update ChangeLog

Change-Id: I32a22786f99d0c239761a362485a4a91c783851b

AUTHORS

@@ -20,5 +20,6 @@ Contributors:
 - 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)
+- Timothy Gu (timothygu99 at gmail dot com)
 - Urvang Joshi (urvang at google dot com)
 - Vikas Arora (vikasa at google dot com)

Android.mk

@@ -33,6 +33,7 @@ LOCAL_SRC_FILES := \
     src/dec/vp8l.c \
     src/dec/webp.c \
     src/dsp/alpha_processing.c \
+    src/dsp/alpha_processing_sse2.c \
     src/dsp/cpu.c \
     src/dsp/dec.c \
     src/dsp/dec_clip_tables.c \

ChangeLog

@@ -1,3 +1,60 @@
+36cad6a bit_reader.h: cosmetics: fix a typo
+e2ecae6 enc_mips32: workaround gcc-4.9 bug
+243e68d update ChangeLog (tag: v0.4.2-rc2)
+eec5f5f enc/vp8enci.h: update version number
+0c1b98d update NEWS
+69b0fc9 update AUTHORS
+857578a bump version to 0.4.2
+9129deb restore encode API compatibility
+f17b95e AssignSegments: quiet -Warray-bounds warning
+9c56c8a enc_neon: initialize vectors w/vdup_n_u32
+a008902 iosbuild: cleanup
+cc6de53 iosbuild: output autoconf req. on failure
+740d765 iosbuild: make iOS 6 the minimum requirement
+403023f iobuild.sh: only install .h files in Headers
+b65727b Premultiply with alpha during U/V downsampling
+8de0deb gif2webp: Background color correction
+f8b7d94 Amend the lossless spec according to issue #205, #206 and #224
+9102a7b Add a WebPExtractAlpha function to dsp
+e407b5d webpmux: simplify InitializeConfig()
+3e70e64 webpmux: fix indent
+be38f1a webpmux: fix exit status on numeric value parse error
+94dadcb webpmux: fix loop_count range check
+40b3a61 examples: warn on invalid numeric parameters
+b7d209a gif2webp: Handle frames with missing  graphic control extension
+bf0eb74 configure: simplify libpng-config invocation
+3740f7d Rectify bug in lossless incremental decoding.
+3ab0a37 make VP8LSetBitPos() set br->eos_ flag
+2e4312b Lossless decoding: fix eos_ flag condition
+e6609ac fix erroneous dec->status_ setting
+5692eae add a fallback to ALPHA_NO_COMPRESSION
+6ecd5bf ExUtilReadFromStdin: (windows) open stdin in bin mode
+4206ac6 webpmux: (windows) open stdout in binary mode
+d40e885 cwebp: (windows) open stdout in binary mode
+4aaf463 example_util: add ExUtilSetBinaryMode
+4c82ff7 webpmux man page: Clarify some title, descriptions and examples
+23d4fb3 dsp/lossless: workaround gcc-4.9 bug on arm
+5af7719 dsp.h: collect gcc/clang version test macros
+90d1124 enc_neon: enable QuantizeBlock for aarch64
+ee78e78 SmartRGBYUV: fix odd-width problem with pixel replication
+c9ac204 fix some MSVC64 warning about float conversion
+f4497a1 cpu: check for _MSC_VER before using msvc inline asm
+e2159fd faster RGB->YUV conversion function (~7% speedup)
+21abaa0 Add smart RGB->YUV conversion option -pre 4
+1a161e2 configure: add work around for gcc-4.9 aarch64 bug
+55b10de MIPS: mips32r2: added optimization for BSwap32
+76d2192 Update PATENTS to reflect s/VP8/WebM/g
+29a9db1 MIPS: detect mips32r6 and disable mips32r1 code
+245c4a6 Correctly use the AC_CANONICAL_* macros
+40aa8b6 cosmetics
+2ddcca5 cosmetics: remove some extraneous 'extern's
+f40dd7c vp8enci.h: cosmetics: fix '*' placement
+4610c9c bit_writer: cosmetics: rename kFlush() -> Flush()
+fc3c175 dsp: detect mips64 & disable mips32 code
+c1a7955 cwebp.1: restore quality description
+57a7e73 correct alpha_dithering_strength ABI check
+6c83157 correct WebPMemoryWriterClear ABI check
+8af2771 update ChangeLog (tag: v0.4.1, origin/0.4.1, 0.4.1)
 f59c0b4 iosbuild.sh: specify optimization flags
 8d34ea3 update ChangeLog (tag: v0.4.1-rc1)
 dbc3da6 makefile.unix: add vwebp.1 to the dist target
@@ -17,7 +74,7 @@ c2fc52e restore encode API compatibility
 793368e restore decode API compatibility
 b8984f3 gif2webp: fix compile with giflib 5.1.0
 222f9b1 gif2webp: simplify giflib version checking
-d2cc61b Extend MakeARGB32() to accept Alpha channel.
+d2cc61b Extend MakeARGB32() to accept Alpha channel. (master)
 4595b62 Merge "use explicit size of kErrorMessages[] arrays"
 157de01 Merge "Actuate memory stats for PRINT_MEMORY_INFO"
 fbda2f4 JPEG decoder: delay conversion to YUV to WebPEncode() call

Makefile.vc

@@ -172,6 +172,7 @@ DEMUX_OBJS = \
 
 DSP_DEC_OBJS = \
     $(DIROBJ)\dsp\alpha_processing.obj \
+    $(DIROBJ)\dsp\alpha_processing_sse2.obj \
     $(DIROBJ)\dsp\cpu.obj \
     $(DIROBJ)\dsp\dec.obj \
     $(DIROBJ)\dsp\dec_clip_tables.obj \

NEWS

@@ -1,3 +1,11 @@
+- 10/13/14: version 0.4.2
+  This is a binary compatible release.
+  * Android / gcc build fixes
+  * (Windows) fix reading from stdin and writing to stdout
+  * gif2webp: miscellaneous fixes
+  * fix 'alpha-leak' with lossy compression (issue #220)
+  * the lossless bitstream spec has been amended to reflect the current code
+
 - 7/24/14: version 0.4.1
   This is a binary compatible release.
   * AArch64 (arm64) & MIPS support/optimizations

PATENTS

@@ -1,22 +1,23 @@
 Additional IP Rights Grant (Patents)
+------------------------------------
 
-"This implementation" means the copyrightable works distributed by
-Google as part of the WebM Project.
+"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 this
-implementation of VP8, 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 this
-implementation of VP8. This grant does not include claims that would be
-infringed only as a consequence of further modification of this
-implementation. If you or your agent or exclusive licensee institute or
-order or agree to the institution of patent litigation against any
-entity (including a cross-claim or counterclaim in a lawsuit) alleging
-that this implementation of VP8 or any code incorporated within this
-implementation of VP8 constitutes direct or contributory patent
-infringement, or inducement of patent infringement, then any patent
-rights granted to you under this License for this implementation of VP8
-shall terminate as of the date such litigation is filed.
+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
+constitutes 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.

README

@@ -4,7 +4,7 @@
           \__\__/\____/\_____/__/ ____  ___
                 / _/ /    \    \ /  _ \/ _/
                /  \_/   / /   \ \   __/  \__
-               \____/____/\_____/_____/____/v0.4.1
+               \____/____/\_____/_____/____/v0.4.2
 
 Description:
 ============

README.mux

@@ -1,7 +1,7 @@
           __   __  ____  ____  ____  __ __  _     __ __
          /  \\/  \/  _ \/  _ \/  _ \/  \  \/ \___/_ / _\
          \       /   __/  _  \   __/      /  /  (_/  /__
-          \__\__/\_____/_____/__/  \__//_/\_____/__/___/v0.2.1
+          \__\__/\_____/_____/__/  \__//_/\_____/__/___/v0.2.2
 
 
 Description:

configure.ac

@@ -1,7 +1,7 @@
-AC_INIT([libwebp], [0.4.1],
+AC_INIT([libwebp], [0.4.2],
         [http://code.google.com/p/webp/issues],,
         [http://developers.google.com/speed/webp])
-AC_CANONICAL_TARGET
+AC_CANONICAL_HOST
 AC_PREREQ([2.60])
 AM_INIT_AUTOMAKE([-Wall foreign subdir-objects])
 
@@ -62,6 +62,19 @@ TEST_AND_ADD_CFLAGS([AM_CFLAGS], [-Wshadow])
 TEST_AND_ADD_CFLAGS([AM_CFLAGS], [-Wunused-but-set-variable])
 TEST_AND_ADD_CFLAGS([AM_CFLAGS], [-Wunused])
 TEST_AND_ADD_CFLAGS([AM_CFLAGS], [-Wvla])
+# https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62040
+# https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61622
+AS_IF([test "$GCC" = "yes" ], [
+       gcc_version=`$CC -dumpversion`
+       gcc_wht_bug=""
+       case "$host_cpu" in
+         aarch64|arm64)
+          case "$gcc_version" in
+            4.9|4.9.0|4.9.1) gcc_wht_bug=yes ;;
+          esac
+       esac
+       AS_IF([test "$gcc_wht_bug" = "yes"], [
+              TEST_AND_ADD_CFLAGS([AM_CFLAGS], [-frename-registers])])])
 AC_SUBST([AM_CFLAGS])
 
 dnl === Check for machine specific flags
@@ -283,15 +296,12 @@ AC_ARG_ENABLE([png], AS_HELP_STRING([--disable-png],
                                      @<:@default=auto@:>@]))
 AS_IF([test "x$enable_png" != "xno"], [
   CLEAR_LIBVARS([PNG])
-  AC_PATH_PROGS(LIBPNG_CONFIG,
-                [libpng-config libpng15-config libpng14-config libpng12-config])
+  AC_PATH_PROGS([LIBPNG_CONFIG],
+                [libpng-config libpng16-config libpng15-config libpng14-config \
+                 libpng12-config])
   if test -n "$LIBPNG_CONFIG"; then
     PNG_INCLUDES=`$LIBPNG_CONFIG --cflags`
-    PNG_PREFIX=`$LIBPNG_CONFIG --prefix`
-    if test "${PNG_PREFIX}/lib" != "/usr/lib" ; then
-      PNG_LIBS="-L${PNG_PREFIX}/lib"
-    fi
-    PNG_LIBS="$PNG_LIBS `$LIBPNG_CONFIG --libs`"
+    PNG_LIBS="`$LIBPNG_CONFIG --ldflags`"
   fi
 
   WITHLIB_OPTION([png], [PNG])
@@ -409,8 +419,9 @@ AC_ARG_ENABLE([wic],
                               @<:@default=auto@:>@]),,
               [enable_wic=yes])
 
-if test \( "$target_os" = "mingw32" -o "$target_os" = "mingw64" \) \
-        -a "$enable_wic" = "yes"; then
+case $host_os in
+mingw*)
+if test "$enable_wic" = "yes"; then
   AC_CHECK_HEADERS([wincodec.h shlwapi.h windows.h])
   if test "$ac_cv_header_wincodec_h" = "yes"; then
     AC_MSG_CHECKING(for Windows Imaging Component support)
@@ -450,6 +461,7 @@ if test \( "$target_os" = "mingw32" -o "$target_os" = "mingw64" \) \
     AC_MSG_RESULT(${wic_support-no})
   fi
 fi
+esac
 
 dnl === If --enable-aligned is defined, define WEBP_FORCE_ALIGNED
 

doc/webp-lossless-bitstream-spec.txt

@@ -14,6 +14,7 @@ Specification for WebP Lossless Bitstream
 
 _Jyrki Alakuijala, Ph.D., Google, Inc., 2012-06-19_
 
+Paragraphs marked as \[AMENDED\] were amended on 2014-09-16.
 
