README.wasm: add instructions for enabling mulhi

Change-Id: I1e9dd737f06ad76f73824352291a6e129ca5ded1

AUTHORS

@@ -35,5 +35,4 @@ Contributors:
 - Urvang Joshi (urvang at google dot com)
 - Vikas Arora (vikasa at google dot com)
 - Vincent Rabaud (vrabaud at google dot com)
-- Vlad Tsyrklevich (vtsyrklevich at chromium dot org)
 - Yang Zhang (yang dot zhang at arm dot com)

Android.mk

@@ -55,6 +55,9 @@ dsp_dec_srcs := \
     src/dsp/alpha_processing_neon.$(NEON) \
     src/dsp/alpha_processing_sse2.c \
     src/dsp/alpha_processing_sse41.c \
+    src/dsp/argb.c \
+    src/dsp/argb_mips_dsp_r2.c \
+    src/dsp/argb_sse2.c \
     src/dsp/cpu.c \
     src/dsp/dec.c \
     src/dsp/dec_clip_tables.c \
@@ -85,13 +88,11 @@ dsp_dec_srcs := \
     src/dsp/upsampling_msa.c \
     src/dsp/upsampling_neon.$(NEON) \
     src/dsp/upsampling_sse2.c \
-    src/dsp/upsampling_sse41.c \
     src/dsp/yuv.c \
     src/dsp/yuv_mips32.c \
     src/dsp/yuv_mips_dsp_r2.c \
     src/dsp/yuv_neon.$(NEON) \
     src/dsp/yuv_sse2.c \
-    src/dsp/yuv_sse41.c \
 
 dsp_enc_srcs := \
     src/dsp/cost.c \
@@ -123,6 +124,7 @@ enc_srcs := \
     src/enc/backward_references_enc.c \
     src/enc/config_enc.c \
     src/enc/cost_enc.c \
+    src/enc/delta_palettization_enc.c \
     src/enc/filter_enc.c \
     src/enc/frame_enc.c \
     src/enc/histogram_enc.c \

CMakeLists.txt

@@ -4,16 +4,17 @@ project(libwebp C)
 
 # Options for coder / decoder executables.
 option(WEBP_ENABLE_SIMD "Enable any SIMD optimization." ON)
+option(WEBP_ENABLE_WASM "Enable WebAssembly optimizations." OFF)
 option(WEBP_BUILD_CWEBP "Build the cwebp command line tool." OFF)
 option(WEBP_BUILD_DWEBP "Build the dwebp command line tool." OFF)
 option(WEBP_BUILD_GIF2WEBP "Build the gif2webp conversion tool." OFF)
 option(WEBP_BUILD_IMG2WEBP "Build the img2webp animation tool." OFF)
 option(WEBP_BUILD_WEBPINFO "Build the webpinfo command line tool." OFF)
 option(WEBP_BUILD_WEBP_JS "Emscripten build of webp.js." OFF)
-option(WEBP_NEAR_LOSSLESS "Enable near-lossless encoding" ON)
+option(WEBP_EXPERIMENTAL_FEATURES "Build with experimental features." OFF)
 option(WEBP_ENABLE_SWAP_16BIT_CSP "Enable byte swap for 16 bit colorspaces." OFF)
 
-if(WEBP_BUILD_WEBP_JS)
+if(WEBP_BUILD_WEBP_JS OR WEBP_ENABLE_WASM)
   set(WEBP_ENABLE_SIMD OFF)
 endif()
 
@@ -26,13 +27,19 @@ if(NOT CMAKE_BUILD_TYPE)
   )
 endif()
 
-# Include dependencies.
+include(cmake/config.h.cmake)
-include(cmake/deps.cmake)
+
+# Extract the version of the library.
+file(READ ${CMAKE_CURRENT_SOURCE_DIR}/configure.ac SOURCE_FILE)
+string(REGEX MATCH "[0-9.]+" WEBP_VERSION ${SOURCE_FILE})
 
 ################################################################################
 # Options.
 if(WEBP_ENABLE_SWAP_16BIT_CSP)
-  add_definitions(-DWEBP_SWAP_16BIT_CSP=1)
+  add_definitions(-DWEBP_SWAP_16BIT_CSP)
+endif()
+if(WEBP_ENABLE_WASM)
+  add_definitions(-DWEBP_USE_WASM)
 endif()
 
 ################################################################################
@@ -47,10 +54,7 @@ if(ANDROID)
   set(WEBP_DEP_INCLUDE_DIRS ${WEBP_DEP_INCLUDE_DIRS}
     ${ANDROID_NDK}/sources/android/cpufeatures
   )
-  add_definitions(-DHAVE_CPU_FEATURES_H=1)
+  add_definitions(-DHAVE_CPU_FEATURES_H)
-  set(HAVE_CPU_FEATURES_H 1)
-else()
-  set(HAVE_CPU_FEATURES_H 0)
 endif()
 
 ################################################################################
@@ -102,13 +106,8 @@ endforeach()
 
 ### Define the mandatory libraries.
 # Build the webpdecoder library.
-if(MSVC)
+add_definitions(-Wall)
-  # avoid security warnings for e.g., fopen() used in the examples.
+include_directories(${CMAKE_CURRENT_SOURCE_DIR}/src/ ${WEBP_DEP_INCLUDE_DIRS})
-  add_definitions(-D_CRT_SECURE_NO_WARNINGS)
-else()
-  add_definitions(-Wall)
-endif()
-include_directories(${CMAKE_CURRENT_SOURCE_DIR} ${WEBP_DEP_INCLUDE_DIRS})
 add_library(webpdecode OBJECT ${WEBP_DEC_SRCS})
 add_library(webpdspdecode OBJECT ${WEBP_DSP_COMMON_SRCS} ${WEBP_DSP_DEC_SRCS})
 add_library(webputilsdecode OBJECT ${WEBP_UTILS_COMMON_SRCS}
@@ -146,13 +145,13 @@ function(parse_version FILE NAME VAR)
   set(${VAR} "${VERSION}" PARENT_SCOPE)
 endfunction()
 parse_version(Makefile.am webp WEBP_WEBP_SOVERSION)
-set_target_properties(webp PROPERTIES VERSION ${PACKAGE_VERSION}
+set_target_properties(webp PROPERTIES VERSION ${WEBP_VERSION}
   SOVERSION ${WEBP_WEBP_SOVERSION})
 parse_version(Makefile.am webpdecoder WEBP_DECODER_SOVERSION)
-set_target_properties(webpdecoder PROPERTIES VERSION ${PACKAGE_VERSION}
+set_target_properties(webpdecoder PROPERTIES VERSION ${WEBP_VERSION}
   SOVERSION ${WEBP_DECODER_SOVERSION})
 parse_version(demux/Makefile.am webpdemux WEBP_DEMUX_SOVERSION)
-set_target_properties(webpdemux PROPERTIES VERSION ${PACKAGE_VERSION}
+set_target_properties(webpdemux PROPERTIES VERSION ${WEBP_VERSION}
   SOVERSION ${WEBP_DEMUX_SOVERSION})
 
 # Define the libraries to install.
@@ -168,9 +167,11 @@ math(EXPR WEBP_SIMD_FILES_TO_INCLUDE_RANGE
 foreach(I_FILE RANGE ${WEBP_SIMD_FILES_TO_INCLUDE_RANGE})
   list(GET WEBP_SIMD_FILES_TO_INCLUDE ${I_FILE} FILE)
   list(GET WEBP_SIMD_FLAGS_TO_INCLUDE ${I_FILE} SIMD_COMPILE_FLAG)
-  set_source_files_properties(${FILE} PROPERTIES
+  if(NOT ${SIMD_COMPILE_FLAG} STREQUAL "NOTFOUND")
-    COMPILE_FLAGS ${SIMD_COMPILE_FLAG}
+    set_source_files_properties(${FILE} PROPERTIES
-  )
+      COMPILE_FLAGS ${SIMD_COMPILE_FLAG}
+    )
+  endif()
 endforeach()
 
 # Build the executables if asked for.
@@ -192,18 +193,13 @@ if(WEBP_BUILD_CWEBP OR WEBP_BUILD_DWEBP OR
   parse_Makefile_am(${CMAKE_CURRENT_SOURCE_DIR}/imageio "IMAGEDEC_SRCS"
     "imagedec_[^ ]*")
   add_library(imagedec ${IMAGEDEC_SRCS})
-  target_link_libraries(imagedec imageioutil webpdemux webp
+  target_link_libraries(imagedec imageioutil webp ${WEBP_DEP_IMG_LIBRARIES})
-    ${WEBP_DEP_IMG_LIBRARIES})
 
   # Image-encoding utility library.
   parse_Makefile_am(${CMAKE_CURRENT_SOURCE_DIR}/imageio "IMAGEENC_SRCS"
     "imageenc_[^ ]*")
   add_library(imageenc ${IMAGEENC_SRCS})
   target_link_libraries(imageenc webp)
-
-  set_property(TARGET exampleutil imageioutil imagedec imageenc
-    PROPERTY INCLUDE_DIRECTORIES
-    ${CMAKE_CURRENT_SOURCE_DIR}/src ${CMAKE_CURRENT_BINARY_DIR}/src)
 endif()
 
 if(WEBP_BUILD_DWEBP)
@@ -214,8 +210,6 @@ if(WEBP_BUILD_DWEBP)
   add_executable(dwebp ${DWEBP_SRCS})
   target_link_libraries(dwebp exampleutil imagedec imageenc webpdecoder)
   install(TARGETS dwebp RUNTIME DESTINATION bin)
-  set_property(TARGET dwebp PROPERTY INCLUDE_DIRECTORIES
-    ${CMAKE_CURRENT_SOURCE_DIR}/src ${CMAKE_CURRENT_BINARY_DIR}/src)
 endif()
 
 if(WEBP_BUILD_CWEBP)
@@ -226,12 +220,6 @@ if(WEBP_BUILD_CWEBP)
   add_executable(cwebp ${CWEBP_SRCS})
   target_link_libraries(cwebp exampleutil imagedec webp)
   install(TARGETS cwebp RUNTIME DESTINATION bin)
-  set_property(TARGET cwebp PROPERTY INCLUDE_DIRECTORIES
-    ${CMAKE_CURRENT_SOURCE_DIR}/src ${CMAKE_CURRENT_BINARY_DIR}/src)
-endif()
-
-if(WEBP_BUILD_GIF2WEBP AND NOT GIF_FOUND)
-  unset(WEBP_BUILD_GIF2WEBP CACHE)
 endif()
 
 if(WEBP_BUILD_GIF2WEBP OR WEBP_BUILD_IMG2WEBP)
@@ -240,7 +228,7 @@ if(WEBP_BUILD_GIF2WEBP OR WEBP_BUILD_IMG2WEBP)
   add_library(webpmux ${WEBP_MUX_SRCS})
   target_link_libraries(webpmux webp)
   parse_version(mux/Makefile.am webpmux WEBP_MUX_SOVERSION)
-  set_target_properties(webpmux PROPERTIES VERSION ${PACKAGE_VERSION}
+  set_target_properties(webpmux PROPERTIES VERSION ${WEBP_VERSION}
     SOVERSION ${WEBP_MUX_SOVERSION})
   list(APPEND INSTALLED_LIBRARIES webpmux)
 endif()
@@ -254,8 +242,6 @@ if(WEBP_BUILD_GIF2WEBP)
   target_link_libraries(gif2webp exampleutil imageioutil webp webpmux
     ${WEBP_DEP_GIF_LIBRARIES})
   install(TARGETS gif2webp RUNTIME DESTINATION bin)
-  set_property(TARGET gif2webp PROPERTY INCLUDE_DIRECTORIES
-    ${CMAKE_CURRENT_SOURCE_DIR}/src ${CMAKE_CURRENT_BINARY_DIR}/src)
 endif()
 
 if(WEBP_BUILD_IMG2WEBP)
@@ -266,8 +252,6 @@ if(WEBP_BUILD_IMG2WEBP)
   add_executable(img2webp ${IMG2WEBP_SRCS})
   target_link_libraries(img2webp exampleutil imagedec imageioutil webp webpmux)
   install(TARGETS img2webp RUNTIME DESTINATION bin)
-  set_property(TARGET img2webp PROPERTY INCLUDE_DIRECTORIES
-    ${CMAKE_CURRENT_SOURCE_DIR}/src ${CMAKE_CURRENT_BINARY_DIR}/src)
 endif()
 
 if (WEBP_BUILD_WEBPINFO)
@@ -278,8 +262,6 @@ if (WEBP_BUILD_WEBPINFO)
   add_executable(webpinfo ${WEBPINFO_SRCS})
   target_link_libraries(webpinfo exampleutil imageioutil)
   install(TARGETS webpinfo RUNTIME DESTINATION bin)
-  set_property(TARGET webpinfo PROPERTY INCLUDE_DIRECTORIES
-    ${CMAKE_CURRENT_SOURCE_DIR}/src ${CMAKE_CURRENT_BINARY_DIR}/src)
 endif()
 
 if(WEBP_BUILD_WEBP_JS)
@@ -287,10 +269,8 @@ if(WEBP_BUILD_WEBP_JS)
   add_executable(webp_js
                  ${CMAKE_CURRENT_SOURCE_DIR}/extras/webp_to_sdl.c)
   target_link_libraries(webp_js webpdecoder SDL)
-  set(WEBP_HAVE_SDL 1)
   set_target_properties(webp_js PROPERTIES LINK_FLAGS
-      "-s EXPORTED_FUNCTIONS='[\"_WebpToSDL\"]' -s INVOKE_RUN=0 \
+      "-s EXPORTED_FUNCTIONS='[\"_WebpToSDL\"]' -s INVOKE_RUN=0")
-       -s EXTRA_EXPORTED_RUNTIME_METHODS='[\"cwrap\"]'")
   set_target_properties(webp_js PROPERTIES OUTPUT_NAME webp)
   target_compile_definitions(webp_js PUBLIC EMSCRIPTEN WEBP_HAVE_SDL)
 
@@ -300,21 +280,12 @@ if(WEBP_BUILD_WEBP_JS)
   target_link_libraries(webp_wasm webpdecoder SDL)
   set_target_properties(webp_wasm PROPERTIES LINK_FLAGS
       "-s WASM=1 -s 'BINARYEN_METHOD=\"native-wasm\"' \
-       -s EXPORTED_FUNCTIONS='[\"_WebpToSDL\"]' -s INVOKE_RUN=0 \
+      -s EXPORTED_FUNCTIONS='[\"_WebpToSDL\"]' -s INVOKE_RUN=0")
-       -s EXTRA_EXPORTED_RUNTIME_METHODS='[\"cwrap\"]'")
   target_compile_definitions(webp_wasm PUBLIC EMSCRIPTEN WEBP_HAVE_SDL)
 
   target_compile_definitions(webpdecoder PUBLIC EMSCRIPTEN)
 endif()
 
-# Generate the config.h file.
-configure_file(${CMAKE_CURRENT_LIST_DIR}/cmake/config.h.in
-  ${CMAKE_CURRENT_BINARY_DIR}/src/webp/config.h)
-add_definitions(-DHAVE_CONFIG_H)
-# The webp folder is included as we reference config.h as
-# ../webp/config.h or webp/config.h
-include_directories(${CMAKE_CURRENT_BINARY_DIR})
-
 # Install the different headers and libraries.
 install(FILES ${CMAKE_CURRENT_SOURCE_DIR}/src/webp/decode.h
               ${CMAKE_CURRENT_SOURCE_DIR}/src/webp/demux.h
@@ -331,7 +302,7 @@ install(TARGETS ${INSTALLED_LIBRARIES}
 include(CMakePackageConfigHelpers)
 write_basic_package_version_file(
   "${CMAKE_CURRENT_BINARY_DIR}/WebPConfigVersion.cmake"
-  VERSION ${PACKAGE_VERSION}
+  VERSION ${WEBP_VERSION}
   COMPATIBILITY AnyNewerVersion
 )
 

