // bits 27 (OSXSAVE) & 28 (256-bit AVX)
-    if ((cpu_info[2] & 0x18000000) == 0x18000000) {
-      // XMM state and YMM state enabled by the OS.
-      return (xgetbv() & 0x6) == 0x6;
-    }
-  }
-  if (feature == kAVX2) {
-    if (x86CPUInfo(kAVX)) {
-      GetCPUInfo(cpu_info, 7);
-      return ((cpu_info[1] & 0x00000020) == 0x00000020);
-    }
-  }
-  return 0;
-}
-VP8CPUInfo VP8GetCPUInfo = x86CPUInfo;
-#elif defined(WEBP_ANDROID_NEON)  // NB: needs to be before generic NEON test.
-static int AndroidCPUInfo(CPUFeature feature) {
-  const AndroidCpuFamily cpu_family = android_getCpuFamily();
-  const uint64_t cpu_features = android_getCpuFeatures();
-  if (feature == kNEON) {
-    return (cpu_family == ANDROID_CPU_FAMILY_ARM &&
-            0 != (cpu_features & ANDROID_CPU_ARM_FEATURE_NEON));
-  }
-  return 0;
-}
-VP8CPUInfo VP8GetCPUInfo = AndroidCPUInfo;
-#elif defined(WEBP_USE_NEON)
-// define a dummy function to enable turning off NEON at runtime by setting
-// VP8DecGetCPUInfo = NULL
-static int armCPUInfo(CPUFeature feature) {
-  (void)feature;
-  return 1;
-}
-VP8CPUInfo VP8GetCPUInfo = armCPUInfo;
-#elif defined(WEBP_USE_MIPS32)
-static int mipsCPUInfo(CPUFeature feature) {
-  (void)feature;
-  return 1;
-}
-VP8CPUInfo VP8GetCPUInfo = mipsCPUInfo;
-#else
-VP8CPUInfo VP8GetCPUInfo = NULL;
-#endif
-
@@ -1,731 +0,0 @@
-// Copyright 2010 Google Inc. All Rights Reserved.
-//
-// Use of this source code is governed by a BSD-style license
-// that can be found in the COPYING file in the root of the source
-// tree. An additional intellectual property rights grant can be found
-// in the file PATENTS. All contributing project authors may
-// be found in the AUTHORS file in the root of the source tree.
-// -----------------------------------------------------------------------------
-//
-// Speed-critical decoding functions.
-//
-// Author: Skal (pascal.massimino@gmail.com)
-
-#include "./dsp.h"
-#include "../dec/vp8i.h"
-
-//------------------------------------------------------------------------------
-
-static WEBP_INLINE uint8_t clip_8b(int v) {
-  return (!(v & ~0xff)) ? v : (v < 0) ? 0 : 255;
-}
-
-//------------------------------------------------------------------------------
-// Transforms (Paragraph 14.4)
-
-#define STORE(x, y, v) \
-  dst[x + y * BPS] = clip_8b(dst[x + y * BPS] + ((v) >> 3))
-
-#define STORE2(y, dc, d, c) do {    \
-  const int DC = (dc);              \
-  STORE(0, y, DC + (d));            \
-  STORE(1, y, DC + (c));            \
-  STORE(2, y, DC - (c));            \
-  STORE(3, y, DC - (d));            \
-} while (0)
-
-static const int kC1 = 20091 + (1 << 16);
-static const int kC2 = 35468;
-#define MUL(a, b) (((a) * (b)) >> 16)
-
-static void TransformOne(const int16_t* in, uint8_t* dst) {
-  int C[4 * 4], *tmp;
-  int i;
-  tmp = C;
-  for (i = 0; i < 4; ++i) {    // vertical pass
-    const int a = in[0] + in[8];    // [-4096, 4094]
-    const int b = in[0] - in[8];    // [-4095, 4095]
-    const int c = MUL(in[4], kC2) - MUL(in[12], kC1);   // [-3783, 3783]
-    const int d = MUL(in[4], kC1) + MUL(in[12], kC2);   // [-3785, 3781]
-    tmp[0] = a + d;   // [-7881, 7875]
-    tmp[1] = b + c;   // [-7878, 7878]
-    tmp[2] = b - c;   // [-7878, 7878]
-    tmp[3] = a - d;   // [-7877, 7879]
-    tmp += 4;
-    in++;
-  }
-  // Each pass is expanding the dynamic range by ~3.85 (upper bound).
-  // The exact value is (2. + (kC1 + kC2) / 65536).
-  // After the second pass, maximum interval is [-3794, 3794], assuming
-  // an input in [-2048, 2047] interval. We then need to add a dst value
-  // in the [0, 255] range.
-  // In the worst case scenario, the input to clip_8b() can be as large as
-  // [-60713, 60968].
-  tmp = C;
-  for (i = 0; i < 4; ++i) {    // horizontal pass
-    const int dc = tmp[0] + 4;
-    const int a =  dc +  tmp[8];
-    const int b =  dc -  tmp[8];
-    const int c = MUL(tmp[4], kC2) - MUL(tmp[12], kC1);
-    const int d = MUL(tmp[4], kC1) + MUL(tmp[12], kC2);
-    STORE(0, 0, a + d);
-    STORE(1, 0, b + c);
-    STORE(2, 0, b - c);
-    STORE(3, 0, a - d);
-    tmp++;
-    dst += BPS;
-  }
-}
-
-// Simplified transform when only in[0], in[1] and in[4] are non-zero
-static void TransformAC3(const int16_t* in, uint8_t* dst) {
-  const int a = in[0] + 4;
-  const int c4 = MUL(in[4], kC2);
-  const int d4 = MUL(in[4], kC1);
-  const int c1 = MUL(in[1], kC2);
-  const int d1 = MUL(in[1], kC1);
-  STORE2(0, a + d4, d1, c1);
-  STORE2(1, a + c4, d1, c1);
-  STORE2(2, a - c4, d1, c1);
-  STORE2(3, a - d4, d1, c1);
-}
-#undef MUL
-#undef STORE2
-
-static void TransformTwo(const int16_t* in, uint8_t* dst, int do_two) {
-  TransformOne(in, dst);
-  if (do_two) {
-    TransformOne(in + 16, dst + 4);
-  }
-}
-
-static void TransformUV(const int16_t* in, uint8_t* dst) {
-  VP8Transform(in + 0 * 16, dst, 1);
-  VP8Transform(in + 2 * 16, dst + 4 * BPS, 1);
-}
-
-static void TransformDC(const int16_t *in, uint8_t* dst) {
-  const int DC = in[0] + 4;
-  int i, j;
-  for (j = 0; j < 4; ++j) {
-    for (i = 0; i < 4; ++i) {
-      STORE(i, j, DC);
-    }
-  }
-}
-
-static void TransformDCUV(const int16_t* in, uint8_t* dst) {
-  if (in[0 * 16]) VP8TransformDC(in + 0 * 16, dst);
-  if (in[1 * 16]) VP8TransformDC(in + 1 * 16, dst + 4);
-  if (in[2 * 16]) VP8TransformDC(in + 2 * 16, dst + 4 * BPS);
-  if (in[3 * 16]) VP8TransformDC(in + 3 * 16, dst + 4 * BPS + 4);
-}
-
-#undef STORE
-
-//------------------------------------------------------------------------------
-// Paragraph 14.3
-
-static void TransformWHT(const int16_t* in, int16_t* out) {
-  int tmp[16];
-  int i;
-  for (i = 0; i < 4; ++i) {
-    const int a0 = in[0 + i] + in[12 + i];
-    const int a1 = in[4 + i] + in[ 8 + i];
-    const int a2 = in[4 + i] - in[ 8 + i];
-    const int a3 = in[0 + i] - in[12 + i];
-    tmp[0  + i] = a0 + a1;
-    tmp[8  + i] = a0 - a1;
-    tmp[4  + i] = a3 + a2;
-    tmp[12 + i] = a3 - a2;
-  }
-  for (i = 0; i < 4; ++i) {
-    const int dc = tmp[0 + i * 4] + 3;    // w/ rounder
-    const int a0 = dc             + tmp[3 + i * 4];
-    const int a1 = tmp[1 + i * 4] + tmp[2 + i * 4];
-    const int a2 = tmp[1 + i * 4] - tmp[2 + i * 4];
-    const int a3 = dc             - tmp[3 + i * 4];
-    out[ 0] = (a0 + a1) >> 3;
-    out[16] = (a3 + a2) >> 3;
-    out[32] = (a0 - a1) >> 3;
-    out[48] = (a3 - a2) >> 3;
-    out += 64;
-  }
-}
-
-void (*VP8TransformWHT)(const int16_t* in, int16_t* out);
-
-//------------------------------------------------------------------------------
-// Intra predictions
-
-#define DST(x, y) dst[(x) + (y) * BPS]
-
-static WEBP_INLINE void TrueMotion(uint8_t *dst, int size) {
-  const uint8_t* top = dst - BPS;
-  const uint8_t* const clip0 = VP8kclip1 - top[-1];
-  int y;
-  for (y = 0; y < size; ++y) {
-    const uint8_t* const clip = clip0 + dst[-1];
-    int x;
-    for (x = 0; x < size; ++x) {
-      dst[x] = clip[top[x]];
-    }
-    dst += BPS;
-  }
-}
-static void TM4(uint8_t *dst)   { TrueMotion(dst, 4); }
-static void TM8uv(uint8_t *dst) { TrueMotion(dst, 8); }
-static void TM16(uint8_t *dst)  { TrueMotion(dst, 16); }
-
-//------------------------------------------------------------------------------
-// 16x16
-
-static void VE16(uint8_t *dst) {     // vertical
-  int j;
-  for (j = 0; j < 16; ++j) {
-    memcpy(dst + j * BPS, dst - BPS, 16);
-  }
-}
-
-static void HE16(uint8_t *dst) {     // horizontal
-  int j;
-  for (j = 16; j > 0; --j) {
-    memset(dst, dst[-1], 16);
-    dst += BPS;
-  }
-}
-
-static WEBP_INLINE void Put16(int v, uint8_t* dst) {
-  int j;
-  for (j = 0; j < 16; ++j) {
-    memset(dst + j * BPS, v, 16);
-  }
-}
-
-static void DC16(uint8_t *dst) {    // DC
-  int DC = 16;
-  int j;
-  for (j = 0; j < 16; ++j) {
-    DC += dst[-1 + j * BPS] + dst[j - BPS];
-  }
-  Put16(DC >> 5, dst);
-}
-
-static void DC16NoTop(uint8_t *dst) {   // DC with top samples not available
-  int DC = 8;
-  int j;
-  for (j = 0; j < 16; ++j) {
-    DC += dst[-1 + j * BPS];
-  }
-  Put16(DC >> 4, dst);
-}
-
-static void DC16NoLeft(uint8_t *dst) {  // DC with left samples not available
-  int DC = 8;
-  int i;
-  for (i = 0; i < 16; ++i) {
-    DC += dst[i - BPS];
-  }
-  Put16(DC >> 4, dst);
-}
-
-static void DC16NoTopLeft(uint8_t *dst) {  // DC with no top and left samples
-  Put16(0x80, dst);
-}
-
-//------------------------------------------------------------------------------
-// 4x4
-
-#define AVG3(a, b, c) (((a) + 2 * (b) + (c) + 2) >> 2)
-#define AVG2(a, b) (((a) + (b) + 1) >> 1)
-
-static void VE4(uint8_t *dst) {    // vertical
-  const uint8_t* top = dst - BPS;
-  const uint8_t vals[4] = {
-    AVG3(top[-1], top[0], top[1]),
-    AVG3(top[ 0], top[1], top[2]),
-    AVG3(top[ 1], top[2], top[3]),
-    AVG3(top[ 2], top[3], top[4])
-  };
-  int i;
-  for (i = 0; i < 4; ++i) {
-    memcpy(dst + i * BPS, vals, sizeof(vals));
-  }
-}
-
-static void HE4(uint8_t *dst) {    // horizontal
-  const int A = dst[-1 - BPS];
-  const int B = dst[-1];
-  const int C = dst[-1 + BPS];
-  const int D = dst[-1 + 2 * BPS];
-  const int E = dst[-1 + 3 * BPS];
-  *(uint32_t*)(dst + 0 * BPS) = 0x01010101U * AVG3(A, B, C);
-  *(uint32_t*)(dst + 1 * BPS) = 0x01010101U * AVG3(B, C, D);
-  *(uint32_t*)(dst + 2 * BPS) = 0x01010101U * AVG3(C, D, E);
-  *(uint32_t*)(dst + 3 * BPS) = 0x01010101U * AVG3(D, E, E);
-}
-
-static void DC4(uint8_t *dst) {   // DC
-  uint32_t dc = 4;
-  int i;
-  for (i = 0; i < 4; ++i) dc += dst[i - BPS] + dst[-1 + i * BPS];
-  dc >>= 3;
-  for (i = 0; i < 4; ++i) memset(dst + i * BPS, dc, 4);
-}
-
-static void RD4(uint8_t *dst) {   // Down-right
-  const int I = dst[-1 + 0 * BPS];
-  const int J = dst[-1 + 1 * BPS];
-  const int K = dst[-1 + 2 * BPS];
-  const int L = dst[-1 + 3 * BPS];
-  const int X = dst[-1 - BPS];
-  const int A = dst[0 - BPS];
-  const int B = dst[1 - BPS];
-  const int C = dst[2 - BPS];
-  const int D = dst[3 - BPS];
-  DST(0, 3)                                     = AVG3(J, K, L);
-  DST(0, 2) = DST(1, 3)                         = AVG3(I, J, K);
-  DST(0, 1) = DST(1, 2) = DST(2, 3)             = AVG3(X, I, J);
-  DST(0, 0) = DST(1, 1) = DST(2, 2) = DST(3, 3) = AVG3(A, X, I);
-  DST(1, 0) = DST(2, 1) = DST(3, 2)             = AVG3(B, A, X);
-  DST(2, 0) = DST(3, 1)                         = AVG3(C, B, A);
-  DST(3, 0)                                     = AVG3(D, C, B);
-}
-
-static void LD4(uint8_t *dst) {   // Down-Left
-  const int A = dst[0 - BPS];
-  const int B = dst[1 - BPS];
-  const int C = dst[2 - BPS];
-  const int D = dst[3 - BPS];
-  const int E = dst[4 - BPS];
-  const int F = dst[5 - BPS];
-  const int G = dst[6 - BPS];
-  const int H = dst[7 - BPS];
-  DST(0, 0)                                     = AVG3(A, B, C);
-  DST(1, 0) = DST(0, 1)                         = AVG3(B, C, D);
-  DST(2, 0) = DST(1, 1) = DST(0, 2)             = AVG3(C, D, E);
-  DST(3, 0) = DST(2, 1) = DST(1, 2) = DST(0, 3) = AVG3(D, E, F);
-  DST(3, 1) = DST(2, 2) = DST(1, 3)             = AVG3(E, F, G);
-  DST(3, 2) = DST(2, 3)                         = AVG3(F, G, H);
-  DST(3, 3)                                     = AVG3(G, H, H);
-}
-
-static void VR4(uint8_t *dst) {   // Vertical-Right
-  const int I = dst[-1 + 0 * BPS];
-  const int J = dst[-1 + 1 * BPS];
-  const int K = dst[-1 + 2 * BPS];
-  const int X = dst[-1 - BPS];
-  const int A = dst[0 - BPS];
-  const int B = dst[1 - BPS];
-  const int C = dst[2 - BPS];
-  const int D = dst[3 - BPS];
-  DST(0, 0) = DST(1, 2) = AVG2(X, A);
-  DST(1, 0) = DST(2, 2) = AVG2(A, B);
-  DST(2, 0) = DST(3, 2) = AVG2(B, C);
-  DST(3, 0)             = AVG2(C, D);
-
-  DST(0, 3) =             AVG3(K, J, I);
-  DST(0, 2) =             AVG3(J, I, X);
-  DST(0, 1) = DST(1, 3) = AVG3(I, X, A);
-  DST(1, 1) = DST(2, 3) = AVG3(X, A, B);
-  DST(2, 1) = DST(3, 3) = AVG3(A, B, C);
-  DST(3, 1) =             AVG3(B, C, D);
-}
-
-static void VL4(uint8_t *dst) {   // Vertical-Left
-  const int A = dst[0 - BPS];
-  const int B = dst[1 - BPS];
-  const int C = dst[2 - BPS];
-  const int D = dst[3 - BPS];
-  const int E = dst[4 - BPS];
-  const int F = dst[5 - BPS];
-  const int G = dst[6 - BPS];
-  const int H = dst[7 - BPS];
-  DST(0, 0) =             AVG2(A, B);
-  DST(1, 0) = DST(0, 2) = AVG2(B, C);
-  DST(2, 0) = DST(1, 2) = AVG2(C, D);
-  DST(3, 0) = DST(2, 2) = AVG2(D, E);
-
-  DST(0, 1) =             AVG3(A, B, C);