 Abstract
 --------
@@ -172,8 +173,8 @@ It should be set to 0 when all alpha values are 255 in the picture, and
 int alpha_is_used = ReadBits(1);
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
-The version_number is a 3 bit code that must be discarded by the decoder
-at this time. Complying encoders write a 3-bit value 0.
+The version_number is a 3 bit code that must be set to 0. Any other value
+should be treated as an error. \[AMENDED\]
 
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 int version_number = ReadBits(3);
@@ -330,7 +331,7 @@ uint32 Select(uint32 L, uint32 T, uint32 TL) {
            abs(pGreen - GREEN(T)) + abs(pBlue - BLUE(T));
 
   // Return either left or top, the one closer to the prediction.
-  if (pL <= pT) {
+  if (pL < pT) {     // \[AMENDED\]
     return L;
   } else {
     return T;
@@ -542,6 +543,9 @@ color.
 argb = color_table[GREEN(argb)];
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
+If the index is equal or larger than color_table_size, the argb color value
+should be set to 0x00000000 (transparent black).  \[AMENDED\]
+
 When the color table is small (equal to or less than 16 colors), several
 pixels are bundled into a single pixel. The pixel bundling packs several
 (2, 4, or 8) pixels into a single pixel, reducing the image width

examples/cwebp.c

@@ -22,6 +22,7 @@
 
 #include "webp/encode.h"
 
+#include "./example_util.h"
 #include "./metadata.h"
 #include "./stopwatch.h"
 
@@ -709,6 +710,7 @@ int main(int argc, const char *argv[]) {
   }
 
   for (c = 1; c < argc; ++c) {
+    int parse_error = 0;
     if (!strcmp(argv[c], "-h") || !strcmp(argv[c], "-help")) {
       HelpShort();
       return 0;
@@ -732,30 +734,31 @@ int main(int argc, const char *argv[]) {
     } else if (!strcmp(argv[c], "-short")) {
       ++short_output;
     } else if (!strcmp(argv[c], "-s") && c < argc - 2) {
-      picture.width = strtol(argv[++c], NULL, 0);
-      picture.height = strtol(argv[++c], NULL, 0);
+      picture.width = ExUtilGetInt(argv[++c], 0, &parse_error);
+      picture.height = ExUtilGetInt(argv[++c], 0, &parse_error);
     } else if (!strcmp(argv[c], "-m") && c < argc - 1) {
-      config.method = strtol(argv[++c], NULL, 0);
+      config.method = ExUtilGetInt(argv[++c], 0, &parse_error);
 #if WEBP_ENCODER_ABI_VERSION > 0x0202
       use_lossless_preset = 0;   // disable -z option
 #endif
     } else if (!strcmp(argv[c], "-q") && c < argc - 1) {
-      config.quality = (float)strtod(argv[++c], NULL);
+      config.quality = ExUtilGetFloat(argv[++c], &parse_error);
 #if WEBP_ENCODER_ABI_VERSION > 0x0202
       use_lossless_preset = 0;   // disable -z option
     } else if (!strcmp(argv[c], "-z") && c < argc - 1) {
-      lossless_preset = strtol(argv[++c], NULL, 0);
+      lossless_preset = ExUtilGetInt(argv[++c], 0, &parse_error);
       if (use_lossless_preset != 0) use_lossless_preset = 1;
 #endif
     } else if (!strcmp(argv[c], "-alpha_q") && c < argc - 1) {
-      config.alpha_quality = strtol(argv[++c], NULL, 0);
+      config.alpha_quality = ExUtilGetInt(argv[++c], 0, &parse_error);
     } else if (!strcmp(argv[c], "-alpha_method") && c < argc - 1) {
-      config.alpha_compression = strtol(argv[++c], NULL, 0);
+      config.alpha_compression = ExUtilGetInt(argv[++c], 0, &parse_error);
     } else if (!strcmp(argv[c], "-alpha_cleanup")) {
       keep_alpha = keep_alpha ? 2 : 0;
     } else if (!strcmp(argv[c], "-blend_alpha") && c < argc - 1) {
       blend_alpha = 1;
-      background_color = strtol(argv[++c], NULL, 16);  // <- parses '0x' prefix
+      // background color is given in hex with an optional '0x' prefix
+      background_color = ExUtilGetInt(argv[++c], 16, &parse_error);
       background_color = background_color & 0x00ffffffu;
     } else if (!strcmp(argv[c], "-alpha_filter") && c < argc - 1) {
       ++c;
@@ -786,13 +789,13 @@ int main(int argc, const char *argv[]) {
         goto Error;
       }
     } else if (!strcmp(argv[c], "-size") && c < argc - 1) {
-      config.target_size = strtol(argv[++c], NULL, 0);
+      config.target_size = ExUtilGetInt(argv[++c], 0, &parse_error);
     } else if (!strcmp(argv[c], "-psnr") && c < argc - 1) {
-      config.target_PSNR = (float)strtod(argv[++c], NULL);
+      config.target_PSNR = ExUtilGetFloat(argv[++c], &parse_error);
     } else if (!strcmp(argv[c], "-sns") && c < argc - 1) {
-      config.sns_strength = strtol(argv[++c], NULL, 0);
+      config.sns_strength = ExUtilGetInt(argv[++c], 0, &parse_error);
     } else if (!strcmp(argv[c], "-f") && c < argc - 1) {
-      config.filter_strength = strtol(argv[++c], NULL, 0);
+      config.filter_strength = ExUtilGetInt(argv[++c], 0, &parse_error);
     } else if (!strcmp(argv[c], "-af")) {
       config.autofilter = 1;
     } else if (!strcmp(argv[c], "-jpeg_like")) {
@@ -806,26 +809,26 @@ int main(int argc, const char *argv[]) {
     } else if (!strcmp(argv[c], "-nostrong")) {
       config.filter_type = 0;
     } else if (!strcmp(argv[c], "-sharpness") && c < argc - 1) {
-      config.filter_sharpness = strtol(argv[++c], NULL, 0);
+      config.filter_sharpness = ExUtilGetInt(argv[++c], 0, &parse_error);
     } else if (!strcmp(argv[c], "-pass") && c < argc - 1) {
-      config.pass = strtol(argv[++c], NULL, 0);
+      config.pass = ExUtilGetInt(argv[++c], 0, &parse_error);
     } else if (!strcmp(argv[c], "-pre") && c < argc - 1) {
-      config.preprocessing = strtol(argv[++c], NULL, 0);
+      config.preprocessing = ExUtilGetInt(argv[++c], 0, &parse_error);
     } else if (!strcmp(argv[c], "-segments") && c < argc - 1) {
-      config.segments = strtol(argv[++c], NULL, 0);
+      config.segments = ExUtilGetInt(argv[++c], 0, &parse_error);
     } else if (!strcmp(argv[c], "-partition_limit") && c < argc - 1) {
-      config.partition_limit = strtol(argv[++c], NULL, 0);
+      config.partition_limit = ExUtilGetInt(argv[++c], 0, &parse_error);
     } else if (!strcmp(argv[c], "-map") && c < argc - 1) {
-      picture.extra_info_type = strtol(argv[++c], NULL, 0);
+      picture.extra_info_type = ExUtilGetInt(argv[++c], 0, &parse_error);
     } else if (!strcmp(argv[c], "-crop") && c < argc - 4) {
       crop = 1;
-      crop_x = strtol(argv[++c], NULL, 0);
-      crop_y = strtol(argv[++c], NULL, 0);
-      crop_w = strtol(argv[++c], NULL, 0);
-      crop_h = strtol(argv[++c], NULL, 0);
+      crop_x = ExUtilGetInt(argv[++c], 0, &parse_error);
+      crop_y = ExUtilGetInt(argv[++c], 0, &parse_error);
+      crop_w = ExUtilGetInt(argv[++c], 0, &parse_error);
+      crop_h = ExUtilGetInt(argv[++c], 0, &parse_error);
     } else if (!strcmp(argv[c], "-resize") && c < argc - 2) {
-      resize_w = strtol(argv[++c], NULL, 0);
-      resize_h = strtol(argv[++c], NULL, 0);
+      resize_w = ExUtilGetInt(argv[++c], 0, &parse_error);
+      resize_h = ExUtilGetInt(argv[++c], 0, &parse_error);
 #ifndef WEBP_DLL
     } else if (!strcmp(argv[c], "-noasm")) {
       VP8GetCPUInfo = NULL;
@@ -920,6 +923,11 @@ int main(int argc, const char *argv[]) {
     } else {
       in_file = argv[c];
     }
+
+    if (parse_error) {
+      HelpLong();
+      return -1;
+    }
   }
   if (in_file == NULL) {
     fprintf(stderr, "No input file specified!\n");
@@ -981,7 +989,7 @@ int main(int argc, const char *argv[]) {
   // Open the output
   if (out_file != NULL) {
     const int use_stdout = !strcmp(out_file, "-");
-    out = use_stdout ? stdout : fopen(out_file, "wb");
+    out = use_stdout ? ExUtilSetBinaryMode(stdout) : fopen(out_file, "wb");
     if (out == NULL) {
       fprintf(stderr, "Error! Cannot open output file '%s'\n", out_file);
       goto Error;
@@ -1126,7 +1134,7 @@ int main(int argc, const char *argv[]) {
   return_value = 0;
 
  Error:
-#if WEBP_ENCODER_ABI_VERSION > 0x0202
+#if WEBP_ENCODER_ABI_VERSION > 0x0203
   WebPMemoryWriterClear(&memory_writer);
 #else
   free(memory_writer.mem);

examples/dwebp.c

@@ -38,11 +38,6 @@
 #include <wincodec.h>
 #endif
 
-#if defined(_WIN32)
-#include <fcntl.h>   // for _O_BINARY
-#include <io.h>      // for _setmode()
-#endif
-
 #include "webp/decode.h"
 #include "./example_util.h"
 #include "./stopwatch.h"
@@ -482,15 +477,8 @@ static int SaveOutput(const WebPDecBuffer* const buffer,
   needs_open_file = (format != PNG);
 #endif
 
-#if defined(_WIN32)
-  if (use_stdout && _setmode(_fileno(stdout), _O_BINARY) == -1) {
-    fprintf(stderr, "Failed to reopen stdout in O_BINARY mode.\n");
-    return -1;
-  }
-#endif
-
   if (needs_open_file) {
-    fout = use_stdout ? stdout : fopen(out_file, "wb");
+    fout = use_stdout ? ExUtilSetBinaryMode(stdout) : fopen(out_file, "wb");
     if (fout == NULL) {
       fprintf(stderr, "Error opening output file %s\n", out_file);
       return 0;
@@ -557,7 +545,7 @@ static void Help(void) {
          "  -nofilter .... disable in-loop filtering\n"
          "  -nodither .... disable dithering\n"
          "  -dither <d> .. dithering strength (in 0..100)\n"
-#if WEBP_DECODER_ABI_VERSION > 0x0203
+#if WEBP_DECODER_ABI_VERSION > 0x0204
          "  -alpha_dither  use alpha-plane dithering if needed\n"
 #endif
          "  -mt .......... use multi-threading\n"
@@ -598,6 +586,7 @@ int main(int argc, const char *argv[]) {
   }
 
   for (c = 1; c < argc; ++c) {
+    int parse_error = 0;
     if (!strcmp(argv[c], "-h") || !strcmp(argv[c], "-help")) {
       Help();
       return 0;
@@ -628,24 +617,25 @@ int main(int argc, const char *argv[]) {
       format = YUV;
     } else if (!strcmp(argv[c], "-mt")) {
       config.options.use_threads = 1;
-#if WEBP_DECODER_ABI_VERSION > 0x0203
+#if WEBP_DECODER_ABI_VERSION > 0x0204
     } else if (!strcmp(argv[c], "-alpha_dither")) {
       config.options.alpha_dithering_strength = 100;
 #endif
     } else if (!strcmp(argv[c], "-nodither")) {
       config.options.dithering_strength = 0;
     } else if (!strcmp(argv[c], "-dither") && c < argc - 1) {
-      config.options.dithering_strength = strtol(argv[++c], NULL, 0);
+      config.options.dithering_strength =
+          ExUtilGetInt(argv[++c], 0, &parse_error);
     } else if (!strcmp(argv[c], "-crop") && c < argc - 4) {
       config.options.use_cropping = 1;
-      config.options.crop_left   = strtol(argv[++c], NULL, 0);
-      config.options.crop_top    = strtol(argv[++c], NULL, 0);
-      config.options.crop_width  = strtol(argv[++c], NULL, 0);
-      config.options.crop_height = strtol(argv[++c], NULL, 0);
+      config.options.crop_left   = ExUtilGetInt(argv[++c], 0, &parse_error);
+      config.options.crop_top    = ExUtilGetInt(argv[++c], 0, &parse_error);
+      config.options.crop_width  = ExUtilGetInt(argv[++c], 0, &parse_error);
+      config.options.crop_height = ExUtilGetInt(argv[++c], 0, &parse_error);
     } else if (!strcmp(argv[c], "-scale") && c < argc - 2) {
       config.options.use_scaling = 1;
-      config.options.scaled_width  = strtol(argv[++c], NULL, 0);
-      config.options.scaled_height = strtol(argv[++c], NULL, 0);
+      config.options.scaled_width  = ExUtilGetInt(argv[++c], 0, &parse_error);
+      config.options.scaled_height = ExUtilGetInt(argv[++c], 0, &parse_error);
 #if WEBP_DECODER_ABI_VERSION > 0x0203
     } else if (!strcmp(argv[c], "-flip")) {
       config.options.flip = 1;
@@ -668,6 +658,11 @@ int main(int argc, const char *argv[]) {
     } else {
       in_file = argv[c];
     }
+
+    if (parse_error) {
+      Help();
+      return -1;
+    }
   }
 
   if (in_file == NULL) {

examples/example_util.c

@@ -11,6 +11,11 @@
 //
 
 #include "./example_util.h"
+
+#if defined(_WIN32)
+#include <fcntl.h>   // for _O_BINARY
+#include <io.h>      // for _setmode()
+#endif
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
@@ -18,12 +23,50 @@
 #include "webp/decode.h"
 #include "./stopwatch.h"
 