ChangeLog

@@ -1,377 +1,9 @@
-8d510751 webp-container-spec: correct frame duration=0 note
-e6b2164e vwebp: Copy Chrome's behavior w/frame duration == 0
-d20b7707 update ChangeLog (tag: v1.0.0-rc3)
-0d5fad46 add WEBP_DSP_INIT / WEBP_DSP_INIT_FUNC
-c1cb86af fix 16b overflow in SSE2
-e577feb7 makefile.unix: add DEBUG flag for compiling w/ debug-symbol
-99be34b3 cwebp,get_disto: fix bpp output
-f5565ca8 cmake: Make sure we use near-lossless by default.
-d898dc14 fix bug in WebPImport565: alpha value was not set
-882784b0 update ChangeLog (tag: v1.0.0-rc2)
-2f930e08 Revert "Use proper targets for CMake."
-8165e8fb Use proper targets for CMake.
-3f157dd5 Remove some very hard TODOs.
-cd758a17 {de,}mux/Makefile.am: add missing headers
-b892b8ba makefile.unix,dist: use ascii for text output
-64a57d05 add -version option to anim_dump,anim_diff and img2webp
-fc1b8e3a webp_js: fix webp_js demo html
-15aa48d9 update ChangeLog (tag: v1.0.0-rc1)
-e607dabc update AUTHORS
-38410c08 [CFI] Remove function pointer casts
-c57b2736 bump version to 1.0.0
-cba28853 update NEWS
-c909d531 Merge "remove some deprecation warning on MacOSX"
-217443c7 remove some deprecation warning on MacOSX
-b672bdfa configure: quiet glut deprecation warnings on OS X
-daa9fcaf configure: use sdl-config if available
-dd174cae Merge "imagedec: support metadata reading for WebP image decoding"
-641cedcc imagedec: support metadata reading for WebP image decoding
-065b2ce1 anim_diff: add a couple missing newlines in Help()
-c4cc1147 Merge "gif2webp: force low duration frames to 100ms"
-09333097 gif2webp: force low duration frames to 100ms
-e03f0ec3 sharp_yuv: use 14b fixed-point precision for gamma
-b2db361c image_enc,WebPWritePNG: move locals after setjmp
-74e82ec6 Merge "WebPPictureDistortion: fix big-endian results order"
-645d04ca Merge "cwebp,get_disto: report bpp"
-120f58c3 Merge "lossless*sse2: improve non-const 16-bit vector creation"
-a7fe9412 WebPPictureDistortion: fix big-endian results order
-e26fe066 cwebp,get_disto: report bpp
-9df64e28 Merge changes Id5b4a1a4,Ia20ce844
-8043504f lossless*sse2: improve non-const 16-bit vector creation
-1e3dfc48 Import: extract condition from loop
-3b07d327 Import,RGBA: fix for BigEndian import
-551948e4 Remove unused argument in VP8LBitsEntropy.
-3005237a ReadWebP: fix for big-endian
-499c395a Merge "anim_diff: expose the -max_diff option"
-f69dcd69 Merge "remove WEBP_EXPERIMENTAL_FEATURES"
-07d884d5 anim_diff: expose the -max_diff option
-f4dd9256 remove WEBP_EXPERIMENTAL_FEATURES
-94a8377b extract the command-line parsing helpers to example_util
-fc09e6e2 PNM decoder: prevent unsupported depth=2 PAM case.
-6de58603 MIPS64: Fix defined-but-not-used errors with WEBP_REDUCE_CSP
-cbde5728 gif2webp: add support for reading from stdin
-cf1c5054 Add an SSE4 version of some lossless color transforms.
-45a8b5eb Fix lint error with man page.
-cff38e8f Merge "PNG decoder: handle gAMA chunk"
-59cb1a48 Merge "enable dc error-diffusion always"
-78318b30 PNG decoder: handle gAMA chunk
-664c21dd Merge "remove some TODOs"
-815652de enable dc error-diffusion always
-aec45cec remove some TODOs
-5715dfce fix block-count[] increment in case of large image
-c2d04f3e enable DC error-diffusion always for multi-pass
-96bf07c5 use DC error diffusion for U/V at low-quality
-1c59020b fix missing sse41 targets in makefile.unix
-7a8e814b cosmetics: s/color_space/colorspace/
-05f6fe24 upsampling: rm asserts w/REDUCE_CSP+OMIT_C_CODE
-b4cf5597 Merge "Upsampling SSE2/SSE4 speedup."
-ccbeb32c Makefile.vc: add missing sse41 files
-55403a9a Upsampling SSE2/SSE4 speedup.
-807b53c4 Implement the upsampling/yuv functions in SSE41
-84101a81 Fix wasm WebP compilation
-8bebd2a3 fix warning on MSVC
-a7f93fe3 webpmux: allow reading argument from a file
-b69f18a7 gif2webp.1: fix -loop_compatibility layout
-72d530c0 Merge "fix lossless decoding w/WEBP_REDUCE_SIZE"
-296c7dc4 fix lossless decoding w/WEBP_REDUCE_SIZE
-0d5d029c Merge "ImgIoUtilReadFile: fix file leak upon error"
-ae568ce7 ImgIoUtilReadFile: fix file leak upon error
-796b5a8a Merge tag 'v0.6.1'
-6b7a95fd update ChangeLog (tag: v0.6.1)
-f66955de WEBP_REDUCE_CSP: restrict colorspace support
-1af0df76 Merge "WEBP_REDUCE_CSP: restrict colorspace support"
-6de20df0 WEBP_REDUCE_CSP: restrict colorspace support
-a289d8e7 update ChangeLog (tag: v0.6.1-rc2)
-c10a493c vwebp: disable double buffering on windows & mac
-0d4466c2 webp_to_sdl.c: fix file mode
-1b27bf8b WEBP_REDUCE_SIZE: disable all rescaler code
-126be109 webpinfo: add -version option
-0df22b9e WEBP_REDUCE_SIZE: disable all rescaler code
-9add62b5 bump version to 0.6.1
-d3e26144 update NEWS
-2edda639 README: add webpinfo section
-9ca568ef Merge "right-size some tables"
-31f1995c Merge "SSE2 implementation of HasAlphaXXX"
-a80c46bd SSE2 implementation of HasAlphaXXX
-083507f2 right-size some tables
-2e5785b2 anim_utils.c: remove warning when !defined(WEBP_HAVE_GIF)
-b299c47e add WEBP_REDUCE_SIZE
-f593d71a enc: disable pic->stats/extra_info w/WEBP_DISABLE_STATS
-541179a9 Merge "predictor_enc: fix build w/--disable-near-lossless"
-5755a7ec predictor_enc: fix build w/--disable-near-lossless
-eab5bab7 add WEBP_DISABLE_STATS
-8052c585 remove some petty TODOs from vwebp.
-c245343d move LOAD8x4 and STORE8x2 closer to their use location
-b9e734fd dec,cosmetics: normalize function naming style
-c188d546 dec: harmonize function suffixes
-28c5ac81 dec_sse41: harmonize function suffixes
-e65b72a3 Merge "introduce WebPHasAlpha8b and WebPHasAlpha32b"
-b94cee98 dec_sse2: remove HE8uv_SSE2
-44a0ee3f introduce WebPHasAlpha8b and WebPHasAlpha32b
-aebf59ac Merge "WebPPictureAllocARGB: align argb allocation"
-c184665e WebPPictureAllocARGB: align argb allocation
-3daf7509 WebPParseHeaders: remove obsolete animation TODO
-80285d97 cmake: avoid security warnings under msvc
-650eac55 cmake: don't set -Wall with MSVC
-c462cd00 Remove useless code.
-01a98217 Merge "remove WebPWorkerImpl declaration from the header"
-3c49fc47 Merge "thread_utils: fix potentially bad call to Execute"
-fde2782e thread_utils: fix potentially bad call to Execute
-2a270c1d remove WebPWorkerImpl declaration from the header
-f1f437cc remove mention of 'lossy-only parameters' from the doc
-3879074d Merge "WebPMemToUint32: remove ptr cast to int"
-04b029d2 WebPMemToUint32: remove ptr cast to int
-b7971d0e dsp: avoid defining _C functions w/NEON builds
-6ba98764 webpdec: correct alloc size check w/use_argb
-5cfb3b0f normalize include guards
-f433205e Merge changes Ia17c7dfc,I75423abb,Ia2f716b4,I161caa14,I4210081a, ...
-8d033b14 {dec,enc}_neon: harmonize function suffixes x2
-0295e981 upsampling_neon: harmonize function suffixes
-d572c4e5 yuv_neon: harmonize function suffixes
-ab9c2500 rescaler_neon: harmonize function suffixes
-93e0ce27 lossless_neon: harmonize function suffixes
-22fbc50e lossless_enc_neon: harmonize function suffixes
-447875b4 filters_neon,cosmetics: fix indent
-e51bdd43 remove unused VP8TokenToStats() function
-785da7ea enc_neon: harmonize function suffixes
-bc1a251f dec_neon: harmonize function suffixes
-61e535f1 dsp/lossless: workaround gcc-4.8 bug on arm
-68b2eab7 cwebp: fix alpha reporting w/lossless & metadata
-30042faa WebPDemuxGetI: add doc details around WebPFormatFeature
-0a17f471 Merge "WIP: list includes as descendants of the project dir"
-a4399721 WIP: list includes as descendants of the project dir
-08275708 Merge "Make sure we reach the full range for alpha blending."
-d361a6a7 yuv_sse2: harmonize function suffixes
-6921aa6f upsampling_sse2: harmonize function suffixes
-08c67d3e ssim_sse2: harmonize function suffixes
-582a1b57 rescaler_sse2: harmonize function suffixes
-2c1b18ba lossless_sse2: harmonize function suffixes
-0ac46e81 lossless_enc_sse2: harmonize function suffixes
-bc634d57 enc_sse2: harmonize function suffixes
-bcb7347c dec_sse2: harmonize function suffixes
-e14ad93c Make sure we reach the full range for alpha blending.
-7038ca8d demux,StoreFrame: restore hdr size check to min req
-fb3daad6 cpu: fix ssse3 check
-be590e06 Merge "Fix CMake redefinition for HAVE_CPU_FEATURES_H"
-35f736e1 Fix CMake redefinition for HAVE_CPU_FEATURES_H
-a5216efc Fix integer overflow warning.
-a9c8916b decode.h,WebPIDecGetRGB: clarify output ptr validity
-3c74c645 gif2webp: handle 1-frame case properly + fix anim_diff
-c7f295d3 Merge "gif2webp: introduce -loop_compatibility option"
-b4e04677 gif2webp: introduce -loop_compatibility option
-f78da3de add LOCAL_CLANG_PREREQ and avoid WORK_AROUND_GCC w/3.8+
-01c426f1 define WEBP_USE_INTRINSICS w/gcc-4.9+
-8635973d use sdl-config (if available) to determine the link flags
-e9459382 use CPPFLAGS before CFLAGS
-4a9d788e Merge "Android.mk,mips: fix clang build with r15"
-4fbdc9fb Android.mk,mips: fix clang build with r15
-a80fcc4a ifdef code not used by Chrome/Android.
-3993af12 Fix signed integer overflows.
-f66f94ef anim_dump: small tool to dump frames from animated WebP
-6eba857b Merge "rationalize the Makefile.am"
-c5e34fba function definition cleanup
-3822762a rationalize the Makefile.am
-501ef6e4 configure style fix: animdiff -> anim_diff
-f8bdc268 Merge "protect against NULL dump_folder[] value in ReadAnimatedImage()"
-23bfc652 protect against NULL dump_folder[] value in ReadAnimatedImage()
-8dc3d71b cosmetics,ReadAnimatedWebP: correct function comment
-5bd40066 Merge changes I66a64a0a,I4d2e520f
-7945575c cosmetics,webpinfo: remove an else after a return
-8729fa11 cosmetics,cwebp: remove an else after a return
-f324b7f9 cosmetics: normalize fn proto & decl param names
-869eb369 CMake cleanups.
-289e62a3 Remove declaration of unimplemented VP8ApplyNearLosslessPredict
-20a94186 pnmdec,PAM: validate depth before calculating bytes_per_px
-34130afe anim_encode: fix integer overflow
-42c79aa6 Merge "Encoder: harmonize function suffixes"
-b09307dc Encoder: harmonize function suffixes
-bed0456d Merge "SSIM: harmonize the function suffix"
-54f6a3cf lossless_sse2.c: fix some missed suffix changes
-088f1dcc SSIM: harmonize the function suffix
-86fc4dd9 webpdec: use ImgIoUtilCheckSizeArgumentsOverflow
-08ea9ecd imageio: add ability restrict max image size
-6f9daa4a jpegdec,ReadError: fix leaks on error
-a0f72a4f VP8LTransformColorFunc: drop an non-respected 'const' from the signature.
-8c934902 Merge "Lossess dec: harmonize the function suffixes"
-622242aa Lossess dec: harmonize the function suffixes
-1411f027 Lossless Enc: harmonize the function suffixes
-24ad2e3c add const to two variables
-46efe062 Merge "Allow the lossless cruncher to work for alpha."
-8c3f9a47 Speed-up LZ77.
-1aef4c71 Allow the lossless cruncher to work for alpha.
-b8821dbd Improve the box LZ77 speed.
-7beed280 add missing ()s to macro parameters
-6473d20b Merge "fix Android standalone toolchain build"
-dcefed95 Merge "build.gradle: fix arm64 build"
-0c83a8bc Merge "yuv: harmonize suffix naming"
-c6d1db4b fix Android standalone toolchain build
-663a6d9d unify the ALTERNATE_CODE flag usage
-73ea9f27 yuv: harmonize suffix naming
-c71b68ac build.gradle: fix arm64 build
-c4568b47 Rescaler: harmonize the suffix naming
-6cb13b05 Merge "alpha_processing: harmonize the naming suffixes to be _C()"
-83a3e69a Merge "simplify WEBP_EXTERN macro"
-7295fde2 Merge "filters: harmonize the suffixes naming to _SSE2(), _C(), etc."
-8e42ba4c simplify WEBP_EXTERN macro
-331ab34b cost*.c: harmonize the suffix namings
-b161f670 filters: harmonize the suffixes naming to _SSE2(), _C(), etc.
-dec5e4d3 alpha_processing: harmonize the naming suffixes to be _C()
-6878d427 fix memory leak in SDL_Init()
-461ae555 Merge "configure: fix warnings in sdl check"
-62486a22 configure: test for -Wundef
-92982609 dsp.h: fix -Wundef w/__mips_dsp_rev
-0265cede configure: fix warnings in sdl check
-88c73d8a backward_references_enc.h: fix WINDOW_SIZE_BITS check
-4ea49f6b rescaler_sse2.c: fix WEBP_RESCALER_FIX -> _RFIX typo
-1b526638 Clean-up some CMake
-87f57a4b Merge "cmake: fix gif lib detection when cross compiling"
-b34a9db1 cosmetics,dec_sse2: remove some redundant comments
-471c5755 cmake: fix gif lib detection when cross compiling
-c793417a cmake: disable gif2webp if gif lib isn't found
-dcbc1c88 cmake: split gif detection from IMG deps
-66ad84f0 Merge "muxread: remove unreachable code"
-50ec3ab7 muxread: remove unreachable code
-7d67a164 Lossy encoding: smoothen transparent areas to improve compression
-e50650c7 Merge "fix signature for DISABLE_TOKEN_BUFFER compilation"
-671d2567 fix signature for DISABLE_TOKEN_BUFFER compilation
-d6755580 cpu.cmake: use unique flag to test simd disable flags
-28914528 Merge "Remove the argb* files."
-8acb4942 Remove the argb* files.
-3b62347b README: correct cmake invocation note
-7ca0df13 Have the SSE2 version of PackARGB use common code.
-7b250459 Merge "Re-use the transformed image when trying several LZ77 in lossless."
-e132072f Re-use the transformed image when trying several LZ77 in lossless.
-5d7a50ef Get code to compile in C++.
-7b012987 configure: test for -Wparentheses-equality
-f0569adb Fix man pages for multi-threading.
-f1d5a397 multithread cruncher: only copy stats when picture->stats != NULL
-f8c2ac15 Multi-thread the lossless cruncher.
-a88c6522 Merge "Integrate a new LZ77 looking for matches in the neighborhood of a pixel only."
-8f6df1d0 Unroll Predictors 10, 11 and 12.
-355c3d1b Integrate a new LZ77 looking for matches in the neighborhood of a pixel only.
-a1779a01 Refactor LZ77 handling in preparation for a new method.
-67de68b5 Android.mk/build.gradle: fix mips build with clang from r14b
-f209a548 Use the plane code and not the distance when computing statistics.
-b903b80c Split cost-based backward references in its own file.
-498cad34 Cosmetic changes in backward reference.
-e4eb4587 lossless, VP8LTransformColor_C: make sure no overflow happens with colors.
-af6deaff webpinfo: handle alpha flag mismatch
-7caef29b Fix typo that creeped in.
-39e19f92 Merge "near lossless: fix unsigned int overflow warnings."
-9bbc0891 near lossless: fix unsigned int overflow warnings.
-e1118d62 Merge "cosmetics,FindClosestDiscretized: use uint in mask creation"
-186bc9b7 Merge "webpinfo: tolerate ALPH+VP8L"
-b5887297 cosmetics,FindClosestDiscretized: use uint in mask creation
-f1784aee near_lossless,FindClosestDiscretized: use unsigned ops
-0d20abb3 webpinfo: tolerate ALPH+VP8L
-972104b3 webpmux: tolerate false positive Alpha flag
-dd7e83cc tiffdec,ReadTIFF: ensure data_size is < tsize_t max
-d988eb7b tiffdec,MyRead: quiet -Wshorten-64-to-32 warning
-dabda707 webpinfo: add support to parse Alpha bitstream
-4c117643 webpinfo: correct background color output, BGRA->ARGB
-defc98d7 Doc: clarify the role of quality in WebPConfig.
-d78ff780 Merge "Fix code to compile with C++."
-c8f14093 Fix code to compile with C++.
-497dc6a7 pnmdec: sanitize invalid header output
-d78e5867 Merge "configure: test for -Wconstant-conversion"
-481e91eb Merge "pnmdec,PAM: set bytes_per_px based on depth when missing"
-93b12753 configure: test for -Wconstant-conversion
-645f0c53 pnmdec,PAM: set bytes_per_px based on depth when missing
-e9154605 Merge "vwebp: activate GLUT double-buffering"
-818d795b vwebp: activate GLUT double-buffering
-d63e6f4b Add a man page for webpinfo
-4d708435 Merge "NEON: implement ConvertRGB24ToY/BGR24/ARGB/RGBA32ToUV/ARGBToUV"
-faf42213 NEON: implement ConvertRGB24ToY/BGR24/ARGB/RGBA32ToUV/ARGBToUV
-b4d576fa Install man pages with CMake.
-cbc1b921 webpinfo: add features to parse bitstream header
-e644c556 Fix bad bit writer initialization.
-b62cdad2 Merge "Implement a cruncher for lossless at method 6."
-da3e4dfb use the exact constant for the gamma transfer function
-a9c701e0 Merge "tiffdec: fix EXTRASAMPLES check"
-adab8ce0 Implement a cruncher for lossless at method 6.
-1b92b237 Merge "Fix VP8ApplyNearLossless to respect const and stride."
-1923ff02 tiffdec: fix EXTRASAMPLES check
-97cce5ba tiffdec: only request EXTRASAMPLES w/> 3 samples/px
-0dcd85b6 Fix VP8ApplyNearLossless to respect const and stride.
-f7682189 yuv: rationalize the C/SSE2 function naming
-52245424 NEON implementation of some Sharp-YUV420 functions
-690efd82 Avoid several backward reference copies.
-4bb1f607 src/dec/vp8_dec.h, cosmetics: fix comments
-285748be cmake: build/install webpinfo
-78fd199c backward_references_enc.c: clear -Wshadow warnings
-ae836410 WebPLog2FloorC: clear -Wshadow warning
-d0b7404e Merge "WASM support"
-134e314f WASM support
-c08adb6f Merge "VP8LEnc: remove use of BitsLog2Ceiling()"
-28c37ebd VP8LEnc: remove use of BitsLog2Ceiling()
-2cb58ab2 webpinfo: output format as a human readable string
-bb175a93 Merge "rename some symbols clashing with MSVC headers"
-39eda658 Remove a duplicated pixel hash implementation.
-36b8274d rename some symbols clashing with MSVC headers
-274daf54 Add webpinfo tool.
-ec5036e4 add explicit reference to /usr/local/{lib,inc}
-18f0dfac Merge "fix TIFF encoder regarding rgbA/RGBA"
-4e2b0b50 Merge "webpdec.h: fix a doc typo"
-e2eeabff Merge "Install binaries, libraries and headers in CMake."
-836607e6 webpdec.h: fix a doc typo
-9273e441 fix TIFF encoder regarding rgbA/RGBA
-17e3c11f Add limited PAM decoding support
-5f624871 Install binaries, libraries and headers in CMake.
-976adac1 Merge "lossless incremental decoding: fix missing eos_ test"
-f8fad4fa lossless incremental decoding: fix missing eos_ test
-27415d41 Merge "vwebp_sdl: fix the makefile.unix"
-49566182 Merge "ImgIoUtilWriteFile(): use ImgIoUtilSetBinaryMode"
-6f75a51b Analyze the transform entropy on the whole image.
-a5e4e3af Use palette only if we can in entropy analysis.
-75a9c3c4 Improve compression by better entropy analysis.
-39cf6f4f vwebp_sdl: fix the makefile.unix
-699b0416 ImgIoUtilWriteFile(): use ImgIoUtilSetBinaryMode
-7d985bd1 Fix small entropy analysis bug.
-6e7caf06 Optimize the color cache size.
-833c9219 More efficient stochastic histogram merge.
-5183326b Refactor the greedy histogram merge.
-99f6f462 Merge "histogram_enc.c,MyRand: s/ul/u/ for unsigned constants"
-80a22186 ssim.c: remove dead include
-a128dfff histogram_enc.c,MyRand: s/ul/u/ for unsigned constants
-693bf74e move the SSIM calculation code in ssim.c / ssim_sse2.c
-10d791ca Merge "Fix the random generator in HistogramCombineStochastic."
-fa63a966 Fix the random generator in HistogramCombineStochastic.
-16be192f VP8LSetBitPos: remove the eos_ setting
-027151ca don't erase the surface before blitting.
-4105d565 disable WEBP_USE_XXX optimisations when EMSCRIPTEN is defined
-9ee32a75 Merge "WebP-JS: emscripten-based Javascript decoder"
-ca9f7b7d WebP-JS: emscripten-based Javascript decoder
-868aa690 Perform greedy histogram merge in a unified way.
-5b393f2d Merge "fix path typo for vwebp_sdl in Makefile.vc"
-e0012bea CMake: only use libwebpdecoder for building dwebp
-84c2a7b0 fix path typo for vwebp_sdl in Makefile.vc
-1b0e4abf Merge "Add a flag to disable SIMD optimizations."
-32263250 Add a flag to disable SIMD optimizations.
-b494fdec optimize the ARGB->ARGB Import to use memcpy
-f1536039 Merge "ReadWebP: decode directly into a pre-allocated buffer"
-e69ed291 ReadWebP: decode directly into a pre-allocated buffer
-57d8de8a Merge "vwebp_sdl: simple viewer based on SDL"
-5cfd4ebc LZ77 interval speedups. Faster, smaller, simpler.
-1e7ad88b PNM header decoder: add some basic numerical validation
-17c7890c Merge "Add a decoder only library for WebP in CMake."
-be733786 Merge "Add clang build fix for MSA"
-03cda0e4 Add a decoder only library for WebP in CMake.
-aa893914 Add clang build fix for MSA
-31a92e97 Merge "imageio: add limited PNM support for reading"
-dcf9d82a imageio: add limited PNM support for reading
-6524fcd6 vwebp_sdl: simple viewer based on SDL
-6cf24a24 get_disto: fix reference file read
-43d472aa Merge tag 'v0.6.0'
-50d1a848 update ChangeLog (tag: v0.6.0, origin/0.6.0, 0.6.0)
 20a7fea0 extras/Makefile.am: fix libwebpextras.la reference
 415f3ffe update ChangeLog (tag: v0.6.0-rc3)
 3c6d1224 update NEWS
 ee4a4141 update AUTHORS
 32ed856f Fix "all|no frames are keyframes" settings.
-1c3190b6 Merge "Fix "all|no frames are keyframes" settings."
 f4dc56fd disable GradientUnfilter_NEON
-4f3e3bbd disable GradientUnfilter_NEON
-2dc0bdca Fix "all|no frames are keyframes" settings.
 0d8e0588 img2webp: treat -loop as a no-op w/single images
 b0450139 ReadImage(): restore size reporting
 0ad3b4ef update ChangeLog (tag: v0.6.0-rc2)

Makefile.vc

@@ -29,7 +29,7 @@ PLATFORM_LDFLAGS = /SAFESEH
 NOLOGO     = /nologo
 CCNODBG    = cl.exe $(NOLOGO) /O2 /DNDEBUG
 CCDEBUG    = cl.exe $(NOLOGO) /Od /Gm /Zi /D_DEBUG /RTC1
-CFLAGS     = /I. /Isrc $(NOLOGO) /W3 /EHsc /c
+CFLAGS     = /Isrc $(NOLOGO) /W3 /EHsc /c
 CFLAGS     = $(CFLAGS) /DWIN32 /D_CRT_SECURE_NO_WARNINGS /DWIN32_LEAN_AND_MEAN
 LDFLAGS    = /LARGEADDRESSAWARE /MANIFEST /NXCOMPAT /DYNAMICBASE
 LDFLAGS    = $(LDFLAGS) $(PLATFORM_LDFLAGS)
@@ -155,7 +155,6 @@ CFGSET = TRUE
 !MESSAGE - all                            - build (de)mux-based targets for CFG
 !MESSAGE - gif2webp                       - requires libgif & >= VS2013
 !MESSAGE - anim_diff                      - requires libgif & >= VS2013
-!MESSAGE - anim_dump
 !MESSAGE
 !MESSAGE RTLIBCFG controls the runtime library linkage - 'static' or 'dynamic'.
 !MESSAGE   'legacy' will produce a Windows 2000 compatible library.
@@ -227,15 +226,16 @@ DSP_DEC_OBJS = \
     $(DIROBJ)\dsp\upsampling_msa.obj \
     $(DIROBJ)\dsp\upsampling_neon.obj \
     $(DIROBJ)\dsp\upsampling_sse2.obj \
-    $(DIROBJ)\dsp\upsampling_sse41.obj \
     $(DIROBJ)\dsp\yuv.obj \
     $(DIROBJ)\dsp\yuv_mips32.obj \
     $(DIROBJ)\dsp\yuv_mips_dsp_r2.obj \
     $(DIROBJ)\dsp\yuv_neon.obj \
     $(DIROBJ)\dsp\yuv_sse2.obj \
-    $(DIROBJ)\dsp\yuv_sse41.obj \
 