-  DST(1, 1) = DST(0, 3) = AVG3(B, C, D);
-  DST(2, 1) = DST(1, 3) = AVG3(C, D, E);
-  DST(3, 1) = DST(2, 3) = AVG3(D, E, F);
-              DST(3, 2) = AVG3(E, F, G);
-              DST(3, 3) = AVG3(F, G, H);
-}
-
-static void HU4(uint8_t *dst) {   // Horizontal-Up
-  const int I = dst[-1 + 0 * BPS];
-  const int J = dst[-1 + 1 * BPS];
-  const int K = dst[-1 + 2 * BPS];
-  const int L = dst[-1 + 3 * BPS];
-  DST(0, 0) =             AVG2(I, J);
-  DST(2, 0) = DST(0, 1) = AVG2(J, K);
-  DST(2, 1) = DST(0, 2) = AVG2(K, L);
-  DST(1, 0) =             AVG3(I, J, K);
-  DST(3, 0) = DST(1, 1) = AVG3(J, K, L);
-  DST(3, 1) = DST(1, 2) = AVG3(K, L, L);
-  DST(3, 2) = DST(2, 2) =
-    DST(0, 3) = DST(1, 3) = DST(2, 3) = DST(3, 3) = L;
-}
-
-static void HD4(uint8_t *dst) {  // Horizontal-Down
-  const int I = dst[-1 + 0 * BPS];
-  const int J = dst[-1 + 1 * BPS];
-  const int K = dst[-1 + 2 * BPS];
-  const int L = dst[-1 + 3 * BPS];
-  const int X = dst[-1 - BPS];
-  const int A = dst[0 - BPS];
-  const int B = dst[1 - BPS];
-  const int C = dst[2 - BPS];
-
-  DST(0, 0) = DST(2, 1) = AVG2(I, X);
-  DST(0, 1) = DST(2, 2) = AVG2(J, I);
-  DST(0, 2) = DST(2, 3) = AVG2(K, J);
-  DST(0, 3)             = AVG2(L, K);
-
-  DST(3, 0)             = AVG3(A, B, C);
-  DST(2, 0)             = AVG3(X, A, B);
-  DST(1, 0) = DST(3, 1) = AVG3(I, X, A);
-  DST(1, 1) = DST(3, 2) = AVG3(J, I, X);
-  DST(1, 2) = DST(3, 3) = AVG3(K, J, I);
-  DST(1, 3)             = AVG3(L, K, J);
-}
-
-#undef DST
-#undef AVG3
-#undef AVG2
-
-//------------------------------------------------------------------------------
-// Chroma
-
-static void VE8uv(uint8_t *dst) {    // vertical
-  int j;
-  for (j = 0; j < 8; ++j) {
-    memcpy(dst + j * BPS, dst - BPS, 8);
-  }
-}
-
-static void HE8uv(uint8_t *dst) {    // horizontal
-  int j;
-  for (j = 0; j < 8; ++j) {
-    memset(dst, dst[-1], 8);
-    dst += BPS;
-  }
-}
-
-// helper for chroma-DC predictions
-static WEBP_INLINE void Put8x8uv(uint8_t value, uint8_t* dst) {
-  int j;
-  for (j = 0; j < 8; ++j) {
-    memset(dst + j * BPS, value, 8);
-  }
-}
-
-static void DC8uv(uint8_t *dst) {     // DC
-  int dc0 = 8;
-  int i;
-  for (i = 0; i < 8; ++i) {
-    dc0 += dst[i - BPS] + dst[-1 + i * BPS];
-  }
-  Put8x8uv(dc0 >> 4, dst);
-}
-
-static void DC8uvNoLeft(uint8_t *dst) {   // DC with no left samples
-  int dc0 = 4;
-  int i;
-  for (i = 0; i < 8; ++i) {
-    dc0 += dst[i - BPS];
-  }
-  Put8x8uv(dc0 >> 3, dst);
-}
-
-static void DC8uvNoTop(uint8_t *dst) {  // DC with no top samples
-  int dc0 = 4;
-  int i;
-  for (i = 0; i < 8; ++i) {
-    dc0 += dst[-1 + i * BPS];
-  }
-  Put8x8uv(dc0 >> 3, dst);
-}
-
-static void DC8uvNoTopLeft(uint8_t *dst) {    // DC with nothing
-  Put8x8uv(0x80, dst);
-}
-
-//------------------------------------------------------------------------------
-// default C implementations
-
-const VP8PredFunc VP8PredLuma4[NUM_BMODES] = {
-  DC4, TM4, VE4, HE4, RD4, VR4, LD4, VL4, HD4, HU4
-};
-
-const VP8PredFunc VP8PredLuma16[NUM_B_DC_MODES] = {
-  DC16, TM16, VE16, HE16,
-  DC16NoTop, DC16NoLeft, DC16NoTopLeft
-};
-
-const VP8PredFunc VP8PredChroma8[NUM_B_DC_MODES] = {
-  DC8uv, TM8uv, VE8uv, HE8uv,
-  DC8uvNoTop, DC8uvNoLeft, DC8uvNoTopLeft
-};
-
-//------------------------------------------------------------------------------
-// Edge filtering functions
-
-// 4 pixels in, 2 pixels out
-static WEBP_INLINE void do_filter2(uint8_t* p, int step) {
-  const int p1 = p[-2*step], p0 = p[-step], q0 = p[0], q1 = p[step];
-  const int a = 3 * (q0 - p0) + VP8ksclip1[p1 - q1];  // in [-893,892]
-  const int a1 = VP8ksclip2[(a + 4) >> 3];            // in [-16,15]
-  const int a2 = VP8ksclip2[(a + 3) >> 3];
-  p[-step] = VP8kclip1[p0 + a2];
-  p[    0] = VP8kclip1[q0 - a1];
-}
-
-// 4 pixels in, 4 pixels out
-static WEBP_INLINE void do_filter4(uint8_t* p, int step) {
-  const int p1 = p[-2*step], p0 = p[-step], q0 = p[0], q1 = p[step];
-  const int a = 3 * (q0 - p0);
-  const int a1 = VP8ksclip2[(a + 4) >> 3];
-  const int a2 = VP8ksclip2[(a + 3) >> 3];
-  const int a3 = (a1 + 1) >> 1;
-  p[-2*step] = VP8kclip1[p1 + a3];
-  p[-  step] = VP8kclip1[p0 + a2];
-  p[      0] = VP8kclip1[q0 - a1];
-  p[   step] = VP8kclip1[q1 - a3];
-}
-
-// 6 pixels in, 6 pixels out
-static WEBP_INLINE void do_filter6(uint8_t* p, int step) {
-  const int p2 = p[-3*step], p1 = p[-2*step], p0 = p[-step];
-  const int q0 = p[0], q1 = p[step], q2 = p[2*step];
-  const int a = VP8ksclip1[3 * (q0 - p0) + VP8ksclip1[p1 - q1]];
-  // a is in [-128,127], a1 in [-27,27], a2 in [-18,18] and a3 in [-9,9]
-  const int a1 = (27 * a + 63) >> 7;  // eq. to ((3 * a + 7) * 9) >> 7
-  const int a2 = (18 * a + 63) >> 7;  // eq. to ((2 * a + 7) * 9) >> 7
-  const int a3 = (9  * a + 63) >> 7;  // eq. to ((1 * a + 7) * 9) >> 7
-  p[-3*step] = VP8kclip1[p2 + a3];
-  p[-2*step] = VP8kclip1[p1 + a2];
-  p[-  step] = VP8kclip1[p0 + a1];
-  p[      0] = VP8kclip1[q0 - a1];
-  p[   step] = VP8kclip1[q1 - a2];
-  p[ 2*step] = VP8kclip1[q2 - a3];
-}
-
-static WEBP_INLINE int hev(const uint8_t* p, int step, int thresh) {
-  const int p1 = p[-2*step], p0 = p[-step], q0 = p[0], q1 = p[step];
-  return (VP8kabs0[p1 - p0] > thresh) || (VP8kabs0[q1 - q0] > thresh);
-}
-
-static WEBP_INLINE int needs_filter(const uint8_t* p, int step, int t) {
-  const int p1 = p[-2 * step], p0 = p[-step], q0 = p[0], q1 = p[step];
-  return ((4 * VP8kabs0[p0 - q0] + VP8kabs0[p1 - q1]) <= t);
-}
-
-static WEBP_INLINE int needs_filter2(const uint8_t* p,
-                                     int step, int t, int it) {
-  const int p3 = p[-4 * step], p2 = p[-3 * step], p1 = p[-2 * step];
-  const int p0 = p[-step], q0 = p[0];
-  const int q1 = p[step], q2 = p[2 * step], q3 = p[3 * step];
-  if ((4 * VP8kabs0[p0 - q0] + VP8kabs0[p1 - q1]) > t) return 0;
-  return VP8kabs0[p3 - p2] <= it && VP8kabs0[p2 - p1] <= it &&
-         VP8kabs0[p1 - p0] <= it && VP8kabs0[q3 - q2] <= it &&
-         VP8kabs0[q2 - q1] <= it && VP8kabs0[q1 - q0] <= it;
-}
-
-//------------------------------------------------------------------------------
-// Simple In-loop filtering (Paragraph 15.2)
-
-static void SimpleVFilter16(uint8_t* p, int stride, int thresh) {
-  int i;
-  const int thresh2 = 2 * thresh + 1;
-  for (i = 0; i < 16; ++i) {
-    if (needs_filter(p + i, stride, thresh2)) {
-      do_filter2(p + i, stride);
-    }
-  }
-}
-
-static void SimpleHFilter16(uint8_t* p, int stride, int thresh) {
-  int i;
-  const int thresh2 = 2 * thresh + 1;
-  for (i = 0; i < 16; ++i) {
-    if (needs_filter(p + i * stride, 1, thresh2)) {
-      do_filter2(p + i * stride, 1);
-    }
-  }
-}
-
-static void SimpleVFilter16i(uint8_t* p, int stride, int thresh) {
-  int k;
-  for (k = 3; k > 0; --k) {
-    p += 4 * stride;
-    SimpleVFilter16(p, stride, thresh);
-  }
-}
-
-static void SimpleHFilter16i(uint8_t* p, int stride, int thresh) {
-  int k;
-  for (k = 3; k > 0; --k) {
-    p += 4;
-    SimpleHFilter16(p, stride, thresh);
-  }
-}
-
-//------------------------------------------------------------------------------
-// Complex In-loop filtering (Paragraph 15.3)
-
-static WEBP_INLINE void FilterLoop26(uint8_t* p,
-                                     int hstride, int vstride, int size,
-                                     int thresh, int ithresh, int hev_thresh) {
-  const int thresh2 = 2 * thresh + 1;
-  while (size-- > 0) {
-    if (needs_filter2(p, hstride, thresh2, ithresh)) {
-      if (hev(p, hstride, hev_thresh)) {
-        do_filter2(p, hstride);
-      } else {
-        do_filter6(p, hstride);
-      }
-    }
-    p += vstride;
-  }
-}
-
-static WEBP_INLINE void FilterLoop24(uint8_t* p,
-                                     int hstride, int vstride, int size,
-                                     int thresh, int ithresh, int hev_thresh) {
-  const int thresh2 = 2 * thresh + 1;
-  while (size-- > 0) {
-    if (needs_filter2(p, hstride, thresh2, ithresh)) {
-      if (hev(p, hstride, hev_thresh)) {
-        do_filter2(p, hstride);
-      } else {
-        do_filter4(p, hstride);
-      }
-    }
-    p += vstride;
-  }
-}
-
-// on macroblock edges
-static void VFilter16(uint8_t* p, int stride,
-                      int thresh, int ithresh, int hev_thresh) {
-  FilterLoop26(p, stride, 1, 16, thresh, ithresh, hev_thresh);
-}
-
-static void HFilter16(uint8_t* p, int stride,
-                      int thresh, int ithresh, int hev_thresh) {
-  FilterLoop26(p, 1, stride, 16, thresh, ithresh, hev_thresh);
-}
-
-// on three inner edges
-static void VFilter16i(uint8_t* p, int stride,
-                       int thresh, int ithresh, int hev_thresh) {
-  int k;
-  for (k = 3; k > 0; --k) {
-    p += 4 * stride;
-    FilterLoop24(p, stride, 1, 16, thresh, ithresh, hev_thresh);
-  }
-}
-
-static void HFilter16i(uint8_t* p, int stride,
-                       int thresh, int ithresh, int hev_thresh) {
-  int k;
-  for (k = 3; k > 0; --k) {
-    p += 4;
-    FilterLoop24(p, 1, stride, 16, thresh, ithresh, hev_thresh);
-  }
-}
-
-// 8-pixels wide variant, for chroma filtering
-static void VFilter8(uint8_t* u, uint8_t* v, int stride,
-                     int thresh, int ithresh, int hev_thresh) {
-  FilterLoop26(u, stride, 1, 8, thresh, ithresh, hev_thresh);
-  FilterLoop26(v, stride, 1, 8, thresh, ithresh, hev_thresh);
-}
-
-static void HFilter8(uint8_t* u, uint8_t* v, int stride,
-                     int thresh, int ithresh, int hev_thresh) {
-  FilterLoop26(u, 1, stride, 8, thresh, ithresh, hev_thresh);
-  FilterLoop26(v, 1, stride, 8, thresh, ithresh, hev_thresh);
-}
-
-static void VFilter8i(uint8_t* u, uint8_t* v, int stride,
-                      int thresh, int ithresh, int hev_thresh) {
-  FilterLoop24(u + 4 * stride, stride, 1, 8, thresh, ithresh, hev_thresh);
-  FilterLoop24(v + 4 * stride, stride, 1, 8, thresh, ithresh, hev_thresh);
-}
-
-static void HFilter8i(uint8_t* u, uint8_t* v, int stride,
-                      int thresh, int ithresh, int hev_thresh) {
-  FilterLoop24(u + 4, 1, stride, 8, thresh, ithresh, hev_thresh);
-  FilterLoop24(v + 4, 1, stride, 8, thresh, ithresh, hev_thresh);
-}
-
-//------------------------------------------------------------------------------
-
-VP8DecIdct2 VP8Transform;
-VP8DecIdct VP8TransformAC3;
-VP8DecIdct VP8TransformUV;
-VP8DecIdct VP8TransformDC;
-VP8DecIdct VP8TransformDCUV;
-
-VP8LumaFilterFunc VP8VFilter16;
-VP8LumaFilterFunc VP8HFilter16;
-VP8ChromaFilterFunc VP8VFilter8;
-VP8ChromaFilterFunc VP8HFilter8;
-VP8LumaFilterFunc VP8VFilter16i;
-VP8LumaFilterFunc VP8HFilter16i;
-VP8ChromaFilterFunc VP8VFilter8i;
-VP8ChromaFilterFunc VP8HFilter8i;
-VP8SimpleFilterFunc VP8SimpleVFilter16;
-VP8SimpleFilterFunc VP8SimpleHFilter16;
-VP8SimpleFilterFunc VP8SimpleVFilter16i;
-VP8SimpleFilterFunc VP8SimpleHFilter16i;
-
-extern void VP8DspInitSSE2(void);
-extern void VP8DspInitNEON(void);
-extern void VP8DspInitMIPS32(void);
-
-void VP8DspInit(void) {
-  VP8InitClipTables();
-
-  VP8TransformWHT = TransformWHT;
-  VP8Transform = TransformTwo;
-  VP8TransformUV = TransformUV;
-  VP8TransformDC = TransformDC;
-  VP8TransformDCUV = TransformDCUV;
-  VP8TransformAC3 = TransformAC3;
-
-  VP8VFilter16 = VFilter16;
-  VP8HFilter16 = HFilter16;
-  VP8VFilter8 = VFilter8;
-  VP8HFilter8 = HFilter8;
-  VP8VFilter16i = VFilter16i;
-  VP8HFilter16i = HFilter16i;
-  VP8VFilter8i = VFilter8i;
-  VP8HFilter8i = HFilter8i;
-  VP8SimpleVFilter16 = SimpleVFilter16;
-  VP8SimpleHFilter16 = SimpleHFilter16;
-  VP8SimpleVFilter16i = SimpleVFilter16i;
-  VP8SimpleHFilter16i = SimpleHFilter16i;
-
-  // If defined, use CPUInfo() to overwrite some pointers with faster versions.
-  if (VP8GetCPUInfo != NULL) {
-#if defined(WEBP_USE_SSE2)
-    if (VP8GetCPUInfo(kSSE2)) {
-      VP8DspInitSSE2();
-    }
-#elif defined(WEBP_USE_NEON)
-    if (VP8GetCPUInfo(kNEON)) {
-      VP8DspInitNEON();
-    }
-#elif defined(WEBP_USE_MIPS32)
-    if (VP8GetCPUInfo(kMIPS32)) {
-      VP8DspInitMIPS32();
-    }
-#endif
-  }
-}
-
@@ -1,366 +0,0 @@
-// Copyright 2014 Google Inc. All Rights Reserved.
-//
-// Use of this source code is governed by a BSD-style license
-// that can be found in the COPYING file in the root of the source
-// tree. An additional intellectual property rights grant can be found
-// in the file PATENTS. All contributing project authors may
-// be found in the AUTHORS file in the root of the source tree.