+//------------------------------------------------------------------------------
+// String parsing
+
+uint32_t ExUtilGetUInt(const char* const v, int base, int* const error) {
+  char* end = NULL;
+  const uint32_t n = (v != NULL) ? (uint32_t)strtoul(v, &end, base) : 0u;
+  if (end == v && error != NULL && !*error) {
+    *error = 1;
+    fprintf(stderr, "Error! '%s' is not an integer.\n",
+            (v != NULL) ? v : "(null)");
+  }
+  return n;
+}
+
+int ExUtilGetInt(const char* const v, int base, int* const error) {
+  return (int)ExUtilGetUInt(v, base, error);
+}
+
+float ExUtilGetFloat(const char* const v, int* const error) {
+  char* end = NULL;
+  const float f = (v != NULL) ? (float)strtod(v, &end) : 0.f;
+  if (end == v && error != NULL && !*error) {
+    *error = 1;
+    fprintf(stderr, "Error! '%s' is not a floating point number.\n",
+            (v != NULL) ? v : "(null)");
+  }
+  return f;
+}
+
 // -----------------------------------------------------------------------------
 // File I/O
 
-static const size_t kBlockSize = 16384;  // default initial size
+FILE* ExUtilSetBinaryMode(FILE* file) {
+#if defined(_WIN32)
+  if (_setmode(_fileno(file), _O_BINARY) == -1) {
+    fprintf(stderr, "Failed to reopen file in O_BINARY mode.\n");
+    return NULL;
+  }
+#endif
+  return file;
+}
 
 int ExUtilReadFromStdin(const uint8_t** data, size_t* data_size) {
+  static const size_t kBlockSize = 16384;  // default initial size
   size_t max_size = 0;
   size_t size = 0;
   uint8_t* input = NULL;
@@ -32,6 +75,8 @@ int ExUtilReadFromStdin(const uint8_t** data, size_t* data_size) {
   *data = NULL;
   *data_size = 0;
 
+  if (!ExUtilSetBinaryMode(stdin)) return 0;
+
   while (!feof(stdin)) {
     // We double the buffer size each time and read as much as possible.
     const size_t extra_size = (max_size == 0) ? kBlockSize : max_size;

examples/example_util.h

@@ -13,12 +13,30 @@
 #ifndef WEBP_EXAMPLES_EXAMPLE_UTIL_H_
 #define WEBP_EXAMPLES_EXAMPLE_UTIL_H_
 
+#include <stdio.h>
 #include "webp/decode.h"
 
 #ifdef __cplusplus
 extern "C" {
 #endif
 
+//------------------------------------------------------------------------------
+// String parsing
+
+// Parses 'v' using strto(ul|l|d)(). If error is non-NULL, '*error' is set to
+// true on failure while on success it is left unmodified to allow chaining of
+// calls. An error is only printed on the first occurrence.
+uint32_t ExUtilGetUInt(const char* const v, int base, int* const error);
+int ExUtilGetInt(const char* const v, int base, int* const error);
+float ExUtilGetFloat(const char* const v, int* const error);
+
+//------------------------------------------------------------------------------
+// File I/O
+
+// Reopen file in binary (O_BINARY) mode.
+// Returns 'file' on success, NULL otherwise.
+FILE* ExUtilSetBinaryMode(FILE* file);
+
 // Allocates storage for entire file 'file_name' and returns contents and size
 // in 'data' and 'data_size'. Returns 1 on success, 0 otherwise. '*data' should
 // be deleted using free().

examples/gif2webp.c

@@ -46,7 +46,18 @@
 
 //------------------------------------------------------------------------------
 
-static int transparent_index = -1;  // Index of transparent color in the map.
+static int transparent_index;  // Index of transparent color in the map.
+
+static void ResetFrameInfo(WebPMuxFrameInfo* const info) {
+  WebPDataInit(&info->bitstream);
+  info->x_offset = 0;
+  info->y_offset = 0;
+  info->duration = 0;
+  info->id = WEBP_CHUNK_ANMF;
+  info->dispose_method = WEBP_MUX_DISPOSE_NONE;
+  info->blend_method = WEBP_MUX_BLEND;
+  transparent_index = -1;  // Opaque frame by default.
+}
 
 static void SanitizeKeyFrameIntervals(size_t* const kmin_ptr,
                                       size_t* const kmax_ptr) {
@@ -163,21 +174,22 @@ static int ReadFrame(GifFileType* const gif, WebPFrameRect* const gif_rect,
   return ok;
 }
 
-static int GetBackgroundColor(const ColorMapObject* const color_map,
+static void GetBackgroundColor(const ColorMapObject* const color_map,
                                int bgcolor_idx, uint32_t* const bgcolor) {
   if (transparent_index != -1 && bgcolor_idx == transparent_index) {
     *bgcolor = WEBP_UTIL_TRANSPARENT_COLOR;  // Special case.
-    return 1;
   } else if (color_map == NULL || color_map->Colors == NULL
              || bgcolor_idx >= color_map->ColorCount) {
-    return 0;  // Invalid color map or index.
+    *bgcolor = WHITE_COLOR;
+    fprintf(stderr,
+            "GIF decode warning: invalid background color index. Assuming "
+            "white background.\n");
   } else {
     const GifColorType color = color_map->Colors[bgcolor_idx];
     *bgcolor = (0xff        << 24)
              | (color.Red   << 16)
              | (color.Green <<  8)
              | (color.Blue  <<  0);
-    return 1;
   }
 }
 
@@ -278,10 +290,7 @@ int main(int argc, const char *argv[]) {
   size_t kmax = 0;
   int allow_mixed = 0;   // If true, each frame can be lossy or lossless.
 
-  memset(&info, 0, sizeof(info));
-  info.id = WEBP_CHUNK_ANMF;
-  info.dispose_method = WEBP_MUX_DISPOSE_BACKGROUND;
-  info.blend_method = WEBP_MUX_BLEND;
+  ResetFrameInfo(&info);
 
   if (!WebPConfigInit(&config) || !WebPPictureInit(&frame)) {
     fprintf(stderr, "Error! Version mismatch!\n");
@@ -296,6 +305,7 @@ int main(int argc, const char *argv[]) {
   }
 
   for (c = 1; c < argc; ++c) {
+    int parse_error = 0;
     if (!strcmp(argv[c], "-h") || !strcmp(argv[c], "-help")) {
       Help();
       return 0;
@@ -307,17 +317,17 @@ int main(int argc, const char *argv[]) {
       allow_mixed = 1;
       config.lossless = 0;
     } else if (!strcmp(argv[c], "-q") && c < argc - 1) {
-      config.quality = (float)strtod(argv[++c], NULL);
+      config.quality = ExUtilGetFloat(argv[++c], &parse_error);
     } else if (!strcmp(argv[c], "-m") && c < argc - 1) {
-      config.method = strtol(argv[++c], NULL, 0);
+      config.method = ExUtilGetInt(argv[++c], 0, &parse_error);
     } else if (!strcmp(argv[c], "-kmax") && c < argc - 1) {
-      kmax = strtoul(argv[++c], NULL, 0);
+      kmax = ExUtilGetUInt(argv[++c], 0, &parse_error);
       default_kmax = 0;
     } else if (!strcmp(argv[c], "-kmin") && c < argc - 1) {
-      kmin = strtoul(argv[++c], NULL, 0);
+      kmin = ExUtilGetUInt(argv[++c], 0, &parse_error);
       default_kmin = 0;
     } else if (!strcmp(argv[c], "-f") && c < argc - 1) {
-      config.filter_strength = strtol(argv[++c], NULL, 0);
+      config.filter_strength = ExUtilGetInt(argv[++c], 0, &parse_error);
     } else if (!strcmp(argv[c], "-metadata") && c < argc - 1) {
       static const struct {
         const char* option;
@@ -381,6 +391,11 @@ int main(int argc, const char *argv[]) {
     } else {
       in_file = argv[c];
     }
+
+    if (parse_error) {
+      Help();
+      return -1;
+    }
   }
 
   // Appropriate default kmin, kmax values for lossy and lossless.
@@ -469,6 +484,10 @@ int main(int argc, const char *argv[]) {
           cache = WebPFrameCacheNew(frame.width, frame.height,
                                     kmin, kmax, allow_mixed);
           if (cache == NULL) goto End;
+
+          // Background color.
+          GetBackgroundColor(gif->SColorMap, gif->SBackGroundColor,
+                             &anim.bgcolor);
         }
         // Some even more broken GIF can have sub-rect with zero width/height.
         if (image_desc->Width == 0 || image_desc->Height == 0) {
@@ -492,6 +511,11 @@ int main(int argc, const char *argv[]) {
           goto End;
         }
         is_first_frame = 0;
+
+        // In GIF, graphic control extensions are optional for a frame, so we
+        // may not get one before reading the next frame. To handle this case,
+        // we reset frame properties to reasonable defaults for the next frame.
+        ResetFrameInfo(&info);
         break;
       }
       case EXTENSION_RECORD_TYPE: {
@@ -525,13 +549,6 @@ int main(int argc, const char *argv[]) {
                                  : WEBP_MUX_DISPOSE_NONE;
             }
             transparent_index = (flags & GIF_TRANSPARENT_MASK) ? data[4] : -1;
-            if (is_first_frame) {
-              if (!GetBackgroundColor(gif->SColorMap, gif->SBackGroundColor,
-                                      &anim.bgcolor)) {
-                fprintf(stderr, "GIF decode warning: invalid background color "
-                                "index. Assuming white background.\n");
-              }
-            }
             break;
           }
           case PLAINTEXT_EXT_FUNC_CODE: {

examples/gif2webp_util.c

@@ -475,7 +475,7 @@ static WebPEncodingError SetFrame(const WebPConfig* const config,
     // TODO(later): Perhaps a rough SSIM/PSNR produced by the encoder should
     // also be a criteria, in addition to sizes.
     if (mem1.size <= mem2.size) {
-#if WEBP_ENCODER_ABI_VERSION > 0x0202
+#if WEBP_ENCODER_ABI_VERSION > 0x0203
       WebPMemoryWriterClear(&mem2);
 #else
       free(mem2.mem);
@@ -483,7 +483,7 @@ static WebPEncodingError SetFrame(const WebPConfig* const config,
 #endif
       GetEncodedData(&mem1, encoded_data);
     } else {
-#if WEBP_ENCODER_ABI_VERSION > 0x0202
+#if WEBP_ENCODER_ABI_VERSION > 0x0203
       WebPMemoryWriterClear(&mem1);
 #else
       free(mem1.mem);
@@ -497,7 +497,7 @@ static WebPEncodingError SetFrame(const WebPConfig* const config,
   return error_code;
 
  Err:
-#if WEBP_ENCODER_ABI_VERSION > 0x0202
+#if WEBP_ENCODER_ABI_VERSION > 0x0203
   WebPMemoryWriterClear(&mem1);
   WebPMemoryWriterClear(&mem2);
 #else

examples/vwebp.c

@@ -377,7 +377,7 @@ static void Help(void) {
          "  -nofancy ..... don't use the fancy YUV420 upscaler\n"
          "  -nofilter .... disable in-loop filtering\n"
          "  -dither <int>  dithering strength (0..100), default=50\n"
-#if WEBP_DECODER_ABI_VERSION > 0x0203
+#if WEBP_DECODER_ABI_VERSION > 0x0204
          "  -noalphadither disable alpha plane dithering\n"
 #endif
          "  -mt .......... use multi-threading\n"
@@ -402,12 +402,13 @@ int main(int argc, char *argv[]) {
     return -1;
   }
   config->options.dithering_strength = 50;
-#if WEBP_DECODER_ABI_VERSION > 0x0203
+#if WEBP_DECODER_ABI_VERSION > 0x0204
   config->options.alpha_dithering_strength = 100;
 #endif
   kParams.use_color_profile = 1;
 
   for (c = 1; c < argc; ++c) {
+    int parse_error = 0;
     if (!strcmp(argv[c], "-h") || !strcmp(argv[c], "-help")) {
       Help();
       return 0;
@@ -417,12 +418,13 @@ int main(int argc, char *argv[]) {
       config->options.no_fancy_upsampling = 1;
     } else if (!strcmp(argv[c], "-nofilter")) {
       config->options.bypass_filtering = 1;
-#if WEBP_DECODER_ABI_VERSION > 0x0203
+#if WEBP_DECODER_ABI_VERSION > 0x0204
     } else if (!strcmp(argv[c], "-noalphadither")) {
       config->options.alpha_dithering_strength = 0;
 #endif
     } else if (!strcmp(argv[c], "-dither") && c + 1 < argc) {
-      config->options.dithering_strength = strtol(argv[++c], NULL, 0);
+      config->options.dithering_strength =
+          ExUtilGetInt(argv[++c], 0, &parse_error);
     } else if (!strcmp(argv[c], "-info")) {
       kParams.print_info = 1;
     } else if (!strcmp(argv[c], "-version")) {
@@ -445,6 +447,11 @@ int main(int argc, char *argv[]) {
     } else {
       kParams.file_name = argv[c];
     }
+
+    if (parse_error) {
+      Help();
+      return -1;
+    }
   }
 
   if (kParams.file_name == NULL) {

examples/webpmux.c

@@ -402,8 +402,9 @@ static int CreateMux(const char* const filename, WebPMux** mux) {
 
 static int WriteData(const char* filename, const WebPData* const webpdata) {
   int ok = 0;
-  FILE* fout = strcmp(filename, "-") ? fopen(filename, "wb") : stdout;
-  if (!fout) {
+  FILE* fout = strcmp(filename, "-") ? fopen(filename, "wb")
+                                     : ExUtilSetBinaryMode(stdout);
+  if (fout == NULL) {
     fprintf(stderr, "Error opening output WebP file %s!\n", filename);
     return 0;
   }
@@ -487,7 +488,7 @@ static int ParseBgcolorArgs(const char* args, uint32_t* const bgcolor) {
 static void DeleteConfig(WebPMuxConfig* config) {
   if (config != NULL) {
     free(config->feature_.args_);
-    free(config);
+    memset(config, 0, sizeof(*config));
   }
 }
 