 DSP_ENC_OBJS = \
+    $(DIROBJ)\dsp\argb.obj \
+    $(DIROBJ)\dsp\argb_mips_dsp_r2.obj \
+    $(DIROBJ)\dsp\argb_sse2.obj \
     $(DIROBJ)\dsp\cost.obj \
     $(DIROBJ)\dsp\cost_mips32.obj \
     $(DIROBJ)\dsp\cost_mips_dsp_r2.obj \
@@ -287,6 +287,7 @@ ENC_OBJS = \
     $(DIROBJ)\enc\backward_references_enc.obj \
     $(DIROBJ)\enc\config_enc.obj \
     $(DIROBJ)\enc\cost_enc.obj \
+    $(DIROBJ)\enc\delta_palettization_enc.obj \
     $(DIROBJ)\enc\filter_enc.obj \
     $(DIROBJ)\enc\frame_enc.obj \
     $(DIROBJ)\enc\histogram_enc.obj \
@@ -357,22 +358,15 @@ all: ex $(EXTRA_EXAMPLES)
 # C99 support which is only available from VS2013 onward.
 gif2webp: $(DIRBIN)\gif2webp.exe
 anim_diff: $(DIRBIN)\anim_diff.exe
-anim_dump: $(DIRBIN)\anim_dump.exe
 
 $(DIRBIN)\anim_diff.exe: $(DIROBJ)\examples\anim_diff.obj $(EX_ANIM_UTIL_OBJS)
 $(DIRBIN)\anim_diff.exe: $(EX_UTIL_OBJS) $(IMAGEIO_UTIL_OBJS)
 $(DIRBIN)\anim_diff.exe: $(EX_GIF_DEC_OBJS) $(LIBWEBPDEMUX) $(LIBWEBP)
-$(DIRBIN)\anim_dump.exe: $(DIROBJ)\examples\anim_dump.obj $(EX_ANIM_UTIL_OBJS)
-$(DIRBIN)\anim_dump.exe: $(EX_UTIL_OBJS) $(IMAGEIO_UTIL_OBJS)
-$(DIRBIN)\anim_dump.exe: $(EX_GIF_DEC_OBJS) $(LIBWEBPDEMUX) $(LIBWEBP)
-$(DIRBIN)\anim_dump.exe: $(IMAGEIO_ENC_OBJS)
 $(DIRBIN)\cwebp.exe: $(DIROBJ)\examples\cwebp.obj $(IMAGEIO_DEC_OBJS)
 $(DIRBIN)\cwebp.exe: $(IMAGEIO_UTIL_OBJS)
-$(DIRBIN)\cwebp.exe: $(LIBWEBPDEMUX)
 $(DIRBIN)\dwebp.exe: $(DIROBJ)\examples\dwebp.obj $(IMAGEIO_DEC_OBJS)
 $(DIRBIN)\dwebp.exe: $(IMAGEIO_ENC_OBJS)
 $(DIRBIN)\dwebp.exe: $(IMAGEIO_UTIL_OBJS)
-$(DIRBIN)\dwebp.exe: $(LIBWEBPDEMUX)
 $(DIRBIN)\gif2webp.exe: $(DIROBJ)\examples\gif2webp.obj $(EX_GIF_DEC_OBJS)
 $(DIRBIN)\gif2webp.exe: $(EX_UTIL_OBJS) $(IMAGEIO_UTIL_OBJS) $(LIBWEBPMUX)
 $(DIRBIN)\gif2webp.exe: $(LIBWEBP)
@@ -385,24 +379,26 @@ $(DIRBIN)\webpmux.exe: $(DIROBJ)\examples\webpmux.obj $(LIBWEBPMUX)
 $(DIRBIN)\webpmux.exe: $(EX_UTIL_OBJS) $(IMAGEIO_UTIL_OBJS) $(LIBWEBP)
 $(DIRBIN)\img2webp.exe: $(DIROBJ)\examples\img2webp.obj $(LIBWEBPMUX)
 $(DIRBIN)\img2webp.exe: $(IMAGEIO_DEC_OBJS)
-$(DIRBIN)\img2webp.exe: $(EX_UTIL_OBJS) $(IMAGEIO_UTIL_OBJS)
+$(DIRBIN)\img2webp.exe: $(EX_UTIL_OBJS) $(IMAGEIO_UTIL_OBJS) $(LIBWEBP)
-$(DIRBIN)\img2webp.exe: $(LIBWEBPDEMUX) $(LIBWEBP)
 $(DIRBIN)\get_disto.exe: $(DIROBJ)\extras\get_disto.obj
-$(DIRBIN)\get_disto.exe: $(IMAGEIO_DEC_OBJS) $(IMAGEIO_UTIL_OBJS)
+$(DIRBIN)\get_disto.exe: $(IMAGEIO_DEC_OBJS) $(IMAGEIO_UTIL_OBJS) $(LIBWEBP)
-$(DIRBIN)\get_disto.exe: $(LIBWEBPDEMUX) $(LIBWEBP)
 $(DIRBIN)\webp_quality.exe: $(DIROBJ)\extras\webp_quality.obj
 $(DIRBIN)\webp_quality.exe: $(IMAGEIO_UTIL_OBJS)
 $(DIRBIN)\webp_quality.exe: $(EXTRAS_OBJS) $(LIBWEBP)
 $(DIRBIN)\webpinfo.exe: $(DIROBJ)\examples\webpinfo.obj
 $(DIRBIN)\webpinfo.exe: $(IMAGEIO_DEC_OBJS)
-$(DIRBIN)\webpinfo.exe: $(EX_UTIL_OBJS) $(IMAGEIO_UTIL_OBJS)
+$(DIRBIN)\webpinfo.exe: $(EX_UTIL_OBJS) $(IMAGEIO_UTIL_OBJS) $(LIBWEBP)
-$(DIRBIN)\webpinfo.exe: $(LIBWEBPDEMUX) $(LIBWEBP)
 
 $(OUT_EXAMPLES): $(EX_UTIL_OBJS) $(LIBWEBP)
 $(EX_UTIL_OBJS) $(IMAGEIO_UTIL_OBJS): $(OUTPUT_DIRS)
 $(IMAGEIO_DEC_OBJS) $(IMAGEIO_ENC_OBJS) $(EXTRAS_OBJS): $(OUTPUT_DIRS)
 !ENDIF  # ARCH == ARM
 
+experimental:
+	$(MAKE) /f Makefile.vc \
+	    CFG=$(CFG) \
+	    CFLAGS="$(CFLAGS) /DWEBP_EXPERIMENTAL_FEATURES" /$(MAKEFLAGS)
+
 $(LIBWEBPDECODER): $(LIBWEBPDECODER_OBJS)
 $(LIBWEBP): $(LIBWEBP_OBJS)
 $(LIBWEBPMUX): $(LIBWEBPMUX_OBJS)
@@ -448,7 +444,7 @@ $(OUTPUT_DIRS):
 $(DIROBJ)\$(DLLINC):
 	@echo #ifndef WEBP_DLL_H_ > $@
 	@echo #define WEBP_DLL_H_ >> $@
-	@echo #define WEBP_EXTERN __declspec(dllexport) >> $@
+	@echo #define WEBP_EXTERN(type) __declspec(dllexport) type >> $@
 	@echo #endif  /* WEBP_DLL_H_ */ >> $@
 
 .SUFFIXES: .c .obj .res .exe
@@ -460,9 +456,6 @@ $(DIROBJ)\dsp\enc_avx2.obj: src\dsp\enc_avx2.c
 $(DIROBJ)\examples\anim_diff.obj: examples\anim_diff.c
 	$(CC) $(CFLAGS) /DWEBP_HAVE_GIF /Fd$(LIBWEBP_PDBNAME) \
 	  /Fo$(DIROBJ)\examples\ examples\$(@B).c
-$(DIROBJ)\examples\anim_dump.obj: examples\anim_dump.c
-	$(CC) $(CFLAGS) /DWEBP_HAVE_GIF /Fd$(LIBWEBP_PDBNAME) \
-	  /Fo$(DIROBJ)\examples\ examples\$(@B).c
 $(DIROBJ)\examples\anim_util.obj: examples\anim_util.c
 	$(CC) $(CFLAGS) /DWEBP_HAVE_GIF /Fd$(LIBWEBP_PDBNAME) \
 	  /Fo$(DIROBJ)\examples\ examples\$(@B).c

NEWS

@@ -1,26 +1,3 @@
-- 4/2/2018: version 1.0.0
-  This is a binary compatible release.
-  * lossy encoder improvements to avoid chroma shifts in various circumstances
-    (issues #308, #340)
-  * big-endian fixes for decode, RGBA import and WebPPictureDistortion
-  Tool updates:
-    gifwebp, anim_diff - default duration behavior (<= 10ms) changed to match
-                         web browsers, transcoding tools (issue #379)
-    img2webp, webpmux - allow options to be passed in via a file (issue #355)
-
-- 11/24/2017: version 0.6.1
-  This is a binary compatible release.
-  * lossless performance and compression improvements + a new 'cruncher' mode
-    (-m 6 -q 100)
-  * ARM performance improvements with clang (15-20% w/ndk r15c, issue #339)
-  * webp-js: emscripten/webassembly based javascript decoder
-  * miscellaneous bug & build fixes (issue #329, #332, #343, #353, #360, #361,
-    #363)
-  Tool updates / additions:
-    added webpinfo - prints file format information (issue #330)
-    gif2webp - loop behavior modified to match Chrome M63+ (crbug.com/649264);
-               '-loop_compatibility' can be used for the old behavior
-
 - 1/26/2017: version 0.6.0
   * lossless performance and compression improvements
   * miscellaneous performance improvements (SSE2, NEON, MSA)

README

@@ -4,7 +4,7 @@
           \__\__/\____/\_____/__/ ____  ___
                 / _/ /    \    \ /  _ \/ _/
                /  \_/   / /   \ \   __/  \__
-               \____/____/\_____/_____/____/v1.0.0
+               \____/____/\_____/_____/____/v0.6.0
 
 Description:
 ============
@@ -113,8 +113,8 @@ make install
 
 CMake:
 ------
-With CMake, you can compile libwebp, cwebp, dwebp, gif2web, img2webp, webpinfo
+With CMake, you can compile libwebp, cwebp, dwebp, gif2web, img2webp and the
-and the JS bindings.
+JS bindings.
 
 Prerequisites:
 A compiler (e.g., gcc with autotools) and CMake.
@@ -367,23 +367,6 @@ Use following options to convert into alternate image formats:
   -quiet ....... quiet mode, don't print anything
   -noasm ....... disable all assembly optimizations
 
-WebP file analysis tool:
-========================
-
-'webpinfo' can be used to print out the chunk level structure and bitstream
-header information of WebP files. It can also check if the files are of valid
-WebP format.
-
-Usage: webpinfo [options] in_files
-Note: there could be multiple input files;
-      options must come before input files.
-Options:
-  -version ........... Print version number and exit.
-  -quiet ............. Do not show chunk parsing information.
-  -diag .............. Show parsing error diagnosis.
-  -summary ........... Show chunk stats summary.
-  -bitstream_info .... Parse bitstream header.
-
 Visualization tool:
 ===================
 
@@ -458,7 +441,6 @@ File-level options (only used at the start of compression):
  -mixed ............... use mixed lossy/lossless automatic mode
  -v ................... verbose mode
  -h ................... this help
- -version ............. print version number and exit
 
 Per-frame options (only used for subsequent images input):
  -d <int> ............. frame duration in ms (default: 100)
@@ -495,8 +477,6 @@ Options:
   -metadata <string> ..... comma separated list of metadata to
                            copy from the input to the output if present
                            Valid values: all, none, icc, xmp (default)
-  -loop_compatibility .... use compatibility mode for Chrome
-                           version prior to M62 (inclusive)
   -mt .................... use multi-threading if available
 
   -version ............... print version number and exit
@@ -525,11 +505,6 @@ Options:
   -min_psnr <float> ... minimum per-frame PSNR
   -raw_comparison ..... if this flag is not used, RGB is
                         premultiplied before comparison
-  -max_diff <int> ..... maximum allowed difference per channel
-                        between corresponding pixels in subsequent
-                        frames
-  -h .................. this help
-  -version ............ print version number and exit
 
 Building:
 ---------

README.mux

@@ -1,7 +1,7 @@
           __   __  ____  ____  ____  __ __  _     __ __
          /  \\/  \/  _ \/  _ \/  _ \/  \  \/ \___/_ / _\
          \       /   __/  _  \   __/      /  /  (_/  /__
-          \__\__/\_____/_____/__/  \__//_/\_____/__/___/v1.0.0
+          \__\__/\_____/_____/__/  \__//_/\_____/__/___/v0.4.0
 
 
 Description:
@@ -33,7 +33,6 @@ Usage: webpmux -get GET_OPTIONS INPUT -o OUTPUT
        webpmux -info INPUT
        webpmux [-h|-help]
        webpmux -version
-       webpmux argument_file_name
 
 GET_OPTIONS:
  Extract relevant data:
@@ -93,9 +92,6 @@ INPUT & OUTPUT are in WebP format.
 Note: The nature of EXIF, XMP and ICC data is not checked and is assumed to be
 valid.
 
-Note: if a single file name is passed as the argument, the arguments will be
-tokenized from this file. The file name must not start with the character '-'.
-
 Visualization tool:
 ===================
 

README.wasm

@@ -0,0 +1,91 @@
+Description:
+============
+
+This file describes the compilation of libwebp using portable intrinsics /
+WebAssembly (wasm) to native targets using clang and CMake.
+
+Prerequisites:
+==============
+
+- cmake 2.8+
+
+- clang 3.9+ for portable intrinsics support; as wasm progresses a tip of tree
+  build may be necessary.
+
+Building:
+=========
+
+ - configure the project with CMake using:
+
+ $ mkdir -p build && \
+   cd build && \
+   cmake -DWEBP_BUILD_DWEBP=1 -DCMAKE_C_COMPILER=clang -DWEBP_ENABLE_WASM=1 ../
+
+ - compile dwebp using 'make'.
+
+ - Note this currently generates native executables only and is incompatible
+   with -DWEBP_BUILD_WEBP_JS.
+
+Build options:
+==============
+
+- platform specific multiply high (mulhi) implementation, disabled by default.
+  arm: -DCMAKE_C_FLAGS='-DENABLE_NEON_BUILTIN_MULHI_INT16X8 ...'
+  x86: -DCMAKE_C_FLAGS='-DENABLE_X86_BUILTIN_MULHI_INT16X8 ...'
+
+Cross compilation:
+==================
+
+ - arm toolchains can be obtained from:
+   http://www.linaro.org/downloads/
+
+ - the android ndk can be obtained from:
+   https://developer.android.com/ndk/downloads/index.html
+
+armv7:
+------
+
+Android:
+ $ ./android-ndk-r15b/build/tools/make_standalone_toolchain.py \
+   --arch arm --api 24 --stl gnustl --install-dir /opt/android-arm-24
+ $ mkdir -p build && cd build
+ $ cmake ../libwebp \
+   -DWEBP_BUILD_DWEBP=1 \
+   -DCMAKE_C_COMPILER=/opt/android-arm-24/bin/clang \
+   -DCMAKE_PREFIX_PATH=/opt/android-arm-24/sysroot/usr/lib \
+   -DCMAKE_C_FLAGS=-fPIE \
+   -DCMAKE_EXE_LINKER_FLAGS=-Wl,-pie \
+   -DCMAKE_BUILD_TYPE=Release \
+   -DWEBP_ENABLE_WASM=1
+
+Linux:
+ $ gcc_arm=/opt/gcc-arm; target=arm-linux-gnueabihf
+ $ mkdir -p build && cd build
+ $ cmake ../libwebp -DWEBP_BUILD_DWEBP=1 -DWEBP_ENABLE_WASM=1 \
+   -DCMAKE_C_COMPILER=clang \
+   -DCMAKE_C_FLAGS="--target=$target --gcc-toolchain=$gcc_arm --sysroot=$gcc_arm/$target/libc -march=armv7-a -mfpu=neon" \
+   -DCMAKE_PREFIX_PATH=$gcc_arm/$target/libc/usr
+
+aarch64 / arm64:
+----------------
+
+Android:
+ $ ./android-ndk-r15b/build/tools/make_standalone_toolchain.py \
+   --arch arm64 --api 24 --stl gnustl --install-dir /opt/android-arm64-24
+ $ mkdir -p build && cd build
+ $ cmake ../libwebp \
+   -DWEBP_BUILD_DWEBP=1 \
+   -DCMAKE_C_COMPILER=/opt/android-arm64-24/bin/clang \
+   -DCMAKE_PREFIX_PATH=/opt/android-arm64-24/sysroot/usr/lib \
+   -DCMAKE_C_FLAGS=-fPIE \
+   -DCMAKE_EXE_LINKER_FLAGS=-Wl,-pie \
+   -DCMAKE_BUILD_TYPE=Release \
+   -DWEBP_ENABLE_WASM=1
+
+Linux:
+ $ gcc_arm=/opt/gcc-aarch64; target=aarch64-linux-gnu
+ $ mkdir -p build && cd build
+ $ cmake ../libwebp -DWEBP_BUILD_DWEBP=1 -DWEBP_ENABLE_WASM=1 \
+   -DCMAKE_C_COMPILER=clang \
+   -DCMAKE_C_FLAGS="--target=$target --gcc-toolchain=$gcc_arm --sysroot=$gcc_arm/$target/libc" \
+   -DCMAKE_PREFIX_PATH=$gcc_arm/$target/libc/usr

README.webp_js

@@ -31,9 +31,13 @@ using Emscripten and CMake.
  - that's it! Upon completion, you should have the webp.js and
    webp.js.mem files generated.
 
+ - Note this generates both webp_js and webp_wasm without any SIMD enabled due
+   to bugs with this toolchain associated with the SSE2 code.
+   -DWEBP_ENABLE_WASM is currently meant to generate native (x86, arm)
+   executables (dwebp, cwebp) and is incompatible with -DWEBP_BUILD_WEBP_JS.
+
 The callable JavaScript function is WebPToSDL(), which decodes a raw WebP
-bitstream into a canvas. See webp_js/index.html for a simple usage sample
+bitstream into a canvas. See webp_js/index.html for a simple usage sample.
-(see below for instructions).
 