-// -----------------------------------------------------------------------------
-//
-// Clipping tables for filtering
-//
-// Author: Skal (pascal.massimino@gmail.com)
-
-#include "./dsp.h"
-
-#define USE_STATIC_TABLES     // undefine to have run-time table initialization
-
-#ifdef USE_STATIC_TABLES
-
-static const uint8_t abs0[255 + 255 + 1] = {
-  0xff, 0xfe, 0xfd, 0xfc, 0xfb, 0xfa, 0xf9, 0xf8, 0xf7, 0xf6, 0xf5, 0xf4,
-  0xf3, 0xf2, 0xf1, 0xf0, 0xef, 0xee, 0xed, 0xec, 0xeb, 0xea, 0xe9, 0xe8,
-  0xe7, 0xe6, 0xe5, 0xe4, 0xe3, 0xe2, 0xe1, 0xe0, 0xdf, 0xde, 0xdd, 0xdc,
-  0xdb, 0xda, 0xd9, 0xd8, 0xd7, 0xd6, 0xd5, 0xd4, 0xd3, 0xd2, 0xd1, 0xd0,
-  0xcf, 0xce, 0xcd, 0xcc, 0xcb, 0xca, 0xc9, 0xc8, 0xc7, 0xc6, 0xc5, 0xc4,
-  0xc3, 0xc2, 0xc1, 0xc0, 0xbf, 0xbe, 0xbd, 0xbc, 0xbb, 0xba, 0xb9, 0xb8,
-  0xb7, 0xb6, 0xb5, 0xb4, 0xb3, 0xb2, 0xb1, 0xb0, 0xaf, 0xae, 0xad, 0xac,
-  0xab, 0xaa, 0xa9, 0xa8, 0xa7, 0xa6, 0xa5, 0xa4, 0xa3, 0xa2, 0xa1, 0xa0,
-  0x9f, 0x9e, 0x9d, 0x9c, 0x9b, 0x9a, 0x99, 0x98, 0x97, 0x96, 0x95, 0x94,
-  0x93, 0x92, 0x91, 0x90, 0x8f, 0x8e, 0x8d, 0x8c, 0x8b, 0x8a, 0x89, 0x88,
-  0x87, 0x86, 0x85, 0x84, 0x83, 0x82, 0x81, 0x80, 0x7f, 0x7e, 0x7d, 0x7c,
-  0x7b, 0x7a, 0x79, 0x78, 0x77, 0x76, 0x75, 0x74, 0x73, 0x72, 0x71, 0x70,
-  0x6f, 0x6e, 0x6d, 0x6c, 0x6b, 0x6a, 0x69, 0x68, 0x67, 0x66, 0x65, 0x64,
-  0x63, 0x62, 0x61, 0x60, 0x5f, 0x5e, 0x5d, 0x5c, 0x5b, 0x5a, 0x59, 0x58,
-  0x57, 0x56, 0x55, 0x54, 0x53, 0x52, 0x51, 0x50, 0x4f, 0x4e, 0x4d, 0x4c,
-  0x4b, 0x4a, 0x49, 0x48, 0x47, 0x46, 0x45, 0x44, 0x43, 0x42, 0x41, 0x40,
-  0x3f, 0x3e, 0x3d, 0x3c, 0x3b, 0x3a, 0x39, 0x38, 0x37, 0x36, 0x35, 0x34,
-  0x33, 0x32, 0x31, 0x30, 0x2f, 0x2e, 0x2d, 0x2c, 0x2b, 0x2a, 0x29, 0x28,
-  0x27, 0x26, 0x25, 0x24, 0x23, 0x22, 0x21, 0x20, 0x1f, 0x1e, 0x1d, 0x1c,
-  0x1b, 0x1a, 0x19, 0x18, 0x17, 0x16, 0x15, 0x14, 0x13, 0x12, 0x11, 0x10,
-  0x0f, 0x0e, 0x0d, 0x0c, 0x0b, 0x0a, 0x09, 0x08, 0x07, 0x06, 0x05, 0x04,
-  0x03, 0x02, 0x01, 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08,
-  0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, 0x10, 0x11, 0x12, 0x13, 0x14,
-  0x15, 0x16, 0x17, 0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f, 0x20,
-  0x21, 0x22, 0x23, 0x24, 0x25, 0x26, 0x27, 0x28, 0x29, 0x2a, 0x2b, 0x2c,
-  0x2d, 0x2e, 0x2f, 0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37, 0x38,
-  0x39, 0x3a, 0x3b, 0x3c, 0x3d, 0x3e, 0x3f, 0x40, 0x41, 0x42, 0x43, 0x44,
-  0x45, 0x46, 0x47, 0x48, 0x49, 0x4a, 0x4b, 0x4c, 0x4d, 0x4e, 0x4f, 0x50,
-  0x51, 0x52, 0x53, 0x54, 0x55, 0x56, 0x57, 0x58, 0x59, 0x5a, 0x5b, 0x5c,
-  0x5d, 0x5e, 0x5f, 0x60, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68,
-  0x69, 0x6a, 0x6b, 0x6c, 0x6d, 0x6e, 0x6f, 0x70, 0x71, 0x72, 0x73, 0x74,
-  0x75, 0x76, 0x77, 0x78, 0x79, 0x7a, 0x7b, 0x7c, 0x7d, 0x7e, 0x7f, 0x80,
-  0x81, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87, 0x88, 0x89, 0x8a, 0x8b, 0x8c,
-  0x8d, 0x8e, 0x8f, 0x90, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97, 0x98,
-  0x99, 0x9a, 0x9b, 0x9c, 0x9d, 0x9e, 0x9f, 0xa0, 0xa1, 0xa2, 0xa3, 0xa4,
-  0xa5, 0xa6, 0xa7, 0xa8, 0xa9, 0xaa, 0xab, 0xac, 0xad, 0xae, 0xaf, 0xb0,
-  0xb1, 0xb2, 0xb3, 0xb4, 0xb5, 0xb6, 0xb7, 0xb8, 0xb9, 0xba, 0xbb, 0xbc,
-  0xbd, 0xbe, 0xbf, 0xc0, 0xc1, 0xc2, 0xc3, 0xc4, 0xc5, 0xc6, 0xc7, 0xc8,
-  0xc9, 0xca, 0xcb, 0xcc, 0xcd, 0xce, 0xcf, 0xd0, 0xd1, 0xd2, 0xd3, 0xd4,
-  0xd5, 0xd6, 0xd7, 0xd8, 0xd9, 0xda, 0xdb, 0xdc, 0xdd, 0xde, 0xdf, 0xe0,
-  0xe1, 0xe2, 0xe3, 0xe4, 0xe5, 0xe6, 0xe7, 0xe8, 0xe9, 0xea, 0xeb, 0xec,
-  0xed, 0xee, 0xef, 0xf0, 0xf1, 0xf2, 0xf3, 0xf4, 0xf5, 0xf6, 0xf7, 0xf8,
-  0xf9, 0xfa, 0xfb, 0xfc, 0xfd, 0xfe, 0xff
-};
-
-static const int8_t sclip1[1020 + 1020 + 1] = {
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
-  0x80, 0x80, 0x80, 0x80, 0x80, 0x81, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87,
-  0x88, 0x89, 0x8a, 0x8b, 0x8c, 0x8d, 0x8e, 0x8f, 0x90, 0x91, 0x92, 0x93,
-  0x94, 0x95, 0x96, 0x97, 0x98, 0x99, 0x9a, 0x9b, 0x9c, 0x9d, 0x9e, 0x9f,
-  0xa0, 0xa1, 0xa2, 0xa3, 0xa4, 0xa5, 0xa6, 0xa7, 0xa8, 0xa9, 0xaa, 0xab,
-  0xac, 0xad, 0xae, 0xaf, 0xb0, 0xb1, 0xb2, 0xb3, 0xb4, 0xb5, 0xb6, 0xb7,
-  0xb8, 0xb9, 0xba, 0xbb, 0xbc, 0xbd, 0xbe, 0xbf, 0xc0, 0xc1, 0xc2, 0xc3,
-  0xc4, 0xc5, 0xc6, 0xc7, 0xc8, 0xc9, 0xca, 0xcb, 0xcc, 0xcd, 0xce, 0xcf,
-  0xd0, 0xd1, 0xd2, 0xd3, 0xd4, 0xd5, 0xd6, 0xd7, 0xd8, 0xd9, 0xda, 0xdb,
-  0xdc, 0xdd, 0xde, 0xdf, 0xe0, 0xe1, 0xe2, 0xe3, 0xe4, 0xe5, 0xe6, 0xe7,
-  0xe8, 0xe9, 0xea, 0xeb, 0xec, 0xed, 0xee, 0xef, 0xf0, 0xf1, 0xf2, 0xf3,
-  0xf4, 0xf5, 0xf6, 0xf7, 0xf8, 0xf9, 0xfa, 0xfb, 0xfc, 0xfd, 0xfe, 0xff,
-  0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x09, 0x0a, 0x0b,
-  0x0c, 0x0d, 0x0e, 0x0f, 0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17,
-  0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f, 0x20, 0x21, 0x22, 0x23,
-  0x24, 0x25, 0x26, 0x27, 0x28, 0x29, 0x2a, 0x2b, 0x2c, 0x2d, 0x2e, 0x2f,
-  0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37, 0x38, 0x39, 0x3a, 0x3b,
-  0x3c, 0x3d, 0x3e, 0x3f, 0x40, 0x41, 0x42, 0x43, 0x44, 0x45, 0x46, 0x47,
-  0x48, 0x49, 0x4a, 0x4b, 0x4c, 0x4d, 0x4e, 0x4f, 0x50, 0x51, 0x52, 0x53,
-  0x54, 0x55, 0x56, 0x57, 0x58, 0x59, 0x5a, 0x5b, 0x5c, 0x5d, 0x5e, 0x5f,
-  0x60, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68, 0x69, 0x6a, 0x6b,
-  0x6c, 0x6d, 0x6e, 0x6f, 0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77,
-  0x78, 0x79, 0x7a, 0x7b, 0x7c, 0x7d, 0x7e, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f,
-  0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f, 0x7f
-};
-
-static const int8_t sclip2[112 + 112 + 1] = {
-  0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0,
-  0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0,
-  0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0,
-  0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0,
-  0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0,
-  0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0,
-  0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0,
-  0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0, 0xf0,
-  0xf0, 0xf1, 0xf2, 0xf3, 0xf4, 0xf5, 0xf6, 0xf7, 0xf8, 0xf9, 0xfa, 0xfb,
-  0xfc, 0xfd, 0xfe, 0xff, 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07,
-  0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f,
-  0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f,
-  0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f,
-  0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f,
-  0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f,
-  0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f,
-  0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f,
-  0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f,
-  0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f
-};
-
-static const uint8_t clip1[255 + 511 + 1] = {
-  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-  0x00, 0x00, 0x00, 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08,
-  0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, 0x10, 0x11, 0x12, 0x13, 0x14,
-  0x15, 0x16, 0x17, 0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f, 0x20,
-  0x21, 0x22, 0x23, 0x24, 0x25, 0x26, 0x27, 0x28, 0x29, 0x2a, 0x2b, 0x2c,
-  0x2d, 0x2e, 0x2f, 0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37, 0x38,
-  0x39, 0x3a, 0x3b, 0x3c, 0x3d, 0x3e, 0x3f, 0x40, 0x41, 0x42, 0x43, 0x44,
-  0x45, 0x46, 0x47, 0x48, 0x49, 0x4a, 0x4b, 0x4c, 0x4d, 0x4e, 0x4f, 0x50,
-  0x51, 0x52, 0x53, 0x54, 0x55, 0x56, 0x57, 0x58, 0x59, 0x5a, 0x5b, 0x5c,
-  0x5d, 0x5e, 0x5f, 0x60, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68,
-  0x69, 0x6a, 0x6b, 0x6c, 0x6d, 0x6e, 0x6f, 0x70, 0x71, 0x72, 0x73, 0x74,
-  0x75, 0x76, 0x77, 0x78, 0x79, 0x7a, 0x7b, 0x7c, 0x7d, 0x7e, 0x7f, 0x80,
-  0x81, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87, 0x88, 0x89, 0x8a, 0x8b, 0x8c,
-  0x8d, 0x8e, 0x8f, 0x90, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97, 0x98,
-  0x99, 0x9a, 0x9b, 0x9c, 0x9d, 0x9e, 0x9f, 0xa0, 0xa1, 0xa2, 0xa3, 0xa4,
-  0xa5, 0xa6, 0xa7, 0xa8, 0xa9, 0xaa, 0xab, 0xac, 0xad, 0xae, 0xaf, 0xb0,
-  0xb1, 0xb2, 0xb3, 0xb4, 0xb5, 0xb6, 0xb7, 0xb8, 0xb9, 0xba, 0xbb, 0xbc,
-  0xbd, 0xbe, 0xbf, 0xc0, 0xc1, 0xc2, 0xc3, 0xc4, 0xc5, 0xc6, 0xc7, 0xc8,
-  0xc9, 0xca, 0xcb, 0xcc, 0xcd, 0xce, 0xcf, 0xd0, 0xd1, 0xd2, 0xd3, 0xd4,
-  0xd5, 0xd6, 0xd7, 0xd8, 0xd9, 0xda, 0xdb, 0xdc, 0xdd, 0xde, 0xdf, 0xe0,
-  0xe1, 0xe2, 0xe3, 0xe4, 0xe5, 0xe6, 0xe7, 0xe8, 0xe9, 0xea, 0xeb, 0xec,
-  0xed, 0xee, 0xef, 0xf0, 0xf1, 0xf2, 0xf3, 0xf4, 0xf5, 0xf6, 0xf7, 0xf8,
-  0xf9, 0xfa, 0xfb, 0xfc, 0xfd, 0xfe, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
-  0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
-  0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
-  0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
-  0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
-  0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
-  0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
-  0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
-  0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
-  0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
-  0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
-  0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
-  0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
-  0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
-  0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
-  0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
-  0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
-  0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
-  0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
-  0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
-  0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
-  0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff
-};
-
-#else
-
-// uninitialized tables
-static uint8_t abs0[255 + 255 + 1];
-static int8_t sclip1[1020 + 1020 + 1];
-static int8_t sclip2[112 + 112 + 1];
-static uint8_t clip1[255 + 511 + 1];
-
-// We declare this variable 'volatile' to prevent instruction reordering
-// and make sure it's set to true _last_ (so as to be thread-safe)
-static volatile int tables_ok = 0;
-
-#endif
-
-const int8_t* const VP8ksclip1 = &sclip1[1020];
-const int8_t* const VP8ksclip2 = &sclip2[112];
-const uint8_t* const VP8kclip1 = &clip1[255];
-const uint8_t* const VP8kabs0 = &abs0[255];
-
-void VP8InitClipTables(void) {
-#if !defined(USE_STATIC_TABLES)
-  int i;
-  if (!tables_ok) {
-    for (i = -255; i <= 255; ++i) {
-      abs0[255 + i] = (i < 0) ? -i : i;
-    }
-    for (i = -1020; i <= 1020; ++i) {
-      sclip1[1020 + i] = (i < -128) ? -128 : (i > 127) ? 127 : i;
-    }
-    for (i = -112; i <= 112; ++i) {
-      sclip2[112 + i] = (i < -16) ? -16 : (i > 15) ? 15 : i;
-    }
-    for (i = -255; i <= 255 + 255; ++i) {
-      clip1[255 + i] = (i < 0) ? 0 : (i > 255) ? 255 : i;
-    }
-    tables_ok = 1;
-  }
-#endif    // USE_STATIC_TABLES
-}
@@ -1,578 +0,0 @@
-// Copyright 2014 Google Inc. All Rights Reserved.
-//
-// Use of this source code is governed by a BSD-style license
-// that can be found in the COPYING file in the root of the source
-// tree. An additional intellectual property rights grant can be found
-// in the file PATENTS. All contributing project authors may
-// be found in the AUTHORS file in the root of the source tree.
-// -----------------------------------------------------------------------------
-//
-// MIPS version of dsp functions
-//
-// Author(s):  Djordje Pesut    (djordje.pesut@imgtec.com)
-//             Jovan Zelincevic (jovan.zelincevic@imgtec.com)
-
-#include "./dsp.h"
-
-#if defined(WEBP_USE_MIPS32)
-
-static const int kC1 = 20091 + (1 << 16);
-static const int kC2 = 35468;
-
-static WEBP_INLINE int abs_mips32(int x) {
-  const int sign = x >> 31;
-  return (x ^ sign) - sign;
-}
-
-// 4 pixels in, 2 pixels out
-static WEBP_INLINE void do_filter2(uint8_t* p, int step) {
-  const int p1 = p[-2 * step], p0 = p[-step], q0 = p[0], q1 = p[step];
-  const int a = 3 * (q0 - p0) + VP8ksclip1[p1 - q1];
-  const int a1 = VP8ksclip2[(a + 4) >> 3];
-  const int a2 = VP8ksclip2[(a + 3) >> 3];
-  p[-step] = VP8kclip1[p0 + a2];
-  p[    0] = VP8kclip1[q0 - a1];
-}
-
-// 4 pixels in, 4 pixels out
-static WEBP_INLINE void do_filter4(uint8_t* p, int step) {
-  const int p1 = p[-2 * step], p0 = p[-step], q0 = p[0], q1 = p[step];
-  const int a = 3 * (q0 - p0);
-  const int a1 = VP8ksclip2[(a + 4) >> 3];
-  const int a2 = VP8ksclip2[(a + 3) >> 3];
-  const int a3 = (a1 + 1) >> 1;
-  p[-2 * step] = VP8kclip1[p1 + a3];
-  p[-    step] = VP8kclip1[p0 + a2];
-  p[        0] = VP8kclip1[q0 - a1];
-  p[     step] = VP8kclip1[q1 - a3];
-}
-
-// 6 pixels in, 6 pixels out
-static WEBP_INLINE void do_filter6(uint8_t* p, int step) {
-  const int p2 = p[-3 * step], p1 = p[-2 * step], p0 = p[-step];
-  const int q0 = p[0], q1 = p[step], q2 = p[2 * step];
-  const int a = VP8ksclip1[3 * (q0 - p0) + VP8ksclip1[p1 - q1]];
-  const int a1 = (27 * a + 63) >> 7;  // eq. to ((3 * a + 7) * 9) >> 7
-  const int a2 = (18 * a + 63) >> 7;  // eq. to ((2 * a + 7) * 9) >> 7
-  const int a3 = (9  * a + 63) >> 7;  // eq. to ((1 * a + 7) * 9) >> 7
-  p[-3 * step] = VP8kclip1[p2 + a3];
-  p[-2 * step] = VP8kclip1[p1 + a2];
-  p[-    step] = VP8kclip1[p0 + a1];
-  p[        0] = VP8kclip1[q0 - a1];
-  p[     step] = VP8kclip1[q1 - a2];
-  p[ 2 * step] = VP8kclip1[q2 - a3];
-}
-
-static WEBP_INLINE int hev(const uint8_t* p, int step, int thresh) {
-  const int p1 = p[-2 * step], p0 = p[-step], q0 = p[0], q1 = p[step];
-  return (abs_mips32(p1 - p0) > thresh) || (abs_mips32(q1 - q0) > thresh);
-}
-
-static WEBP_INLINE int needs_filter(const uint8_t* p, int step, int thresh) {
-  const int p1 = p[-2 * step], p0 = p[-step], q0 = p[0], q1 = p[step];
-  return ((2 * abs_mips32(p0 - q0) + (abs_mips32(p1 - q1) >> 1)) <= thresh);
-}
-
-static WEBP_INLINE int needs_filter2(const uint8_t* p,
-                                     int step, int t, int it) {
-  const int p3 = p[-4 * step], p2 = p[-3 * step];
-  const int p1 = p[-2 * step], p0 = p[-step];
-  const int q0 = p[0], q1 = p[step], q2 = p[2 * step], q3 = p[3 * step];
-  if ((2 * abs_mips32(p0 - q0) + (abs_mips32(p1 - q1) >> 1)) > t) {
-    return 0;
-  }
-  return abs_mips32(p3 - p2) <= it && abs_mips32(p2 - p1) <= it &&
-         abs_mips32(p1 - p0) <= it && abs_mips32(q3 - q2) <= it &&
-         abs_mips32(q2 - q1) <= it && abs_mips32(q1 - q0) <= it;
-}
-
-static WEBP_INLINE void FilterLoop26(uint8_t* p,
-                                     int hstride, int vstride, int size,
-                                     int thresh, int ithresh, int hev_thresh) {
-  while (size-- > 0) {
-    if (needs_filter2(p, hstride, thresh, ithresh)) {
-      if (hev(p, hstride, hev_thresh)) {
-        do_filter2(p, hstride);
-      } else {
-        do_filter6(p, hstride);
-      }
-    }
-    p += vstride;
-  }
-}
-
-static WEBP_INLINE void FilterLoop24(uint8_t* p,
-                                     int hstride, int vstride, int size,
-                                     int thresh, int ithresh, int hev_thresh) {
-  while (size-- > 0) {
-    if (needs_filter2(p, hstride, thresh, ithresh)) {
-      if (hev(p, hstride, hev_thresh)) {
-        do_filter2(p, hstride);
-      } else {
-        do_filter4(p, hstride);
-      }
-    }
-    p += vstride;
-  }
-}
-
-// on macroblock edges
-static void VFilter16(uint8_t* p, int stride,
-                      int thresh, int ithresh, int hev_thresh) {
-  FilterLoop26(p, stride, 1, 16, thresh, ithresh, hev_thresh);
-}
-
-static void HFilter16(uint8_t* p, int stride,
-                      int thresh, int ithresh, int hev_thresh) {
-  FilterLoop26(p, 1, stride, 16, thresh, ithresh, hev_thresh);
-}
-
-// 8-pixels wide variant, for chroma filtering
-static void VFilter8(uint8_t* u, uint8_t* v, int stride,
-                     int thresh, int ithresh, int hev_thresh) {
-  FilterLoop26(u, stride, 1, 8, thresh, ithresh, hev_thresh);
-  FilterLoop26(v, stride, 1, 8, thresh, ithresh, hev_thresh);
-}
-