@@ -790,33 +791,27 @@ static int ValidateConfig(WebPMuxConfig* config) {
 
 // Create config object from command-line arguments.
 static int InitializeConfig(int argc, const char* argv[],
-                            WebPMuxConfig** config) {
+                            WebPMuxConfig* config) {
   int num_feature_args = 0;
   int ok = 1;
 
   assert(config != NULL);
-  *config = NULL;
+  memset(config, 0, sizeof(*config));
 
   // Validate command-line arguments.
   if (!ValidateCommandLine(argc, argv, &num_feature_args)) {
     ERROR_GOTO1("Exiting due to command-line parsing error.\n", Err1);
   }
 
-  // Allocate memory.
-  *config = (WebPMuxConfig*)calloc(1, sizeof(**config));
-  if (*config == NULL) {
-    ERROR_GOTO1("ERROR: Memory allocation error.\n", Err1);
-  }
-  (*config)->feature_.arg_count_ = num_feature_args;
-  (*config)->feature_.args_ =
-      (FeatureArg*)calloc(num_feature_args, sizeof(FeatureArg));
-  if ((*config)->feature_.args_ == NULL) {
+  config->feature_.arg_count_ = num_feature_args;
+  config->feature_.args_ =
+      (FeatureArg*)calloc(num_feature_args, sizeof(*config->feature_.args_));
+  if (config->feature_.args_ == NULL) {
     ERROR_GOTO1("ERROR: Memory allocation error.\n", Err1);
   }
 
   // Parse command-line.
-  if (!ParseCommandLine(argc, argv, *config) ||
-      !ValidateConfig(*config)) {
+  if (!ParseCommandLine(argc, argv, config) || !ValidateConfig(config)) {
     ERROR_GOTO1("Exiting due to command-line parsing error.\n", Err1);
   }
 
@@ -838,14 +833,16 @@ static int GetFrameFragment(const WebPMux* mux,
   WebPMux* mux_single = NULL;
   long num = 0;
   int ok = 1;
+  int parse_error = 0;
   const WebPChunkId id = is_frame ? WEBP_CHUNK_ANMF : WEBP_CHUNK_FRGM;
   WebPMuxFrameInfo info;
   WebPDataInit(&info.bitstream);
 
-  num = strtol(config->feature_.args_[0].params_, NULL, 10);
+  num = ExUtilGetInt(config->feature_.args_[0].params_, 10, &parse_error);
   if (num < 0) {
     ERROR_GOTO1("ERROR: Frame/Fragment index must be non-negative.\n", ErrGet);
   }
+  if (parse_error) goto ErrGet;
 
   err = WebPMuxGetFrame(mux, num, &info);
   if (err == WEBP_MUX_OK && info.id != id) err = WEBP_MUX_NOT_FOUND;
@@ -871,7 +868,7 @@ static int GetFrameFragment(const WebPMux* mux,
  ErrGet:
   WebPDataClear(&info.bitstream);
   WebPMuxDelete(mux_single);
-  return ok;
+  return ok && !parse_error;
 }
 
 // Read and process config.
@@ -933,16 +930,19 @@ static int Process(const WebPMuxConfig* config) {
                 break;
               }
               case SUBTYPE_LOOP: {
-                const long loop_count =
-                    strtol(feature->args_[i].params_, NULL, 10);
-                if (loop_count != (int)loop_count) {
+                int parse_error = 0;
+                const int loop_count =
+                    ExUtilGetInt(feature->args_[i].params_, 10, &parse_error);
+                if (loop_count < 0 || loop_count > 65535) {
                   // Note: This is only a 'necessary' condition for loop_count
                   // to be valid. The 'sufficient' conditioned in checked in
                   // WebPMuxSetAnimationParams() method called later.
                   ERROR_GOTO1("ERROR: Loop count must be in the range 0 to "
                               "65535.\n", Err2);
                 }
-                params.loop_count = (int)loop_count;
+                ok = !parse_error;
+                if (!ok) goto Err2;
+                params.loop_count = loop_count;
                 break;
               }
               case SUBTYPE_ANMF: {
@@ -1069,14 +1069,14 @@ static int Process(const WebPMuxConfig* config) {
 // Main.
 
 int main(int argc, const char* argv[]) {
-  WebPMuxConfig* config;
+  WebPMuxConfig config;
   int ok = InitializeConfig(argc - 1, argv + 1, &config);
   if (ok) {
-    ok = Process(config);
+    ok = Process(&config);
   } else {
     PrintHelp();
   }
-  DeleteConfig(config);
+  DeleteConfig(&config);
   return !ok;
 }
 

iosbuild.sh

@@ -12,31 +12,35 @@
 set -e
 
 # Extract the latest SDK version from the final field of the form: iphoneosX.Y
-declare -r SDK=$(xcodebuild -showsdks \
+readonly SDK=$(xcodebuild -showsdks \
   | grep iphoneos | sort | tail -n 1 | awk '{print substr($NF, 9)}'
 )
 # Extract Xcode version.
-declare -r XCODE=$(xcodebuild -version | grep Xcode | cut -d " " -f2)
+readonly XCODE=$(xcodebuild -version | grep Xcode | cut -d " " -f2)
+if [[ -z "${XCODE}" ]]; then
+  echo "Xcode not available"
+  exit 1
+fi
 
-declare -r OLDPATH=${PATH}
+readonly OLDPATH=${PATH}
 
 # Add iPhoneOS-V6 to the list of platforms below if you need armv6 support.
 # Note that iPhoneOS-V6 support is not available with the iOS6 SDK.
-declare -r PLATFORMS="iPhoneSimulator iPhoneOS-V7 iPhoneOS-V7s iPhoneOS-V7-arm64"
-declare -r SRCDIR=$(dirname $0)
-declare -r TOPDIR=$(pwd)
-declare -r BUILDDIR="${TOPDIR}/iosbuild"
-declare -r TARGETDIR="${TOPDIR}/WebP.framework"
-declare -r DEVELOPER=$(xcode-select --print-path)
-declare -r PLATFORMSROOT="${DEVELOPER}/Platforms"
-declare -r LIPO=$(xcrun -sdk iphoneos${SDK} -find lipo)
+readonly PLATFORMS="iPhoneSimulator iPhoneOS-V7 iPhoneOS-V7s iPhoneOS-V7-arm64"
+readonly SRCDIR=$(dirname $0)
+readonly TOPDIR=$(pwd)
+readonly BUILDDIR="${TOPDIR}/iosbuild"
+readonly TARGETDIR="${TOPDIR}/WebP.framework"
+readonly DEVELOPER=$(xcode-select --print-path)
+readonly PLATFORMSROOT="${DEVELOPER}/Platforms"
+readonly LIPO=$(xcrun -sdk iphoneos${SDK} -find lipo)
 LIBLIST=''
 
 if [[ -z "${SDK}" ]]; then
   echo "iOS SDK not available"
   exit 1
-elif [[ ${SDK} < 4.0 ]]; then
-  echo "You need iOS SDK version 4.0 or above"
+elif [[ ${SDK} < 6.0 ]]; then
+  echo "You need iOS SDK version 6.0 or above"
   exit 1
 else
   echo "iOS SDK Version ${SDK}"
@@ -47,7 +51,17 @@ rm -rf ${TARGETDIR}
 mkdir -p ${BUILDDIR}
 mkdir -p ${TARGETDIR}/Headers/
 
-[[ -e ${SRCDIR}/configure ]] || (cd ${SRCDIR} && sh autogen.sh)
+if [[ ! -e ${SRCDIR}/configure ]]; then
+  if ! (cd ${SRCDIR} && sh autogen.sh); then
+    cat <<EOT
+Error creating configure script!
+This script requires the autoconf/automake and libtool to build. MacPorts can
+be used to obtain these:
+http://www.macports.org/install.php
+EOT
+    exit 1
+  fi
+fi
 
 for PLATFORM in ${PLATFORMS}; do
   ARCH2=""
@@ -71,30 +85,20 @@ for PLATFORM in ${PLATFORMS}; do
   ROOTDIR="${BUILDDIR}/${PLATFORM}-${SDK}-${ARCH}"
   mkdir -p "${ROOTDIR}"
 
-  SDKROOT="${PLATFORMSROOT}/${PLATFORM}.platform/Developer/SDKs/${PLATFORM}${SDK}.sdk/"
-  CFLAGS="-arch ${ARCH2:-${ARCH}} -pipe -isysroot ${SDKROOT} -O3 -DNDEBUG"
-  LDFLAGS="-arch ${ARCH2:-${ARCH}} -pipe -isysroot ${SDKROOT}"
-
-  if [[ -z "${XCODE}" ]]; then
-    echo "XCODE not available"
-    exit 1
-  elif [[ ${SDK} < 5.0.0 ]]; then
-    DEVROOT="${PLATFORMSROOT}/${PLATFORM}.platform/Developer/"
-  else
   DEVROOT="${DEVELOPER}/Toolchains/XcodeDefault.xctoolchain"
-    CFLAGS+=" -miphoneos-version-min=5.0"
-    LDFLAGS+=" -miphoneos-version-min=5.0"
-  fi
+  SDKROOT="${PLATFORMSROOT}/"
+  SDKROOT+="${PLATFORM}.platform/Developer/SDKs/${PLATFORM}${SDK}.sdk/"
+  CFLAGS="-arch ${ARCH2:-${ARCH}} -pipe -isysroot ${SDKROOT} -O3 -DNDEBUG"
+  CFLAGS+=" -miphoneos-version-min=6.0"
 
-  export CFLAGS
-  export LDFLAGS
-  export CXXFLAGS=${CFLAGS}
+  set -x
   export PATH="${DEVROOT}/usr/bin:${OLDPATH}"
-
   ${SRCDIR}/configure --host=${ARCH}-apple-darwin --prefix=${ROOTDIR} \
     --build=$(${SRCDIR}/config.guess) \
     --disable-shared --enable-static \
-    --enable-libwebpdecoder --enable-swap-16bit-csp
+    --enable-libwebpdecoder --enable-swap-16bit-csp \
+    CFLAGS="${CFLAGS}"
+  set +x
 
   # run make only in the src/ directory to create libwebpdecoder.a
   cd src/
@@ -109,5 +113,5 @@ for PLATFORM in ${PLATFORMS}; do
   export PATH=${OLDPATH}
 done
 