 Demo HTML page:
 ===============

build.gradle

@@ -82,14 +82,12 @@ model {
         }
       }
       // Check for NEON usage.
-      if (getTargetPlatform() == "arm") {
+      if (getTargetPlatform() == "arm" || getTargetPlatform() == "arm64") {
         NEON = "c.neon"
         cCompiler.define "HAVE_CPU_FEATURES_H"
       } else {
         NEON = "c"
       }
-
-      cCompiler.args "-I" + file(".").absolutePath
     }
     // Link to pthread for shared libraries.
     withType(SharedLibraryBinarySpec) {
@@ -122,6 +120,9 @@ model {
             include "alpha_processing_neon.$NEON"
             include "alpha_processing_sse2.c"
             include "alpha_processing_sse41.c"
+            include "argb.c"
+            include "argb_mips_dsp_r2.c"
+            include "argb_sse2.c"
             include "cpu.c"
             include "dec.c"
             include "dec_clip_tables.c"
@@ -152,13 +153,11 @@ model {
             include "upsampling_msa.c"
             include "upsampling_neon.$NEON"
             include "upsampling_sse2.c"
-            include "upsampling_sse41.c"
             include "yuv.c"
             include "yuv_mips32.c"
             include "yuv_mips_dsp_r2.c"
             include "yuv_neon.$NEON"
             include "yuv_sse2.c"
-            include "yuv_sse41.c"
             srcDir "src/utils"
             include "bit_reader_utils.c"
             include "color_cache_utils.c"
@@ -198,6 +197,7 @@ model {
             include "backward_references_enc.c"
             include "config_enc.c"
             include "cost_enc.c"
+            include "delta_palettization_enc.c"
             include "filter_enc.c"
             include "frame_enc.c"
             include "histogram_enc.c"
@@ -289,7 +289,6 @@ model {
     imagedec(NativeLibrarySpec) {
       binaries {
         all {
-          lib library: "webpdemux", linkage: "static"
           lib library: "webp", linkage: "static"
         }
       }
@@ -332,7 +331,6 @@ model {
           lib library: "example_util", linkage: "static"
           lib library: "imagedec", linkage: "static"
           lib library: "imageio_util", linkage: "static"
-          lib library: "webpdemux", linkage: "static"
           lib library: "webp", linkage: "static"
         }
       }
@@ -353,7 +351,6 @@ model {
           lib library: "imagedec", linkage: "static"
           lib library: "imageenc", linkage: "static"
           lib library: "imageio_util", linkage: "static"
-          lib library: "webpdemux", linkage: "static"
           lib library: "webp"
         }
       }
@@ -393,7 +390,6 @@ model {
           lib library: "imagedec", linkage: "static"
           lib library: "imageio_util", linkage: "static"
           lib library: "webpmux", linkage: "static"
-          lib library: "webpdemux", linkage: "static"
           lib library: "webp"
         }
       }

cmake/config.h.cmake

@@ -70,43 +70,18 @@ foreach(I_LIB PNG JPEG TIFF)
   set(WEBP_HAVE_${I_LIB} ${${I_LIB}_FOUND})
   if(${I_LIB}_FOUND)
     list(APPEND WEBP_DEP_IMG_LIBRARIES ${${I_LIB}_LIBRARIES})
-    list(APPEND WEBP_DEP_IMG_INCLUDE_DIRS
+    list(APPEND WEBP_DEP_IMG_INCLUDE_DIRS ${${I_LIB}_INCLUDE_DIRS})
-         ${${I_LIB}_INCLUDE_DIR} ${${I_LIB}_INCLUDE_DIRS})
   endif()
 endforeach()
-if(WEBP_DEP_IMG_INCLUDE_DIRS)
-  list(REMOVE_DUPLICATES WEBP_DEP_IMG_INCLUDE_DIRS)
-endif()
 
 # GIF detection, gifdec isn't part of the imageio lib.
-include(CMakePushCheckState)
 set(WEBP_DEP_GIF_LIBRARIES)
 set(WEBP_DEP_GIF_INCLUDE_DIRS)
 find_package(GIF)
 set(WEBP_HAVE_GIF ${GIF_FOUND})
 if(GIF_FOUND)
-  # GIF find_package only locates the header and library, it doesn't fail
+  list(APPEND WEBP_DEP_GIF_LIBRARIES ${GIF_LIBRARIES})
-  # compile tests when detecting the version, but falls back to 3 (as of at
+  list(APPEND WEBP_DEP_GIF_INCLUDE_DIRS ${GIF_INCLUDE_DIR})
-  # least cmake 3.7.2). Make sure the library links to avoid incorrect
-  # detection when cross compiling.
-  cmake_push_check_state()
-  set(CMAKE_REQUIRED_LIBRARIES ${GIF_LIBRARIES})
-  set(CMAKE_REQUIRED_INCLUDES ${GIF_INCLUDE_DIR})
-  check_c_source_compiles("
-      #include <gif_lib.h>
-      int main(void) {
-        (void)DGifOpenFileHandle;
-        return 0;
-      }
-      " GIF_COMPILES
-  )
-  cmake_pop_check_state()
-  if(GIF_COMPILES)
-    list(APPEND WEBP_DEP_GIF_LIBRARIES ${GIF_LIBRARIES})
-    list(APPEND WEBP_DEP_GIF_INCLUDE_DIRS ${GIF_INCLUDE_DIR})
-  else()
-    unset(GIF_FOUND)
-  endif()
 endif()
 
 ## Check for specific headers.
@@ -164,3 +139,13 @@ strip_bracket(PACKAGE_URL)
 set(PACKAGE_STRING "${PACKAGE_NAME} ${PACKAGE_VERSION}")
 set(PACKAGE_TARNAME ${PACKAGE_NAME})
 set(VERSION ${PACKAGE_VERSION})
+
+## Generate the config.h header.
+configure_file(${CMAKE_CURRENT_LIST_DIR}/config.h.in
+  ${CMAKE_CURRENT_BINARY_DIR}/include/webp/config.h)
+add_definitions(-DHAVE_CONFIG_H)
+# The webp folder is included as we reference config.h as
+# ../webp/config.h or webp/config.h
+include_directories(${CMAKE_CURRENT_BINARY_DIR}/include
+  ${CMAKE_CURRENT_BINARY_DIR}/include/webp
+)

cmake/config.h.in

@@ -13,9 +13,6 @@
 /* Set to 1 if __builtin_bswap64 is available */
 #cmakedefine HAVE_BUILTIN_BSWAP64 1
 
-/* Define to 1 if you have the <cpu-features.h> header file. */
-#cmakedefine HAVE_CPU_FEATURES_H 1
-
 /* Define to 1 if you have the <dlfcn.h> header file. */
 #cmakedefine HAVE_DLFCN_H 1
 
@@ -103,6 +100,9 @@
 /* Version number of package */
 #cmakedefine VERSION "@VERSION@"
 
+/* Enable experimental code */
+#cmakedefine WEBP_EXPERIMENTAL_FEATURES 1
+
 /* Set to 1 if AVX2 is supported */
 #cmakedefine WEBP_HAVE_AVX2 1
 
@@ -115,19 +115,9 @@
 /* Set to 1 if JPEG library is installed */
 #cmakedefine WEBP_HAVE_JPEG 1
 
-/* Set to 1 if NEON is supported */
-#cmakedefine WEBP_HAVE_NEON
-
-/* Set to 1 if runtime detection of NEON is enabled */
-/* TODO: handle properly in CMake */
-#cmakedefine WEBP_HAVE_NEON_RTCD
-
 /* Set to 1 if PNG library is installed */
 #cmakedefine WEBP_HAVE_PNG 1
 
-/* Set to 1 if SDL library is installed */
-#cmakedefine WEBP_HAVE_SDL 1
-
 /* Set to 1 if SSE2 is supported */
 #cmakedefine WEBP_HAVE_SSE2 1
 
@@ -137,9 +127,6 @@
 /* Set to 1 if TIFF library is installed */
 #cmakedefine WEBP_HAVE_TIFF 1
 
-/* Enable near lossless encoding */
-#cmakedefine WEBP_NEAR_LOSSLESS 1
-
 /* Undefine this to disable thread support. */
 #cmakedefine WEBP_USE_THREAD 1
 

cmake/cpu.cmake

@@ -1,5 +1,4 @@
 ## Check for SIMD extensions.
-include(CMakePushCheckState)
 
 function(webp_check_compiler_flag WEBP_SIMD_FLAG ENABLE_SIMD)
   if(NOT ENABLE_SIMD)
@@ -8,8 +7,6 @@ function(webp_check_compiler_flag WEBP_SIMD_FLAG ENABLE_SIMD)
     return()
   endif()
   unset(WEBP_HAVE_FLAG_${WEBP_SIMD_FLAG} CACHE)
-  cmake_push_check_state()
-  set(CMAKE_REQUIRED_INCLUDES ${CMAKE_CURRENT_SOURCE_DIR})
   check_c_source_compiles("
       #include \"${CMAKE_CURRENT_LIST_DIR}/../src/dsp/dsp.h\"
       int main(void) {
@@ -20,7 +17,6 @@ function(webp_check_compiler_flag WEBP_SIMD_FLAG ENABLE_SIMD)
       }
     " WEBP_HAVE_FLAG_${WEBP_SIMD_FLAG}
   )
-  cmake_pop_check_state()
   if(WEBP_HAVE_FLAG_${WEBP_SIMD_FLAG})
     set(WEBP_HAVE_${WEBP_SIMD_FLAG} 1 PARENT_SCOPE)
   else()
@@ -64,7 +60,6 @@ foreach(I_SIMD RANGE ${WEBP_SIMD_FLAGS_RANGE})
   # First try with no extra flag added as the compiler might have default flags
   # (especially on Android).
   unset(WEBP_HAVE_${WEBP_SIMD_FLAG} CACHE)
-  cmake_push_check_state()
   set(CMAKE_REQUIRED_FLAGS)
   webp_check_compiler_flag(${WEBP_SIMD_FLAG} ${WEBP_ENABLE_SIMD})
   if(NOT WEBP_HAVE_${WEBP_SIMD_FLAG})
@@ -90,8 +85,11 @@ foreach(I_SIMD RANGE ${WEBP_SIMD_FLAGS_RANGE})
     foreach(FILE ${SIMD_FILES})
       list(APPEND WEBP_SIMD_FILES_NOT_TO_INCLUDE ${FILE})
     endforeach()
-    # Explicitly disable SIMD.
+    # Explicitly disable SIMD. Avoid this with WASM to avoid an ICE with clang:
-    if(SIMD_DISABLE_FLAGS)
+    # https://bugs.chromium.org/p/webp/issues/detail?id=350
+    # WASM overrides the native SIMD so building it in is harmless aside from
+    # binary size.
+    if(NOT WEBP_ENABLE_WASM AND SIMD_DISABLE_FLAGS)
       list(GET SIMD_DISABLE_FLAGS ${I_SIMD} SIMD_COMPILE_FLAG)
       include(CheckCCompilerFlag)
       if(SIMD_COMPILE_FLAG)
@@ -106,12 +104,11 @@ foreach(I_SIMD RANGE ${WEBP_SIMD_FLAGS_RANGE})
             set(COMMON_PATTERNS)
           endif()
           set(CMAKE_REQUIRED_DEFINITIONS ${SIMD_COMPILE_FLAG})
-          check_c_source_compiles("int main(void) {return 0;}"
+          check_c_source_compiles("int main(void) {return 0;}" FLAG2
-            FLAG_${SIMD_COMPILE_FLAG}
             FAIL_REGEX "warning: argument unused during compilation:"
             ${COMMON_PATTERNS}
           )
-          if(NOT FLAG_${SIMD_COMPILE_FLAG})
+          if(NOT FLAG2)
             unset(HAS_COMPILE_FLAG CACHE)
           endif()
         endif()
@@ -121,5 +118,14 @@ foreach(I_SIMD RANGE ${WEBP_SIMD_FLAGS_RANGE})
       endif()
     endif()
   endif()
-  cmake_pop_check_state()
 endforeach()
+
+## Add *_wasm.c files if enabled.
+if(WEBP_ENABLE_WASM)
+  file(GLOB SIMD_FILES "${CMAKE_CURRENT_LIST_DIR}/../"
+    "src/dsp/*_wasm.c"
+  )
+  foreach(FILE ${SIMD_FILES})
+    list(APPEND WEBP_SIMD_FILES_TO_INCLUDE ${FILE})
+  endforeach()
+endif()

configure.ac

@@ -1,4 +1,4 @@
-AC_INIT([libwebp], [1.0.0],
+AC_INIT([libwebp], [0.6.0],
         [https://bugs.chromium.org/p/webp],,
         [http://developers.google.com/speed/webp])
 AC_CANONICAL_HOST
@@ -79,7 +79,6 @@ TEST_AND_ADD_CFLAGS([AM_CFLAGS], [-Wold-style-definition])
 TEST_AND_ADD_CFLAGS([AM_CFLAGS], [-Wparentheses-equality])
 TEST_AND_ADD_CFLAGS([AM_CFLAGS], [-Wshadow])
 TEST_AND_ADD_CFLAGS([AM_CFLAGS], [-Wshorten-64-to-32])
-TEST_AND_ADD_CFLAGS([AM_CFLAGS], [-Wundef])
 TEST_AND_ADD_CFLAGS([AM_CFLAGS], [-Wunreachable-code])
 TEST_AND_ADD_CFLAGS([AM_CFLAGS], [-Wunused-but-set-variable])
 TEST_AND_ADD_CFLAGS([AM_CFLAGS], [-Wunused])
@@ -347,8 +346,6 @@ AS_IF([test "x$enable_gl" != "xno"], [
       # override with --with-gl*
       glut_cflags="$glut_cflags|-framework GLUT -framework OpenGL"
       glut_ldflags="$glut_ldflags|-framework GLUT -framework OpenGL"
-      # quiet deprecation warnings for glut
-      TEST_AND_ADD_CFLAGS([AM_CFLAGS], [-Wno-deprecated-declarations])
       ;;
   esac
 
@@ -445,54 +442,31 @@ AC_ARG_ENABLE([sdl],
                               @<:@default=auto@:>@]))
 AS_IF([test "x$enable_sdl" != "xno"], [
   CLEAR_LIBVARS([SDL])
-  AC_PATH_PROGS([LIBSDL_CONFIG], [sdl-config])
-  if test -n "$LIBSDL_CONFIG"; then
-    SDL_INCLUDES=`$LIBSDL_CONFIG --cflags`
-    SDL_LIBS="`$LIBSDL_CONFIG --libs`"
-  fi
-
   WITHLIB_OPTION([sdl], [SDL])
 
-  sdl_header="no"
+  $sdl_header = "no";
   LIBCHECK_PROLOGUE([SDL])
-  AC_CHECK_HEADER([SDL/SDL.h], [sdl_header="SDL/SDL.h"],
+  AC_CHECK_HEADER([SDL/SDL.h], [sdl_header="SDL_SDL.h"],
                   [AC_CHECK_HEADER([SDL.h], [sdl_header="SDL.h"],
                   [AC_MSG_WARN(SDL library not available - no sdl.h)])])
-  if test x"$sdl_header" != "xno"; then
+  if test x"$sdl_header" != "xno" ; then
-    AC_LANG_PUSH(C)
+    AC_CHECK_LIB(SDL, SDL_Init,
-    SDL_SAVED_LIBS="$LIBS"
+                 [SDL_LIBS="-lSDL"
-    for lib in "" "-lSDL" "-lSDLmain -lSDL"; do
+                  SDL_INCLUDES="-DWEBP_HAVE_SDL"
-      LIBS="$SDL_SAVED_LIBS $lib"
+                  AC_DEFINE(WEBP_HAVE_SDL, [1],
-      # Perform a full link to ensure SDL_main is resolved if needed.
+                            [Set to 1 if SDL library is installed])
-      AC_LINK_IFELSE(
+                  sdl_support=yes
-        [AC_LANG_SOURCE([
+                 ],
-           #include <$sdl_header>
+                 AC_MSG_WARN(Optional SDL library not found),
-           int main(int argc, char** argv) {
+                 [$MATH_LIBS]),
-             SDL_Init(0);
+    if test x"$sdl_header" == "xSDL.h" ; then
-             return 0;
-           }])],
-        [SDL_LIBS="$LDFLAGS $LIBS"
-         SDL_INCLUDES="$SDL_INCLUDES -DWEBP_HAVE_SDL"
-         AC_DEFINE(WEBP_HAVE_SDL, [1],
-                   [Set to 1 if SDL library is installed])
-         sdl_support=yes]
-      )
-      if test x"$sdl_support" = "xyes"; then
-        break
-      fi
-    done
-    # LIBS is restored by LIBCHECK_EPILOGUE
-    AC_LANG_POP
-    if test x"$sdl_header" = "xSDL.h"; then
       SDL_INCLUDES="$SDL_INCLUDES -DWEBP_HAVE_JUST_SDL_H"
     fi
   fi
   LIBCHECK_EPILOGUE([SDL])
 
-  if test x"$sdl_support" = "xyes"; then
+  if test "$sdl_support" = "yes" ; then
     build_vwebp_sdl=yes
-  else
-    AC_MSG_WARN(Optional SDL library not found)
   fi
 ])
 
@@ -615,7 +589,7 @@ AS_IF([test "x$enable_gif" != "xno"], [
 
   if test "$gif_support" = "yes" -a \
           "$enable_libwebpdemux" = "yes"; then
-    build_anim_diff=yes
+    build_animdiff=yes
   fi
 
   if test "$gif_support" = "yes" -a \
@@ -623,10 +597,10 @@ AS_IF([test "x$enable_gif" != "xno"], [
     build_gif2webp=yes
   fi
 ])
-AM_CONDITIONAL([BUILD_ANIMDIFF], [test "${build_anim_diff}" = "yes"])
+AM_CONDITIONAL([BUILD_ANIMDIFF], [test "${build_animdiff}" = "yes"])
 AM_CONDITIONAL([BUILD_GIF2WEBP], [test "${build_gif2webp}" = "yes"])
 
-if test "$enable_libwebpdemux" = "yes" -a "$enable_libwebpmux" = "yes"; then
+if test "$enable_libwebpmux" = "yes"; then
   build_img2webp=yes
 fi
 AM_CONDITIONAL([BUILD_IMG2WEBP], [test "${build_img2webp}" = "yes"])
@@ -688,7 +662,7 @@ if test "$enable_wic" = "yes"; then
 fi
 esac
 
-dnl === If --enable-swap-16bit-csp is defined, add -DWEBP_SWAP_16BIT_CSP=1
+dnl === If --enable-swap-16bit-csp is defined, add -DWEBP_SWAP_16BIT_CSP
 
 USE_SWAP_16BIT_CSP=""
 AC_MSG_CHECKING(if --enable-swap-16bit-csp option is specified)
@@ -696,25 +670,23 @@ AC_ARG_ENABLE([swap-16bit-csp],
               AS_HELP_STRING([--enable-swap-16bit-csp],
                              [Enable byte swap for 16 bit colorspaces]))
 if test "$enable_swap_16bit_csp" = "yes"; then
-  USE_SWAP_16BIT_CSP="-DWEBP_SWAP_16BIT_CSP=1"
+  USE_SWAP_16BIT_CSP="-DWEBP_SWAP_16BIT_CSP"
 fi
 AC_MSG_RESULT(${enable_swap_16bit_csp-no})
 AC_SUBST(USE_SWAP_16BIT_CSP)
 
-dnl === If --disable-near-lossless is defined, add -DWEBP_NEAR_LOSSLESS=0
+dnl === If --enable-experimental is defined, add -DWEBP_EXPERIMENTAL_FEATURES
 
-AC_DEFINE(WEBP_NEAR_LOSSLESS, [1], [Enable near lossless encoding])
+USE_EXPERIMENTAL_CODE=""
-AC_MSG_CHECKING(if --disable-near-lossless option is specified)
+AC_MSG_CHECKING(if --enable-experimental option is specified)
-AC_ARG_ENABLE([near_lossless],
+AC_ARG_ENABLE([experimental], AS_HELP_STRING([--enable-experimental],
-              AS_HELP_STRING([--disable-near-lossless],
+                                             [Activate experimental features]))
-                             [Disable near lossless encoding]),
+if test "$enable_experimental" = "yes"; then
-              [], [enable_near_lossless=yes])
+  AC_DEFINE(WEBP_EXPERIMENTAL_FEATURES, [1], [Enable experimental code])
-if test "$enable_near_lossless" = "no"; then
+  USE_EXPERIMENTAL_CODE="-DWEBP_EXPERIMENTAL_FEATURES"
-  AC_DEFINE(WEBP_NEAR_LOSSLESS, [0], [Enable near lossless encoding])
-  AC_MSG_RESULT([yes])
-else
-  AC_MSG_RESULT([no])
 fi
+AC_MSG_RESULT(${enable_experimental-no})
+AC_SUBST(USE_EXPERIMENTAL_CODE)
 
 dnl === Check whether libwebpmux should be built
 AC_MSG_CHECKING(whether libwebpmux is to be built)
@@ -727,9 +699,8 @@ AM_CONDITIONAL([WANT_MUX], [test "$enable_libwebpmux" = "yes"])
 dnl === Check whether libwebpdemux should be built
 AC_MSG_CHECKING(whether libwebpdemux is to be built)
 AC_ARG_ENABLE([libwebpdemux],
-              AS_HELP_STRING([--disable-libwebpdemux],
+              AS_HELP_STRING([--enable-libwebpdemux],
-                             [Disable libwebpdemux @<:@default=no@:>@]),
+                             [Build libwebpdemux @<:@default=no@:>@]))
-              [], [enable_libwebpdemux=yes])
 AC_MSG_RESULT(${enable_libwebpdemux-no})
 AM_CONDITIONAL([WANT_DEMUX], [test "$enable_libwebpdemux" = "yes"])
 
@@ -778,20 +749,20 @@ libwebpmux: ${enable_libwebpmux-no}
 libwebpextras: ${enable_libwebpextras-no}
 
 Tools:
-cwebp : ${enable_libwebpdemux-no}
+cwebp : yes
   Input format support
   ====================
   JPEG : ${jpeg_support-no}
   PNG  : ${png_support-no}
   TIFF : ${tiff_support-no}
   WIC  : ${wic_support-no}
-dwebp : ${enable_libwebpdemux-no}
+dwebp : yes
   Output format support
   =====================
   PNG  : ${png_support-no}
   WIC  : ${wic_support-no}
 GIF support : ${gif_support-no}
-anim_diff   : ${build_anim_diff-no}
+anim_diff   : ${build_animdiff-no}
 gif2webp    : ${build_gif2webp-no}
 img2webp    : ${build_img2webp-no}
 webpmux     : ${enable_libwebpmux-no}

doc/webp-container-spec.txt

@@ -446,9 +446,8 @@ Frame Height Minus One: 24 bits (_uint24_)
 Frame Duration: 24 bits (_uint24_)
 
 : The time to wait before displaying the next frame, in 1 millisecond units.
-  Note the interpretation of frame duration of 0 (and often <= 10) is
+  In particular, frame duration of 0 is useful when one wants to update
-  implementation defined. Many tools and browsers assign a minimum duration
+  multiple areas of the canvas at once during the animation.
-  similar to GIF.
 