-static void HFilter8(uint8_t* u, uint8_t* v, int stride,
-                     int thresh, int ithresh, int hev_thresh) {
-  FilterLoop26(u, 1, stride, 8, thresh, ithresh, hev_thresh);
-  FilterLoop26(v, 1, stride, 8, thresh, ithresh, hev_thresh);
-}
-
-static void VFilter8i(uint8_t* u, uint8_t* v, int stride,
-                      int thresh, int ithresh, int hev_thresh) {
-  FilterLoop24(u + 4 * stride, stride, 1, 8, thresh, ithresh, hev_thresh);
-  FilterLoop24(v + 4 * stride, stride, 1, 8, thresh, ithresh, hev_thresh);
-}
-
-static void HFilter8i(uint8_t* u, uint8_t* v, int stride,
-                      int thresh, int ithresh, int hev_thresh) {
-  FilterLoop24(u + 4, 1, stride, 8, thresh, ithresh, hev_thresh);
-  FilterLoop24(v + 4, 1, stride, 8, thresh, ithresh, hev_thresh);
-}
-
-// on three inner edges
-static void VFilter16i(uint8_t* p, int stride,
-                       int thresh, int ithresh, int hev_thresh) {
-  int k;
-  for (k = 3; k > 0; --k) {
-    p += 4 * stride;
-    FilterLoop24(p, stride, 1, 16, thresh, ithresh, hev_thresh);
-  }
-}
-
-static void HFilter16i(uint8_t* p, int stride,
-                       int thresh, int ithresh, int hev_thresh) {
-  int k;
-  for (k = 3; k > 0; --k) {
-    p += 4;
-    FilterLoop24(p, 1, stride, 16, thresh, ithresh, hev_thresh);
-  }
-}
-
-//------------------------------------------------------------------------------
-// Simple In-loop filtering (Paragraph 15.2)
-
-static void SimpleVFilter16(uint8_t* p, int stride, int thresh) {
-  int i;
-  for (i = 0; i < 16; ++i) {
-    if (needs_filter(p + i, stride, thresh)) {
-      do_filter2(p + i, stride);
-    }
-  }
-}
-
-static void SimpleHFilter16(uint8_t* p, int stride, int thresh) {
-  int i;
-  for (i = 0; i < 16; ++i) {
-    if (needs_filter(p + i * stride, 1, thresh)) {
-      do_filter2(p + i * stride, 1);
-    }
-  }
-}
-
-static void SimpleVFilter16i(uint8_t* p, int stride, int thresh) {
-  int k;
-  for (k = 3; k > 0; --k) {
-    p += 4 * stride;
-    SimpleVFilter16(p, stride, thresh);
-  }
-}
-
-static void SimpleHFilter16i(uint8_t* p, int stride, int thresh) {
-  int k;
-  for (k = 3; k > 0; --k) {
-    p += 4;
-    SimpleHFilter16(p, stride, thresh);
-  }
-}
-
-static void TransformOne(const int16_t* in, uint8_t* dst) {
-  int temp0, temp1, temp2, temp3, temp4;
-  int temp5, temp6, temp7, temp8, temp9;
-  int temp10, temp11, temp12, temp13, temp14;
-  int temp15, temp16, temp17, temp18;
-  int16_t* p_in = (int16_t*)in;
-
-  // loops unrolled and merged to avoid usage of tmp buffer
-  // and to reduce number of stalls. MUL macro is written
-  // in assembler and inlined
-  __asm__ volatile(
-    "lh       %[temp0],  0(%[in])                      \n\t"
-    "lh       %[temp8],  16(%[in])                     \n\t"
-    "lh       %[temp4],  8(%[in])                      \n\t"
-    "lh       %[temp12], 24(%[in])                     \n\t"
-    "addu     %[temp16], %[temp0],  %[temp8]           \n\t"
-    "subu     %[temp0],  %[temp0],  %[temp8]           \n\t"
-    "mul      %[temp8],  %[temp4],  %[kC2]             \n\t"
-    "mul      %[temp17], %[temp12], %[kC1]             \n\t"
-    "mul      %[temp4],  %[temp4],  %[kC1]             \n\t"
-    "mul      %[temp12], %[temp12], %[kC2]             \n\t"
-    "lh       %[temp1],  2(%[in])                      \n\t"
-    "lh       %[temp5],  10(%[in])                     \n\t"
-    "lh       %[temp9],  18(%[in])                     \n\t"
-    "lh       %[temp13], 26(%[in])                     \n\t"
-    "sra      %[temp8],  %[temp8],  16                 \n\t"
-    "sra      %[temp17], %[temp17], 16                 \n\t"
-    "sra      %[temp4],  %[temp4],  16                 \n\t"
-    "sra      %[temp12], %[temp12], 16                 \n\t"
-    "lh       %[temp2],  4(%[in])                      \n\t"
-    "lh       %[temp6],  12(%[in])                     \n\t"
-    "lh       %[temp10], 20(%[in])                     \n\t"
-    "lh       %[temp14], 28(%[in])                     \n\t"
-    "subu     %[temp17], %[temp8],  %[temp17]          \n\t"
-    "addu     %[temp4],  %[temp4],  %[temp12]          \n\t"
-    "addu     %[temp8],  %[temp16], %[temp4]           \n\t"
-    "subu     %[temp4],  %[temp16], %[temp4]           \n\t"
-    "addu     %[temp16], %[temp1],  %[temp9]           \n\t"
-    "subu     %[temp1],  %[temp1],  %[temp9]           \n\t"
-    "lh       %[temp3],  6(%[in])                      \n\t"
-    "lh       %[temp7],  14(%[in])                     \n\t"
-    "lh       %[temp11], 22(%[in])                     \n\t"
-    "lh       %[temp15], 30(%[in])                     \n\t"
-    "addu     %[temp12], %[temp0],  %[temp17]          \n\t"
-    "subu     %[temp0],  %[temp0],  %[temp17]          \n\t"
-    "mul      %[temp9],  %[temp5],  %[kC2]             \n\t"
-    "mul      %[temp17], %[temp13], %[kC1]             \n\t"
-    "mul      %[temp5],  %[temp5],  %[kC1]             \n\t"
-    "mul      %[temp13], %[temp13], %[kC2]             \n\t"
-    "sra      %[temp9],  %[temp9],  16                 \n\t"
-    "sra      %[temp17], %[temp17], 16                 \n\t"
-    "subu     %[temp17], %[temp9],  %[temp17]          \n\t"
-    "sra      %[temp5],  %[temp5],  16                 \n\t"
-    "sra      %[temp13], %[temp13], 16                 \n\t"
-    "addu     %[temp5],  %[temp5],  %[temp13]          \n\t"
-    "addu     %[temp13], %[temp1],  %[temp17]          \n\t"
-    "subu     %[temp1],  %[temp1],  %[temp17]          \n\t"
-    "mul      %[temp17], %[temp14], %[kC1]             \n\t"
-    "mul      %[temp14], %[temp14], %[kC2]             \n\t"
-    "addu     %[temp9],  %[temp16], %[temp5]           \n\t"
-    "subu     %[temp5],  %[temp16], %[temp5]           \n\t"
-    "addu     %[temp16], %[temp2],  %[temp10]          \n\t"
-    "subu     %[temp2],  %[temp2],  %[temp10]          \n\t"
-    "mul      %[temp10], %[temp6],  %[kC2]             \n\t"
-    "mul      %[temp6],  %[temp6],  %[kC1]             \n\t"
-    "sra      %[temp17], %[temp17], 16                 \n\t"
-    "sra      %[temp14], %[temp14], 16                 \n\t"
-    "sra      %[temp10], %[temp10], 16                 \n\t"
-    "sra      %[temp6],  %[temp6],  16                 \n\t"
-    "subu     %[temp17], %[temp10], %[temp17]          \n\t"
-    "addu     %[temp6],  %[temp6],  %[temp14]          \n\t"
-    "addu     %[temp10], %[temp16], %[temp6]           \n\t"
-    "subu     %[temp6],  %[temp16], %[temp6]           \n\t"
-    "addu     %[temp14], %[temp2],  %[temp17]          \n\t"
-    "subu     %[temp2],  %[temp2],  %[temp17]          \n\t"
-    "mul      %[temp17], %[temp15], %[kC1]             \n\t"
-    "mul      %[temp15], %[temp15], %[kC2]             \n\t"
-    "addu     %[temp16], %[temp3],  %[temp11]          \n\t"
-    "subu     %[temp3],  %[temp3],  %[temp11]          \n\t"
-    "mul      %[temp11], %[temp7],  %[kC2]             \n\t"
-    "mul      %[temp7],  %[temp7],  %[kC1]             \n\t"
-    "addiu    %[temp8],  %[temp8],  4                  \n\t"
-    "addiu    %[temp12], %[temp12], 4                  \n\t"
-    "addiu    %[temp0],  %[temp0],  4                  \n\t"
-    "addiu    %[temp4],  %[temp4],  4                  \n\t"
-    "sra      %[temp17], %[temp17], 16                 \n\t"
-    "sra      %[temp15], %[temp15], 16                 \n\t"
-    "sra      %[temp11], %[temp11], 16                 \n\t"
-    "sra      %[temp7],  %[temp7],  16                 \n\t"
-    "subu     %[temp17], %[temp11], %[temp17]          \n\t"
-    "addu     %[temp7],  %[temp7],  %[temp15]          \n\t"
-    "addu     %[temp15], %[temp3],  %[temp17]          \n\t"
-    "subu     %[temp3],  %[temp3],  %[temp17]          \n\t"
-    "addu     %[temp11], %[temp16], %[temp7]           \n\t"
-    "subu     %[temp7],  %[temp16], %[temp7]           \n\t"
-    "addu     %[temp16], %[temp8],  %[temp10]          \n\t"
-    "subu     %[temp8],  %[temp8],  %[temp10]          \n\t"
-    "mul      %[temp10], %[temp9],  %[kC2]             \n\t"
-    "mul      %[temp17], %[temp11], %[kC1]             \n\t"
-    "mul      %[temp9],  %[temp9],  %[kC1]             \n\t"
-    "mul      %[temp11], %[temp11], %[kC2]             \n\t"
-    "sra      %[temp10], %[temp10], 16                 \n\t"
-    "sra      %[temp17], %[temp17], 16                 \n\t"
-    "sra      %[temp9],  %[temp9],  16                 \n\t"
-    "sra      %[temp11], %[temp11], 16                 \n\t"
-    "subu     %[temp17], %[temp10], %[temp17]          \n\t"
-    "addu     %[temp11], %[temp9],  %[temp11]          \n\t"
-    "addu     %[temp10], %[temp12], %[temp14]          \n\t"
-    "subu     %[temp12], %[temp12], %[temp14]          \n\t"
-    "mul      %[temp14], %[temp13], %[kC2]             \n\t"
-    "mul      %[temp9],  %[temp15], %[kC1]             \n\t"
-    "mul      %[temp13], %[temp13], %[kC1]             \n\t"
-    "mul      %[temp15], %[temp15], %[kC2]             \n\t"
-    "sra      %[temp14], %[temp14], 16                 \n\t"
-    "sra      %[temp9],  %[temp9],  16                 \n\t"
-    "sra      %[temp13], %[temp13], 16                 \n\t"
-    "sra      %[temp15], %[temp15], 16                 \n\t"
-    "subu     %[temp9],  %[temp14], %[temp9]           \n\t"
-    "addu     %[temp15], %[temp13], %[temp15]          \n\t"
-    "addu     %[temp14], %[temp0],  %[temp2]           \n\t"
-    "subu     %[temp0],  %[temp0],  %[temp2]           \n\t"
-    "mul      %[temp2],  %[temp1],  %[kC2]             \n\t"
-    "mul      %[temp13], %[temp3],  %[kC1]             \n\t"
-    "mul      %[temp1],  %[temp1],  %[kC1]             \n\t"
-    "mul      %[temp3],  %[temp3],  %[kC2]             \n\t"
-    "sra      %[temp2],  %[temp2],  16                 \n\t"
-    "sra      %[temp13], %[temp13], 16                 \n\t"
-    "sra      %[temp1],  %[temp1],  16                 \n\t"
-    "sra      %[temp3],  %[temp3],  16                 \n\t"
-    "subu     %[temp13], %[temp2],  %[temp13]          \n\t"
-    "addu     %[temp3],  %[temp1],  %[temp3]           \n\t"
-    "addu     %[temp2],  %[temp4],  %[temp6]           \n\t"
-    "subu     %[temp4],  %[temp4],  %[temp6]           \n\t"
-    "mul      %[temp6],  %[temp5],  %[kC2]             \n\t"
-    "mul      %[temp1],  %[temp7],  %[kC1]             \n\t"
-    "mul      %[temp5],  %[temp5],  %[kC1]             \n\t"
-    "mul      %[temp7],  %[temp7],  %[kC2]             \n\t"
-    "sra      %[temp6],  %[temp6],  16                 \n\t"
-    "sra      %[temp1],  %[temp1],  16                 \n\t"
-    "sra      %[temp5],  %[temp5],  16                 \n\t"
-    "sra      %[temp7],  %[temp7],  16                 \n\t"
-    "subu     %[temp1],  %[temp6],  %[temp1]           \n\t"
-    "addu     %[temp7],  %[temp5],  %[temp7]           \n\t"
-    "addu     %[temp5],  %[temp16], %[temp11]          \n\t"
-    "subu     %[temp16], %[temp16], %[temp11]          \n\t"
-    "addu     %[temp11], %[temp8],  %[temp17]          \n\t"
-    "subu     %[temp8],  %[temp8],  %[temp17]          \n\t"
-    "sra      %[temp5],  %[temp5],  3                  \n\t"
-    "sra      %[temp16], %[temp16], 3                  \n\t"
-    "sra      %[temp11], %[temp11], 3                  \n\t"
-    "sra      %[temp8],  %[temp8],  3                  \n\t"
-    "addu     %[temp17], %[temp10], %[temp15]          \n\t"
-    "subu     %[temp10], %[temp10], %[temp15]          \n\t"
-    "addu     %[temp15], %[temp12], %[temp9]           \n\t"
-    "subu     %[temp12], %[temp12], %[temp9]           \n\t"
-    "sra      %[temp17], %[temp17], 3                  \n\t"
-    "sra      %[temp10], %[temp10], 3                  \n\t"
-    "sra      %[temp15], %[temp15], 3                  \n\t"
-    "sra      %[temp12], %[temp12], 3                  \n\t"
-    "addu     %[temp9],  %[temp14], %[temp3]           \n\t"
-    "subu     %[temp14], %[temp14], %[temp3]           \n\t"
-    "addu     %[temp3],  %[temp0],  %[temp13]          \n\t"
-    "subu     %[temp0],  %[temp0],  %[temp13]          \n\t"
-    "sra      %[temp9],  %[temp9],  3                  \n\t"
-    "sra      %[temp14], %[temp14], 3                  \n\t"
-    "sra      %[temp3],  %[temp3],  3                  \n\t"
-    "sra      %[temp0],  %[temp0],  3                  \n\t"
-    "addu     %[temp13], %[temp2],  %[temp7]           \n\t"
-    "subu     %[temp2],  %[temp2],  %[temp7]           \n\t"
-    "addu     %[temp7],  %[temp4],  %[temp1]           \n\t"
-    "subu     %[temp4],  %[temp4],  %[temp1]           \n\t"
-    "sra      %[temp13], %[temp13], 3                  \n\t"
-    "sra      %[temp2],  %[temp2],  3                  \n\t"
-    "sra      %[temp7],  %[temp7],  3                  \n\t"
-    "sra      %[temp4],  %[temp4],  3                  \n\t"
-    "addiu    %[temp6],  $zero,     255                \n\t"
-    "lbu      %[temp1],  0(%[dst])                     \n\t"
-    "addu     %[temp1],  %[temp1],  %[temp5]           \n\t"
-    "sra      %[temp5],  %[temp1],  8                  \n\t"
-    "sra      %[temp18], %[temp1],  31                 \n\t"
-    "beqz     %[temp5],  1f                            \n\t"
-    "xor      %[temp1],  %[temp1],  %[temp1]           \n\t"
-    "movz     %[temp1],  %[temp6],  %[temp18]          \n\t"
-  "1:                                                  \n\t"
-    "lbu      %[temp18], 1(%[dst])                     \n\t"
-    "sb       %[temp1],  0(%[dst])                     \n\t"
-    "addu     %[temp18], %[temp18], %[temp11]          \n\t"
-    "sra      %[temp11], %[temp18], 8                  \n\t"
-    "sra      %[temp1],  %[temp18], 31                 \n\t"
-    "beqz     %[temp11], 2f                            \n\t"
-    "xor      %[temp18], %[temp18], %[temp18]          \n\t"
-    "movz     %[temp18], %[temp6],  %[temp1]           \n\t"
-  "2:                                                  \n\t"
-    "lbu      %[temp1],  2(%[dst])                     \n\t"
-    "sb       %[temp18], 1(%[dst])                     \n\t"
-    "addu     %[temp1],  %[temp1],  %[temp8]           \n\t"
-    "sra      %[temp8],  %[temp1],  8                  \n\t"
-    "sra      %[temp18], %[temp1],  31                 \n\t"
-    "beqz     %[temp8],  3f                            \n\t"
-    "xor      %[temp1],  %[temp1],  %[temp1]           \n\t"
-    "movz     %[temp1],  %[temp6],  %[temp18]          \n\t"
-  "3:                                                  \n\t"
-    "lbu      %[temp18], 3(%[dst])                     \n\t"
-    "sb       %[temp1],  2(%[dst])                     \n\t"
-    "addu     %[temp18], %[temp18], %[temp16]          \n\t"
-    "sra      %[temp16], %[temp18], 8                  \n\t"
-    "sra      %[temp1],  %[temp18], 31                 \n\t"
-    "beqz     %[temp16], 4f                            \n\t"
-    "xor      %[temp18], %[temp18], %[temp18]          \n\t"
-    "movz     %[temp18], %[temp6],  %[temp1]           \n\t"
-  "4:                                                  \n\t"
-    "sb       %[temp18], 3(%[dst])                     \n\t"
-    "lbu      %[temp5],  32(%[dst])                    \n\t"
-    "lbu      %[temp8],  33(%[dst])                    \n\t"
-    "lbu      %[temp11], 34(%[dst])                    \n\t"
-    "lbu      %[temp16], 35(%[dst])                    \n\t"
-    "addu     %[temp5],  %[temp5],  %[temp17]          \n\t"
-    "addu     %[temp8],  %[temp8],  %[temp15]          \n\t"
-    "addu     %[temp11], %[temp11], %[temp12]          \n\t"
-    "addu     %[temp16], %[temp16], %[temp10]          \n\t"
-    "sra      %[temp18], %[temp5],  8                  \n\t"
-    "sra      %[temp1],  %[temp5],  31                 \n\t"
-    "beqz     %[temp18], 5f                            \n\t"
-    "xor      %[temp5],  %[temp5],  %[temp5]           \n\t"
-    "movz     %[temp5],  %[temp6],  %[temp1]           \n\t"
-  "5:                                                  \n\t"
-    "sra      %[temp18], %[temp8],  8                  \n\t"
-    "sra      %[temp1],  %[temp8],  31                 \n\t"