-cp -a ${SRCDIR}/src/webp/* ${TARGETDIR}/Headers/
+cp -a ${SRCDIR}/src/webp/*.h ${TARGETDIR}/Headers/
 ${LIPO} -create ${LIBLIST} -output ${TARGETDIR}/WebP

makefile.unix

@@ -108,6 +108,7 @@ DEMUX_OBJS = \
 
 DSP_DEC_OBJS = \
     src/dsp/alpha_processing.o \
+    src/dsp/alpha_processing_sse2.o \
     src/dsp/cpu.o \
     src/dsp/dec.o \
     src/dsp/dec_clip_tables.o \

man/cwebp.1

@@ -1,5 +1,5 @@
 .\"                                      Hey, EMACS: -*- nroff -*-
-.TH CWEBP 1 "July 22, 2014"
+.TH CWEBP 1 "Oct 13, 2014"
 .SH NAME
 cwebp \- compress an image file to a WebP file
 .SH SYNOPSIS
@@ -38,6 +38,13 @@ Print the version number (as major.minor.revision) and exit.
 .BI \-q " float
 Specify the compression factor for RGB channels between 0 and 100. The default
 is 75.
+.br
+In case of lossy compression (default), a small factor produces a smaller file
+with lower quality. Best quality is achieved by using a value of 100.
+.br
+In case of lossless compression (specified by the \-lossless option), a small
+factor enables faster compression speed, but produces a larger file. Maximum
+compression is achieved by using a value of 100.
 .\" TODO(jzern): restore post-v0.4.1
 .\" .TP
 .\" .BI \-z " int
@@ -47,13 +54,6 @@ is 75.
 .\" This option is actually a shortcut for some predefined settings for quality
 .\" and method. If options \-q  or \-m are subsequently used, they will invalidate
 .\" the effect of this \-z option.
-.\" .br
-.\" In case of lossy compression (default), a small factor produces a smaller file
-.\" with lower quality. Best quality is achieved by using a value of 100.
-.\" .br
-.\" In case of lossless compression (specified by the \-lossless option), a small
-.\" factor enables faster compression speed, but produces a larger file. Maximum
-.\" compression is achieved by using a value of 100.
 .TP
 .BI \-alpha_q " int
 Specify the compression factor for alpha compression between 0 and 100.

man/webpmux.1

@@ -1,7 +1,8 @@
 .\"                                      Hey, EMACS: -*- nroff -*-
-.TH WEBPMUX 1 "December 17, 2013"
+.TH WEBPMUX 1 "August 28, 2014"
 .SH NAME
-webpmux \- command line tool to create WebP Mux/container file.
+webpmux \- create animated WebP files from non\-animated WebP images, extract
+frames from animated WebP images, and manage XMP/EXIF metadata and ICC profile.
 .SH SYNOPSIS
 .B webpmux \-get
 .I GET_OPTIONS
@@ -45,8 +46,8 @@ This manual page documents the
 .B webpmux
 command.
 .PP
-\fBwebpmux\fP can be used to create a WebP container file
-and extract/strip relevant data from the container file.
+\fBwebpmux\fP can be used to create/extract from animated WebP files, as well as
+to add/extract/strip XMP/EXIF metadata and ICC profile.
 .SH OPTIONS
 .SS GET_OPTIONS (\-get):
 .TP
@@ -60,7 +61,7 @@ Get EXIF metadata.
 Get XMP metadata.
 .TP
 .BI frame " n
-Get nth frame.
+Get nth frame from an animated image. (n = 0 has a special meaning: last frame).
 
 .SS SET_OPTIONS (\-set)
 .TP
@@ -91,12 +92,13 @@ Strip EXIF metadata.
 Strip XMP metadata.
 
 .SS FRAME_OPTIONS (\-frame)
+Create an animated WebP file from multiple (non\-animated) WebP images.
 .TP
 .I file_i +di[+xi+yi[+mi[bi]]]
 Where: 'file_i' is the i'th frame (WebP format), 'xi','yi' specify the image
 offset for this frame, 'di' is the pause duration before next frame, 'mi' is
 the dispose method for this frame (0 for NONE or 1 for BACKGROUND) and 'bi' is
-the blending method for this frame (+b for BLEND or -b for NO_BLEND).
+the blending method for this frame (+b for BLEND or \-b for NO_BLEND).
 Argument 'bi' can be omitted and will default to +b (BLEND).
 Also, 'mi' can be omitted if 'bi' is omitted and will default to 0 (NONE).
 Finally, if 'mi' and 'bi' are omitted then 'xi' and 'yi' can be omitted and will
@@ -130,37 +132,61 @@ Please report all bugs to our issue tracker:
 http://code.google.com/p/webp/issues
 .br
 Patches welcome! See this page to get started:
-http://www.webmproject.org/code/contribute/submitting-patches/
+http://www.webmproject.org/code/contribute/submitting\-patches/
 
 .SH EXAMPLES
+.P
+Add ICC profile:
+.br
 webpmux \-set icc image_profile.icc in.webp \-o icc_container.webp
+.P
+Extract ICC profile:
 .br
 webpmux \-get icc icc_container.webp \-o image_profile.icc
+.P
+Strip ICC profile:
 .br
 webpmux \-strip icc icc_container.webp \-o without_icc.webp
+.P
+Add XMP metadata:
 .br
 webpmux \-set xmp image_metadata.xmp in.webp \-o xmp_container.webp
+.P
+Extract XMP metadata:
 .br
 webpmux \-get xmp xmp_container.webp \-o image_metadata.xmp
+.P
+Strip XMP metadata:
 .br
 webpmux \-strip xmp xmp_container.webp \-o without_xmp.webp
+.P
+Add EXIF metadata:
 .br
 webpmux \-set exif image_metadata.exif in.webp \-o exif_container.webp
+.P
+Extract EXIF metadata:
 .br
 webpmux \-get exif exif_container.webp \-o image_metadata.exif
+.P
+Strip EXIF metadata:
 .br
 webpmux \-strip exif exif_container.webp \-o without_exif.webp
+.P
+Create an animated WebP file from 3 (non\-animated) WebP images:
 .br
-webpmux \-frame anim_1.webp +100 \-frame anim_2.webp +100+50+50
+webpmux \-frame 1.webp +100 \-frame 2.webp +100+50+50
 .br
 .RS 8
-\-frame anim_2.webp +100+50+50+1+b \-loop 10 \-bgcolor 255,255,255,255
+\-frame 3.webp +100+50+50+1+b \-loop 10 \-bgcolor 255,255,255,255
 .br
-.RS 8
 \-o anim_container.webp
 .RE
+.P
+Get the 2nd frame from an animated WebP file:
 .br
 webpmux \-get frame 2 anim_container.webp \-o frame_2.webp
+.P
+Using \-get/\-set/\-strip with input file name starting with '\-':
 .br
 webpmux \-set icc image_profile.icc \-o icc_container.webp \-\- \-\-\-in.webp
 .br

src/Makefile.am

@@ -35,7 +35,7 @@ libwebp_la_LIBADD += utils/libwebputils.la
 # other than the ones listed on the command line, i.e., after linking, it will
 # not have unresolved symbols. Some platforms (Windows among them) require all
 # symbols in shared libraries to be resolved at library creation.
-libwebp_la_LDFLAGS = -no-undefined -version-info 5:1:0
+libwebp_la_LDFLAGS = -no-undefined -version-info 5:2:0
 libwebpincludedir = $(includedir)/webp
 pkgconfig_DATA = libwebp.pc
 
@@ -47,7 +47,7 @@ if BUILD_LIBWEBPDECODER
   libwebpdecoder_la_LIBADD += dsp/libwebpdspdecode.la
   libwebpdecoder_la_LIBADD += utils/libwebputilsdecode.la
 
-  libwebpdecoder_la_LDFLAGS = -no-undefined -version-info 1:1:0
+  libwebpdecoder_la_LDFLAGS = -no-undefined -version-info 1:2:0
   pkgconfig_DATA += libwebpdecoder.pc
 endif
 

src/dec/frame.c

@@ -177,7 +177,7 @@ void VP8InitDithering(const WebPDecoderOptions* const options,
         dec->dither_ = 1;
       }
     }
-#if WEBP_DECODER_ABI_VERSION > 0x0203
+#if WEBP_DECODER_ABI_VERSION > 0x0204
     // potentially allow alpha dithering
     dec->alpha_dithering_ = options->alpha_dithering_strength;
     if (dec->alpha_dithering_ > 100) {

src/dec/idec.c

@@ -529,6 +529,12 @@ static VP8StatusCode DecodeVP8LData(WebPIDecoder* const idec) {
   }
 
   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_);
   }
 

src/dec/tree.c

@@ -11,7 +11,7 @@
 //
 // Author: Skal (pascal.massimino@gmail.com)
 
-#include "vp8i.h"
+#include "./vp8i.h"
 #include "../utils/bit_reader_inl.h"
 
 #define USE_GENERIC_TREE

src/dec/vp8i.h

@@ -31,7 +31,7 @@ extern "C" {
 // version numbers
 #define DEC_MAJ_VERSION 0
 #define DEC_MIN_VERSION 4
-#define DEC_REV_VERSION 1
+#define DEC_REV_VERSION 2
 
 // intra prediction modes
 enum { B_DC_PRED = 0,   // 4x4 modes

src/dec/vp8l.c

@@ -234,6 +234,7 @@ static int ReadHuffmanCodeLengths(
 
  End:
   VP8LHuffmanTreeFree(&tree);
+  if (!ok) dec->status_ = VP8_STATUS_BITSTREAM_ERROR;
   return ok;
 }
 
@@ -801,6 +802,7 @@ static int DecodeAlphaData(VP8LDecoder* const dec, uint8_t* const data,
       ok = 0;
       goto End;
     }
+    assert(br->eos_ == VP8LIsEndOfStream(br));
     ok = !br->error_;
     if (!ok) goto End;
   }
@@ -918,6 +920,7 @@ static int DecodeImageData(VP8LDecoder* const dec, uint32_t* const data,
       ok = 0;
       goto End;
     }
+    assert(br->eos_ == VP8LIsEndOfStream(br));
     ok = !br->error_;
     if (!ok) goto End;
   }
@@ -1354,6 +1357,10 @@ int VP8LDecodeImage(VP8LDecoder* const dec) {
   // Sanity checks.
   if (dec == NULL) return 0;
 
+  dec->status_ = VP8_STATUS_BITSTREAM_ERROR;
+  assert(dec->hdr_.htree_groups_ != NULL);
+  assert(dec->hdr_.num_htree_groups_ > 0);
+
   io = dec->io_;
   assert(io != NULL);
   params = (WebPDecParams*)io->opaque;

src/demux/Makefile.am

@@ -9,6 +9,6 @@ libwebpdemuxinclude_HEADERS += ../webp/mux_types.h
 libwebpdemuxinclude_HEADERS += ../webp/types.h
 
 libwebpdemux_la_LIBADD = ../libwebp.la
-libwebpdemux_la_LDFLAGS = -no-undefined -version-info 1:1:0
+libwebpdemux_la_LDFLAGS = -no-undefined -version-info 1:2:0
 libwebpdemuxincludedir = $(includedir)/webp
 pkgconfig_DATA = libwebpdemux.pc

src/demux/demux.c

@@ -25,7 +25,7 @@
 
 #define DMUX_MAJ_VERSION 0
 #define DMUX_MIN_VERSION 2
-#define DMUX_REV_VERSION 1
+#define DMUX_REV_VERSION 2
 
 typedef struct {
   size_t start_;        // start location of the data

src/dsp/Makefile.am

@@ -38,6 +38,7 @@ libwebpdsp_avx2_la_CPPFLAGS = $(libwebpdsp_la_CPPFLAGS)
 libwebpdsp_avx2_la_CFLAGS = $(AM_CFLAGS) $(AVX2_FLAGS)
 
 libwebpdspdecode_sse2_la_SOURCES =
+libwebpdspdecode_sse2_la_SOURCES += alpha_processing_sse2.c
 libwebpdspdecode_sse2_la_SOURCES += dec_sse2.c
 libwebpdspdecode_sse2_la_SOURCES += lossless_sse2.c
 libwebpdspdecode_sse2_la_SOURCES += upsampling_sse2.c

src/dsp/alpha_processing.c

@@ -284,15 +284,46 @@ static void ApplyAlphaMultiply_16b(uint8_t* rgba4444,
 #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
+  }
 }

src/dsp/alpha_processing_sse2.c

@@ -0,0 +1,77 @@
+// 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
+}

src/dsp/cpu.c

@@ -57,7 +57,7 @@ static WEBP_INLINE uint64_t xgetbv(void) {
 }
 #elif defined(_MSC_FULL_VER) && _MSC_FULL_VER >= 160040219  // >= VS2010 SP1
 #define xgetbv() _xgetbv(0)
-#elif defined(_M_IX86)
+#elif defined(_MSC_VER) && defined(_M_IX86)
 static WEBP_INLINE uint64_t xgetbv(void) {
   uint32_t eax_, edx_;
   __asm {
@@ -118,7 +118,7 @@ static int armCPUInfo(CPUFeature feature) {
   return 1;
 }
 VP8CPUInfo VP8GetCPUInfo = armCPUInfo;
-#elif defined(__mips__)
+#elif defined(WEBP_USE_MIPS32)
 static int mipsCPUInfo(CPUFeature feature) {
   (void)feature;
   return 1;

src/dsp/dsp.h

@@ -32,9 +32,19 @@ extern "C" {
 # 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
@@ -60,8 +70,11 @@ extern "C" {
 #define WEBP_USE_NEON
 #endif
 
-#if defined(__mips__)
+#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 {
@@ -244,6 +257,13 @@ extern void (*WebPApplyAlphaMultiply)(
 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.
 