 Reserved: 6 bits
 

examples/Android.mk

@@ -27,7 +27,7 @@ LOCAL_SRC_FILES := \
 
 LOCAL_CFLAGS := $(WEBP_CFLAGS)
 LOCAL_C_INCLUDES := $(LOCAL_PATH)/../src
-LOCAL_STATIC_LIBRARIES := example_util imageio_util imagedec webpdemux webp
+LOCAL_STATIC_LIBRARIES := example_util imageio_util imagedec webp
 
 LOCAL_MODULE := cwebp
 
@@ -43,7 +43,8 @@ LOCAL_SRC_FILES := \
 
 LOCAL_CFLAGS := $(WEBP_CFLAGS)
 LOCAL_C_INCLUDES := $(LOCAL_PATH)/../src
-LOCAL_STATIC_LIBRARIES := example_util imagedec imageenc webpdemux webp
+LOCAL_STATIC_LIBRARIES := example_util imagedec imageenc webp
+
 LOCAL_MODULE := dwebp
 
 include $(BUILD_EXECUTABLE)
@@ -74,8 +75,7 @@ LOCAL_SRC_FILES := \
 
 LOCAL_CFLAGS := $(WEBP_CFLAGS)
 LOCAL_C_INCLUDES := $(LOCAL_PATH)/../src
-LOCAL_STATIC_LIBRARIES := example_util imageio_util imagedec webpmux webpdemux \
+LOCAL_STATIC_LIBRARIES := example_util imageio_util imagedec webpmux webp
-                          webp
 
 LOCAL_MODULE := img2webp_example
 

examples/Makefile.am

@@ -1,11 +1,8 @@
 AM_CPPFLAGS += -I$(top_builddir)/src -I$(top_srcdir)/src
 
-bin_PROGRAMS =
+bin_PROGRAMS = dwebp cwebp
-if WANT_DEMUX
-  bin_PROGRAMS += dwebp cwebp
-endif
 if BUILD_ANIMDIFF
-  noinst_PROGRAMS = anim_diff anim_dump
+  noinst_PROGRAMS = anim_diff
 endif
 if BUILD_GIF2WEBP
   bin_PROGRAMS += gif2webp
@@ -29,37 +26,21 @@ libexample_util_la_SOURCES = example_util.c example_util.h
 libexample_util_la_LIBADD = ../src/libwebp.la
 
 anim_diff_SOURCES = anim_diff.c anim_util.c anim_util.h
-anim_diff_CPPFLAGS = $(AM_CPPFLAGS) $(GIF_INCLUDES)
+anim_diff_CPPFLAGS = $(AM_CPPFLAGS) $(USE_EXPERIMENTAL_CODE) $(GIF_INCLUDES)
-anim_diff_LDADD  =
+anim_diff_LDADD  = ../src/demux/libwebpdemux.la
-anim_diff_LDADD += ../src/demux/libwebpdemux.la
+anim_diff_LDADD += libexample_util.la ../imageio/libimageio_util.la
-anim_diff_LDADD += libexample_util.la
-anim_diff_LDADD += ../imageio/libimageio_util.la
 anim_diff_LDADD += $(GIF_LIBS) -lm
 
-anim_dump_SOURCES = anim_dump.c anim_util.c anim_util.h
-anim_dump_CPPFLAGS = $(AM_CPPFLAGS) $(PNG_INCLUDES)
-anim_dump_CPPFLAGS += $(GIF_INCLUDES)
-anim_dump_LDADD  =
-anim_dump_LDADD += ../src/demux/libwebpdemux.la
-anim_dump_LDADD += libexample_util.la
-anim_dump_LDADD += ../imageio/libimageio_util.la
-anim_dump_LDADD += ../imageio/libimageenc.la
-anim_dump_LDADD += $(PNG_LIBS) $(GIF_LIBS) $(TIFF_LIBS) -lm
-
 cwebp_SOURCES  = cwebp.c stopwatch.h
-cwebp_CPPFLAGS  = $(AM_CPPFLAGS)
+cwebp_CPPFLAGS  = $(AM_CPPFLAGS) $(USE_EXPERIMENTAL_CODE)
-cwebp_LDADD  =
+cwebp_LDADD  = libexample_util.la ../imageio/libimageio_util.la
-cwebp_LDADD += libexample_util.la
+cwebp_LDADD += ../imageio/libimagedec.la ../src/libwebp.la
-cwebp_LDADD += ../imageio/libimageio_util.la
-cwebp_LDADD += ../imageio/libimagedec.la
-cwebp_LDADD += ../src/libwebp.la
 cwebp_LDADD += $(JPEG_LIBS) $(PNG_LIBS) $(TIFF_LIBS)
 
 dwebp_SOURCES = dwebp.c stopwatch.h
-dwebp_CPPFLAGS  = $(AM_CPPFLAGS)
+dwebp_CPPFLAGS  = $(AM_CPPFLAGS) $(USE_EXPERIMENTAL_CODE)
 dwebp_CPPFLAGS += $(JPEG_INCLUDES) $(PNG_INCLUDES)
-dwebp_LDADD  =
+dwebp_LDADD  = libexample_util.la
-dwebp_LDADD += libexample_util.la
 dwebp_LDADD += ../imageio/libimagedec.la
 dwebp_LDADD += ../imageio/libimageenc.la
 dwebp_LDADD += ../imageio/libimageio_util.la
@@ -67,53 +48,36 @@ dwebp_LDADD += ../src/libwebp.la
 dwebp_LDADD +=$(PNG_LIBS) $(JPEG_LIBS)
 
 gif2webp_SOURCES = gif2webp.c gifdec.c gifdec.h
-gif2webp_CPPFLAGS = $(AM_CPPFLAGS) $(GIF_INCLUDES)
+gif2webp_CPPFLAGS = $(AM_CPPFLAGS) $(USE_EXPERIMENTAL_CODE) $(GIF_INCLUDES)
-gif2webp_LDADD  =
+gif2webp_LDADD  = libexample_util.la ../imageio/libimageio_util.la
-gif2webp_LDADD += libexample_util.la
+gif2webp_LDADD += ../src/mux/libwebpmux.la ../src/libwebp.la $(GIF_LIBS)
-gif2webp_LDADD += ../imageio/libimageio_util.la
-gif2webp_LDADD += ../src/mux/libwebpmux.la
-gif2webp_LDADD += ../src/libwebp.la
-gif2webp_LDADD += $(GIF_LIBS)
 
 vwebp_SOURCES = vwebp.c
-vwebp_CPPFLAGS = $(AM_CPPFLAGS) $(GL_INCLUDES)
+vwebp_CPPFLAGS = $(AM_CPPFLAGS) $(USE_EXPERIMENTAL_CODE) $(GL_INCLUDES)
-vwebp_LDADD  =
+vwebp_LDADD  = libexample_util.la ../imageio/libimageio_util.la
-vwebp_LDADD += libexample_util.la
+vwebp_LDADD += ../src/demux/libwebpdemux.la $(GL_LIBS)
-vwebp_LDADD += ../imageio/libimageio_util.la
-vwebp_LDADD += ../src/demux/libwebpdemux.la
-vwebp_LDADD += $(GL_LIBS)
 
 webpmux_SOURCES = webpmux.c
-webpmux_CPPFLAGS = $(AM_CPPFLAGS)
+webpmux_CPPFLAGS = $(AM_CPPFLAGS) $(USE_EXPERIMENTAL_CODE)
-webpmux_LDADD  =
+webpmux_LDADD  = libexample_util.la ../imageio/libimageio_util.la
-webpmux_LDADD += libexample_util.la
+webpmux_LDADD += ../src/mux/libwebpmux.la ../src/libwebp.la
-webpmux_LDADD += ../imageio/libimageio_util.la
-webpmux_LDADD += ../src/mux/libwebpmux.la
-webpmux_LDADD += ../src/libwebp.la
 
 img2webp_SOURCES = img2webp.c
-img2webp_CPPFLAGS = $(AM_CPPFLAGS)
+img2webp_CPPFLAGS = $(AM_CPPFLAGS) $(USE_EXPERIMENTAL_CODE)
-img2webp_LDADD  =
+img2webp_LDADD  = libexample_util.la ../imageio/libimageio_util.la
-img2webp_LDADD += libexample_util.la
-img2webp_LDADD += ../imageio/libimageio_util.la
 img2webp_LDADD += ../imageio/libimagedec.la
-img2webp_LDADD += ../src/mux/libwebpmux.la
+img2webp_LDADD += ../src/mux/libwebpmux.la ../src/libwebp.la
-img2webp_LDADD += ../src/libwebp.la
 img2webp_LDADD += $(PNG_LIBS) $(JPEG_LIBS) $(TIFF_LIBS)
 
 webpinfo_SOURCES = webpinfo.c
-webpinfo_CPPFLAGS = $(AM_CPPFLAGS)
+webpinfo_CPPFLAGS = $(AM_CPPFLAGS) $(USE_EXPERIMENTAL_CODE)
-webpinfo_LDADD  =
+webpinfo_LDADD  = libexample_util.la ../imageio/libimageio_util.la
-webpinfo_LDADD += libexample_util.la
-webpinfo_LDADD += ../imageio/libimageio_util.la
 webpinfo_LDADD += ../src/libwebp.la
 
 if BUILD_LIBWEBPDECODER
   anim_diff_LDADD += ../src/libwebpdecoder.la
-  anim_dump_LDADD += ../src/libwebpdecoder.la
   vwebp_LDADD += ../src/libwebpdecoder.la
 else
   anim_diff_LDADD += ../src/libwebp.la
-  anim_dump_LDADD += ../src/libwebp.la
   vwebp_LDADD += ../src/libwebp.la
 endif

examples/anim_diff.c

@@ -187,11 +187,11 @@ static void Help(void) {
   printf("  -min_psnr <float> ... minimum per-frame PSNR\n");
   printf("  -raw_comparison ..... if this flag is not used, RGB is\n");
   printf("                        premultiplied before comparison\n");
-  printf("  -max_diff <int> ..... maximum allowed difference per channel\n"
+#ifdef WEBP_EXPERIMENTAL_FEATURES
-         "                        between corresponding pixels in subsequent\n"
+  printf("  -max_diff <int> ..... maximum allowed difference per channel "
+         "                        between corresponding pixels in subsequent"
          "                        frames\n");
-  printf("  -h .................. this help\n");
+#endif
-  printf("  -version ............ print version number and exit\n");
 }
 
 int main(int argc, const char* argv[]) {
@@ -207,6 +207,11 @@ int main(int argc, const char* argv[]) {
   const char* files[2] = { NULL, NULL };
   AnimatedImage images[2];
 
+  if (argc < 3) {
+    Help();
+    return -1;
+  }
+
   for (c = 1; c < argc; ++c) {
     int parse_error = 0;
     if (!strcmp(argv[c], "-dump_frames")) {
@@ -231,6 +236,7 @@ int main(int argc, const char* argv[]) {
       }
     } else if (!strcmp(argv[c], "-raw_comparison")) {
       premultiply = 0;
+#ifdef WEBP_EXPERIMENTAL_FEATURES
     } else if (!strcmp(argv[c], "-max_diff")) {
       if (c < argc - 1) {
         const char* const v = argv[++c];
@@ -244,18 +250,7 @@ int main(int argc, const char* argv[]) {
       } else {
         parse_error = 1;
       }
-    } else if (!strcmp(argv[c], "-h") || !strcmp(argv[c], "-help")) {
+#endif
-      Help();
-      return 0;
-    } else if (!strcmp(argv[c], "-version")) {
-      int dec_version, demux_version;
-      GetAnimatedImageVersions(&dec_version, &demux_version);
-      printf("WebP Decoder version: %d.%d.%d\nWebP Demux version: %d.%d.%d\n",
-             (dec_version >> 16) & 0xff, (dec_version >> 8) & 0xff,
-             (dec_version >> 0) & 0xff,
-             (demux_version >> 16) & 0xff, (demux_version >> 8) & 0xff,
-             (demux_version >> 0) & 0xff);
-      return 0;
     } else {
       if (!got_input1) {
         files[0] = argv[c];
@@ -272,12 +267,6 @@ int main(int argc, const char* argv[]) {
       return -1;
     }
   }
-  if (argc < 3) {
-    Help();
-    return -1;
-  }
-
-
   if (!got_input2) {
     Help();
     return -1;

examples/anim_dump.c

@@ -1,118 +0,0 @@
-// Copyright 2017 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.
-// -----------------------------------------------------------------------------
-//
-// Decodes an animated WebP file and dumps the decoded frames as PNG or TIFF.
-//
-// Author: Skal (pascal.massimino@gmail.com)
-
-#include <stdio.h>
-#include <string.h>  // for 'strcmp'.
-
-#include "./anim_util.h"
-#include "webp/decode.h"
-#include "../imageio/image_enc.h"
-
-#if defined(_MSC_VER) && _MSC_VER < 1900
-#define snprintf _snprintf
-#endif
-
-static void Help(void) {
-  printf("Usage: anim_dump [options] files...\n");
-  printf("\nOptions:\n");
-  printf("  -folder <string> .... dump folder (default: '.')\n");
-  printf("  -prefix <string> .... prefix for dumped frames "
-                                  "(default: 'dump_')\n");
-  printf("  -tiff ............... save frames as TIFF\n");
-  printf("  -pam ................ save frames as PAM\n");
-  printf("  -h .................. this help\n");
-  printf("  -version ............ print version number and exit\n");
-}
-
-int main(int argc, const char* argv[]) {
-  int error = 0;
-  const char* dump_folder = ".";
-  const char* prefix = "dump_";
-  const char* suffix = "png";
-  WebPOutputFileFormat format = PNG;
-  int c;
-
-  if (argc < 2) {
-    Help();
-    return -1;
-  }
-
-  for (c = 1; !error && c < argc; ++c) {
-    if (!strcmp(argv[c], "-folder")) {
-      if (c + 1 == argc) {
-        fprintf(stderr, "missing argument after option '%s'\n", argv[c]);
-        error = 1;
-        break;
-      }
-      dump_folder = argv[++c];
-    } else if (!strcmp(argv[c], "-prefix")) {
-      if (c + 1 == argc) {
-        fprintf(stderr, "missing argument after option '%s'\n", argv[c]);
-        error = 1;
-        break;
-      }
-      prefix = argv[++c];
-    } else if (!strcmp(argv[c], "-tiff")) {
-      format = TIFF;
-      suffix = "tiff";
-    } else if (!strcmp(argv[c], "-pam")) {
-      format = PAM;
-      suffix = "pam";
-    } else if (!strcmp(argv[c], "-h") || !strcmp(argv[c], "-help")) {
-      Help();
-      return 0;
-    } else if (!strcmp(argv[c], "-version")) {
-      int dec_version, demux_version;
-      GetAnimatedImageVersions(&dec_version, &demux_version);
-      printf("WebP Decoder version: %d.%d.%d\nWebP Demux version: %d.%d.%d\n",
-             (dec_version >> 16) & 0xff, (dec_version >> 8) & 0xff,
-             (dec_version >> 0) & 0xff,
-             (demux_version >> 16) & 0xff, (demux_version >> 8) & 0xff,
-             (demux_version >> 0) & 0xff);
-      return 0;
-    } else {
-      uint32_t i;
-      AnimatedImage image;
-      const char* const file = argv[c];
-      memset(&image, 0, sizeof(image));
-      printf("Decoding file: %s as %s/%sxxxx.%s\n",
-             file, dump_folder, prefix, suffix);
-      if (!ReadAnimatedImage(file, &image, 0, NULL)) {
-        fprintf(stderr, "Error decoding file: %s\n Aborting.\n", file);
-        error = 1;
-        break;
-      }
-      for (i = 0; !error && i < image.num_frames; ++i) {
-        char out_file[1024];
-        WebPDecBuffer buffer;
-        WebPInitDecBuffer(&buffer);
-        buffer.colorspace = MODE_RGBA;
-        buffer.is_external_memory = 1;
-        buffer.width = image.canvas_width;
-        buffer.height = image.canvas_height;
-        buffer.u.RGBA.rgba = image.frames[i].rgba;
-        buffer.u.RGBA.stride = buffer.width * sizeof(uint32_t);
-        buffer.u.RGBA.size = buffer.u.RGBA.stride * buffer.height;
-        snprintf(out_file, sizeof(out_file), "%s/%s%.4d.%s",
-                 dump_folder, prefix, i, suffix);
-        if (!WebPSaveImage(&buffer, format, out_file)) {
-          fprintf(stderr, "Error while saving image '%s'\n", out_file);
-          error = 1;
-        }
-        WebPFreeDecBuffer(&buffer);
-      }
-      ClearAnimatedImage(&image);
-    }
-  }
-  return error ? 1 : 0;
-}

examples/anim_util.c

@@ -16,7 +16,7 @@
 #include <stdio.h>
 #include <string.h>
 
-#if defined(WEBP_HAVE_GIF)
+#ifdef WEBP_HAVE_GIF
 #include <gif_lib.h>
 #endif
 #include "webp/format_constants.h"
@@ -33,13 +33,11 @@ static const int kNumChannels = 4;
 // -----------------------------------------------------------------------------
 // Common utilities.
 
-#if defined(WEBP_HAVE_GIF)
 // Returns true if the frame covers the full canvas.
 static int IsFullFrame(int width, int height,
                        int canvas_width, int canvas_height) {
   return (width == canvas_width && height == canvas_height);
 }
-#endif // WEBP_HAVE_GIF
 
 static int CheckSizeForOverflow(uint64_t size) {
   return (size == (size_t)size);
@@ -87,7 +85,6 @@ void ClearAnimatedImage(AnimatedImage* const image) {
   }
 }
 
-#if defined(WEBP_HAVE_GIF)
 // Clear the canvas to transparent.
 static void ZeroFillCanvas(uint8_t* rgba,
                            uint32_t canvas_width, uint32_t canvas_height) {
@@ -129,7 +126,6 @@ static void CopyFrameRectangle(const uint8_t* src, uint8_t* dst, int stride,
     dst += stride;
   }
 }
-#endif // WEBP_HAVE_GIF
 
 // Canonicalize all transparent pixels to transparent black to aid comparison.
 static void CleanupTransparentPixels(uint32_t* rgba,
@@ -156,8 +152,6 @@ static int DumpFrame(const char filename[], const char dump_folder[],
   FILE* f = NULL;
   const char* row;
 
-  if (dump_folder == NULL) dump_folder = ".";
-
   base_name = strrchr(filename, '/');
   base_name = (base_name == NULL) ? filename : base_name + 1;
   max_len = strlen(dump_folder) + 1 + strlen(base_name)
@@ -206,7 +200,7 @@ static int IsWebP(const WebPData* const webp_data) {
   return (WebPGetInfo(webp_data->bytes, webp_data->size, NULL, NULL) != 0);
 }
 
-// Read animated WebP bitstream 'webp_data' into 'AnimatedImage' struct.
+// Read animated WebP bitstream 'file_str' into 'AnimatedImage' struct.
 static int ReadAnimatedWebP(const char filename[],
                             const WebPData* const webp_data,
                             AnimatedImage* const image, int dump_frames,
@@ -284,7 +278,7 @@ static int ReadAnimatedWebP(const char filename[],
 // -----------------------------------------------------------------------------
 // GIF Decoding.
 