-    "beqz     %[temp18], 6f                            \n\t"
-    "xor      %[temp8],  %[temp8],  %[temp8]           \n\t"
-    "movz     %[temp8],  %[temp6],  %[temp1]           \n\t"
-  "6:                                                  \n\t"
-    "sra      %[temp18], %[temp11], 8                  \n\t"
-    "sra      %[temp1],  %[temp11], 31                 \n\t"
-    "sra      %[temp17], %[temp16], 8                  \n\t"
-    "sra      %[temp15], %[temp16], 31                 \n\t"
-    "beqz     %[temp18], 7f                            \n\t"
-    "xor      %[temp11], %[temp11], %[temp11]          \n\t"
-    "movz     %[temp11], %[temp6],  %[temp1]           \n\t"
-  "7:                                                  \n\t"
-    "beqz     %[temp17], 8f                            \n\t"
-    "xor      %[temp16], %[temp16], %[temp16]          \n\t"
-    "movz     %[temp16], %[temp6],  %[temp15]          \n\t"
-  "8:                                                  \n\t"
-    "sb       %[temp5],  32(%[dst])                    \n\t"
-    "sb       %[temp8],  33(%[dst])                    \n\t"
-    "sb       %[temp11], 34(%[dst])                    \n\t"
-    "sb       %[temp16], 35(%[dst])                    \n\t"
-    "lbu      %[temp5],  64(%[dst])                    \n\t"
-    "lbu      %[temp8],  65(%[dst])                    \n\t"
-    "lbu      %[temp11], 66(%[dst])                    \n\t"
-    "lbu      %[temp16], 67(%[dst])                    \n\t"
-    "addu     %[temp5],  %[temp5],  %[temp9]           \n\t"
-    "addu     %[temp8],  %[temp8],  %[temp3]           \n\t"
-    "addu     %[temp11], %[temp11], %[temp0]           \n\t"
-    "addu     %[temp16], %[temp16], %[temp14]          \n\t"
-    "sra      %[temp18], %[temp5],  8                  \n\t"
-    "sra      %[temp1],  %[temp5],  31                 \n\t"
-    "sra      %[temp17], %[temp8],  8                  \n\t"
-    "sra      %[temp15], %[temp8],  31                 \n\t"
-    "sra      %[temp12], %[temp11], 8                  \n\t"
-    "sra      %[temp10], %[temp11], 31                 \n\t"
-    "sra      %[temp9],  %[temp16], 8                  \n\t"
-    "sra      %[temp3],  %[temp16], 31                 \n\t"
-    "beqz     %[temp18], 9f                            \n\t"
-    "xor      %[temp5],  %[temp5],  %[temp5]           \n\t"
-    "movz     %[temp5],  %[temp6],  %[temp1]           \n\t"
-  "9:                                                  \n\t"
-    "beqz     %[temp17], 10f                           \n\t"
-    "xor      %[temp8],  %[temp8],  %[temp8]           \n\t"
-    "movz     %[temp8],  %[temp6],  %[temp15]          \n\t"
-  "10:                                                 \n\t"
-    "beqz     %[temp12], 11f                           \n\t"
-    "xor      %[temp11], %[temp11], %[temp11]          \n\t"
-    "movz     %[temp11], %[temp6],  %[temp10]          \n\t"
-  "11:                                                 \n\t"
-    "beqz     %[temp9],  12f                           \n\t"
-    "xor      %[temp16], %[temp16], %[temp16]          \n\t"
-    "movz     %[temp16], %[temp6],  %[temp3]           \n\t"
-  "12:                                                 \n\t"
-    "sb       %[temp5],  64(%[dst])                    \n\t"
-    "sb       %[temp8],  65(%[dst])                    \n\t"
-    "sb       %[temp11], 66(%[dst])                    \n\t"
-    "sb       %[temp16], 67(%[dst])                    \n\t"
-    "lbu      %[temp5],  96(%[dst])                    \n\t"
-    "lbu      %[temp8],  97(%[dst])                    \n\t"
-    "lbu      %[temp11], 98(%[dst])                    \n\t"
-    "lbu      %[temp16], 99(%[dst])                    \n\t"
-    "addu     %[temp5],  %[temp5],  %[temp13]          \n\t"
-    "addu     %[temp8],  %[temp8],  %[temp7]           \n\t"
-    "addu     %[temp11], %[temp11], %[temp4]           \n\t"
-    "addu     %[temp16], %[temp16], %[temp2]           \n\t"
-    "sra      %[temp18], %[temp5],  8                  \n\t"
-    "sra      %[temp1],  %[temp5],  31                 \n\t"
-    "sra      %[temp17], %[temp8],  8                  \n\t"
-    "sra      %[temp15], %[temp8],  31                 \n\t"
-    "sra      %[temp12], %[temp11], 8                  \n\t"
-    "sra      %[temp10], %[temp11], 31                 \n\t"
-    "sra      %[temp9],  %[temp16], 8                  \n\t"
-    "sra      %[temp3],  %[temp16], 31                 \n\t"
-    "beqz     %[temp18], 13f                           \n\t"
-    "xor      %[temp5],  %[temp5],  %[temp5]           \n\t"
-    "movz     %[temp5],  %[temp6],  %[temp1]           \n\t"
-  "13:                                                 \n\t"
-    "beqz     %[temp17], 14f                           \n\t"
-    "xor      %[temp8],  %[temp8],  %[temp8]           \n\t"
-    "movz     %[temp8],  %[temp6],  %[temp15]          \n\t"
-  "14:                                                 \n\t"
-    "beqz     %[temp12], 15f                           \n\t"
-    "xor      %[temp11], %[temp11], %[temp11]          \n\t"
-    "movz     %[temp11], %[temp6],  %[temp10]          \n\t"
-  "15:                                                 \n\t"
-    "beqz     %[temp9],  16f                           \n\t"
-    "xor      %[temp16], %[temp16], %[temp16]          \n\t"
-    "movz     %[temp16], %[temp6],  %[temp3]           \n\t"
-  "16:                                                 \n\t"
-    "sb       %[temp5],  96(%[dst])                    \n\t"
-    "sb       %[temp8],  97(%[dst])                    \n\t"
-    "sb       %[temp11], 98(%[dst])                    \n\t"
-    "sb       %[temp16], 99(%[dst])                    \n\t"
-
-    : [temp0]"=&r"(temp0), [temp1]"=&r"(temp1), [temp2]"=&r"(temp2),
-      [temp3]"=&r"(temp3), [temp4]"=&r"(temp4), [temp5]"=&r"(temp5),
-      [temp6]"=&r"(temp6), [temp7]"=&r"(temp7), [temp8]"=&r"(temp8),
-      [temp9]"=&r"(temp9), [temp10]"=&r"(temp10), [temp11]"=&r"(temp11),
-      [temp12]"=&r"(temp12), [temp13]"=&r"(temp13), [temp14]"=&r"(temp14),
-      [temp15]"=&r"(temp15), [temp16]"=&r"(temp16), [temp17]"=&r"(temp17),
-      [temp18]"=&r"(temp18)
-    : [in]"r"(p_in), [kC1]"r"(kC1), [kC2]"r"(kC2), [dst]"r"(dst)
-    : "memory", "hi", "lo"
-  );
-}
-
-static void TransformTwo(const int16_t* in, uint8_t* dst, int do_two) {
-  TransformOne(in, dst);
-  if (do_two) {
-    TransformOne(in + 16, dst + 4);
-  }
-}
-
-#endif  // WEBP_USE_MIPS32
-
-//------------------------------------------------------------------------------
-// Entry point
-
-extern void VP8DspInitMIPS32(void);
-
-void VP8DspInitMIPS32(void) {
-#if defined(WEBP_USE_MIPS32)
-  VP8InitClipTables();
-
-  VP8Transform = TransformTwo;
-
-  VP8VFilter16 = VFilter16;
-  VP8HFilter16 = HFilter16;
-  VP8VFilter8 = VFilter8;
-  VP8HFilter8 = HFilter8;
-  VP8VFilter16i = VFilter16i;
-  VP8HFilter16i = HFilter16i;
-  VP8VFilter8i = VFilter8i;
-  VP8HFilter8i = HFilter8i;
-
-  VP8SimpleVFilter16 = SimpleVFilter16;
-  VP8SimpleHFilter16 = SimpleHFilter16;
-  VP8SimpleVFilter16i = SimpleVFilter16i;
-  VP8SimpleHFilter16i = SimpleHFilter16i;
-#endif  // WEBP_USE_MIPS32
-}
@@ -1,978 +0,0 @@
-// Copyright 2011 Google Inc. All Rights Reserved.
-//
-// Use of this source code is governed by a BSD-style license
-// that can be found in the COPYING file in the root of the source
-// tree. An additional intellectual property rights grant can be found
-// in the file PATENTS. All contributing project authors may
-// be found in the AUTHORS file in the root of the source tree.
-// -----------------------------------------------------------------------------
-//
-// SSE2 version of some decoding functions (idct, loop filtering).
-//
-// Author: somnath@google.com (Somnath Banerjee)
-//         cduvivier@google.com (Christian Duvivier)
-
-#include "./dsp.h"
-
-#if defined(WEBP_USE_SSE2)
-
-// The 3-coeff sparse transform in SSE2 is not really faster than the plain-C
-// one it seems => disable it by default. Uncomment the following to enable:
-// #define USE_TRANSFORM_AC3
-
-#include <emmintrin.h>
-#include "../dec/vp8i.h"
-
-//------------------------------------------------------------------------------
-// Transforms (Paragraph 14.4)
-
-static void Transform(const int16_t* in, uint8_t* dst, int do_two) {
-  // This implementation makes use of 16-bit fixed point versions of two
-  // multiply constants:
-  //    K1 = sqrt(2) * cos (pi/8) ~= 85627 / 2^16
-  //    K2 = sqrt(2) * sin (pi/8) ~= 35468 / 2^16
-  //
-  // To be able to use signed 16-bit integers, we use the following trick to
-  // have constants within range:
-  // - Associated constants are obtained by subtracting the 16-bit fixed point
-  //   version of one:
-  //      k = K - (1 << 16)  =>  K = k + (1 << 16)
-  //      K1 = 85267  =>  k1 =  20091
-  //      K2 = 35468  =>  k2 = -30068
-  // - The multiplication of a variable by a constant become the sum of the
-  //   variable and the multiplication of that variable by the associated
-  //   constant:
-  //      (x * K) >> 16 = (x * (k + (1 << 16))) >> 16 = ((x * k ) >> 16) + x
-  const __m128i k1 = _mm_set1_epi16(20091);
-  const __m128i k2 = _mm_set1_epi16(-30068);
-  __m128i T0, T1, T2, T3;
-
-  // Load and concatenate the transform coefficients (we'll do two transforms
-  // in parallel). In the case of only one transform, the second half of the
-  // vectors will just contain random value we'll never use nor store.
-  __m128i in0, in1, in2, in3;
-  {
-    in0 = _mm_loadl_epi64((__m128i*)&in[0]);
-    in1 = _mm_loadl_epi64((__m128i*)&in[4]);
-    in2 = _mm_loadl_epi64((__m128i*)&in[8]);
-    in3 = _mm_loadl_epi64((__m128i*)&in[12]);
-    // a00 a10 a20 a30   x x x x
-    // a01 a11 a21 a31   x x x x
-    // a02 a12 a22 a32   x x x x
-    // a03 a13 a23 a33   x x x x
-    if (do_two) {
-      const __m128i inB0 = _mm_loadl_epi64((__m128i*)&in[16]);
-      const __m128i inB1 = _mm_loadl_epi64((__m128i*)&in[20]);
-      const __m128i inB2 = _mm_loadl_epi64((__m128i*)&in[24]);
-      const __m128i inB3 = _mm_loadl_epi64((__m128i*)&in[28]);
-      in0 = _mm_unpacklo_epi64(in0, inB0);
-      in1 = _mm_unpacklo_epi64(in1, inB1);
-      in2 = _mm_unpacklo_epi64(in2, inB2);
-      in3 = _mm_unpacklo_epi64(in3, inB3);
-      // a00 a10 a20 a30   b00 b10 b20 b30
-      // a01 a11 a21 a31   b01 b11 b21 b31
-      // a02 a12 a22 a32   b02 b12 b22 b32
-      // a03 a13 a23 a33   b03 b13 b23 b33
-    }
-  }
-
-  // Vertical pass and subsequent transpose.
-  {
-    // First pass, c and d calculations are longer because of the "trick"
-    // multiplications.
-    const __m128i a = _mm_add_epi16(in0, in2);
-    const __m128i b = _mm_sub_epi16(in0, in2);
-    // c = MUL(in1, K2) - MUL(in3, K1) = MUL(in1, k2) - MUL(in3, k1) + in1 - in3
-    const __m128i c1 = _mm_mulhi_epi16(in1, k2);
-    const __m128i c2 = _mm_mulhi_epi16(in3, k1);
-    const __m128i c3 = _mm_sub_epi16(in1, in3);
-    const __m128i c4 = _mm_sub_epi16(c1, c2);
-    const __m128i c = _mm_add_epi16(c3, c4);
-    // d = MUL(in1, K1) + MUL(in3, K2) = MUL(in1, k1) + MUL(in3, k2) + in1 + in3
-    const __m128i d1 = _mm_mulhi_epi16(in1, k1);
-    const __m128i d2 = _mm_mulhi_epi16(in3, k2);
-    const __m128i d3 = _mm_add_epi16(in1, in3);
-    const __m128i d4 = _mm_add_epi16(d1, d2);
-    const __m128i d = _mm_add_epi16(d3, d4);
-
-    // Second pass.
-    const __m128i tmp0 = _mm_add_epi16(a, d);
-    const __m128i tmp1 = _mm_add_epi16(b, c);
-    const __m128i tmp2 = _mm_sub_epi16(b, c);
-    const __m128i tmp3 = _mm_sub_epi16(a, d);
-
-    // Transpose the two 4x4.
-    // a00 a01 a02 a03   b00 b01 b02 b03
-    // a10 a11 a12 a13   b10 b11 b12 b13
-    // a20 a21 a22 a23   b20 b21 b22 b23
-    // a30 a31 a32 a33   b30 b31 b32 b33
-    const __m128i transpose0_0 = _mm_unpacklo_epi16(tmp0, tmp1);
-    const __m128i transpose0_1 = _mm_unpacklo_epi16(tmp2, tmp3);
-    const __m128i transpose0_2 = _mm_unpackhi_epi16(tmp0, tmp1);
-    const __m128i transpose0_3 = _mm_unpackhi_epi16(tmp2, tmp3);
-    // a00 a10 a01 a11   a02 a12 a03 a13
-    // a20 a30 a21 a31   a22 a32 a23 a33
-    // b00 b10 b01 b11   b02 b12 b03 b13
-    // b20 b30 b21 b31   b22 b32 b23 b33
-    const __m128i transpose1_0 = _mm_unpacklo_epi32(transpose0_0, transpose0_1);
-    const __m128i transpose1_1 = _mm_unpacklo_epi32(transpose0_2, transpose0_3);
-    const __m128i transpose1_2 = _mm_unpackhi_epi32(transpose0_0, transpose0_1);
-    const __m128i transpose1_3 = _mm_unpackhi_epi32(transpose0_2, transpose0_3);
-    // a00 a10 a20 a30 a01 a11 a21 a31
-    // b00 b10 b20 b30 b01 b11 b21 b31
-    // a02 a12 a22 a32 a03 a13 a23 a33
-    // b02 b12 a22 b32 b03 b13 b23 b33
-    T0 = _mm_unpacklo_epi64(transpose1_0, transpose1_1);
-    T1 = _mm_unpackhi_epi64(transpose1_0, transpose1_1);
-    T2 = _mm_unpacklo_epi64(transpose1_2, transpose1_3);
-    T3 = _mm_unpackhi_epi64(transpose1_2, transpose1_3);
-    // a00 a10 a20 a30   b00 b10 b20 b30
-    // a01 a11 a21 a31   b01 b11 b21 b31
-    // a02 a12 a22 a32   b02 b12 b22 b32
-    // a03 a13 a23 a33   b03 b13 b23 b33
-  }
-
-  // Horizontal pass and subsequent transpose.
-  {
-    // First pass, c and d calculations are longer because of the "trick"
-    // multiplications.
-    const __m128i four = _mm_set1_epi16(4);
-    const __m128i dc = _mm_add_epi16(T0, four);
-    const __m128i a =  _mm_add_epi16(dc, T2);
-    const __m128i b =  _mm_sub_epi16(dc, T2);
-    // c = MUL(T1, K2) - MUL(T3, K1) = MUL(T1, k2) - MUL(T3, k1) + T1 - T3
-    const __m128i c1 = _mm_mulhi_epi16(T1, k2);
-    const __m128i c2 = _mm_mulhi_epi16(T3, k1);
-    const __m128i c3 = _mm_sub_epi16(T1, T3);
-    const __m128i c4 = _mm_sub_epi16(c1, c2);
-    const __m128i c = _mm_add_epi16(c3, c4);
-    // d = MUL(T1, K1) + MUL(T3, K2) = MUL(T1, k1) + MUL(T3, k2) + T1 + T3
-    const __m128i d1 = _mm_mulhi_epi16(T1, k1);
-    const __m128i d2 = _mm_mulhi_epi16(T3, k2);
-    const __m128i d3 = _mm_add_epi16(T1, T3);
-    const __m128i d4 = _mm_add_epi16(d1, d2);
-    const __m128i d = _mm_add_epi16(d3, d4);
-
-    // Second pass.
-    const __m128i tmp0 = _mm_add_epi16(a, d);
-    const __m128i tmp1 = _mm_add_epi16(b, c);
-    const __m128i tmp2 = _mm_sub_epi16(b, c);
-    const __m128i tmp3 = _mm_sub_epi16(a, d);
-    const __m128i shifted0 = _mm_srai_epi16(tmp0, 3);
-    const __m128i shifted1 = _mm_srai_epi16(tmp1, 3);
-    const __m128i shifted2 = _mm_srai_epi16(tmp2, 3);
-    const __m128i shifted3 = _mm_srai_epi16(tmp3, 3);
-
-    // Transpose the two 4x4.
-    // a00 a01 a02 a03   b00 b01 b02 b03
-    // a10 a11 a12 a13   b10 b11 b12 b13
-    // a20 a21 a22 a23   b20 b21 b22 b23
-    // a30 a31 a32 a33   b30 b31 b32 b33
-    const __m128i transpose0_0 = _mm_unpacklo_epi16(shifted0, shifted1);
-    const __m128i transpose0_1 = _mm_unpacklo_epi16(shifted2, shifted3);
-    const __m128i transpose0_2 = _mm_unpackhi_epi16(shifted0, shifted1);
-    const __m128i transpose0_3 = _mm_unpackhi_epi16(shifted2, shifted3);
-    // a00 a10 a01 a11   a02 a12 a03 a13
-    // a20 a30 a21 a31   a22 a32 a23 a33
-    // b00 b10 b01 b11   b02 b12 b03 b13
-    // b20 b30 b21 b31   b22 b32 b23 b33
-    const __m128i transpose1_0 = _mm_unpacklo_epi32(transpose0_0, transpose0_1);
-    const __m128i transpose1_1 = _mm_unpacklo_epi32(transpose0_2, transpose0_3);
-    const __m128i transpose1_2 = _mm_unpackhi_epi32(transpose0_0, transpose0_1);
-    const __m128i transpose1_3 = _mm_unpackhi_epi32(transpose0_2, transpose0_3);
-    // a00 a10 a20 a30 a01 a11 a21 a31
-    // b00 b10 b20 b30 b01 b11 b21 b31
-    // a02 a12 a22 a32 a03 a13 a23 a33
-    // b02 b12 a22 b32 b03 b13 b23 b33
-    T0 = _mm_unpacklo_epi64(transpose1_0, transpose1_1);
-    T1 = _mm_unpackhi_epi64(transpose1_0, transpose1_1);
-    T2 = _mm_unpacklo_epi64(transpose1_2, transpose1_3);
-    T3 = _mm_unpackhi_epi64(transpose1_2, transpose1_3);
-    // a00 a10 a20 a30   b00 b10 b20 b30
-    // a01 a11 a21 a31   b01 b11 b21 b31
-    // a02 a12 a22 a32   b02 b12 b22 b32
-    // a03 a13 a23 a33   b03 b13 b23 b33
-  }
-
-  // Add inverse transform to 'dst' and store.