src/dsp/enc_mips32.c

@@ -20,6 +20,10 @@
 #include "../enc/vp8enci.h"
 #include "../enc/cost.h"
 
+#if defined(__GNUC__) && defined(__ANDROID__) && LOCAL_GCC_VERSION == 0x409
+#define WORK_AROUND_GCC
+#endif
+
 static const int kC1 = 20091 + (1 << 16);
 static const int kC2 = 35468;
 
@@ -641,6 +645,7 @@ int VP8GetResidualCostMIPS32(int ctx0, const VP8Residual* const res) {
   GET_SSE_INNER(C, C + 1, C + 2, C + 3)   \
   GET_SSE_INNER(D, D + 1, D + 2, D + 3)
 
+#if !defined(WORK_AROUND_GCC)
 static int SSE16x16(const uint8_t* a, const uint8_t* b) {
   int count;
   int temp0, temp1, temp2, temp3, temp4, temp5, temp6, temp7;
@@ -742,6 +747,8 @@ static int SSE4x4(const uint8_t* a, const uint8_t* b) {
   return count;
 }
 
+#endif  // WORK_AROUND_GCC
+
 #undef GET_SSE_MIPS32
 #undef GET_SSE_MIPS32_INNER
 
@@ -759,9 +766,11 @@ void VP8EncDspInitMIPS32(void) {
   VP8TDisto4x4 = Disto4x4;
   VP8TDisto16x16 = Disto16x16;
   VP8FTransform = FTransform;
+#if !defined(WORK_AROUND_GCC)
   VP8SSE16x16 = SSE16x16;
   VP8SSE8x8 = SSE8x8;
   VP8SSE16x8 = SSE16x8;
   VP8SSE4x4 = SSE4x4;
+#endif
 #endif  // WEBP_USE_MIPS32
 }

src/dsp/enc_neon.c

@@ -253,7 +253,7 @@ static void ITransform(const uint8_t* ref,
 
 // Load all 4x4 pixels into a single uint8x16_t variable.
 static uint8x16_t Load4x4(const uint8_t* src) {
-  uint32x4_t out = { 0, 0, 0, 0 };
+  uint32x4_t out = vdupq_n_u32(0);
   out = vld1q_lane_u32((const uint32_t*)(src + 0 * BPS), out, 0);
   out = vld1q_lane_u32((const uint32_t*)(src + 1 * BPS), out, 1);
   out = vld1q_lane_u32((const uint32_t*)(src + 2 * BPS), out, 2);
@@ -929,7 +929,7 @@ static int SumToInt(uint32x4_t sum) {
 }
 
 static int SSE16x16(const uint8_t* a, const uint8_t* b) {
-  uint32x4_t sum = { 0, 0, 0, 0 };
+  uint32x4_t sum = vdupq_n_u32(0);
   int y;
   for (y = 0; y < 16; ++y) {
     AccumulateSSE16(a + y * BPS, b + y * BPS, &sum);
@@ -938,7 +938,7 @@ static int SSE16x16(const uint8_t* a, const uint8_t* b) {
 }
 
 static int SSE16x8(const uint8_t* a, const uint8_t* b) {
-  uint32x4_t sum = { 0, 0, 0, 0 };
+  uint32x4_t sum = vdupq_n_u32(0);
   int y;
   for (y = 0; y < 8; ++y) {
     AccumulateSSE16(a + y * BPS, b + y * BPS, &sum);
@@ -947,7 +947,7 @@ static int SSE16x8(const uint8_t* a, const uint8_t* b) {
 }
 
 static int SSE8x8(const uint8_t* a, const uint8_t* b) {
-  uint32x4_t sum = { 0, 0, 0, 0 };
+  uint32x4_t sum = vdupq_n_u32(0);
   int y;
   for (y = 0; y < 8; ++y) {
     const uint8x8_t a0 = vld1_u8(a + y * BPS);
@@ -970,9 +970,8 @@ static int SSE4x4(const uint8_t* a, const uint8_t* b) {
 
 //------------------------------------------------------------------------------
 
-// Compilation with gcc-4.6.x is problematic for now and vtbl? are unavailable
-// in iOS/arm64 builds. Disable this function in those cases.
-#if !(defined(WORK_AROUND_GCC) || defined(__aarch64__))
+// Compilation with gcc-4.6.x is problematic for now.
+#if !defined(WORK_AROUND_GCC)
 
 static int16x8_t Quantize(int16_t* const in,
                           const VP8Matrix* const mtx, int offset) {
@@ -1012,17 +1011,34 @@ static int QuantizeBlock(int16_t in[16], int16_t out[16],
                          const VP8Matrix* const mtx) {
   const int16x8_t out0 = Quantize(in, mtx, 0);
   const int16x8_t out1 = Quantize(in, mtx, 8);
+  uint8x8x4_t shuffles;
+  // vtbl4_u8 is marked unavailable for iOS arm64, use wider versions there.
+#if defined(__APPLE__) && defined(__aarch64__)
+  uint8x16x2_t all_out;
+  INIT_VECTOR2(all_out, vreinterpretq_u8_s16(out0), vreinterpretq_u8_s16(out1));
+  INIT_VECTOR4(shuffles,
+               vtbl2q_u8(all_out, vld1_u8(kShuffles[0])),
+               vtbl2q_u8(all_out, vld1_u8(kShuffles[1])),
+               vtbl2q_u8(all_out, vld1_u8(kShuffles[2])),
+               vtbl2q_u8(all_out, vld1_u8(kShuffles[3])));
+#else
   uint8x8x4_t all_out;
   INIT_VECTOR4(all_out,
                vreinterpret_u8_s16(vget_low_s16(out0)),
                vreinterpret_u8_s16(vget_high_s16(out0)),
                vreinterpret_u8_s16(vget_low_s16(out1)),
                vreinterpret_u8_s16(vget_high_s16(out1)));
+  INIT_VECTOR4(shuffles,
+               vtbl4_u8(all_out, vld1_u8(kShuffles[0])),
+               vtbl4_u8(all_out, vld1_u8(kShuffles[1])),
+               vtbl4_u8(all_out, vld1_u8(kShuffles[2])),
+               vtbl4_u8(all_out, vld1_u8(kShuffles[3])));
+#endif
   // Zigzag reordering
-  vst1_u8((uint8_t*)(out +  0), vtbl4_u8(all_out, vld1_u8(kShuffles[0])));
-  vst1_u8((uint8_t*)(out +  4), vtbl4_u8(all_out, vld1_u8(kShuffles[1])));
-  vst1_u8((uint8_t*)(out +  8), vtbl4_u8(all_out, vld1_u8(kShuffles[2])));
-  vst1_u8((uint8_t*)(out + 12), vtbl4_u8(all_out, vld1_u8(kShuffles[3])));
+  vst1_u8((uint8_t*)(out +  0), shuffles.val[0]);
+  vst1_u8((uint8_t*)(out +  4), shuffles.val[1]);
+  vst1_u8((uint8_t*)(out +  8), shuffles.val[2]);
+  vst1_u8((uint8_t*)(out + 12), shuffles.val[3]);
   // test zeros
   if (*(uint64_t*)(out +  0) != 0) return 1;
   if (*(uint64_t*)(out +  4) != 0) return 1;
@@ -1031,7 +1047,7 @@ static int QuantizeBlock(int16_t in[16], int16_t out[16],
   return 0;
 }
 
-#endif   // !WORK_AROUND_GCC && !__aarch64__
+#endif   // !WORK_AROUND_GCC
 
 #endif   // WEBP_USE_NEON
 
@@ -1054,7 +1070,7 @@ void VP8EncDspInitNEON(void) {
   VP8SSE16x8 = SSE16x8;
   VP8SSE8x8 = SSE8x8;
   VP8SSE4x4 = SSE4x4;
-#if !(defined(WORK_AROUND_GCC) || defined(__aarch64__))
+#if !defined(WORK_AROUND_GCC)
   VP8EncQuantizeBlock = QuantizeBlock;
 #endif
 #endif   // WEBP_USE_NEON

src/dsp/lossless.c

@@ -450,12 +450,21 @@ static WEBP_INLINE uint32_t ClampedAddSubtractHalf(uint32_t c0, uint32_t c1,
   return ((uint32_t)a << 24) | (r << 16) | (g << 8) | b;
 }
 
-static WEBP_INLINE int Sub3(int a, int b, int c) {
+// gcc-4.9 on ARM generates incorrect code in Select() when Sub3() is inlined.
+#if defined(__arm__) && LOCAL_GCC_VERSION == 0x409
+# define LOCAL_INLINE __attribute__ ((noinline))
+#else
+# define LOCAL_INLINE WEBP_INLINE
+#endif
+
+static LOCAL_INLINE int Sub3(int a, int b, int c) {
   const int pb = b - c;
   const int pa = a - c;
   return abs(pb) - abs(pa);
 }
 
+#undef LOCAL_INLINE
+
 static WEBP_INLINE uint32_t Select(uint32_t a, uint32_t b, uint32_t c) {
   const int pa_minus_pb =
       Sub3((a >> 24)       , (b >> 24)       , (c >> 24)       ) +
@@ -1169,7 +1178,7 @@ static void ColorSpaceInverseTransform(const VP8LTransform* const transform,
       data += remaining_width;
     }
     ++y;
-    if ((y & mask) == 0) pred_row += tiles_per_row;;
+    if ((y & mask) == 0) pred_row += tiles_per_row;
   }
 }
 

src/dsp/lossless.h

@@ -56,24 +56,20 @@ extern VP8LConvertFunc VP8LConvertBGRAToRGB565;
 extern VP8LConvertFunc VP8LConvertBGRAToBGR;
 
 // Expose some C-only fallback functions
-extern void VP8LTransformColor_C(const VP8LMultipliers* const m,
+void VP8LTransformColor_C(const VP8LMultipliers* const m,
                           uint32_t* data, int num_pixels);
-extern void VP8LTransformColorInverse_C(const VP8LMultipliers* const m,
+void VP8LTransformColorInverse_C(const VP8LMultipliers* const m,
                                  uint32_t* data, int num_pixels);
 
-extern void VP8LConvertBGRAToRGB_C(const uint32_t* src,
+void VP8LConvertBGRAToRGB_C(const uint32_t* src, int num_pixels, uint8_t* dst);
+void VP8LConvertBGRAToRGBA_C(const uint32_t* src, int num_pixels, uint8_t* dst);
+void VP8LConvertBGRAToRGBA4444_C(const uint32_t* src,
                                  int num_pixels, uint8_t* dst);
-extern void VP8LConvertBGRAToRGBA_C(const uint32_t* src,
+void VP8LConvertBGRAToRGB565_C(const uint32_t* src,
                                int num_pixels, uint8_t* dst);
-extern void VP8LConvertBGRAToRGBA4444_C(const uint32_t* src,
-                                        int num_pixels, uint8_t* dst);
-extern void VP8LConvertBGRAToRGB565_C(const uint32_t* src,
-                                      int num_pixels, uint8_t* dst);
-extern void VP8LConvertBGRAToBGR_C(const uint32_t* src,
-                                   int num_pixels, uint8_t* dst);
-extern void VP8LSubtractGreenFromBlueAndRed_C(uint32_t* argb_data,
-                                              int num_pixels);
-extern void VP8LAddGreenToBlueAndRed_C(uint32_t* data, int num_pixels);
+void VP8LConvertBGRAToBGR_C(const uint32_t* src, int num_pixels, uint8_t* dst);
+void VP8LSubtractGreenFromBlueAndRed_C(uint32_t* argb_data, int num_pixels);
+void VP8LAddGreenToBlueAndRed_C(uint32_t* data, int num_pixels);
 
 // Must be called before calling any of the above methods.
 void VP8LDspInit(void);

src/enc/alpha.c

@@ -47,12 +47,11 @@
 
 static int EncodeLossless(const uint8_t* const data, int width, int height,
                           int effort_level,  // in [0..6] range
-                          VP8BitWriter* const bw,
+                          VP8LBitWriter* const bw,
                           WebPAuxStats* const stats) {
   int ok = 0;
   WebPConfig config;
   WebPPicture picture;
-  VP8LBitWriter tmp_bw;
 
   WebPPictureInit(&picture);
   picture.width = width;
@@ -84,16 +83,15 @@ static int EncodeLossless(const uint8_t* const data, int width, int height,
   config.quality = 8.f * effort_level;
   assert(config.quality >= 0 && config.quality <= 100.f);
 
-  ok = VP8LBitWriterInit(&tmp_bw, (width * height) >> 3);
-  ok = ok && (VP8LEncodeStream(&config, &picture, &tmp_bw) == VP8_ENC_OK);
+  ok = (VP8LEncodeStream(&config, &picture, bw) == VP8_ENC_OK);
   WebPPictureFree(&picture);
-  if (ok) {
-    const uint8_t* const buffer = VP8LBitWriterFinish(&tmp_bw);
-    const size_t buffer_size = VP8LBitWriterNumBytes(&tmp_bw);
-    VP8BitWriterAppend(bw, buffer, buffer_size);
+  ok = ok && !bw->error_;
+  if (!ok) {
+    VP8LBitWriterDestroy(bw);
+    return 0;
   }
-  VP8LBitWriterDestroy(&tmp_bw);
-  return ok && !bw->error_;
+  return 1;
+
 }
 
 // -----------------------------------------------------------------------------
@@ -115,8 +113,10 @@ static int EncodeAlphaInternal(const uint8_t* const data, int width, int height,
   const uint8_t* alpha_src;
   WebPFilterFunc filter_func;
   uint8_t header;
-  size_t expected_size;
   const size_t data_size = width * height;
+  const uint8_t* output = NULL;
+  size_t output_size = 0;
+  VP8LBitWriter tmp_bw;
 
   assert((uint64_t)data_size == (uint64_t)width * height);  // as per spec
   assert(filter >= 0 && filter < WEBP_FILTER_LAST);
@@ -125,15 +125,6 @@ static int EncodeAlphaInternal(const uint8_t* const data, int width, int height,
   assert(sizeof(header) == ALPHA_HEADER_LEN);
   // TODO(skal): have a common function and #define's to validate alpha params.
 