-#if defined(WEBP_HAVE_GIF)
+#ifdef WEBP_HAVE_GIF
 
 // Returns true if this is a valid GIF bitstream.
 static int IsGIF(const WebPData* const data) {
@@ -429,11 +423,6 @@ static uint32_t GetBackgroundColorGIF(GifFileType* gif) {
 }
 
 // Find appropriate app extension and get loop count from the next extension.
-// We use Chrome's interpretation of the 'loop_count' semantics:
-//   if not present -> loop once
-//   if present and loop_count == 0, return 0 ('infinite').
-//   if present and loop_count != 0, it's the number of *extra* loops
-//     so we need to return loop_count + 1 as total loop number.
 static uint32_t GetLoopCountGIF(const GifFileType* const gif) {
   int i;
   for (i = 0; i < gif->ImageCount; ++i) {
@@ -451,13 +440,12 @@ static uint32_t GetLoopCountGIF(const GifFileType* const gif) {
       if (signature_is_ok &&
           eb2->Function == CONTINUE_EXT_FUNC_CODE && eb2->ByteCount >= 3 &&
           eb2->Bytes[0] == 1) {
-        const uint32_t extra_loop = ((uint32_t)(eb2->Bytes[2]) << 8) +
+        return ((uint32_t)(eb2->Bytes[2]) << 8) +
-                                    ((uint32_t)(eb2->Bytes[1]) << 0);
+               ((uint32_t)(eb2->Bytes[1]) << 0);
-        return (extra_loop > 0) ? extra_loop + 1 : 0;
       }
     }
   }
-  return 1;  // Default.
+  return 0;  // Default.
 }
 
 // Get duration of 'n'th frame in milliseconds.
@@ -593,9 +581,6 @@ static int ReadAnimatedGIF(const char filename[], AnimatedImage* const image,
     curr_frame = &image->frames[i];
     curr_rgba = curr_frame->rgba;
     curr_frame->duration = GetFrameDurationGIF(gif, i);
-    // Force frames with a small or no duration to 100ms to be consistent
-    // with web browsers and other transcoding tools (like gif2webp itself).
-    if (curr_frame->duration <= 10) curr_frame->duration = 100;
 
     if (i == 0) {  // Initialize as transparent.
       curr_frame->is_key_frame = 1;
@@ -786,9 +771,3 @@ void GetDiffAndPSNR(const uint8_t rgba1[], const uint8_t rgba2[],
     *psnr = 4.3429448 * log(255. * 255. / sse);
   }
 }
-
-void GetAnimatedImageVersions(int* const decoder_version,
-                              int* const demux_version) {
-  *decoder_version = WebPGetDecoderVersion();
-  *demux_version = WebPGetDemuxVersion();
-}

examples/anim_util.h

@@ -56,10 +56,6 @@ void GetDiffAndPSNR(const uint8_t rgba1[], const uint8_t rgba2[],
                     uint32_t width, uint32_t height, int premultiply,
                     int* const max_diff, double* const psnr);
 
-// Return library versions used by anim_util.
-void GetAnimatedImageVersions(int* const decoder_version,
-                              int* const demux_version);
-
 #ifdef __cplusplus
 }    // extern "C"
 #endif

examples/cwebp.c

@@ -140,11 +140,10 @@ static void PrintByteCount(const int bytes[4], int total_size,
   fprintf(stderr, "| %7d  (%.1f%%)\n", total, 100.f * total / total_size);
 }
 
-static void PrintPercents(const int counts[4]) {
+static void PrintPercents(const int counts[4], int total) {
   int s;
-  const int total = counts[0] + counts[1] + counts[2] + counts[3];
   for (s = 0; s < 4; ++s) {
-    fprintf(stderr, "|      %2d%%", (int)(100. * counts[s] / total + .5));
+    fprintf(stderr, "|      %2d%%", 100 * counts[s] / total);
   }
   fprintf(stderr, "| %7d\n", total);
 }
@@ -187,8 +186,7 @@ static void PrintExtraInfoLossless(const WebPPicture* const pic,
   } else {
     fprintf(stderr, "File:      %s\n", file_name);
     fprintf(stderr, "Dimension: %d x %d\n", pic->width, pic->height);
-    fprintf(stderr, "Output:    %d bytes (%.2f bpp)\n", stats->coded_size,
+    fprintf(stderr, "Output:    %d bytes\n", stats->coded_size);
-            8.f * stats->coded_size / pic->width / pic->height);
     PrintFullLosslessInfo(stats, "ARGB");
   }
 }
@@ -209,18 +207,15 @@ static void PrintExtraInfoLossy(const WebPPicture* const pic, int short_output,
             pic->width, pic->height,
             stats->alpha_data_size ? " (with alpha)" : "");
     fprintf(stderr, "Output:    "
-            "%d bytes Y-U-V-All-PSNR %2.2f %2.2f %2.2f   %2.2f dB\n"
+            "%d bytes Y-U-V-All-PSNR %2.2f %2.2f %2.2f   %2.2f dB\n",
-            "           (%.2f bpp)\n",
             stats->coded_size,
-            stats->PSNR[0], stats->PSNR[1], stats->PSNR[2], stats->PSNR[3],
+            stats->PSNR[0], stats->PSNR[1], stats->PSNR[2], stats->PSNR[3]);
-            8.f * stats->coded_size / pic->width / pic->height);
     if (total > 0) {
       int totals[4] = { 0, 0, 0, 0 };
-      fprintf(stderr, "block count:  intra4:     %6d  (%.2f%%)\n"
+      fprintf(stderr, "block count:  intra4: %d\n"
-                      "              intra16:    %6d  (%.2f%%)\n"
+                      "              intra16: %d  (-> %.2f%%)\n",
-                      "              skipped:    %6d  (%.2f%%)\n",
+              num_i4, num_i16, 100.f * num_i16 / total);
-              num_i4, 100.f * num_i4 / total,
+      fprintf(stderr, "              skipped block: %d (%.2f%%)\n",
-              num_i16, 100.f * num_i16 / total,
               num_skip, 100.f * num_skip / total);
       fprintf(stderr, "bytes used:  header:         %6d  (%.1f%%)\n"
                       "             mode-partition: %6d  (%.1f%%)\n",
@@ -244,7 +239,7 @@ static void PrintExtraInfoLossy(const WebPPicture* const pic, int short_output,
         PrintByteCount(stats->residual_bytes[2], stats->coded_size, totals);
       }
       fprintf(stderr, "    macroblocks:  ");
-      PrintPercents(stats->segment_size);
+      PrintPercents(stats->segment_size, total);
       fprintf(stderr, "      quantizer:  ");
       PrintValues(stats->segment_quant);
       fprintf(stderr, "   filter level:  ");
@@ -468,9 +463,8 @@ static int WriteWebPWithMetadata(FILE* const out,
     } else {
       const int is_lossless = !memcmp(webp, "VP8L", kTagSize);
       if (is_lossless) {
-        // Presence of alpha is stored in the 37th bit (29th after the
+        // Presence of alpha is stored in the 29th bit of VP8L data.
-        // signature) of VP8L data.
+        if (webp[kChunkHeaderSize + 3] & (1 << 5)) flags |= kAlphaFlag;
-        if (webp[kChunkHeaderSize + 4] & (1 << 4)) flags |= kAlphaFlag;
       }
       ok = ok && (fwrite(kVP8XHeader, kChunkHeaderSize, 1, out) == 1);
       ok = ok && WriteLE32(out, flags);
@@ -492,10 +486,10 @@ static int WriteWebPWithMetadata(FILE* const out,
       *metadata_written |= METADATA_XMP;
     }
     return ok;
+  } else {
+    // No metadata, just write the original image file.
+    return (fwrite(webp, webp_size, 1, out) == 1);
   }
-
-  // No metadata, just write the original image file.
-  return (fwrite(webp, webp_size, 1, out) == 1);
 }
 
 //------------------------------------------------------------------------------
@@ -585,6 +579,9 @@ static void HelpLong(void) {
   printf("  -near_lossless <int> ... use near-lossless image\n"
          "                           preprocessing (0..100=off), "
          "default=100\n");
+#ifdef WEBP_EXPERIMENTAL_FEATURES  /* not documented yet */
+  printf("  -delta_palette ......... use delta palettization\n");
+#endif  // WEBP_EXPERIMENTAL_FEATURES
   printf("  -hint <string> ......... specify image characteristics hint,\n");
   printf("                           one of: photo, picture or graph\n");
 
@@ -753,6 +750,11 @@ int main(int argc, const char *argv[]) {
     } else if (!strcmp(argv[c], "-near_lossless") && c < argc - 1) {
       config.near_lossless = ExUtilGetInt(argv[++c], 0, &parse_error);
       config.lossless = 1;  // use near-lossless only with lossless
+#ifdef WEBP_EXPERIMENTAL_FEATURES
+    } else if (!strcmp(argv[c], "-delta_palette")) {
+      config.use_delta_palette = 1;
+      config.lossless = 1;  // delta-palette is for lossless only
+#endif  // WEBP_EXPERIMENTAL_FEATURES
     } else if (!strcmp(argv[c], "-hint") && c < argc - 1) {
       ++c;
       if (!strcmp(argv[c], "photo")) {

examples/example_util.c

@@ -12,14 +12,10 @@
 
 #include "./example_util.h"
 
-#include <assert.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
 
-#include "webp/mux_types.h"
-#include "../imageio/imageio_util.h"
-
 //------------------------------------------------------------------------------
 // String parsing
 
@@ -60,68 +56,3 @@ float ExUtilGetFloat(const char* const v, int* const error) {
   }
   return f;
 }
-
-//------------------------------------------------------------------------------
-
-static void ResetCommandLineArguments(int argc, const char* argv[],
-                                      CommandLineArguments* const args) {
-  assert(args != NULL);
-  args->argc_ = argc;
-  args->argv_ = argv;
-  args->own_argv_ = 0;
-  WebPDataInit(&args->argv_data_);
-}
-
-void ExUtilDeleteCommandLineArguments(CommandLineArguments* const args) {
-  if (args != NULL) {
-    if (args->own_argv_) {
-      free((void*)args->argv_);
-      WebPDataClear(&args->argv_data_);
-    }
-    ResetCommandLineArguments(0, NULL, args);
-  }
-}
-
-#define MAX_ARGC 16384
-int ExUtilInitCommandLineArguments(int argc, const char* argv[],
-                                   CommandLineArguments* const args) {
-  if (args == NULL || argv == NULL) return 0;
-  ResetCommandLineArguments(argc, argv, args);
-  if (argc == 1 && argv[0][0] != '-') {
-    char* cur;
-    const char sep[] = " \t\r\n\f\v";
-    if (!ExUtilReadFileToWebPData(argv[0], &args->argv_data_)) {
-      return 0;
-    }
-    args->own_argv_ = 1;
-    args->argv_ = (const char**)malloc(MAX_ARGC * sizeof(*args->argv_));
-    if (args->argv_ == NULL) return 0;
-
-    argc = 0;
-    for (cur = strtok((char*)args->argv_data_.bytes, sep);
-         cur != NULL;
-         cur = strtok(NULL, sep)) {
-      if (argc == MAX_ARGC) {
-        fprintf(stderr, "ERROR: Arguments limit %d reached\n", MAX_ARGC);
-        return 0;
-      }
-      assert(strlen(cur) != 0);
-      args->argv_[argc++] = cur;
-    }
-    args->argc_ = argc;
-  }
-  return 1;
-}
-
-//------------------------------------------------------------------------------
-
-int ExUtilReadFileToWebPData(const char* const filename,
-                             WebPData* const webp_data) {
-  const uint8_t* data;
-  size_t size;
-  if (webp_data == NULL) return 0;
-  if (!ImgIoUtilReadFile(filename, &data, &size)) return 0;
-  webp_data->bytes = data;
-  webp_data->size = size;
-  return 1;
-}

examples/example_util.h

@@ -14,7 +14,6 @@
 #define WEBP_EXAMPLES_EXAMPLE_UTIL_H_
 
 #include "webp/types.h"
-#include "webp/mux_types.h"
 
 #ifdef __cplusplus
 extern "C" {
@@ -36,33 +35,6 @@ float ExUtilGetFloat(const char* const v, int* const error);
 // actually parsed is returned, or -1 if an error occurred.
 int ExUtilGetInts(const char* v, int base, int max_output, int output[]);
 
-// Reads a file named 'filename' into a WebPData structure. The content of
-// webp_data is overwritten. Returns false in case of error.
-int ExUtilReadFileToWebPData(const char* const filename,
-                             WebPData* const webp_data);
-
-//------------------------------------------------------------------------------
-// Command-line arguments
-
-typedef struct {
-  int argc_;
-  const char** argv_;
-  WebPData argv_data_;
-  int own_argv_;
-} CommandLineArguments;
-
-// Initializes the structure from the command-line parameters. If there is
-// only one parameter and it does not start with a '-', then it is assumed to
-// be a file name. This file will be read and tokenized into command-line
-// arguments. The content of 'args' is overwritten.
-// Returns false in case of error (memory allocation failure, non
-// existing file, too many arguments, ...).
-int ExUtilInitCommandLineArguments(int argc, const char* argv[],
-                                   CommandLineArguments* const args);
-
-// Deallocate all memory and reset 'args'.
-void ExUtilDeleteCommandLineArguments(CommandLineArguments* const args);
-
 #ifdef __cplusplus
 }    // extern "C"
 #endif

examples/gif2webp.c

@@ -23,10 +23,6 @@
 
 #ifdef WEBP_HAVE_GIF
 
-#if defined(HAVE_UNISTD_H) && HAVE_UNISTD_H
-#include <unistd.h>
-#endif
-
 #include <gif_lib.h>
 #include "webp/encode.h"
 #include "webp/mux.h"
@@ -34,10 +30,6 @@
 #include "../imageio/imageio_util.h"
 #include "./gifdec.h"
 
-#if !defined(STDIN_FILENO)
-#define STDIN_FILENO 0
-#endif
-
 //------------------------------------------------------------------------------
 
 static int transparent_index = GIF_INDEX_INVALID;  // Opaque by default.
@@ -80,10 +72,8 @@ static void Help(void) {
   printf("  -metadata <string> ..... comma separated list of metadata to\n");
   printf("                           ");
   printf("copy from the input to the output if present\n");
-  printf("                           ");
+  printf("                           "
-  printf("Valid values: all, none, icc, xmp (default)\n");
+         "Valid values: all, none, icc, xmp (default)\n");
-  printf("  -loop_compatibility .... use compatibility mode for Chrome\n");
-  printf("                           version prior to M62 (inclusive)\n");
   printf("  -mt .................... use multi-threading if available\n");
   printf("\n");
   printf("  -version ............... print version number and exit\n");
@@ -114,7 +104,7 @@ int main(int argc, const char *argv[]) {
   WebPAnimEncoderOptions enc_options;
   WebPConfig config;
 
-  int frame_number = 0;     // Whether we are processing the first frame.
+  int is_first_frame = 1;     // Whether we are processing the first frame.
   int done;
   int c;
   int quiet = 0;
@@ -125,9 +115,8 @@ int main(int argc, const char *argv[]) {
   int stored_icc = 0;         // Whether we have already stored an ICC profile.
   WebPData xmp_data;
   int stored_xmp = 0;         // Whether we have already stored an XMP profile.
-  int loop_count = 0;         // default: infinite
+  int loop_count = 0;
   int stored_loop_count = 0;  // Whether we have found an explicit loop count.
-  int loop_compatibility = 0;
   WebPMux* mux = NULL;
 
   int default_kmin = 1;  // Whether to use default kmin value.
@@ -162,8 +151,6 @@ int main(int argc, const char *argv[]) {
     } else if (!strcmp(argv[c], "-mixed")) {
       enc_options.allow_mixed = 1;
       config.lossless = 0;
-    } else if (!strcmp(argv[c], "-loop_compatibility")) {
-      loop_compatibility = 1;
     } else if (!strcmp(argv[c], "-q") && c < argc - 1) {
       config.quality = ExUtilGetFloat(argv[++c], &parse_error);
     } else if (!strcmp(argv[c], "-m") && c < argc - 1) {
@@ -271,11 +258,9 @@ int main(int argc, const char *argv[]) {
 
   // Start the decoder object
 #if LOCAL_GIF_PREREQ(5,0)
-  gif = !strcmp(in_file, "-") ? DGifOpenFileHandle(STDIN_FILENO, &gif_error)
+  gif = DGifOpenFileName(in_file, &gif_error);
-                              : DGifOpenFileName(in_file, &gif_error);
 #else
-  gif = !strcmp(in_file, "-") ? DGifOpenFileHandle(STDIN_FILENO)
+  gif = DGifOpenFileName(in_file);
-                              : DGifOpenFileName(in_file);
 #endif
   if (gif == NULL) goto End;
 
@@ -292,7 +277,7 @@ int main(int argc, const char *argv[]) {
 
         if (!DGifGetImageDesc(gif)) goto End;
 
-        if (frame_number == 0) {
+        if (is_first_frame) {
           if (verbose) {
             printf("Canvas screen: %d x %d\n", gif->SWidth, gif->SHeight);
           }
@@ -334,6 +319,7 @@ int main(int argc, const char *argv[]) {
                     "a memory error.\n");
             goto End;
           }
+          is_first_frame = 0;
         }
 
         // Some even more broken GIF can have sub-rect with zero width/height.
@@ -350,25 +336,13 @@ int main(int argc, const char *argv[]) {
         GIFBlendFrames(&frame, &gif_rect, &curr_canvas);
 
         if (!WebPAnimEncoderAdd(enc, &curr_canvas, frame_timestamp, &config)) {
-          fprintf(stderr, "Error while adding frame #%d: %s\n", frame_number,
+          fprintf(stderr, "%s\n", WebPAnimEncoderGetError(enc));
-                  WebPAnimEncoderGetError(enc));
-          goto End;
-        } else {
-          ++frame_number;
         }
 
         // Update canvases.
         GIFDisposeFrame(orig_dispose, &gif_rect, &prev_canvas, &curr_canvas);
         GIFCopyPixels(&curr_canvas, &prev_canvas);
 
-        // Force frames with a small or no duration to 100ms to be consistent
-        // with web browsers and other transcoding tools. This also avoids
-        // incorrect durations between frames when padding frames are
-        // discarded.
-        if (frame_duration <= 10) {
-          frame_duration = 100;
-        }
-
         // Update timestamp (for next frame).
         frame_timestamp += frame_duration;
 
@@ -412,7 +386,7 @@ int main(int argc, const char *argv[]) {
               if (verbose) {
                 fprintf(stderr, "Loop count: %d\n", loop_count);
               }
-              stored_loop_count = loop_compatibility ? (loop_count != 0) : 1;
+              stored_loop_count = (loop_count != 0);
             } else {  // An extension containing metadata.
               // We only store the first encountered chunk of each type, and
               // only if requested by the user.
@@ -469,23 +443,6 @@ int main(int argc, const char *argv[]) {
     goto End;
   }
 
-  if (!loop_compatibility) {
-    if (!stored_loop_count) {
-      // if no loop-count element is seen, the default is '1' (loop-once)
-      // and we need to signal it explicitly in WebP. Note however that
-      // in case there's a single frame, we still don't need to store it.
-      if (frame_number > 1) {
-        stored_loop_count = 1;
-        loop_count = 1;
-      }
-    } else if (loop_count > 0) {
-      // adapt GIF's semantic to WebP's (except in the infinite-loop case)
-      loop_count += 1;
-    }
-  }
-  // loop_count of 0 is the default (infinite), so no need to signal it
-  if (loop_count == 0) stored_loop_count = 0;
-
   if (stored_loop_count || stored_icc || stored_xmp) {
     // Re-mux to add loop count and/or metadata as needed.
     mux = WebPMuxCreate(&webp_data, 1);
@@ -550,13 +507,8 @@ int main(int argc, const char *argv[]) {
       goto End;
     }
     if (!quiet) {
-      if (!strcmp(out_file, "-")) {
+      fprintf(stderr, "Saved output file (%d bytes): %s\n",
-        fprintf(stderr, "Saved %d bytes to STDIO\n",
+              (int)webp_data.size, out_file);
-                (int)webp_data.size);
-      } else {
-        fprintf(stderr, "Saved output file (%d bytes): %s\n",
-                (int)webp_data.size, out_file);
-      }
     }
   } else {
     if (!quiet) {

examples/img2webp.c

@@ -48,7 +48,6 @@ static void Help(void) {
   printf(" -mixed ............... use mixed lossy/lossless automatic mode\n");
   printf(" -v ................... verbose mode\n");
   printf(" -h ................... this help\n");
-  printf(" -version ............. print version number and exit\n");
   printf("\n");
 
   printf("Per-frame options (only used for subsequent images input):\n");
@@ -118,13 +117,14 @@ static int SetLoopCount(int loop_count, WebPData* const webp_data) {
 
 //------------------------------------------------------------------------------
 
-int main(int argc, const char* argv[]) {
+int main(int argc, char* argv[]) {
   const char* output = NULL;
   WebPAnimEncoder* enc = NULL;
   int verbose = 0;
   int pic_num = 0;
   int duration = 100;
   int timestamp_ms = 0;
+  int ok = 1;
   int loop_count = 0;
   int width = 0, height = 0;
   WebPAnimEncoderOptions anim_config;
@@ -133,23 +133,17 @@ int main(int argc, const char* argv[]) {
   WebPData webp_data;
   int c;
   int have_input = 0;
-  CommandLineArguments cmd_args;
-  int ok = ExUtilInitCommandLineArguments(argc - 1, argv + 1, &cmd_args);
-  if (!ok) return 1;
-  argc = cmd_args.argc_;
-  argv = cmd_args.argv_;
 
   WebPDataInit(&webp_data);
   if (!WebPAnimEncoderOptionsInit(&anim_config) ||
       !WebPConfigInit(&config) ||
       !WebPPictureInit(&pic)) {
     fprintf(stderr, "Library version mismatch!\n");
-    ok = 0;
+    return 1;
-    goto End;
   }
 
   // 1st pass of option parsing
-  for (c = 0; ok && c < argc; ++c) {
+  for (c = 1; ok && c < argc; ++c) {
     if (argv[c][0] == '-') {
       int parse_error = 0;
       if (!strcmp(argv[c], "-o") && c + 1 < argc) {
@@ -177,15 +171,7 @@ int main(int argc, const char* argv[]) {
         verbose = 1;
       } else if (!strcmp(argv[c], "-h") || !strcmp(argv[c], "-help")) {
         Help();
-        goto End;
+        return 0;
-      } else if (!strcmp(argv[c], "-version")) {
-        const int enc_version = WebPGetEncoderVersion();
-        const int mux_version = WebPGetMuxVersion();
-        printf("WebP Encoder version: %d.%d.%d\nWebP Mux version: %d.%d.%d\n",
-               (enc_version >> 16) & 0xff, (enc_version >> 8) & 0xff,
-               enc_version & 0xff, (mux_version >> 16) & 0xff,
-               (mux_version >> 8) & 0xff, mux_version & 0xff);
-        goto End;
       } else {
         continue;
       }
@@ -198,13 +184,13 @@ int main(int argc, const char* argv[]) {
   }
   if (!have_input) {
     fprintf(stderr, "No input file(s) for generating animation!\n");
-    goto End;
+    return 0;
   }
 