-  {
-    const __m128i zero = _mm_setzero_si128();
-    // Load the reference(s).
-    __m128i dst0, dst1, dst2, dst3;
-    if (do_two) {
-      // Load eight bytes/pixels per line.
-      dst0 = _mm_loadl_epi64((__m128i*)(dst + 0 * BPS));
-      dst1 = _mm_loadl_epi64((__m128i*)(dst + 1 * BPS));
-      dst2 = _mm_loadl_epi64((__m128i*)(dst + 2 * BPS));
-      dst3 = _mm_loadl_epi64((__m128i*)(dst + 3 * BPS));
-    } else {
-      // Load four bytes/pixels per line.
-      dst0 = _mm_cvtsi32_si128(*(int*)(dst + 0 * BPS));
-      dst1 = _mm_cvtsi32_si128(*(int*)(dst + 1 * BPS));
-      dst2 = _mm_cvtsi32_si128(*(int*)(dst + 2 * BPS));
-      dst3 = _mm_cvtsi32_si128(*(int*)(dst + 3 * BPS));
-    }
-    // Convert to 16b.
-    dst0 = _mm_unpacklo_epi8(dst0, zero);
-    dst1 = _mm_unpacklo_epi8(dst1, zero);
-    dst2 = _mm_unpacklo_epi8(dst2, zero);
-    dst3 = _mm_unpacklo_epi8(dst3, zero);
-    // Add the inverse transform(s).
-    dst0 = _mm_add_epi16(dst0, T0);
-    dst1 = _mm_add_epi16(dst1, T1);
-    dst2 = _mm_add_epi16(dst2, T2);
-    dst3 = _mm_add_epi16(dst3, T3);
-    // Unsigned saturate to 8b.
-    dst0 = _mm_packus_epi16(dst0, dst0);
-    dst1 = _mm_packus_epi16(dst1, dst1);
-    dst2 = _mm_packus_epi16(dst2, dst2);
-    dst3 = _mm_packus_epi16(dst3, dst3);
-    // Store the results.
-    if (do_two) {
-      // Store eight bytes/pixels per line.
-      _mm_storel_epi64((__m128i*)(dst + 0 * BPS), dst0);
-      _mm_storel_epi64((__m128i*)(dst + 1 * BPS), dst1);
-      _mm_storel_epi64((__m128i*)(dst + 2 * BPS), dst2);
-      _mm_storel_epi64((__m128i*)(dst + 3 * BPS), dst3);
-    } else {
-      // Store four bytes/pixels per line.
-      *(int*)(dst + 0 * BPS) = _mm_cvtsi128_si32(dst0);
-      *(int*)(dst + 1 * BPS) = _mm_cvtsi128_si32(dst1);
-      *(int*)(dst + 2 * BPS) = _mm_cvtsi128_si32(dst2);
-      *(int*)(dst + 3 * BPS) = _mm_cvtsi128_si32(dst3);
-    }
-  }
-}
-
-#if defined(USE_TRANSFORM_AC3)
-#define MUL(a, b) (((a) * (b)) >> 16)
-static void TransformAC3(const int16_t* in, uint8_t* dst) {
-  static const int kC1 = 20091 + (1 << 16);
-  static const int kC2 = 35468;
-  const __m128i A = _mm_set1_epi16(in[0] + 4);
-  const __m128i c4 = _mm_set1_epi16(MUL(in[4], kC2));
-  const __m128i d4 = _mm_set1_epi16(MUL(in[4], kC1));
-  const int c1 = MUL(in[1], kC2);
-  const int d1 = MUL(in[1], kC1);
-  const __m128i CD = _mm_set_epi16(0, 0, 0, 0, -d1, -c1, c1, d1);
-  const __m128i B = _mm_adds_epi16(A, CD);
-  const __m128i m0 = _mm_adds_epi16(B, d4);
-  const __m128i m1 = _mm_adds_epi16(B, c4);
-  const __m128i m2 = _mm_subs_epi16(B, c4);
-  const __m128i m3 = _mm_subs_epi16(B, d4);
-  const __m128i zero = _mm_setzero_si128();
-  // Load the source pixels.
-  __m128i dst0 = _mm_cvtsi32_si128(*(int*)(dst + 0 * BPS));
-  __m128i dst1 = _mm_cvtsi32_si128(*(int*)(dst + 1 * BPS));
-  __m128i dst2 = _mm_cvtsi32_si128(*(int*)(dst + 2 * BPS));
-  __m128i dst3 = _mm_cvtsi32_si128(*(int*)(dst + 3 * BPS));
-  // Convert to 16b.
-  dst0 = _mm_unpacklo_epi8(dst0, zero);
-  dst1 = _mm_unpacklo_epi8(dst1, zero);
-  dst2 = _mm_unpacklo_epi8(dst2, zero);
-  dst3 = _mm_unpacklo_epi8(dst3, zero);
-  // Add the inverse transform.
-  dst0 = _mm_adds_epi16(dst0, _mm_srai_epi16(m0, 3));
-  dst1 = _mm_adds_epi16(dst1, _mm_srai_epi16(m1, 3));
-  dst2 = _mm_adds_epi16(dst2, _mm_srai_epi16(m2, 3));
-  dst3 = _mm_adds_epi16(dst3, _mm_srai_epi16(m3, 3));
-  // Unsigned saturate to 8b.
-  dst0 = _mm_packus_epi16(dst0, dst0);
-  dst1 = _mm_packus_epi16(dst1, dst1);
-  dst2 = _mm_packus_epi16(dst2, dst2);
-  dst3 = _mm_packus_epi16(dst3, dst3);
-  // Store the results.
-  *(int*)(dst + 0 * BPS) = _mm_cvtsi128_si32(dst0);
-  *(int*)(dst + 1 * BPS) = _mm_cvtsi128_si32(dst1);
-  *(int*)(dst + 2 * BPS) = _mm_cvtsi128_si32(dst2);
-  *(int*)(dst + 3 * BPS) = _mm_cvtsi128_si32(dst3);
-}
-#undef MUL
-#endif   // USE_TRANSFORM_AC3
-
-//------------------------------------------------------------------------------
-// Loop Filter (Paragraph 15)
-
-// Compute abs(p - q) = subs(p - q) OR subs(q - p)
-#define MM_ABS(p, q)  _mm_or_si128(                                            \
-    _mm_subs_epu8((q), (p)),                                                   \
-    _mm_subs_epu8((p), (q)))
-
-// Shift each byte of "x" by 3 bits while preserving by the sign bit.
-static WEBP_INLINE void SignedShift8b(__m128i* const x) {
-  const __m128i zero = _mm_setzero_si128();
-  const __m128i signs = _mm_cmpgt_epi8(zero, *x);
-  const __m128i lo_0 = _mm_unpacklo_epi8(*x, signs);  // s8 -> s16 sign extend
-  const __m128i hi_0 = _mm_unpackhi_epi8(*x, signs);
-  const __m128i lo_1 = _mm_srai_epi16(lo_0, 3);
-  const __m128i hi_1 = _mm_srai_epi16(hi_0, 3);
-  *x = _mm_packs_epi16(lo_1, hi_1);
-}
-
-#define FLIP_SIGN_BIT2(a, b) {                                                 \
-  a = _mm_xor_si128(a, sign_bit);                                              \
-  b = _mm_xor_si128(b, sign_bit);                                              \
-}
-
-#define FLIP_SIGN_BIT4(a, b, c, d) {                                           \
-  FLIP_SIGN_BIT2(a, b);                                                        \
-  FLIP_SIGN_BIT2(c, d);                                                        \
-}
-
-// input/output is uint8_t
-static WEBP_INLINE void GetNotHEV(const __m128i* const p1,
-                                  const __m128i* const p0,
-                                  const __m128i* const q0,
-                                  const __m128i* const q1,
-                                  int hev_thresh, __m128i* const not_hev) {
-  const __m128i zero = _mm_setzero_si128();
-  const __m128i t_1 = MM_ABS(*p1, *p0);
-  const __m128i t_2 = MM_ABS(*q1, *q0);
-
-  const __m128i h = _mm_set1_epi8(hev_thresh);
-  const __m128i t_3 = _mm_subs_epu8(t_1, h);  // abs(p1 - p0) - hev_tresh
-  const __m128i t_4 = _mm_subs_epu8(t_2, h);  // abs(q1 - q0) - hev_tresh
-
-  *not_hev = _mm_or_si128(t_3, t_4);
-  *not_hev = _mm_cmpeq_epi8(*not_hev, zero);  // not_hev <= t1 && not_hev <= t2
-}
-
-// input pixels are int8_t
-static WEBP_INLINE void GetBaseDelta(const __m128i* const p1,
-                                     const __m128i* const p0,
-                                     const __m128i* const q0,
-                                     const __m128i* const q1,
-                                     __m128i* const delta) {
-  // beware of addition order, for saturation!
-  const __m128i p1_q1 = _mm_subs_epi8(*p1, *q1);   // p1 - q1
-  const __m128i q0_p0 = _mm_subs_epi8(*q0, *p0);   // q0 - p0
-  const __m128i s1 = _mm_adds_epi8(p1_q1, q0_p0);  // p1 - q1 + 1 * (q0 - p0)
-  const __m128i s2 = _mm_adds_epi8(q0_p0, s1);     // p1 - q1 + 2 * (q0 - p0)
-  const __m128i s3 = _mm_adds_epi8(q0_p0, s2);     // p1 - q1 + 3 * (q0 - p0)
-  *delta = s3;
-}
-
-// input and output are int8_t
-static WEBP_INLINE void DoSimpleFilter(__m128i* const p0, __m128i* const q0,
-                                       const __m128i* const fl) {
-  const __m128i k3 = _mm_set1_epi8(3);
-  const __m128i k4 = _mm_set1_epi8(4);
-  __m128i v3 = _mm_adds_epi8(*fl, k3);
-  __m128i v4 = _mm_adds_epi8(*fl, k4);
-
-  SignedShift8b(&v4);                  // v4 >> 3
-  SignedShift8b(&v3);                  // v3 >> 3
-  *q0 = _mm_subs_epi8(*q0, v4);        // q0 -= v4
-  *p0 = _mm_adds_epi8(*p0, v3);        // p0 += v3
-}
-
-// Updates values of 2 pixels at MB edge during complex filtering.
-// Update operations:
-// q = q - delta and p = p + delta; where delta = [(a_hi >> 7), (a_lo >> 7)]
-// Pixels 'pi' and 'qi' are int8_t on input, uint8_t on output (sign flip).
-static WEBP_INLINE void Update2Pixels(__m128i* const pi, __m128i* const qi,
-                                      const __m128i* const a0_lo,
-                                      const __m128i* const a0_hi) {
-  const __m128i a1_lo = _mm_srai_epi16(*a0_lo, 7);
-  const __m128i a1_hi = _mm_srai_epi16(*a0_hi, 7);
-  const __m128i delta = _mm_packs_epi16(a1_lo, a1_hi);
-  const __m128i sign_bit = _mm_set1_epi8(0x80);
-  *pi = _mm_adds_epi8(*pi, delta);
-  *qi = _mm_subs_epi8(*qi, delta);
-  FLIP_SIGN_BIT2(*pi, *qi);
-}
-
-// input pixels are uint8_t
-static WEBP_INLINE void NeedsFilter(const __m128i* const p1,
-                                    const __m128i* const p0,
-                                    const __m128i* const q0,
-                                    const __m128i* const q1,
-                                    int thresh, __m128i* const mask) {
-  const __m128i m_thresh = _mm_set1_epi8(thresh);
-  const __m128i t1 = MM_ABS(*p1, *q1);        // abs(p1 - q1)
-  const __m128i kFE = _mm_set1_epi8(0xFE);
-  const __m128i t2 = _mm_and_si128(t1, kFE);  // set lsb of each byte to zero
-  const __m128i t3 = _mm_srli_epi16(t2, 1);   // abs(p1 - q1) / 2
-
-  const __m128i t4 = MM_ABS(*p0, *q0);        // abs(p0 - q0)
-  const __m128i t5 = _mm_adds_epu8(t4, t4);   // abs(p0 - q0) * 2
-  const __m128i t6 = _mm_adds_epu8(t5, t3);   // abs(p0-q0)*2 + abs(p1-q1)/2
-
-  const __m128i t7 = _mm_subs_epu8(t6, m_thresh);  // mask <= m_thresh
-  *mask = _mm_cmpeq_epi8(t7, _mm_setzero_si128());
-}
-
-//------------------------------------------------------------------------------
-// Edge filtering functions
-
-// Applies filter on 2 pixels (p0 and q0)
-static WEBP_INLINE void DoFilter2(__m128i* const p1, __m128i* const p0,
-                                  __m128i* const q0, __m128i* const q1,
-                                  int thresh) {
-  __m128i a, mask;
-  const __m128i sign_bit = _mm_set1_epi8(0x80);
-  // convert p1/q1 to int8_t (for GetBaseDelta)
-  const __m128i p1s = _mm_xor_si128(*p1, sign_bit);
-  const __m128i q1s = _mm_xor_si128(*q1, sign_bit);
-
-  NeedsFilter(p1, p0, q0, q1, thresh, &mask);
-
-  FLIP_SIGN_BIT2(*p0, *q0);
-  GetBaseDelta(&p1s, p0, q0, &q1s, &a);
-  a = _mm_and_si128(a, mask);     // mask filter values we don't care about
-  DoSimpleFilter(p0, q0, &a);
-  FLIP_SIGN_BIT2(*p0, *q0);
-}
-
-// Applies filter on 4 pixels (p1, p0, q0 and q1)
-static WEBP_INLINE void DoFilter4(__m128i* const p1, __m128i* const p0,
-                                  __m128i* const q0, __m128i* const q1,
-                                  const __m128i* const mask, int hev_thresh) {
-  const __m128i sign_bit = _mm_set1_epi8(0x80);
-  const __m128i k64 = _mm_set1_epi8(0x40);
-  const __m128i zero = _mm_setzero_si128();
-  __m128i not_hev;
-  __m128i t1, t2, t3;
-
-  // compute hev mask
-  GetNotHEV(p1, p0, q0, q1, hev_thresh, &not_hev);
-
-  // convert to signed values
-  FLIP_SIGN_BIT4(*p1, *p0, *q0, *q1);
-
-  t1 = _mm_subs_epi8(*p1, *q1);        // p1 - q1
-  t1 = _mm_andnot_si128(not_hev, t1);  // hev(p1 - q1)
-  t2 = _mm_subs_epi8(*q0, *p0);        // q0 - p0
-  t1 = _mm_adds_epi8(t1, t2);          // hev(p1 - q1) + 1 * (q0 - p0)
-  t1 = _mm_adds_epi8(t1, t2);          // hev(p1 - q1) + 2 * (q0 - p0)
-  t1 = _mm_adds_epi8(t1, t2);          // hev(p1 - q1) + 3 * (q0 - p0)
-  t1 = _mm_and_si128(t1, *mask);       // mask filter values we don't care about
-
-  t2 = _mm_set1_epi8(3);
-  t3 = _mm_set1_epi8(4);
-  t2 = _mm_adds_epi8(t1, t2);        // 3 * (q0 - p0) + (p1 - q1) + 3
-  t3 = _mm_adds_epi8(t1, t3);        // 3 * (q0 - p0) + (p1 - q1) + 4
-  SignedShift8b(&t2);                // (3 * (q0 - p0) + hev(p1 - q1) + 3) >> 3
-  SignedShift8b(&t3);                // (3 * (q0 - p0) + hev(p1 - q1) + 4) >> 3
-  *p0 = _mm_adds_epi8(*p0, t2);      // p0 += t2
-  *q0 = _mm_subs_epi8(*q0, t3);      // q0 -= t3
-  FLIP_SIGN_BIT2(*p0, *q0);
-
-  // this is equivalent to signed (a + 1) >> 1 calculation
-  t2 = _mm_add_epi8(t3, sign_bit);
-  t3 = _mm_avg_epu8(t2, zero);
-  t3 = _mm_sub_epi8(t3, k64);
-
-  t3 = _mm_and_si128(not_hev, t3);   // if !hev
-  *q1 = _mm_subs_epi8(*q1, t3);      // q1 -= t3
-  *p1 = _mm_adds_epi8(*p1, t3);      // p1 += t3
-  FLIP_SIGN_BIT2(*p1, *q1);
-}
-
-// Applies filter on 6 pixels (p2, p1, p0, q0, q1 and q2)
-static WEBP_INLINE void DoFilter6(__m128i* const p2, __m128i* const p1,
-                                  __m128i* const p0, __m128i* const q0,
-                                  __m128i* const q1, __m128i* const q2,
-                                  const __m128i* const mask, int hev_thresh) {
-  const __m128i zero = _mm_setzero_si128();
-  const __m128i sign_bit = _mm_set1_epi8(0x80);
-  __m128i a, not_hev;
-
-  // compute hev mask
-  GetNotHEV(p1, p0, q0, q1, hev_thresh, &not_hev);
-
-  FLIP_SIGN_BIT4(*p1, *p0, *q0, *q1);
-  FLIP_SIGN_BIT2(*p2, *q2);
-  GetBaseDelta(p1, p0, q0, q1, &a);
-
-  { // do simple filter on pixels with hev
-    const __m128i m = _mm_andnot_si128(not_hev, *mask);
-    const __m128i f = _mm_and_si128(a, m);
-    DoSimpleFilter(p0, q0, &f);
-  }
-
-  { // do strong filter on pixels with not hev
-    const __m128i k9 = _mm_set1_epi16(0x0900);
-    const __m128i k63 = _mm_set1_epi16(63);
-
-    const __m128i m = _mm_and_si128(not_hev, *mask);
-    const __m128i f = _mm_and_si128(a, m);
-
-    const __m128i f_lo = _mm_unpacklo_epi8(zero, f);
-    const __m128i f_hi = _mm_unpackhi_epi8(zero, f);
-
-    const __m128i f9_lo = _mm_mulhi_epi16(f_lo, k9);    // Filter (lo) * 9
-    const __m128i f9_hi = _mm_mulhi_epi16(f_hi, k9);    // Filter (hi) * 9
-
-    const __m128i a2_lo = _mm_add_epi16(f9_lo, k63);    // Filter * 9 + 63
-    const __m128i a2_hi = _mm_add_epi16(f9_hi, k63);    // Filter * 9 + 63
-
-    const __m128i a1_lo = _mm_add_epi16(a2_lo, f9_lo);  // Filter * 18 + 63
-    const __m128i a1_hi = _mm_add_epi16(a2_hi, f9_hi);  // Filter * 18 + 63
-
-    const __m128i a0_lo = _mm_add_epi16(a1_lo, f9_lo);  // Filter * 27 + 63
-    const __m128i a0_hi = _mm_add_epi16(a1_hi, f9_hi);  // Filter * 27 + 63
-
-    Update2Pixels(p2, q2, &a2_lo, &a2_hi);
-    Update2Pixels(p1, q1, &a1_lo, &a1_hi);
-    Update2Pixels(p0, q0, &a0_lo, &a0_hi);
-  }
-}
-
-// reads 8 rows across a vertical edge.