-  expected_size =
-      (method == ALPHA_NO_COMPRESSION) ? (ALPHA_HEADER_LEN + data_size)
-                                       : (data_size >> 5);
-  header = method | (filter << 2);
-  if (reduce_levels) header |= ALPHA_PREPROCESSED_LEVELS << 4;
-
-  VP8BitWriterInit(&result->bw, expected_size);
-  VP8BitWriterAppend(&result->bw, &header, ALPHA_HEADER_LEN);
-
   filter_func = WebPFilters[filter];
   if (filter_func != NULL) {
     filter_func(data, width, height, width, tmp_alpha);
@@ -142,14 +133,42 @@ static int EncodeAlphaInternal(const uint8_t* const data, int width, int height,
     alpha_src = data;
   }
 
-  if (method == ALPHA_NO_COMPRESSION) {
-    ok = VP8BitWriterAppend(&result->bw, alpha_src, width * height);
-    ok = ok && !result->bw.error_;
-  } else {
-    ok = EncodeLossless(alpha_src, width, height, effort_level,
-                        &result->bw, &result->stats);
-    VP8BitWriterFinish(&result->bw);
+  if (method != ALPHA_NO_COMPRESSION) {
+    ok = VP8LBitWriterInit(&tmp_bw, data_size >> 3);
+    ok = ok && EncodeLossless(alpha_src, width, height, effort_level,
+                              &tmp_bw, &result->stats);
+    if (ok) {
+      output = VP8LBitWriterFinish(&tmp_bw);
+      output_size = VP8LBitWriterNumBytes(&tmp_bw);
+      if (output_size > data_size) {
+        // compressed size is larger than source! Revert to uncompressed mode.
+        method = ALPHA_NO_COMPRESSION;
+        VP8LBitWriterDestroy(&tmp_bw);
       }
+    } else {
+      VP8LBitWriterDestroy(&tmp_bw);
+      return 0;
+    }
+  }
+
+  if (method == ALPHA_NO_COMPRESSION) {
+    output = alpha_src;
+    output_size = data_size;
+    ok = 1;
+  }
+
+  // Emit final result.
+  header = method | (filter << 2);
+  if (reduce_levels) header |= ALPHA_PREPROCESSED_LEVELS << 4;
+
+  VP8BitWriterInit(&result->bw, ALPHA_HEADER_LEN + output_size);
+  ok = ok && VP8BitWriterAppend(&result->bw, &header, ALPHA_HEADER_LEN);
+  ok = ok && VP8BitWriterAppend(&result->bw, output, output_size);
+
+  if (method != ALPHA_NO_COMPRESSION) {
+    VP8LBitWriterDestroy(&tmp_bw);
+  }
+  ok = ok && !result->bw.error_;
   result->score = VP8BitWriterSize(&result->bw);
   return ok;
 }

src/enc/analysis.c

@@ -141,7 +141,11 @@ static void MergeHistograms(const VP8Histogram* const in,
 
 static void AssignSegments(VP8Encoder* const enc,
                            const int alphas[MAX_ALPHA + 1]) {
-  const int nb = enc->segment_hdr_.num_segments_;
+  // 'num_segments_' is previously validated and <= NUM_MB_SEGMENTS, but an
+  // explicit check is needed to avoid spurious warning about 'n + 1' exceeding
+  // array bounds of 'centers' with some compilers (noticed with gcc-4.9).
+  const int nb = (enc->segment_hdr_.num_segments_ < NUM_MB_SEGMENTS) ?
+                 enc->segment_hdr_.num_segments_ : NUM_MB_SEGMENTS;
   int centers[NUM_MB_SEGMENTS];
   int weighted_average = 0;
   int map[MAX_ALPHA + 1];

src/enc/config.c

@@ -111,7 +111,11 @@ int WebPValidateConfig(const WebPConfig* config) {
     return 0;
   if (config->show_compressed < 0 || config->show_compressed > 1)
     return 0;
+#if WEBP_ENCODER_ABI_VERSION > 0x0204
+  if (config->preprocessing < 0 || config->preprocessing > 7)
+#else
   if (config->preprocessing < 0 || config->preprocessing > 3)
+#endif
     return 0;
   if (config->partitions < 0 || config->partitions > 3)
     return 0;

src/enc/cost.h

@@ -42,7 +42,7 @@ typedef void (*VP8SetResidualCoeffsFunc)(const int16_t* const coeffs,
                                          VP8Residual* const res);
 extern VP8SetResidualCoeffsFunc VP8SetResidualCoeffs;
 
-extern void VP8SetResidualCoeffsInit(void);  // must be called first
+void VP8SetResidualCoeffsInit(void);  // must be called first
 
 int VP8RecordCoeffs(int ctx, const VP8Residual* const res);
 
@@ -59,7 +59,7 @@ static WEBP_INLINE int VP8BitCost(int bit, uint8_t proba) {
 typedef int (*VP8GetResidualCostFunc)(int ctx0, const VP8Residual* const res);
 extern VP8GetResidualCostFunc VP8GetResidualCost;
 
-extern void VP8GetResidualCostInit(void);  // must be called first
+void VP8GetResidualCostInit(void);  // must be called first
 
 // Level cost calculations
 extern const uint16_t VP8LevelCodes[MAX_VARIABLE_LEVEL][2];

src/enc/frame.c

@@ -508,7 +508,7 @@ static void StoreSideInfo(const VP8EncIterator* const it) {
       }
       case 7: *info = mb->alpha_; break;
       default: *info = 0; break;
-    };
+    }
   }
 #if SEGMENT_VISU  // visualize segments and prediction modes
   SetBlock(it->yuv_out_ + Y_OFF, mb->segment_ * 64, 16);

src/enc/vp8enci.h

@@ -30,7 +30,7 @@ extern "C" {
 // version numbers
 #define ENC_MAJ_VERSION 0
 #define ENC_MIN_VERSION 4
-#define ENC_REV_VERSION 1
+#define ENC_REV_VERSION 2
 
 // intra prediction modes
 enum { B_DC_PRED = 0,   // 4x4 modes
@@ -571,7 +571,7 @@ int WebPPictureAllocYUVA(WebPPicture* const picture, int width, int height);
 
 //------------------------------------------------------------------------------
 
-#if WEBP_ENCODER_ABI_VERSION <= 0x0202
+#if WEBP_ENCODER_ABI_VERSION <= 0x0203
 void WebPMemoryWriterClear(WebPMemoryWriter* writer);
 #endif
 

src/enc/webpenc.c

@@ -328,6 +328,13 @@ int WebPEncode(const WebPConfig* config, WebPPicture* pic) {
     VP8Encoder* enc = NULL;
     if (pic->y == NULL || pic->u == NULL || pic->v == NULL) {
       // Make sure we have YUVA samples.
+      if (config->preprocessing & 4) {
+#if WEBP_ENCODER_ABI_VERSION > 0x0204
+        if (!WebPPictureSmartARGBToYUVA(pic)) {
+          return 0;
+        }
+#endif
+      } else {
         float dithering = 0.f;
         if (config->preprocessing & 2) {
           const float x = config->quality / 100.f;
@@ -340,6 +347,7 @@ int WebPEncode(const WebPConfig* config, WebPPicture* pic) {
           return 0;
         }
       }
+    }
 
     enc = InitVP8Encoder(config, pic);
     if (enc == NULL) return 0;  // pic->error is already set.

src/mux/Makefile.am

@@ -12,6 +12,6 @@ libwebpmuxinclude_HEADERS += ../webp/mux_types.h
 libwebpmuxinclude_HEADERS += ../webp/types.h
 
 libwebpmux_la_LIBADD = ../libwebp.la
-libwebpmux_la_LDFLAGS = -no-undefined -version-info 1:1:0
+libwebpmux_la_LDFLAGS = -no-undefined -version-info 1:2:0
 libwebpmuxincludedir = $(includedir)/webp
 pkgconfig_DATA = libwebpmux.pc

src/mux/muxi.h

@@ -28,7 +28,7 @@ extern "C" {
 
 #define MUX_MAJ_VERSION 0
 #define MUX_MIN_VERSION 2
-#define MUX_REV_VERSION 1
+#define MUX_REV_VERSION 2
 
 // Chunk object.
 typedef struct WebPChunk WebPChunk;

src/utils/bit_reader.c

@@ -105,9 +105,7 @@ int32_t VP8GetSignedValue(VP8BitReader* const br, int bits) {
 //------------------------------------------------------------------------------
 // VP8LBitReader
 
-#define LBITS 64      // Number of bits prefetched.
-#define WBITS 32      // Minimum number of bytes needed after VP8LFillBitWindow.
-#define LOG8_WBITS 4  // Number of bytes needed to store WBITS bits.
+#define VP8L_LOG8_WBITS 4  // Number of bytes needed to store VP8L_WBITS bits.
 
 #if !defined(WEBP_FORCE_ALIGNED) && \
     (defined(__arm__) || defined(_M_ARM) || defined(__aarch64__) || \
@@ -151,16 +149,6 @@ void VP8LInitBitReader(VP8LBitReader* const br, const uint8_t* const start,
   br->buf_ = start;
 }
 
-// Special version that assumes br->pos_ <= br_len_.
-static int IsEndOfStreamSpecial(const VP8LBitReader* const br) {
-  assert(br->pos_ <= br->len_);
-  return br->pos_ == br->len_ && br->bit_pos_ >= LBITS;
-}
-
-static int IsEndOfStream(const VP8LBitReader* const br) {
-  return (br->pos_ > br->len_) || IsEndOfStreamSpecial(br);
-}
-
 void VP8LBitReaderSetBuffer(VP8LBitReader* const br,
                             const uint8_t* const buf, size_t len) {
   assert(br != NULL);
@@ -168,38 +156,39 @@ void VP8LBitReaderSetBuffer(VP8LBitReader* const br,
   assert(len < 0xfffffff8u);   // can't happen with a RIFF chunk.
   br->buf_ = buf;
   br->len_ = len;
-  br->eos_ = IsEndOfStream(br);
+  // pos_ > len_ should be considered a param error.
+  br->error_ = (br->pos_ > br->len_);
+  br->eos_ = br->error_ || VP8LIsEndOfStream(br);
 }
 
-// If not at EOS, reload up to LBITS byte-by-byte
+// If not at EOS, reload up to VP8L_LBITS byte-by-byte
 static void ShiftBytes(VP8LBitReader* const br) {
   while (br->bit_pos_ >= 8 && br->pos_ < br->len_) {
     br->val_ >>= 8;
-    br->val_ |= ((vp8l_val_t)br->buf_[br->pos_]) << (LBITS - 8);
+    br->val_ |= ((vp8l_val_t)br->buf_[br->pos_]) << (VP8L_LBITS - 8);
     ++br->pos_;
     br->bit_pos_ -= 8;
   }
+  br->eos_ = VP8LIsEndOfStream(br);
 }
 