   // image-reading pass
   pic_num = 0;
   config.lossless = 1;
-  for (c = 0; ok && c < argc; ++c) {
+  for (c = 1; ok && c < argc; ++c) {
     if (argv[c] == NULL) continue;
     if (argv[c][0] == '-') {    // parse local options
       int parse_error = 0;
@@ -308,7 +294,7 @@ int main(int argc, const char* argv[]) {
     fprintf(stderr, "[%d frames, %u bytes].\n",
             pic_num, (unsigned int)webp_data.size);
   }
+
   WebPDataClear(&webp_data);
-  ExUtilDeleteCommandLineArguments(&cmd_args);
   return ok ? 0 : 1;
 }

examples/vwebp.c

@@ -205,11 +205,6 @@ static void decode_callback(int what) {
         }
       }
       duration = curr->duration;
-      // Behavior copied from Chrome, cf:
-      // https://cs.chromium.org/chromium/src/third_party/WebKit/Source/
-      // platform/graphics/DeferredImageDecoder.cpp?
-      // rcl=b4c33049f096cd283f32be9a58b9a9e768227c26&l=246
-      if (duration <= 10) duration = 100;
     }
     if (!Decode()) {
       kParams.decoding_error = 1;
@@ -253,9 +248,9 @@ static void HandleKey(unsigned char key, int pos_x, int pos_y) {
       }
     }
   } else if (key == 'i') {
-    // Note: doesn't handle refresh of animation's last-frame (it's quite
-    // more involved to do, since you need to save the previous frame).
     kParams.print_info = 1 - kParams.print_info;
+    // TODO(skal): handle refresh of animation's last-frame too. It's quite
+    // more involved though (need to save the previous frame).
     if (!kParams.has_animation) ClearPreviousFrame();
     glutPostRedisplay();
   } else if (key == 'd') {
@@ -265,8 +260,8 @@ static void HandleKey(unsigned char key, int pos_x, int pos_y) {
 }
 
 static void HandleReshape(int width, int height) {
-  // Note: reshape doesn't preserve aspect ratio, and might
+  // TODO(skal): should we preserve aspect ratio?
-  // be handling larger-than-screen pictures incorrectly.
+  // Also: handle larger-than-screen pictures correctly.
   glViewport(0, 0, width, height);
   glMatrixMode(GL_PROJECTION);
   glLoadIdentity();
@@ -383,23 +378,13 @@ static void HandleDisplay(void) {
     }
   }
   glPopMatrix();
-#if defined(__APPLE__) || defined(_WIN32)
-  glFlush();
-#else
   glutSwapBuffers();
-#endif
 }
 
 static void StartDisplay(void) {
   const int width = kParams.canvas_width;
   const int height = kParams.canvas_height;
-  // TODO(webp:365) GLUT_DOUBLE results in flickering / old frames to be
-  // partially displayed with animated webp + alpha.
-#if defined(__APPLE__) || defined(_WIN32)
-  glutInitDisplayMode(GLUT_RGBA);
-#else
   glutInitDisplayMode(GLUT_DOUBLE | GLUT_RGBA);
-#endif
   glutInitWindowSize(width, height);
   glutCreateWindow("WebP viewer");
   glutDisplayFunc(HandleDisplay);

examples/webpinfo.c

@@ -233,20 +233,20 @@ static int GetSignedBits(const uint8_t* const data, size_t data_size, size_t nb,
   return 1;
 }
 
-#define GET_BITS(v, n)                                 \
+#define GET_BITS(v, n)                               \
-  do {                                                 \
+  do {                                               \
-    if (!GetBits(data, data_size, n, &(v), bit_pos)) { \
+    if (!GetBits(data, data_size, n, &v, bit_pos)) { \
-      LOG_ERROR("Truncated lossy bitstream.");         \
+      LOG_ERROR("Truncated lossy bitstream.");       \
-      return WEBP_INFO_TRUNCATED_DATA;                 \
+      return WEBP_INFO_TRUNCATED_DATA;               \
-    }                                                  \
+    }                                                \
   } while (0)
 
-#define GET_SIGNED_BITS(v, n)                                \
+#define GET_SIGNED_BITS(v, n)                              \
-  do {                                                       \
+  do {                                                     \
-    if (!GetSignedBits(data, data_size, n, &(v), bit_pos)) { \
+    if (!GetSignedBits(data, data_size, n, &v, bit_pos)) { \
-      LOG_ERROR("Truncated lossy bitstream.");               \
+      LOG_ERROR("Truncated lossy bitstream.");             \
-      return WEBP_INFO_TRUNCATED_DATA;                       \
+      return WEBP_INFO_TRUNCATED_DATA;                     \
-    }                                                        \
+    }                                                      \
   } while (0)
 
 static WebPInfoStatus ParseLossySegmentHeader(const WebPInfo* const webp_info,
@@ -340,7 +340,7 @@ static WebPInfoStatus ParseLossyHeader(const ChunkData* const chunk_data,
   WebPInfoStatus status = WEBP_INFO_OK;
   uint64_t bit_position = 0;
   uint64_t* const bit_pos = &bit_position;
-  int colorspace, clamp_type;
+  int color_space, clamp_type;
   printf("  Parsing lossy bitstream...\n");
   // Calling WebPGetFeatures() in ProcessImageChunk() should ensure this.
   assert(chunk_data->size_ >= CHUNK_HEADER_SIZE + 10);
@@ -381,9 +381,9 @@ static WebPInfoStatus ParseLossyHeader(const ChunkData* const chunk_data,
     LOG_ERROR("Bad partition length.");
     return WEBP_INFO_BITSTREAM_ERROR;
   }
-  GET_BITS(colorspace, 1);
+  GET_BITS(color_space, 1);
   GET_BITS(clamp_type, 1);
-  printf("  Color space:      %d\n", colorspace);
+  printf("  Color space:      %d\n", color_space);
   printf("  Clamp type:       %d\n", clamp_type);
   status = ParseLossySegmentHeader(webp_info, data, data_size, bit_pos);
   if (status != WEBP_INFO_OK) return status;
@@ -462,12 +462,12 @@ static int LLGetBits(const uint8_t* const data, size_t data_size, size_t nb,
   return 1;
 }
 
-#define LL_GET_BITS(v, n)                                \
+#define LL_GET_BITS(v, n)                              \
-  do {                                                   \
+  do {                                                 \
-    if (!LLGetBits(data, data_size, n, &(v), bit_pos)) { \
+    if (!LLGetBits(data, data_size, n, &v, bit_pos)) { \
-      LOG_ERROR("Truncated lossless bitstream.");        \
+      LOG_ERROR("Truncated lossless bitstream.");      \
-      return WEBP_INFO_TRUNCATED_DATA;                   \
+      return WEBP_INFO_TRUNCATED_DATA;                 \
-    }                                                    \
+    }                                                  \
   } while (0)
 
 static WebPInfoStatus ParseLosslessTransform(WebPInfo* const webp_info,
@@ -817,8 +817,9 @@ static WebPInfoStatus ProcessImageChunk(const ChunkData* const chunk_data,
     if (webp_info->seen_image_subchunk_) {
       LOG_ERROR("Consecutive VP8/VP8L sub-chunks in an ANMF chunk.");
       return WEBP_INFO_PARSE_ERROR;
+    } else {
+      webp_info->seen_image_subchunk_ = 1;
     }
-    webp_info->seen_image_subchunk_ = 1;
   } else {
     if (webp_info->chunk_counts_[CHUNK_VP8] ||
         webp_info->chunk_counts_[CHUNK_VP8L]) {
@@ -872,9 +873,9 @@ static WebPInfoStatus ProcessALPHChunk(const ChunkData* const chunk_data,
     if (webp_info->seen_alpha_subchunk_) {
       LOG_ERROR("Consecutive ALPH sub-chunks in an ANMF chunk.");
       return WEBP_INFO_PARSE_ERROR;
+    } else {
+      webp_info->seen_alpha_subchunk_ = 1;
     }
-    webp_info->seen_alpha_subchunk_ = 1;
-
     if (webp_info->seen_image_subchunk_) {
       LOG_ERROR("ALPHA sub-chunk detected after VP8 sub-chunk "
                 "in an ANMF chunk.");
@@ -1106,7 +1107,6 @@ static void HelpLong(void) {
          "Note: there could be multiple input files;\n"
          "      options must come before input files.\n"
          "Options:\n"
-         "  -version ........... Print version number and exit.\n"
          "  -quiet ............. Do not show chunk parsing information.\n"
          "  -diag .............. Show parsing error diagnosis.\n"
          "  -summary ........... Show chunk stats summary.\n"
@@ -1140,11 +1140,6 @@ int main(int argc, const char* argv[]) {
       show_summary = 1;
     } else if (!strcmp(argv[c], "-bitstream_info")) {
       parse_bitstream = 1;
-    } else if (!strcmp(argv[c], "-version")) {
-      const int version = WebPGetDecoderVersion();
-      printf("WebP Decoder version: %d.%d.%d\n",
-             (version >> 16) & 0xff, (version >> 8) & 0xff, version & 0xff);
-      return 0;
     } else {  // Assume the remaining are all input files.
       break;
     }

examples/webpmux.c

@@ -47,7 +47,6 @@
     webpmux -info in.webp
     webpmux [ -h | -help ]
     webpmux -version
-    webpmux argument_file_name
 */
 
 #ifdef HAVE_CONFIG_H
@@ -109,26 +108,28 @@ static const char* const kDescriptions[LAST_FEATURE] = {
 };
 
 typedef struct {
-  CommandLineArguments cmd_args_;
-
-  ActionType action_type_;
-  const char* input_;
-  const char* output_;
   FeatureType type_;
   FeatureArg* args_;
   int arg_count_;
-} Config;
+} Feature;
+
+typedef struct {
+  ActionType action_type_;
+  const char* input_;
+  const char* output_;
+  Feature feature_;
+} WebPMuxConfig;
 
 //------------------------------------------------------------------------------
 // Helper functions.
 
-static int CountOccurrences(const CommandLineArguments* const args,
+static int CountOccurrences(const char* arglist[], int list_length,
-                            const char* const arg) {
+                            const char* arg) {
   int i;
   int num_occurences = 0;
 
-  for (i = 0; i < args->argc_; ++i) {
+  for (i = 0; i < list_length; ++i) {
-    if (!strcmp(args->argv_[i], arg)) {
+    if (!strcmp(arglist[i], arg)) {
       ++num_occurences;
     }
   }
@@ -300,7 +301,6 @@ static void PrintHelp(void) {
   printf("       webpmux -info INPUT\n");
   printf("       webpmux [-h|-help]\n");
   printf("       webpmux -version\n");
-  printf("       webpmux argument_file_name\n");
 
   printf("\n");
   printf("GET_OPTIONS:\n");
@@ -369,10 +369,6 @@ static void PrintHelp(void) {
 
   printf("\nNote: The nature of EXIF, XMP and ICC data is not checked");
   printf(" and is assumed to be\nvalid.\n");
-  printf("\nNote: if a single file name is passed as the argument, the "
-         "arguments will be\n");
-  printf("tokenized from this file. The file name must not start with "
-         "the character '-'.\n");
 }
 
 static void WarnAboutOddOffset(const WebPMuxFrameInfo* const info) {
@@ -383,12 +379,22 @@ static void WarnAboutOddOffset(const WebPMuxFrameInfo* const info) {
   }
 }
 
+static int ReadFileToWebPData(const char* const filename,
+                              WebPData* const webp_data) {
+  const uint8_t* data;
+  size_t size;
+  if (!ImgIoUtilReadFile(filename, &data, &size)) return 0;
+  webp_data->bytes = data;
+  webp_data->size = size;
+  return 1;
+}
+
 static int CreateMux(const char* const filename, WebPMux** mux) {
   WebPData bitstream;
   assert(mux != NULL);
-  if (!ExUtilReadFileToWebPData(filename, &bitstream)) return 0;
+  if (!ReadFileToWebPData(filename, &bitstream)) return 0;
   *mux = WebPMuxCreate(&bitstream, 1);
-  WebPDataClear(&bitstream);
+  free((void*)bitstream.bytes);
   if (*mux != NULL) return 1;
   fprintf(stderr, "Failed to create mux object from file %s.\n", filename);
   return 0;
@@ -511,10 +517,9 @@ static int ParseBgcolorArgs(const char* args, uint32_t* const bgcolor) {
 //------------------------------------------------------------------------------
 // Clean-up.
 
-static void DeleteConfig(Config* const config) {
+static void DeleteConfig(WebPMuxConfig* config) {
   if (config != NULL) {
-    free(config->args_);
+    free(config->feature_.args_);
-    ExUtilDeleteCommandLineArguments(&config->cmd_args_);
     memset(config, 0, sizeof(*config));
   }
 }
@@ -526,7 +531,7 @@ static void DeleteConfig(Config* const config) {
 // Returns 1 on valid, 0 otherwise.
 // Also fills up num_feature_args to be number of feature arguments given.
 // (e.g. if there are 4 '-frame's and 1 '-loop', then num_feature_args = 5).
-static int ValidateCommandLine(const CommandLineArguments* const cmd_args,
+static int ValidateCommandLine(int argc, const char* argv[],
                                int* num_feature_args) {
   int num_frame_args;
   int num_loop_args;
@@ -538,27 +543,27 @@ static int ValidateCommandLine(const CommandLineArguments* const cmd_args,
   *num_feature_args = 0;
 
   // Simple checks.
-  if (CountOccurrences(cmd_args, "-get") > 1) {
+  if (CountOccurrences(argv, argc, "-get") > 1) {
     ERROR_GOTO1("ERROR: Multiple '-get' arguments specified.\n", ErrValidate);
   }
-  if (CountOccurrences(cmd_args, "-set") > 1) {
+  if (CountOccurrences(argv, argc, "-set") > 1) {
     ERROR_GOTO1("ERROR: Multiple '-set' arguments specified.\n", ErrValidate);
   }
-  if (CountOccurrences(cmd_args, "-strip") > 1) {
+  if (CountOccurrences(argv, argc, "-strip") > 1) {
     ERROR_GOTO1("ERROR: Multiple '-strip' arguments specified.\n", ErrValidate);
   }
-  if (CountOccurrences(cmd_args, "-info") > 1) {
+  if (CountOccurrences(argv, argc, "-info") > 1) {
     ERROR_GOTO1("ERROR: Multiple '-info' arguments specified.\n", ErrValidate);
   }
-  if (CountOccurrences(cmd_args, "-o") > 1) {
+  if (CountOccurrences(argv, argc, "-o") > 1) {
     ERROR_GOTO1("ERROR: Multiple output files specified.\n", ErrValidate);
   }
 
   // Compound checks.
-  num_frame_args = CountOccurrences(cmd_args, "-frame");
+  num_frame_args = CountOccurrences(argv, argc, "-frame");
-  num_loop_args = CountOccurrences(cmd_args, "-loop");
+  num_loop_args = CountOccurrences(argv, argc, "-loop");
-  num_bgcolor_args = CountOccurrences(cmd_args, "-bgcolor");
+  num_bgcolor_args = CountOccurrences(argv, argc, "-bgcolor");
-  num_durations_args = CountOccurrences(cmd_args, "-duration");
+  num_durations_args = CountOccurrences(argv, argc, "-duration");
 
   if (num_loop_args > 1) {
     ERROR_GOTO1("ERROR: Multiple loop counts specified.\n", ErrValidate);
@@ -593,7 +598,7 @@ static int ValidateCommandLine(const CommandLineArguments* const cmd_args,
 
 #define ACTION_IS_NIL (config->action_type_ == NIL_ACTION)
 
-#define FEATURETYPE_IS_NIL (config->type_ == NIL_FEATURE)
+#define FEATURETYPE_IS_NIL (feature->type_ == NIL_FEATURE)
 
 #define CHECK_NUM_ARGS_LESS(NUM, LABEL)                                  \
   if (argc < i + (NUM)) {                                                \
@@ -609,15 +614,15 @@ static int ValidateCommandLine(const CommandLineArguments* const cmd_args,
 
 // Parses command-line arguments to fill up config object. Also performs some
 // semantic checks.
-static int ParseCommandLine(Config* config) {
+static int ParseCommandLine(int argc, const char* argv[],
+                            WebPMuxConfig* config) {
   int i = 0;
   int feature_arg_index = 0;
   int ok = 1;
-  int argc = config->cmd_args_.argc_;
-  const char* const* argv = config->cmd_args_.argv_;
 
   while (i < argc) {
-    FeatureArg* const arg = &config->args_[feature_arg_index];
+    Feature* const feature = &config->feature_;
+    FeatureArg* const arg = &feature->args_[feature_arg_index];
     if (argv[i][0] == '-') {  // One of the action types or output.
       if (!strcmp(argv[i], "-set")) {
         if (ACTION_IS_NIL) {
@@ -633,8 +638,8 @@ static int ParseCommandLine(Config* config) {
         } else {
           ERROR_GOTO1("ERROR: Multiple actions specified.\n", ErrParse);
         }
-        if (FEATURETYPE_IS_NIL || config->type_ == FEATURE_DURATION) {
+        if (FEATURETYPE_IS_NIL || feature->type_ == FEATURE_DURATION) {
-          config->type_ = FEATURE_DURATION;
+          feature->type_ = FEATURE_DURATION;
         } else {
           ERROR_GOTO1("ERROR: Multiple features specified.\n", ErrParse);
         }
@@ -651,7 +656,7 @@ static int ParseCommandLine(Config* config) {
       } else if (!strcmp(argv[i], "-strip")) {
         if (ACTION_IS_NIL) {
           config->action_type_ = ACTION_STRIP;
-          config->arg_count_ = 0;
+          feature->arg_count_ = 0;
         } else {
           ERROR_GOTO1("ERROR: Multiple actions specified.\n", ErrParse);
         }
@@ -663,8 +668,8 @@ static int ParseCommandLine(Config* config) {
         } else {
           ERROR_GOTO1("ERROR: Multiple actions specified.\n", ErrParse);
         }
-        if (FEATURETYPE_IS_NIL || config->type_ == FEATURE_ANMF) {
+        if (FEATURETYPE_IS_NIL || feature->type_ == FEATURE_ANMF) {
-          config->type_ = FEATURE_ANMF;
+          feature->type_ = FEATURE_ANMF;
         } else {
           ERROR_GOTO1("ERROR: Multiple features specified.\n", ErrParse);
         }
@@ -680,8 +685,8 @@ static int ParseCommandLine(Config* config) {
         } else {
           ERROR_GOTO1("ERROR: Multiple actions specified.\n", ErrParse);
         }
-        if (FEATURETYPE_IS_NIL || config->type_ == FEATURE_ANMF) {
+        if (FEATURETYPE_IS_NIL || feature->type_ == FEATURE_ANMF) {
-          config->type_ = FEATURE_ANMF;
+          feature->type_ = FEATURE_ANMF;
         } else {
           ERROR_GOTO1("ERROR: Multiple features specified.\n", ErrParse);
         }
@@ -700,7 +705,7 @@ static int ParseCommandLine(Config* config) {
           ERROR_GOTO1("ERROR: Multiple actions specified.\n", ErrParse);
         } else {
           config->action_type_ = ACTION_INFO;
-          config->arg_count_ = 0;
+          feature->arg_count_ = 0;
           config->input_ = argv[i + 1];
         }
         i += 2;
@@ -736,7 +741,7 @@ static int ParseCommandLine(Config* config) {
       if (!strcmp(argv[i], "icc") || !strcmp(argv[i], "exif") ||
           !strcmp(argv[i], "xmp")) {
         if (FEATURETYPE_IS_NIL) {
-          config->type_ = (!strcmp(argv[i], "icc")) ? FEATURE_ICCP :
+          feature->type_ = (!strcmp(argv[i], "icc")) ? FEATURE_ICCP :
               (!strcmp(argv[i], "exif")) ? FEATURE_EXIF : FEATURE_XMP;
         } else {
           ERROR_GOTO1("ERROR: Multiple features specified.\n", ErrParse);
@@ -752,7 +757,7 @@ static int ParseCommandLine(Config* config) {
       } else if (!strcmp(argv[i], "frame") &&
                  (config->action_type_ == ACTION_GET)) {
         CHECK_NUM_ARGS_LESS(2, ErrParse);
-        config->type_ = FEATURE_ANMF;
+        feature->type_ = FEATURE_ANMF;
         arg->params_ = argv[i + 1];
         ++feature_arg_index;
         i += 2;
@@ -772,8 +777,9 @@ static int ParseCommandLine(Config* config) {
 }
 
 // Additional checks after config is filled.
-static int ValidateConfig(Config* const config) {
+static int ValidateConfig(WebPMuxConfig* config) {
   int ok = 1;
+  Feature* const feature = &config->feature_;
 
   // Action.
   if (ACTION_IS_NIL) {
@@ -789,7 +795,7 @@ static int ValidateConfig(Config* const config) {
   if (config->input_ == NULL) {
     if (config->action_type_ != ACTION_SET) {
       ERROR_GOTO1("ERROR: No input file specified.\n", ErrValidate2);
-    } else if (config->type_ != FEATURE_ANMF) {
+    } else if (feature->type_ != FEATURE_ANMF) {
       ERROR_GOTO1("ERROR: No input file specified.\n", ErrValidate2);
     }
   }
@@ -805,28 +811,27 @@ static int ValidateConfig(Config* const config) {
 
 // Create config object from command-line arguments.
 static int InitializeConfig(int argc, const char* argv[],
-                            Config* const config) {
+                            WebPMuxConfig* config) {
   int num_feature_args = 0;
-  int ok;
+  int ok = 1;
 
+  assert(config != NULL);
   memset(config, 0, sizeof(*config));
 
-  ok = ExUtilInitCommandLineArguments(argc, argv, &config->cmd_args_);
-  if (!ok) return 0;
-
   // Validate command-line arguments.
-  if (!ValidateCommandLine(&config->cmd_args_, &num_feature_args)) {
+  if (!ValidateCommandLine(argc, argv, &num_feature_args)) {
     ERROR_GOTO1("Exiting due to command-line parsing error.\n", Err1);
   }
 
-  config->arg_count_ = num_feature_args;
+  config->feature_.arg_count_ = num_feature_args;
-  config->args_ = (FeatureArg*)calloc(num_feature_args, sizeof(*config->args_));
+  config->feature_.args_ =
-  if (config->args_ == NULL) {
+      (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(config) || !ValidateConfig(config)) {
+  if (!ParseCommandLine(argc, argv, config) || !ValidateConfig(config)) {
     ERROR_GOTO1("Exiting due to command-line parsing error.\n", Err1);
   }
 
@@ -842,7 +847,7 @@ static int InitializeConfig(int argc, const char* argv[],
 //------------------------------------------------------------------------------
 // Processing.
 