-//
-// TODO(somnath): Investigate _mm_shuffle* also see if it can be broken into
-// two Load4x4() to avoid code duplication.
-static WEBP_INLINE void Load8x4(const uint8_t* const b, int stride,
-                                __m128i* const p, __m128i* const q) {
-  __m128i t1, t2;
-
-  // Load 0th, 1st, 4th and 5th rows
-  __m128i r0 =  _mm_cvtsi32_si128(*((int*)&b[0 * stride]));  // 03 02 01 00
-  __m128i r1 =  _mm_cvtsi32_si128(*((int*)&b[1 * stride]));  // 13 12 11 10
-  __m128i r4 =  _mm_cvtsi32_si128(*((int*)&b[4 * stride]));  // 43 42 41 40
-  __m128i r5 =  _mm_cvtsi32_si128(*((int*)&b[5 * stride]));  // 53 52 51 50
-
-  r0 = _mm_unpacklo_epi32(r0, r4);               // 43 42 41 40 03 02 01 00
-  r1 = _mm_unpacklo_epi32(r1, r5);               // 53 52 51 50 13 12 11 10
-
-  // t1 = 53 43 52 42 51 41 50 40 13 03 12 02 11 01 10 00
-  t1 = _mm_unpacklo_epi8(r0, r1);
-
-  // Load 2nd, 3rd, 6th and 7th rows
-  r0 =  _mm_cvtsi32_si128(*((int*)&b[2 * stride]));          // 23 22 21 22
-  r1 =  _mm_cvtsi32_si128(*((int*)&b[3 * stride]));          // 33 32 31 30
-  r4 =  _mm_cvtsi32_si128(*((int*)&b[6 * stride]));          // 63 62 61 60
-  r5 =  _mm_cvtsi32_si128(*((int*)&b[7 * stride]));          // 73 72 71 70
-
-  r0 = _mm_unpacklo_epi32(r0, r4);               // 63 62 61 60 23 22 21 20
-  r1 = _mm_unpacklo_epi32(r1, r5);               // 73 72 71 70 33 32 31 30
-
-  // t2 = 73 63 72 62 71 61 70 60 33 23 32 22 31 21 30 20
-  t2 = _mm_unpacklo_epi8(r0, r1);
-
-  // t1 = 33 23 13 03 32 22 12 02 31 21 11 01 30 20 10 00
-  // t2 = 73 63 53 43 72 62 52 42 71 61 51 41 70 60 50 40
-  r0 = t1;
-  t1 = _mm_unpacklo_epi16(t1, t2);
-  t2 = _mm_unpackhi_epi16(r0, t2);
-
-  // *p = 71 61 51 41 31 21 11 01 70 60 50 40 30 20 10 00
-  // *q = 73 63 53 43 33 23 13 03 72 62 52 42 32 22 12 02
-  *p = _mm_unpacklo_epi32(t1, t2);
-  *q = _mm_unpackhi_epi32(t1, t2);
-}
-
-static WEBP_INLINE void Load16x4(const uint8_t* const r0,
-                                 const uint8_t* const r8,
-                                 int stride,
-                                 __m128i* const p1, __m128i* const p0,
-                                 __m128i* const q0, __m128i* const q1) {
-  __m128i t1, t2;
-  // Assume the pixels around the edge (|) are numbered as follows
-  //                00 01 | 02 03
-  //                10 11 | 12 13
-  //                 ...  |  ...
-  //                e0 e1 | e2 e3
-  //                f0 f1 | f2 f3
-  //
-  // r0 is pointing to the 0th row (00)
-  // r8 is pointing to the 8th row (80)
-
-  // Load
-  // p1 = 71 61 51 41 31 21 11 01 70 60 50 40 30 20 10 00
-  // q0 = 73 63 53 43 33 23 13 03 72 62 52 42 32 22 12 02
-  // p0 = f1 e1 d1 c1 b1 a1 91 81 f0 e0 d0 c0 b0 a0 90 80
-  // q1 = f3 e3 d3 c3 b3 a3 93 83 f2 e2 d2 c2 b2 a2 92 82
-  Load8x4(r0, stride, p1, q0);
-  Load8x4(r8, stride, p0, q1);
-
-  t1 = *p1;
-  t2 = *q0;
-  // p1 = f0 e0 d0 c0 b0 a0 90 80 70 60 50 40 30 20 10 00
-  // p0 = f1 e1 d1 c1 b1 a1 91 81 71 61 51 41 31 21 11 01
-  // q0 = f2 e2 d2 c2 b2 a2 92 82 72 62 52 42 32 22 12 02
-  // q1 = f3 e3 d3 c3 b3 a3 93 83 73 63 53 43 33 23 13 03
-  *p1 = _mm_unpacklo_epi64(t1, *p0);
-  *p0 = _mm_unpackhi_epi64(t1, *p0);
-  *q0 = _mm_unpacklo_epi64(t2, *q1);
-  *q1 = _mm_unpackhi_epi64(t2, *q1);
-}
-
-static WEBP_INLINE void Store4x4(__m128i* const x, uint8_t* dst, int stride) {
-  int i;
-  for (i = 0; i < 4; ++i, dst += stride) {
-    *((int32_t*)dst) = _mm_cvtsi128_si32(*x);
-    *x = _mm_srli_si128(*x, 4);
-  }
-}
-
-// Transpose back and store
-static WEBP_INLINE void Store16x4(const __m128i* const p1,
-                                  const __m128i* const p0,
-                                  const __m128i* const q0,
-                                  const __m128i* const q1,
-                                  uint8_t* r0, uint8_t* r8,
-                                  int stride) {
-  __m128i t1, p1_s, p0_s, q0_s, q1_s;
-
-  // p0 = 71 70 61 60 51 50 41 40 31 30 21 20 11 10 01 00
-  // p1 = f1 f0 e1 e0 d1 d0 c1 c0 b1 b0 a1 a0 91 90 81 80
-  t1 = *p0;
-  p0_s = _mm_unpacklo_epi8(*p1, t1);
-  p1_s = _mm_unpackhi_epi8(*p1, t1);
-
-  // q0 = 73 72 63 62 53 52 43 42 33 32 23 22 13 12 03 02
-  // q1 = f3 f2 e3 e2 d3 d2 c3 c2 b3 b2 a3 a2 93 92 83 82
-  t1 = *q0;
-  q0_s = _mm_unpacklo_epi8(t1, *q1);
-  q1_s = _mm_unpackhi_epi8(t1, *q1);
-
-  // p0 = 33 32 31 30 23 22 21 20 13 12 11 10 03 02 01 00
-  // q0 = 73 72 71 70 63 62 61 60 53 52 51 50 43 42 41 40
-  t1 = p0_s;
-  p0_s = _mm_unpacklo_epi16(t1, q0_s);
-  q0_s = _mm_unpackhi_epi16(t1, q0_s);
-
-  // p1 = b3 b2 b1 b0 a3 a2 a1 a0 93 92 91 90 83 82 81 80
-  // q1 = f3 f2 f1 f0 e3 e2 e1 e0 d3 d2 d1 d0 c3 c2 c1 c0
-  t1 = p1_s;
-  p1_s = _mm_unpacklo_epi16(t1, q1_s);
-  q1_s = _mm_unpackhi_epi16(t1, q1_s);
-
-  Store4x4(&p0_s, r0, stride);
-  r0 += 4 * stride;
-  Store4x4(&q0_s, r0, stride);
-
-  Store4x4(&p1_s, r8, stride);
-  r8 += 4 * stride;
-  Store4x4(&q1_s, r8, stride);
-}
-
-//------------------------------------------------------------------------------
-// Simple In-loop filtering (Paragraph 15.2)
-
-static void SimpleVFilter16(uint8_t* p, int stride, int thresh) {
-  // Load
-  __m128i p1 = _mm_loadu_si128((__m128i*)&p[-2 * stride]);
-  __m128i p0 = _mm_loadu_si128((__m128i*)&p[-stride]);
-  __m128i q0 = _mm_loadu_si128((__m128i*)&p[0]);
-  __m128i q1 = _mm_loadu_si128((__m128i*)&p[stride]);
-
-  DoFilter2(&p1, &p0, &q0, &q1, thresh);
-
-  // Store
-  _mm_storeu_si128((__m128i*)&p[-stride], p0);
-  _mm_storeu_si128((__m128i*)&p[0], q0);
-}
-
-static void SimpleHFilter16(uint8_t* p, int stride, int thresh) {
-  __m128i p1, p0, q0, q1;
-
-  p -= 2;  // beginning of p1
-
-  Load16x4(p, p + 8 * stride, stride, &p1, &p0, &q0, &q1);
-  DoFilter2(&p1, &p0, &q0, &q1, thresh);
-  Store16x4(&p1, &p0, &q0, &q1, p, p + 8 * stride, stride);
-}
-
-static void SimpleVFilter16i(uint8_t* p, int stride, int thresh) {
-  int k;
-  for (k = 3; k > 0; --k) {
-    p += 4 * stride;
-    SimpleVFilter16(p, stride, thresh);
-  }
-}
-
-static void SimpleHFilter16i(uint8_t* p, int stride, int thresh) {
-  int k;
-  for (k = 3; k > 0; --k) {
-    p += 4;
-    SimpleHFilter16(p, stride, thresh);
-  }
-}
-
-//------------------------------------------------------------------------------
-// Complex In-loop filtering (Paragraph 15.3)
-
-#define MAX_DIFF1(p3, p2, p1, p0, m) do {                                      \
-  m = MM_ABS(p1, p0);                                                          \
-  m = _mm_max_epu8(m, MM_ABS(p3, p2));                                         \
-  m = _mm_max_epu8(m, MM_ABS(p2, p1));                                         \
-} while (0)
-
-#define MAX_DIFF2(p3, p2, p1, p0, m) do {                                      \
-  m = _mm_max_epu8(m, MM_ABS(p1, p0));                                         \
-  m = _mm_max_epu8(m, MM_ABS(p3, p2));                                         \
-  m = _mm_max_epu8(m, MM_ABS(p2, p1));                                         \
-} while (0)
-
-#define LOAD_H_EDGES4(p, stride, e1, e2, e3, e4) {                             \
-  e1 = _mm_loadu_si128((__m128i*)&(p)[0 * stride]);                            \
-  e2 = _mm_loadu_si128((__m128i*)&(p)[1 * stride]);                            \
-  e3 = _mm_loadu_si128((__m128i*)&(p)[2 * stride]);                            \
-  e4 = _mm_loadu_si128((__m128i*)&(p)[3 * stride]);                            \
-}
-
-#define LOADUV_H_EDGE(p, u, v, stride) do {                                    \
-  const __m128i U = _mm_loadl_epi64((__m128i*)&(u)[(stride)]);                 \
-  const __m128i V = _mm_loadl_epi64((__m128i*)&(v)[(stride)]);                 \
-  p = _mm_unpacklo_epi64(U, V);                                                \
-} while (0)
-
-#define LOADUV_H_EDGES4(u, v, stride, e1, e2, e3, e4) {                        \
-  LOADUV_H_EDGE(e1, u, v, 0 * stride);                                         \
-  LOADUV_H_EDGE(e2, u, v, 1 * stride);                                         \
-  LOADUV_H_EDGE(e3, u, v, 2 * stride);                                         \
-  LOADUV_H_EDGE(e4, u, v, 3 * stride);                                         \
-}
-
-#define STOREUV(p, u, v, stride) {                                             \
-  _mm_storel_epi64((__m128i*)&u[(stride)], p);                                 \
-  p = _mm_srli_si128(p, 8);                                                    \
-  _mm_storel_epi64((__m128i*)&v[(stride)], p);                                 \
-}
-
-static WEBP_INLINE void ComplexMask(const __m128i* const p1,
-                                    const __m128i* const p0,
-                                    const __m128i* const q0,
-                                    const __m128i* const q1,
-                                    int thresh, int ithresh,
-                                    __m128i* const mask) {
-  const __m128i it = _mm_set1_epi8(ithresh);
-  const __m128i diff = _mm_subs_epu8(*mask, it);
-  const __m128i thresh_mask = _mm_cmpeq_epi8(diff, _mm_setzero_si128());
-  __m128i filter_mask;
-  NeedsFilter(p1, p0, q0, q1, thresh, &filter_mask);
-  *mask = _mm_and_si128(thresh_mask, filter_mask);
-}
-
-// on macroblock edges
-static void VFilter16(uint8_t* p, int stride,
-                      int thresh, int ithresh, int hev_thresh) {
-  __m128i t1;
-  __m128i mask;
-  __m128i p2, p1, p0, q0, q1, q2;
-
-  // Load p3, p2, p1, p0
-  LOAD_H_EDGES4(p - 4 * stride, stride, t1, p2, p1, p0);
-  MAX_DIFF1(t1, p2, p1, p0, mask);
-
-  // Load q0, q1, q2, q3
-  LOAD_H_EDGES4(p, stride, q0, q1, q2, t1);
-  MAX_DIFF2(t1, q2, q1, q0, mask);
-
-  ComplexMask(&p1, &p0, &q0, &q1, thresh, ithresh, &mask);
-  DoFilter6(&p2, &p1, &p0, &q0, &q1, &q2, &mask, hev_thresh);
-
-  // Store
-  _mm_storeu_si128((__m128i*)&p[-3 * stride], p2);
-  _mm_storeu_si128((__m128i*)&p[-2 * stride], p1);
-  _mm_storeu_si128((__m128i*)&p[-1 * stride], p0);
-  _mm_storeu_si128((__m128i*)&p[+0 * stride], q0);
-  _mm_storeu_si128((__m128i*)&p[+1 * stride], q1);
-  _mm_storeu_si128((__m128i*)&p[+2 * stride], q2);
-}
-
-static void HFilter16(uint8_t* p, int stride,
-                      int thresh, int ithresh, int hev_thresh) {
-  __m128i mask;
-  __m128i p3, p2, p1, p0, q0, q1, q2, q3;
-
-  uint8_t* const b = p - 4;
-  Load16x4(b, b + 8 * stride, stride, &p3, &p2, &p1, &p0);  // p3, p2, p1, p0
-  MAX_DIFF1(p3, p2, p1, p0, mask);
-
-  Load16x4(p, p + 8 * stride, stride, &q0, &q1, &q2, &q3);  // q0, q1, q2, q3
-  MAX_DIFF2(q3, q2, q1, q0, mask);
-
-  ComplexMask(&p1, &p0, &q0, &q1, thresh, ithresh, &mask);
-  DoFilter6(&p2, &p1, &p0, &q0, &q1, &q2, &mask, hev_thresh);
-
-  Store16x4(&p3, &p2, &p1, &p0, b, b + 8 * stride, stride);
-  Store16x4(&q0, &q1, &q2, &q3, p, p + 8 * stride, stride);
-}
-
-// on three inner edges
-static void VFilter16i(uint8_t* p, int stride,
-                       int thresh, int ithresh, int hev_thresh) {
-  int k;
-  __m128i p3, p2, p1, p0;   // loop invariants
-
-  LOAD_H_EDGES4(p, stride, p3, p2, p1, p0);  // prologue
-
-  for (k = 3; k > 0; --k) {
-    __m128i mask, tmp1, tmp2;
-    uint8_t* const b = p + 2 * stride;   // beginning of p1
-    p += 4 * stride;
-
-    MAX_DIFF1(p3, p2, p1, p0, mask);   // compute partial mask
-    LOAD_H_EDGES4(p, stride, p3, p2, tmp1, tmp2);
-    MAX_DIFF2(p3, p2, tmp1, tmp2, mask);
-
-    // p3 and p2 are not just temporary variables here: they will be
-    // re-used for next span. And q2/q3 will become p1/p0 accordingly.