-void VP8LFillBitWindow(VP8LBitReader* const br) {
-  if (br->bit_pos_ >= WBITS) {
+void VP8LDoFillBitWindow(VP8LBitReader* const br) {
+  assert(br->bit_pos_ >= VP8L_WBITS);
   // TODO(jzern): given the fixed read size it may be possible to force
   //              alignment in this block.
 #if defined(VP8L_USE_UNALIGNED_LOAD)
   if (br->pos_ + sizeof(br->val_) < br->len_) {
-      br->val_ >>= WBITS;
-      br->bit_pos_ -= WBITS;
+    br->val_ >>= VP8L_WBITS;
+    br->bit_pos_ -= VP8L_WBITS;
     // The expression below needs a little-endian arch to work correctly.
     // This gives a large speedup for decoding speed.
     br->val_ |= (vp8l_val_t)*(const uint32_t*)(br->buf_ + br->pos_) <<
-                  (LBITS - WBITS);
-      br->pos_ += LOG8_WBITS;
+                (VP8L_LBITS - VP8L_WBITS);
+    br->pos_ += VP8L_LOG8_WBITS;
     return;
   }
 #endif
   ShiftBytes(br);       // Slow path.
-    br->eos_ = IsEndOfStreamSpecial(br);
-  }
 }
 
 uint32_t VP8LReadBits(VP8LBitReader* const br, int n_bits) {
@@ -210,8 +199,6 @@ uint32_t VP8LReadBits(VP8LBitReader* const br, int n_bits) {
         (uint32_t)(br->val_ >> br->bit_pos_) & kBitMask[n_bits];
     const int new_bits = br->bit_pos_ + n_bits;
     br->bit_pos_ = new_bits;
-    // If this read is going to cross the read buffer, set the eos flag.
-    br->eos_ = IsEndOfStreamSpecial(br);
     ShiftBytes(br);
     return val;
   } else {

src/utils/bit_reader.h

@@ -107,6 +107,9 @@ int32_t VP8GetSignedValue(VP8BitReader* const br, int num_bits);
 // maximum number of bits (inclusive) the bit-reader can handle:
 #define VP8L_MAX_NUM_BIT_READ 24
 
+#define VP8L_LBITS 64  // Number of bits prefetched.
+#define VP8L_WBITS 32  // Minimum number of bytes ready after VP8LFillBitWindow.
+
 typedef uint64_t vp8l_val_t;  // right now, this bit-reader can only use 64bit.
 
 typedef struct {
@@ -138,14 +141,26 @@ static WEBP_INLINE uint32_t VP8LPrefetchBits(VP8LBitReader* const br) {
   return (uint32_t)(br->val_ >> br->bit_pos_);
 }
 
+// Returns true if there was an attempt at reading bit past the end of
+// the buffer. Doesn't set br->eos_ flag.
+static WEBP_INLINE int VP8LIsEndOfStream(const VP8LBitReader* const br) {
+  assert(br->pos_ <= br->len_);
+  return (br->pos_ == br->len_) && (br->bit_pos_ > VP8L_LBITS);
+}
+
 // For jumping over a number of bits in the bit stream when accessed with
 // VP8LPrefetchBits and VP8LFillBitWindow.
 static WEBP_INLINE void VP8LSetBitPos(VP8LBitReader* const br, int val) {
   br->bit_pos_ = val;
+  br->eos_ = VP8LIsEndOfStream(br);
 }
 
 // Advances the read buffer by 4 bytes to make room for reading next 32 bits.
-void VP8LFillBitWindow(VP8LBitReader* const br);
+// Speed critical, but infrequent part of the code can be non-inlined.
+extern void VP8LDoFillBitWindow(VP8LBitReader* const br);
+static WEBP_INLINE void VP8LFillBitWindow(VP8LBitReader* const br) {
+  if (br->bit_pos_ >= VP8L_WBITS) VP8LDoFillBitWindow(br);
+}
 
 #ifdef __cplusplus
 }    // extern "C"

src/utils/bit_reader_inl.h

@@ -24,6 +24,7 @@
 #include <string.h>  // memcpy
 #endif
 
+#include "../dsp/dsp.h"
 #include "./bit_reader.h"
 #include "./endian_inl.h"
 
@@ -63,7 +64,7 @@ static WEBP_INLINE void VP8LoadNewBytes(VP8BitReader* const br) {
 #if defined(WEBP_FORCE_ALIGNED)
     lbit_t in_bits;
     memcpy(&in_bits, br->buf_, sizeof(in_bits));
-#elif defined(__mips__)                        // MIPS
+#elif defined(WEBP_USE_MIPS32)
     // This is needed because of un-aligned read.
     lbit_t in_bits;
     lbit_t* p_buf_ = (lbit_t*)br->buf_;

src/utils/bit_writer.c

@@ -52,7 +52,7 @@ static int BitWriterResize(VP8BitWriter* const bw, size_t extra_size) {
   return 1;
 }
 
-static void kFlush(VP8BitWriter* const bw) {
+static void Flush(VP8BitWriter* const bw) {
   const int s = 8 + bw->nb_bits_;
   const int32_t bits = bw->value_ >> s;
   assert(bw->nb_bits_ >= 0);
@@ -118,7 +118,7 @@ int VP8PutBit(VP8BitWriter* const bw, int bit, int prob) {
     bw->range_ = kNewRange[bw->range_];
     bw->value_ <<= shift;
     bw->nb_bits_ += shift;
-    if (bw->nb_bits_ > 0) kFlush(bw);
+    if (bw->nb_bits_ > 0) Flush(bw);
   }
   return bit;
 }
@@ -135,7 +135,7 @@ int VP8PutBitUniform(VP8BitWriter* const bw, int bit) {
     bw->range_ = kNewRange[bw->range_];
     bw->value_ <<= 1;
     bw->nb_bits_ += 1;
-    if (bw->nb_bits_ > 0) kFlush(bw);
+    if (bw->nb_bits_ > 0) Flush(bw);
   }
   return bit;
 }
@@ -173,14 +173,14 @@ int VP8BitWriterInit(VP8BitWriter* const bw, size_t expected_size) {
 uint8_t* VP8BitWriterFinish(VP8BitWriter* const bw) {
   VP8PutValue(bw, 0, 9 - bw->nb_bits_);
   bw->nb_bits_ = 0;   // pad with zeroes
-  kFlush(bw);
+  Flush(bw);
   return bw->buf_;
 }
 
 int VP8BitWriterAppend(VP8BitWriter* const bw,
                        const uint8_t* data, size_t size) {
   assert(data != NULL);
-  if (bw->nb_bits_ != -8) return 0;   // kFlush() must have been called
+  if (bw->nb_bits_ != -8) return 0;   // Flush() must have been called
   if (!BitWriterResize(bw, size)) return 0;
   memcpy(bw->buf_ + bw->pos_, data, size);
   bw->pos_ += size;

src/utils/endian_inl.h

@@ -16,6 +16,7 @@
 #include "../webp/config.h"
 #endif
 
+#include "../dsp/dsp.h"
 #include "../webp/types.h"
 
 // some endian fix (e.g.: mips-gcc doesn't define __BIG_ENDIAN__)
@@ -34,25 +35,13 @@
 #endif
 
 #if !defined(HAVE_CONFIG_H)
-#ifdef __GNUC__
-# define LOCAL_GCC_VERSION ((__GNUC__ << 8) | __GNUC_MINOR__)
-#else
-# define LOCAL_GCC_VERSION 0
-#endif  // __GNUC__
-
-#ifdef __clang__
-# define LOCAL_CLANG_VERSION ((__clang_major__ << 8) | __clang_minor__)
-#else
-# define LOCAL_CLANG_VERSION 0
-#endif  // __clang__
-
 // clang-3.3 and gcc-4.3 have builtin functions for swap32/swap64
-#if LOCAL_GCC_VERSION >= 0x403 || LOCAL_CLANG_VERSION >= 0x303
+#if LOCAL_GCC_PREREQ(4,3) || LOCAL_CLANG_PREREQ(3,3)
 #define HAVE_BUILTIN_BSWAP32
 #define HAVE_BUILTIN_BSWAP64
 #endif
 // clang-3.3 and gcc-4.8 have a builtin function for swap16
-#if LOCAL_GCC_VERSION >= 0x408 || LOCAL_CLANG_VERSION >= 0x303
+#if LOCAL_GCC_PREREQ(4,8) || LOCAL_CLANG_PREREQ(3,3)
 #define HAVE_BUILTIN_BSWAP16
 #endif
 #endif  // !HAVE_CONFIG_H
@@ -69,7 +58,16 @@ static WEBP_INLINE uint16_t BSwap16(uint16_t x) {
 }
 
 static WEBP_INLINE uint32_t BSwap32(uint32_t x) {
-#if defined(HAVE_BUILTIN_BSWAP32)
+#if defined(WEBP_USE_MIPS32_R2)
+  uint32_t ret;
+  __asm__ volatile (
+    "wsbh   %[ret], %[x]          \n\t"
+    "rotr   %[ret], %[ret],  16   \n\t"
+    : [ret]"=r"(ret)
+    : [x]"r"(x)
+  );
+  return ret;
+#elif defined(HAVE_BUILTIN_BSWAP32)
   return __builtin_bswap32(x);
 #elif defined(__i386__) || defined(__x86_64__)
   uint32_t swapped_bytes;

src/webp/decode.h

@@ -444,16 +444,20 @@ struct WebPDecoderOptions {
   int dithering_strength;             // dithering strength (0=Off, 100=full)
 #if WEBP_DECODER_ABI_VERSION > 0x0203
   int flip;                           // flip output vertically
+#endif
+#if WEBP_DECODER_ABI_VERSION > 0x0204
   int alpha_dithering_strength;       // alpha dithering strength in [0..100]
 #endif
 
   // Unused for now:
   int force_rotation;                 // forced rotation (to be applied _last_)
   int no_enhancement;                 // if true, discard enhancement layer
-#if WEBP_DECODER_ABI_VERSION > 0x0203
-  uint32_t pad[3];                    // padding for later use
-#else
+#if WEBP_DECODER_ABI_VERSION < 0x0203
   uint32_t pad[5];                    // padding for later use
+#elif WEBP_DECODER_ABI_VERSION < 0x0204
+  uint32_t pad[4];                    // padding for later use
+#else
+  uint32_t pad[3];                    // padding for later use
 #endif
 };
 

src/webp/encode.h

@@ -231,14 +231,14 @@ struct WebPMemoryWriter {
 // The following must be called first before any use.
 WEBP_EXTERN(void) WebPMemoryWriterInit(WebPMemoryWriter* writer);
 
-#if WEBP_ENCODER_ABI_VERSION > 0x0202
+#if WEBP_ENCODER_ABI_VERSION > 0x0203
 // The following must be called to deallocate writer->mem memory. The 'writer'
 // object itself is not deallocated.
 WEBP_EXTERN(void) WebPMemoryWriterClear(WebPMemoryWriter* writer);
 #endif
 // The custom writer to be used with WebPMemoryWriter as custom_ptr. Upon
 // completion, writer.mem and writer.size will hold the coded data.
-#if WEBP_ENCODER_ABI_VERSION > 0x0202
+#if WEBP_ENCODER_ABI_VERSION > 0x0203
 // writer.mem must be freed by calling WebPMemoryWriterClear.
 #else
 // writer.mem must be freed by calling 'free(writer.mem)'.
@@ -446,13 +446,14 @@ WEBP_EXTERN(int) WebPPictureImportBGRA(
 WEBP_EXTERN(int) WebPPictureImportBGRX(
     WebPPicture* picture, const uint8_t* bgrx, int bgrx_stride);
 
-// Converts picture->argb data to the YUVA format specified by 'colorspace'.
+// Converts picture->argb data to the YUV420A format. The 'colorspace'
+// parameter is deprecated and should be equal to WEBP_YUV420.
 // Upon return, picture->use_argb is set to false. The presence of real
 // non-opaque transparent values is detected, and 'colorspace' will be
 // adjusted accordingly. Note that this method is lossy.
 // Returns false in case of error.
 WEBP_EXTERN(int) WebPPictureARGBToYUVA(WebPPicture* picture,
-                                       WebPEncCSP colorspace);
+                                       WebPEncCSP /*colorspace = WEBP_YUV420*/);
 
 // Same as WebPPictureARGBToYUVA(), but the conversion is done using
 // pseudo-random dithering with a strength 'dithering' between
@@ -461,6 +462,15 @@ WEBP_EXTERN(int) WebPPictureARGBToYUVA(WebPPicture* picture,
 WEBP_EXTERN(int) WebPPictureARGBToYUVADithered(
     WebPPicture* picture, WebPEncCSP colorspace, float dithering);
 
+#if WEBP_ENCODER_ABI_VERSION > 0x0204
+// Performs 'smart' RGBA->YUVA420 downsampling and colorspace conversion.
+// Downsampling is handled with extra care in case of color clipping. This
+// method is roughly 2x slower than WebPPictureARGBToYUVA() but produces better
+// YUV representation.
+// Returns false in case of error.
+WEBP_EXTERN(int) WebPPictureSmartARGBToYUVA(WebPPicture* picture);
+#endif
+
 // Converts picture->yuv to picture->argb and sets picture->use_argb to true.
 // The input format must be YUV_420 or YUV_420A.
 // Note that the use of this method is discouraged if one has access to the