-    ComplexMask(&p1, &p0, &p3, &p2, thresh, ithresh, &mask);
-    DoFilter4(&p1, &p0, &p3, &p2, &mask, hev_thresh);
-
-    // Store
-    _mm_storeu_si128((__m128i*)&b[0 * stride], p1);
-    _mm_storeu_si128((__m128i*)&b[1 * stride], p0);
-    _mm_storeu_si128((__m128i*)&b[2 * stride], p3);
-    _mm_storeu_si128((__m128i*)&b[3 * stride], p2);
-
-    // rotate samples
-    p1 = tmp1;
-    p0 = tmp2;
-  }
-}
-
-static void HFilter16i(uint8_t* p, int stride,
-                       int thresh, int ithresh, int hev_thresh) {
-  int k;
-  __m128i p3, p2, p1, p0;   // loop invariants
-
-  Load16x4(p, p + 8 * stride, stride, &p3, &p2, &p1, &p0);  // prologue
-
-  for (k = 3; k > 0; --k) {
-    __m128i mask, tmp1, tmp2;
-    uint8_t* const b = p + 2;   // beginning of p1
-
-    p += 4;  // beginning of q0 (and next span)
-
-    MAX_DIFF1(p3, p2, p1, p0, mask);   // compute partial mask
-    Load16x4(p, p + 8 * stride, stride, &p3, &p2, &tmp1, &tmp2);
-    MAX_DIFF2(p3, p2, tmp1, tmp2, mask);
-
-    ComplexMask(&p1, &p0, &p3, &p2, thresh, ithresh, &mask);
-    DoFilter4(&p1, &p0, &p3, &p2, &mask, hev_thresh);
-
-    Store16x4(&p1, &p0, &p3, &p2, b, b + 8 * stride, stride);
-
-    // rotate samples
-    p1 = tmp1;
-    p0 = tmp2;
-  }
-}
-
-// 8-pixels wide variant, for chroma filtering
-static void VFilter8(uint8_t* u, uint8_t* v, int stride,
-                     int thresh, int ithresh, int hev_thresh) {
-  __m128i mask;
-  __m128i t1, p2, p1, p0, q0, q1, q2;
-
-  // Load p3, p2, p1, p0
-  LOADUV_H_EDGES4(u - 4 * stride, v - 4 * stride, stride, t1, p2, p1, p0);
-  MAX_DIFF1(t1, p2, p1, p0, mask);
-
-  // Load q0, q1, q2, q3
-  LOADUV_H_EDGES4(u, v, stride, q0, q1, q2, t1);
-  MAX_DIFF2(t1, q2, q1, q0, mask);
-
-  ComplexMask(&p1, &p0, &q0, &q1, thresh, ithresh, &mask);
-  DoFilter6(&p2, &p1, &p0, &q0, &q1, &q2, &mask, hev_thresh);
-
-  // Store
-  STOREUV(p2, u, v, -3 * stride);
-  STOREUV(p1, u, v, -2 * stride);
-  STOREUV(p0, u, v, -1 * stride);
-  STOREUV(q0, u, v, 0 * stride);
-  STOREUV(q1, u, v, 1 * stride);
-  STOREUV(q2, u, v, 2 * stride);
-}
-
-static void HFilter8(uint8_t* u, uint8_t* v, int stride,
-                     int thresh, int ithresh, int hev_thresh) {
-  __m128i mask;
-  __m128i p3, p2, p1, p0, q0, q1, q2, q3;
-
-  uint8_t* const tu = u - 4;
-  uint8_t* const tv = v - 4;
-  Load16x4(tu, tv, stride, &p3, &p2, &p1, &p0);  // p3, p2, p1, p0
-  MAX_DIFF1(p3, p2, p1, p0, mask);
-
-  Load16x4(u, v, stride, &q0, &q1, &q2, &q3);    // q0, q1, q2, q3
-  MAX_DIFF2(q3, q2, q1, q0, mask);
-
-  ComplexMask(&p1, &p0, &q0, &q1, thresh, ithresh, &mask);
-  DoFilter6(&p2, &p1, &p0, &q0, &q1, &q2, &mask, hev_thresh);
-
-  Store16x4(&p3, &p2, &p1, &p0, tu, tv, stride);
-  Store16x4(&q0, &q1, &q2, &q3, u, v, stride);
-}
-
-static void VFilter8i(uint8_t* u, uint8_t* v, int stride,
-                      int thresh, int ithresh, int hev_thresh) {
-  __m128i mask;
-  __m128i t1, t2, p1, p0, q0, q1;
-
-  // Load p3, p2, p1, p0
-  LOADUV_H_EDGES4(u, v, stride, t2, t1, p1, p0);
-  MAX_DIFF1(t2, t1, p1, p0, mask);
-
-  u += 4 * stride;
-  v += 4 * stride;
-
-  // Load q0, q1, q2, q3
-  LOADUV_H_EDGES4(u, v, stride, q0, q1, t1, t2);
-  MAX_DIFF2(t2, t1, q1, q0, mask);
-
-  ComplexMask(&p1, &p0, &q0, &q1, thresh, ithresh, &mask);
-  DoFilter4(&p1, &p0, &q0, &q1, &mask, hev_thresh);
-
-  // Store
-  STOREUV(p1, u, v, -2 * stride);
-  STOREUV(p0, u, v, -1 * stride);
-  STOREUV(q0, u, v, 0 * stride);
-  STOREUV(q1, u, v, 1 * stride);
-}
-
-static void HFilter8i(uint8_t* u, uint8_t* v, int stride,
-                      int thresh, int ithresh, int hev_thresh) {
-  __m128i mask;
-  __m128i t1, t2, p1, p0, q0, q1;
-  Load16x4(u, v, stride, &t2, &t1, &p1, &p0);   // p3, p2, p1, p0
-  MAX_DIFF1(t2, t1, p1, p0, mask);
-
-  u += 4;  // beginning of q0
-  v += 4;
-  Load16x4(u, v, stride, &q0, &q1, &t1, &t2);  // q0, q1, q2, q3
-  MAX_DIFF2(t2, t1, q1, q0, mask);
-
-  ComplexMask(&p1, &p0, &q0, &q1, thresh, ithresh, &mask);
-  DoFilter4(&p1, &p0, &q0, &q1, &mask, hev_thresh);
-
-  u -= 2;  // beginning of p1
-  v -= 2;
-  Store16x4(&p1, &p0, &q0, &q1, u, v, stride);
-}
-
-#endif   // WEBP_USE_SSE2
-
-//------------------------------------------------------------------------------
-// Entry point
-
-extern void VP8DspInitSSE2(void);
-
-void VP8DspInitSSE2(void) {
-#if defined(WEBP_USE_SSE2)
-  VP8Transform = Transform;
-#if defined(USE_TRANSFORM_AC3)
-  VP8TransformAC3 = TransformAC3;
-#endif
-
-  VP8VFilter16 = VFilter16;
-  VP8HFilter16 = HFilter16;
-  VP8VFilter8 = VFilter8;
-  VP8HFilter8 = HFilter8;
-  VP8VFilter16i = VFilter16i;
-  VP8HFilter16i = HFilter16i;
-  VP8VFilter8i = VFilter8i;
-  VP8HFilter8i = HFilter8i;
-
-  VP8SimpleVFilter16 = SimpleVFilter16;
-  VP8SimpleHFilter16 = SimpleHFilter16;
-  VP8SimpleVFilter16i = SimpleVFilter16i;
-  VP8SimpleHFilter16i = SimpleHFilter16i;
-#endif   // WEBP_USE_SSE2
-}
@@ -1,293 +0,0 @@
-// Copyright 2011 Google Inc. All Rights Reserved.
-//
-// Use of this source code is governed by a BSD-style license
-// that can be found in the COPYING file in the root of the source
-// tree. An additional intellectual property rights grant can be found
-// in the file PATENTS. All contributing project authors may
-// be found in the AUTHORS file in the root of the source tree.
-// -----------------------------------------------------------------------------
-//
-//   Speed-critical functions.
-//
-// Author: Skal (pascal.massimino@gmail.com)
-
-#ifndef WEBP_DSP_DSP_H_
-#define WEBP_DSP_DSP_H_
-
-#ifdef HAVE_CONFIG_H
-#include "../webp/config.h"
-#endif
-
-#include "../webp/types.h"
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-//------------------------------------------------------------------------------
-// CPU detection
-
-#if defined(__GNUC__)
-# define LOCAL_GCC_VERSION ((__GNUC__ << 8) | __GNUC_MINOR__)
-# define LOCAL_GCC_PREREQ(maj, min) \
-    (LOCAL_GCC_VERSION >= (((maj) << 8) | (min)))
-#else
-# define LOCAL_GCC_VERSION 0
-# define LOCAL_GCC_PREREQ(maj, min) 0
-#endif
-
-#ifdef __clang__
-# define LOCAL_CLANG_VERSION ((__clang_major__ << 8) | __clang_minor__)
-# define LOCAL_CLANG_PREREQ(maj, min) \
-    (LOCAL_CLANG_VERSION >= (((maj) << 8) | (min)))
-#else
-# define LOCAL_CLANG_VERSION 0
-# define LOCAL_CLANG_PREREQ(maj, min) 0
-#endif  // __clang__
-
-#if defined(_MSC_VER) && _MSC_VER > 1310 && \
-    (defined(_M_X64) || defined(_M_IX86))
-#define WEBP_MSC_SSE2  // Visual C++ SSE2 targets
-#endif
-
-// WEBP_HAVE_* are used to indicate the presence of the instruction set in dsp
-// files without intrinsics, allowing the corresponding Init() to be called.
-// Files containing intrinsics will need to be built targeting the instruction
-// set so should succeed on one of the earlier tests.
-#if defined(__SSE2__) || defined(WEBP_MSC_SSE2) || defined(WEBP_HAVE_SSE2)
-#define WEBP_USE_SSE2
-#endif
-
-#if defined(__AVX2__) || defined(WEBP_HAVE_AVX2)
-#define WEBP_USE_AVX2
-#endif
-
-#if defined(__ANDROID__) && defined(__ARM_ARCH_7A__)
-#define WEBP_ANDROID_NEON  // Android targets that might support NEON
-#endif
-
-#if defined(__ARM_NEON__) || defined(WEBP_ANDROID_NEON) || defined(__aarch64__)
-#define WEBP_USE_NEON
-#endif
-
-#if defined(__mips__) && !defined(__mips64) && (__mips_isa_rev < 6)
-#define WEBP_USE_MIPS32
-#if (__mips_isa_rev >= 2)
-#define WEBP_USE_MIPS32_R2
-#endif
-#endif
-
-typedef enum {
-  kSSE2,
-  kSSE3,
-  kAVX,
-  kAVX2,
-  kNEON,
-  kMIPS32
-} CPUFeature;
-// returns true if the CPU supports the feature.
-typedef int (*VP8CPUInfo)(CPUFeature feature);
-extern VP8CPUInfo VP8GetCPUInfo;
-
-//------------------------------------------------------------------------------
-// Encoding
-
-// Transforms
-// VP8Idct: Does one of two inverse transforms. If do_two is set, the transforms
-//          will be done for (ref, in, dst) and (ref + 4, in + 16, dst + 4).
-typedef void (*VP8Idct)(const uint8_t* ref, const int16_t* in, uint8_t* dst,
-                        int do_two);
-typedef void (*VP8Fdct)(const uint8_t* src, const uint8_t* ref, int16_t* out);
-typedef void (*VP8WHT)(const int16_t* in, int16_t* out);
-extern VP8Idct VP8ITransform;
-extern VP8Fdct VP8FTransform;
-extern VP8WHT VP8FTransformWHT;
-// Predictions
-// *dst is the destination block. *top and *left can be NULL.
-typedef void (*VP8IntraPreds)(uint8_t *dst, const uint8_t* left,
-                              const uint8_t* top);
-typedef void (*VP8Intra4Preds)(uint8_t *dst, const uint8_t* top);
-extern VP8Intra4Preds VP8EncPredLuma4;
-extern VP8IntraPreds VP8EncPredLuma16;
-extern VP8IntraPreds VP8EncPredChroma8;
-
-typedef int (*VP8Metric)(const uint8_t* pix, const uint8_t* ref);
-extern VP8Metric VP8SSE16x16, VP8SSE16x8, VP8SSE8x8, VP8SSE4x4;
-typedef int (*VP8WMetric)(const uint8_t* pix, const uint8_t* ref,
-                          const uint16_t* const weights);
-extern VP8WMetric VP8TDisto4x4, VP8TDisto16x16;
-
-typedef void (*VP8BlockCopy)(const uint8_t* src, uint8_t* dst);
-extern VP8BlockCopy VP8Copy4x4;
-// Quantization
-struct VP8Matrix;   // forward declaration
-typedef int (*VP8QuantizeBlock)(int16_t in[16], int16_t out[16],
-                                const struct VP8Matrix* const mtx);
-extern VP8QuantizeBlock VP8EncQuantizeBlock;
-
-// specific to 2nd transform:
-typedef int (*VP8QuantizeBlockWHT)(int16_t in[16], int16_t out[16],
-                                   const struct VP8Matrix* const mtx);
-extern VP8QuantizeBlockWHT VP8EncQuantizeBlockWHT;
-
-// Collect histogram for susceptibility calculation and accumulate in histo[].
-struct VP8Histogram;
-typedef void (*VP8CHisto)(const uint8_t* ref, const uint8_t* pred,
-                          int start_block, int end_block,
-                          struct VP8Histogram* const histo);
-extern const int VP8DspScan[16 + 4 + 4];
-extern VP8CHisto VP8CollectHistogram;
-
-void VP8EncDspInit(void);   // must be called before using any of the above
-
-//------------------------------------------------------------------------------
-// Decoding
-
-typedef void (*VP8DecIdct)(const int16_t* coeffs, uint8_t* dst);
-// when doing two transforms, coeffs is actually int16_t[2][16].
-typedef void (*VP8DecIdct2)(const int16_t* coeffs, uint8_t* dst, int do_two);
-extern VP8DecIdct2 VP8Transform;
-extern VP8DecIdct VP8TransformAC3;
-extern VP8DecIdct VP8TransformUV;
-extern VP8DecIdct VP8TransformDC;
-extern VP8DecIdct VP8TransformDCUV;
-extern VP8WHT VP8TransformWHT;
-
-// *dst is the destination block, with stride BPS. Boundary samples are
-// assumed accessible when needed.
-typedef void (*VP8PredFunc)(uint8_t* dst);
-extern const VP8PredFunc VP8PredLuma16[/* NUM_B_DC_MODES */];
-extern const VP8PredFunc VP8PredChroma8[/* NUM_B_DC_MODES */];
-extern const VP8PredFunc VP8PredLuma4[/* NUM_BMODES */];
-
-// clipping tables (for filtering)
-extern const int8_t* const VP8ksclip1;  // clips [-1020, 1020] to [-128, 127]
-extern const int8_t* const VP8ksclip2;  // clips [-112, 112] to [-16, 15]
-extern const uint8_t* const VP8kclip1;  // clips [-255,511] to [0,255]
-extern const uint8_t* const VP8kabs0;   // abs(x) for x in [-255,255]
-void VP8InitClipTables(void);           // must be called first
-
-// simple filter (only for luma)
-typedef void (*VP8SimpleFilterFunc)(uint8_t* p, int stride, int thresh);
-extern VP8SimpleFilterFunc VP8SimpleVFilter16;
-extern VP8SimpleFilterFunc VP8SimpleHFilter16;
-extern VP8SimpleFilterFunc VP8SimpleVFilter16i;  // filter 3 inner edges
-extern VP8SimpleFilterFunc VP8SimpleHFilter16i;
-
-// regular filter (on both macroblock edges and inner edges)
-typedef void (*VP8LumaFilterFunc)(uint8_t* luma, int stride,
-                                  int thresh, int ithresh, int hev_t);
-typedef void (*VP8ChromaFilterFunc)(uint8_t* u, uint8_t* v, int stride,
-                                    int thresh, int ithresh, int hev_t);
-// on outer edge
-extern VP8LumaFilterFunc VP8VFilter16;
-extern VP8LumaFilterFunc VP8HFilter16;
-extern VP8ChromaFilterFunc VP8VFilter8;
-extern VP8ChromaFilterFunc VP8HFilter8;
-
-// on inner edge
-extern VP8LumaFilterFunc VP8VFilter16i;   // filtering 3 inner edges altogether
-extern VP8LumaFilterFunc VP8HFilter16i;
-extern VP8ChromaFilterFunc VP8VFilter8i;  // filtering u and v altogether
-extern VP8ChromaFilterFunc VP8HFilter8i;
-
-// must be called before anything using the above
-void VP8DspInit(void);
-
-//------------------------------------------------------------------------------
-// WebP I/O
-
-#define FANCY_UPSAMPLING   // undefined to remove fancy upsampling support
-
-// Convert a pair of y/u/v lines together to the output rgb/a colorspace.
-// bottom_y can be NULL if only one line of output is needed (at top/bottom).
-typedef void (*WebPUpsampleLinePairFunc)(
-    const uint8_t* top_y, const uint8_t* bottom_y,
-    const uint8_t* top_u, const uint8_t* top_v,
-    const uint8_t* cur_u, const uint8_t* cur_v,
-    uint8_t* top_dst, uint8_t* bottom_dst, int len);
-
-#ifdef FANCY_UPSAMPLING
-
-// Fancy upsampling functions to convert YUV to RGB(A) modes
-extern WebPUpsampleLinePairFunc WebPUpsamplers[/* MODE_LAST */];
-
-#endif    // FANCY_UPSAMPLING
-
-// Per-row point-sampling methods.
-typedef void (*WebPSamplerRowFunc)(const uint8_t* y,
-                                   const uint8_t* u, const uint8_t* v,
-                                   uint8_t* dst, int len);
-// Generic function to apply 'WebPSamplerRowFunc' to the whole plane:
-void WebPSamplerProcessPlane(const uint8_t* y, int y_stride,
-                             const uint8_t* u, const uint8_t* v, int uv_stride,
-                             uint8_t* dst, int dst_stride,
-                             int width, int height, WebPSamplerRowFunc func);
-
-// Sampling functions to convert rows of YUV to RGB(A)
-extern WebPSamplerRowFunc WebPSamplers[/* MODE_LAST */];
-
-// General function for converting two lines of ARGB or RGBA.
-// 'alpha_is_last' should be true if 0xff000000 is stored in memory as
-// as 0x00, 0x00, 0x00, 0xff (little endian).
-WebPUpsampleLinePairFunc WebPGetLinePairConverter(int alpha_is_last);
-
-// YUV444->RGB converters
-typedef void (*WebPYUV444Converter)(const uint8_t* y,
-                                    const uint8_t* u, const uint8_t* v,
-                                    uint8_t* dst, int len);
-
-extern const WebPYUV444Converter WebPYUV444Converters[/* MODE_LAST */];
-
-// Must be called before using the WebPUpsamplers[] (and for premultiplied
-// colorspaces like rgbA, rgbA4444, etc)
-void WebPInitUpsamplers(void);
-// Must be called before using WebPSamplers[]
-void WebPInitSamplers(void);
-
-//------------------------------------------------------------------------------
-// Utilities for processing transparent channel.
-
-// Apply alpha pre-multiply on an rgba, bgra or argb plane of size w * h.
-// alpha_first should be 0 for argb, 1 for rgba or bgra (where alpha is last).
-extern void (*WebPApplyAlphaMultiply)(
-    uint8_t* rgba, int alpha_first, int w, int h, int stride);
-
-// Same, buf specifically for RGBA4444 format
-extern void (*WebPApplyAlphaMultiply4444)(
-    uint8_t* rgba4444, int w, int h, int stride);
-
-// Extract the alpha values from 32b values in argb[] and pack them into alpha[]
-// (this is the opposite of WebPDispatchAlpha).
-// Returns true if there's only trivial 0xff alpha values.
-extern int (*WebPExtractAlpha)(const uint8_t* argb, int argb_stride,
-                               int width, int height,
-                               uint8_t* alpha, int alpha_stride);
-
-// Pre-Multiply operation transforms x into x * A / 255  (where x=Y,R,G or B).
-// Un-Multiply operation transforms x into x * 255 / A.
-
-// Pre-Multiply or Un-Multiply (if 'inverse' is true) argb values in a row.
-extern void (*WebPMultARGBRow)(uint32_t* const ptr, int width, int inverse);
-
-// Same a WebPMultARGBRow(), but for several rows.
-void WebPMultARGBRows(uint8_t* ptr, int stride, int width, int num_rows,
-                      int inverse);
-
-// Same for a row of single values, with side alpha values.
-extern void (*WebPMultRow)(uint8_t* const ptr, const uint8_t* const alpha,
-                           int width, int inverse);
-
-// Same a WebPMultRow(), but for several 'num_rows' rows.
-void WebPMultRows(uint8_t* ptr, int stride,
-                  const uint8_t* alpha, int alpha_stride,
-                  int width, int num_rows, int inverse);
-
-// To be called first before using the above.
-void WebPInitAlphaProcessing(void);
-
-#ifdef __cplusplus
-}    // extern "C"
-#endif
-
-#endif  /* WEBP_DSP_DSP_H_ */
--- a/Show More
+++ b/Show More