-static int GetFrame(const WebPMux* mux, const Config* config) {
+static int GetFrame(const WebPMux* mux, const WebPMuxConfig* config) {
   WebPMuxError err = WEBP_MUX_OK;
   WebPMux* mux_single = NULL;
   int num = 0;
@@ -852,7 +857,7 @@ static int GetFrame(const WebPMux* mux, const Config* config) {
   WebPMuxFrameInfo info;
   WebPDataInit(&info.bitstream);
 
-  num = ExUtilGetInt(config->args_[0].params_, 10, &parse_error);
+  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);
   }
@@ -886,17 +891,18 @@ static int GetFrame(const WebPMux* mux, const Config* config) {
 }
 
 // Read and process config.
-static int Process(const Config* config) {
+static int Process(const WebPMuxConfig* config) {
   WebPMux* mux = NULL;
   WebPData chunk;
   WebPMuxError err = WEBP_MUX_OK;
   int ok = 1;
+  const Feature* const feature = &config->feature_;
 
   switch (config->action_type_) {
     case ACTION_GET: {
       ok = CreateMux(config->input_, &mux);
       if (!ok) goto Err2;
-      switch (config->type_) {
+      switch (feature->type_) {
         case FEATURE_ANMF:
           ok = GetFrame(mux, config);
           break;
@@ -904,10 +910,10 @@ static int Process(const Config* config) {
         case FEATURE_ICCP:
         case FEATURE_EXIF:
         case FEATURE_XMP:
-          err = WebPMuxGetChunk(mux, kFourccList[config->type_], &chunk);
+          err = WebPMuxGetChunk(mux, kFourccList[feature->type_], &chunk);
           if (err != WEBP_MUX_OK) {
             ERROR_GOTO3("ERROR (%s): Could not get the %s.\n",
-                        ErrorString(err), kDescriptions[config->type_], Err2);
+                        ErrorString(err), kDescriptions[feature->type_], Err2);
           }
           ok = WriteData(config->output_, &chunk);
           break;
@@ -919,7 +925,7 @@ static int Process(const Config* config) {
       break;
     }
     case ACTION_SET: {
-      switch (config->type_) {
+      switch (feature->type_) {
         case FEATURE_ANMF: {
           int i;
           WebPMuxAnimParams params = { 0xFFFFFFFF, 0 };
@@ -928,11 +934,11 @@ static int Process(const Config* config) {
             ERROR_GOTO2("ERROR (%s): Could not allocate a mux object.\n",
                         ErrorString(WEBP_MUX_MEMORY_ERROR), Err2);
           }
-          for (i = 0; i < config->arg_count_; ++i) {
+          for (i = 0; i < feature->arg_count_; ++i) {
-            switch (config->args_[i].subtype_) {
+            switch (feature->args_[i].subtype_) {
               case SUBTYPE_BGCOLOR: {
                 uint32_t bgcolor;
-                ok = ParseBgcolorArgs(config->args_[i].params_, &bgcolor);
+                ok = ParseBgcolorArgs(feature->args_[i].params_, &bgcolor);
                 if (!ok) {
                   ERROR_GOTO1("ERROR: Could not parse the background color \n",
                               Err2);
@@ -943,7 +949,7 @@ static int Process(const Config* config) {
               case SUBTYPE_LOOP: {
                 int parse_error = 0;
                 const int loop_count =
-                    ExUtilGetInt(config->args_[i].params_, 10, &parse_error);
+                    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
@@ -959,10 +965,10 @@ static int Process(const Config* config) {
               case SUBTYPE_ANMF: {
                 WebPMuxFrameInfo frame;
                 frame.id = WEBP_CHUNK_ANMF;
-                ok = ExUtilReadFileToWebPData(config->args_[i].filename_,
+                ok = ReadFileToWebPData(feature->args_[i].filename_,
-                                              &frame.bitstream);
+                                        &frame.bitstream);
                 if (!ok) goto Err2;
-                ok = ParseFrameArgs(config->args_[i].params_, &frame);
+                ok = ParseFrameArgs(feature->args_[i].params_, &frame);
                 if (!ok) {
                   WebPDataClear(&frame.bitstream);
                   ERROR_GOTO1("ERROR: Could not parse frame properties.\n",
@@ -995,13 +1001,13 @@ static int Process(const Config* config) {
         case FEATURE_XMP: {
           ok = CreateMux(config->input_, &mux);
           if (!ok) goto Err2;
-          ok = ExUtilReadFileToWebPData(config->args_[0].filename_, &chunk);
+          ok = ReadFileToWebPData(feature->args_[0].filename_, &chunk);
           if (!ok) goto Err2;
-          err = WebPMuxSetChunk(mux, kFourccList[config->type_], &chunk, 1);
+          err = WebPMuxSetChunk(mux, kFourccList[feature->type_], &chunk, 1);
           free((void*)chunk.bytes);
           if (err != WEBP_MUX_OK) {
             ERROR_GOTO3("ERROR (%s): Could not set the %s.\n",
-                        ErrorString(err), kDescriptions[config->type_], Err2);
+                        ErrorString(err), kDescriptions[feature->type_], Err2);
           }
           break;
         }
@@ -1037,11 +1043,11 @@ static int Process(const Config* config) {
         for (i = 0; i < num_frames; ++i) durations[i] = -1;
 
         // Parse intervals to process.
-        for (i = 0; i < config->arg_count_; ++i) {
+        for (i = 0; i < feature->arg_count_; ++i) {
           int k;
           int args[3];
           int duration, start, end;
-          const int nb_args = ExUtilGetInts(config->args_[i].params_,
+          const int nb_args = ExUtilGetInts(feature->args_[i].params_,
                                             10, 3, args);
           ok = (nb_args >= 1);
           if (!ok) goto Err3;
@@ -1099,12 +1105,12 @@ static int Process(const Config* config) {
     case ACTION_STRIP: {
       ok = CreateMux(config->input_, &mux);
       if (!ok) goto Err2;
-      if (config->type_ == FEATURE_ICCP || config->type_ == FEATURE_EXIF ||
+      if (feature->type_ == FEATURE_ICCP || feature->type_ == FEATURE_EXIF ||
-          config->type_ == FEATURE_XMP) {
+          feature->type_ == FEATURE_XMP) {
-        err = WebPMuxDeleteChunk(mux, kFourccList[config->type_]);
+        err = WebPMuxDeleteChunk(mux, kFourccList[feature->type_]);
         if (err != WEBP_MUX_OK) {
           ERROR_GOTO3("ERROR (%s): Could not strip the %s.\n",
-                      ErrorString(err), kDescriptions[config->type_], Err2);
+                      ErrorString(err), kDescriptions[feature->type_], Err2);
         }
       } else {
         ERROR_GOTO1("ERROR: Invalid feature for action 'strip'.\n", Err2);
@@ -1134,7 +1140,7 @@ static int Process(const Config* config) {
 // Main.
 
 int main(int argc, const char* argv[]) {
-  Config config;
+  WebPMuxConfig config;
   int ok = InitializeConfig(argc - 1, argv + 1, &config);
   if (ok) {
     ok = Process(&config);

extras/Makefile.am

@@ -1,4 +1,3 @@
-AM_CPPFLAGS += -I$(top_builddir) -I$(top_srcdir)
 AM_CPPFLAGS += -I$(top_builddir)/src -I$(top_srcdir)/src
 noinst_LTLIBRARIES = libwebpextras.la
 
@@ -13,32 +12,25 @@ libwebpextras_la_LDFLAGS = -lm
 libwebpextras_la_LIBADD = ../src/libwebp.la
 
 noinst_PROGRAMS =
-noinst_PROGRAMS += webp_quality
+noinst_PROGRAMS += get_disto webp_quality
-if WANT_DEMUX
-  noinst_PROGRAMS += get_disto
-endif
 if BUILD_VWEBP_SDL
   noinst_PROGRAMS += vwebp_sdl
 endif
 
 get_disto_SOURCES  = get_disto.c
 get_disto_CPPFLAGS = $(AM_CPPFLAGS)
-get_disto_LDADD =
+get_disto_LDADD = ../imageio/libimageio_util.la ../imageio/libimagedec.la
-get_disto_LDADD += ../imageio/libimageio_util.la
-get_disto_LDADD += ../imageio/libimagedec.la
 get_disto_LDADD += ../src/libwebp.la
 get_disto_LDADD += $(PNG_LIBS) $(JPEG_LIBS) $(TIFF_LIBS)
 
 webp_quality_SOURCES  = webp_quality.c
-webp_quality_CPPFLAGS = $(AM_CPPFLAGS)
+webp_quality_CPPFLAGS = $(AM_CPPFLAGS) $(USE_EXPERIMENTAL_CODE)
-webp_quality_LDADD =
+webp_quality_LDADD  = ../imageio/libimageio_util.la
-webp_quality_LDADD += ../imageio/libimageio_util.la
 webp_quality_LDADD += libwebpextras.la
 webp_quality_LDADD += ../src/libwebp.la
 
 vwebp_sdl_SOURCES  = vwebp_sdl.c webp_to_sdl.c webp_to_sdl.h
 vwebp_sdl_CPPFLAGS = $(AM_CPPFLAGS) $(SDL_INCLUDES)
-vwebp_sdl_LDADD =
+vwebp_sdl_LDADD = ../imageio/libimageio_util.la
-vwebp_sdl_LDADD += ../imageio/libimageio_util.la
 vwebp_sdl_LDADD += ../src/libwebp.la
 vwebp_sdl_LDADD += $(SDL_LIBS)

extras/extras.c

@@ -10,14 +10,14 @@
 //  Additional WebP utilities.
 //
 
-#include "extras/extras.h"
+#include "./extras.h"
 #include "webp/format_constants.h"
 
 #include <assert.h>
 #include <string.h>
 
-#define XTRA_MAJ_VERSION 1
+#define XTRA_MAJ_VERSION 0
-#define XTRA_MIN_VERSION 0
+#define XTRA_MIN_VERSION 1
 #define XTRA_REV_VERSION 0
 
 //------------------------------------------------------------------------------
@@ -48,14 +48,13 @@ int WebPImportGray(const uint8_t* gray_data, WebPPicture* pic) {
 
 int WebPImportRGB565(const uint8_t* rgb565, WebPPicture* pic) {
   int x, y;
-  uint32_t* dst;
   if (pic == NULL || rgb565 == NULL) return 0;
   pic->colorspace = WEBP_YUV420;
   pic->use_argb = 1;
   if (!WebPPictureAlloc(pic)) return 0;
-  dst = pic->argb;
   for (y = 0; y < pic->height; ++y) {
     const int width = pic->width;
+    uint32_t* dst = pic->argb + y * pic->argb_stride;
     for (x = 0; x < width; ++x) {
 #ifdef WEBP_SWAP_16BIT_CSP
       const uint32_t rg = rgb565[2 * x + 1];
@@ -71,24 +70,22 @@ int WebPImportRGB565(const uint8_t* rgb565, WebPPicture* pic) {
       r = r | (r >> 5);
       g = g | (g >> 6);
       b = b | (b >> 5);
-      dst[x] = (0xffu << 24) | (r << 16) | (g << 8) | b;
+      dst[x] = (r << 16) | (g << 8) | b;
     }
     rgb565 += 2 * width;
-    dst += pic->argb_stride;
   }
   return 1;
 }
 
 int WebPImportRGB4444(const uint8_t* rgb4444, WebPPicture* pic) {
   int x, y;
-  uint32_t* dst;
   if (pic == NULL || rgb4444 == NULL) return 0;
   pic->colorspace = WEBP_YUV420;
   pic->use_argb = 1;
   if (!WebPPictureAlloc(pic)) return 0;
-  dst = pic->argb;
   for (y = 0; y < pic->height; ++y) {
     const int width = pic->width;
+    uint32_t* dst = pic->argb + y * pic->argb_stride;
     for (x = 0; x < width; ++x) {
 #ifdef WEBP_SWAP_16BIT_CSP
       const uint32_t rg = rgb4444[2 * x + 1];
@@ -109,7 +106,6 @@ int WebPImportRGB4444(const uint8_t* rgb4444, WebPPicture* pic) {
       dst[x] = (a << 24) | (r << 16) | (g << 8) | b;
     }
     rgb4444 += 2 * width;
-    dst += pic->argb_stride;
   }
   return 1;
 }

extras/extras.h

@@ -25,28 +25,28 @@ extern "C" {
 
 // Returns the version number of the extras library, packed in hexadecimal using
 // 8bits for each of major/minor/revision. E.g: v2.5.7 is 0x020507.
-WEBP_EXTERN int WebPGetExtrasVersion(void);
+WEBP_EXTERN(int) WebPGetExtrasVersion(void);
 
 //------------------------------------------------------------------------------
 // Ad-hoc colorspace importers.
 
 // Import luma sample (gray scale image) into 'picture'. The 'picture'
 // width and height must be set prior to calling this function.
-WEBP_EXTERN int WebPImportGray(const uint8_t* gray, WebPPicture* picture);
+WEBP_EXTERN(int) WebPImportGray(const uint8_t* gray, WebPPicture* picture);
 
 // Import rgb sample in RGB565 packed format into 'picture'. The 'picture'
 // width and height must be set prior to calling this function.
-WEBP_EXTERN int WebPImportRGB565(const uint8_t* rgb565, WebPPicture* pic);
+WEBP_EXTERN(int) WebPImportRGB565(const uint8_t* rgb565, WebPPicture* pic);
 
 // Import rgb sample in RGB4444 packed format into 'picture'. The 'picture'
 // width and height must be set prior to calling this function.
-WEBP_EXTERN int WebPImportRGB4444(const uint8_t* rgb4444, WebPPicture* pic);
+WEBP_EXTERN(int) WebPImportRGB4444(const uint8_t* rgb4444, WebPPicture* pic);
 
 // Import a color mapped image. The number of colors is less or equal to
 // MAX_PALETTE_SIZE. 'pic' must have been initialized. Its content, if any,
 // will be discarded. Returns 'false' in case of error, or if indexed[] contains
 // invalid indices.
-WEBP_EXTERN int
+WEBP_EXTERN(int)
 WebPImportColorMappedARGB(const uint8_t* indexed, int indexed_stride,
                           const uint32_t palette[], int palette_size,
                           WebPPicture* pic);
@@ -59,7 +59,7 @@ WebPImportColorMappedARGB(const uint8_t* indexed, int indexed_stride,
 // Otherwise (lossy bitstream), the returned value is in the range [0..100].
 // Any error (invalid bitstream, animated WebP, incomplete header, etc.)
 // will return a value of -1.
-WEBP_EXTERN int VP8EstimateQuality(const uint8_t* const data, size_t size);
+WEBP_EXTERN(int) VP8EstimateQuality(const uint8_t* const data, size_t size);
 
 //------------------------------------------------------------------------------
 

extras/get_disto.c

@@ -24,8 +24,8 @@
 #include <string.h>
 
 #include "webp/encode.h"
-#include "imageio/image_dec.h"
+#include "../imageio/image_dec.h"
-#include "imageio/imageio_util.h"
+#include "../imageio/imageio_util.h"
 
 static size_t ReadPicture(const char* const filename, WebPPicture* const pic,
                           int keep_alpha) {
@@ -290,10 +290,9 @@ int main(int argc, const char *argv[]) {
     fprintf(stderr, "Error while computing the distortion.\n");
     goto End;
   }
-  printf("%u %.2f    %.2f %.2f %.2f %.2f [ %.2f bpp ]\n",
+  printf("%u %.2f    %.2f %.2f %.2f %.2f\n",
          (unsigned int)size1,
-         disto[4], disto[0], disto[1], disto[2], disto[3],
+         disto[4], disto[0], disto[1], disto[2], disto[3]);
-         8.f * size1 / pic1.width / pic1.height);
 
   if (output != NULL) {
     uint8_t* data = NULL;
@@ -323,7 +322,6 @@ int main(int argc, const char *argv[]) {
       fprintf(stderr, "Can only compute the difference map in ARGB format.\n");
       goto End;
     }
-#if !defined(WEBP_REDUCE_CSP)
     data_size = WebPEncodeLosslessBGRA((const uint8_t*)pic1.argb,
                                        pic1.width, pic1.height,
                                        pic1.argb_stride * 4,
@@ -335,12 +333,6 @@ int main(int argc, const char *argv[]) {
     ret = ImgIoUtilWriteFile(output, data, data_size) ? 0 : 1;
     WebPFree(data);
     if (ret) goto End;
-#else
-    (void)data;
-    (void)data_size;
-    fprintf(stderr, "Cannot save the difference map. Please recompile "
-                    "without the WEBP_REDUCE_CSP flag.\n");
-#endif  // WEBP_REDUCE_CSP
   }
   ret = 0;
 

extras/quality_estimate.c

@@ -11,7 +11,7 @@
 //
 // Author: Skal (pascal.massimino@gmail.com)
 
-#include "extras/extras.h"
+#include "./extras.h"
 #include "webp/decode.h"
 
 #include <math.h>
@@ -73,7 +73,7 @@ int VP8EstimateQuality(const uint8_t* const data, size_t size) {
   pos += 4;
   bit_pos = pos * 8;
 
-  GET_BIT(2);  // colorspace + clamp type
+  GET_BIT(2);  // color_space + clamp type
 
   // Segment header
   if (GET_BIT(1)) {       // use_segment_

extras/vwebp_sdl.c

@@ -24,7 +24,7 @@
 
 #include "webp_to_sdl.h"
 #include "webp/decode.h"
-#include "imageio/imageio_util.h"
+#include "../imageio/imageio_util.h"
 
 #if defined(WEBP_HAVE_JUST_SDL_H)
 #include <SDL.h>

extras/webp_quality.c

@@ -11,8 +11,8 @@
 #include <stdlib.h>
 #include <string.h>
 
-#include "extras/extras.h"
+#include "./extras.h"
-#include "imageio/imageio_util.h"
+#include "../imageio/imageio_util.h"
 
 int main(int argc, const char *argv[]) {
   int c;

extras/webp_to_sdl.c

@@ -12,7 +12,7 @@
 // Author: James Zern (jzern@google.com)
 
 #ifdef HAVE_CONFIG_H
-#include "src/webp/config.h"
+#include "webp/config.h"
 #endif
 
 #if defined(WEBP_HAVE_SDL)
@@ -20,7 +20,7 @@
 #include "webp_to_sdl.h"
 
 #include <stdio.h>
-#include "src/webp/decode.h"
+#include "webp/decode.h"
 
 #if defined(WEBP_HAVE_JUST_SDL_H)
 #include <SDL.h>
@@ -28,7 +28,6 @@
 #include <SDL/SDL.h>
 #endif
 
-static int init_ok = 0;
 int WebpToSDL(const char* data, unsigned int data_size) {
   int ok = 0;
   VP8StatusCode status;
@@ -43,10 +42,7 @@ int WebpToSDL(const char* data, unsigned int data_size) {
     return 1;
   }
 
-  if (!init_ok) {
+  SDL_Init(SDL_INIT_VIDEO);
-    SDL_Init(SDL_INIT_VIDEO);
-    init_ok = 1;
-  }
 
   status = WebPGetFeatures((uint8_t*)data, (size_t)data_size, &config.input);
   if (status != VP8_STATUS_OK) goto Error;
@@ -101,7 +97,6 @@ int WebpToSDL(const char* data, unsigned int data_size) {
  Error:
   SDL_FreeSurface(surface);
   SDL_FreeSurface(screen);
-  WebPFreeDecBuffer(output);
   return ok;
 }
 

imageio/Makefile.am

@@ -1,20 +1,13 @@
 AM_CPPFLAGS += -I$(top_builddir)/src -I$(top_srcdir)/src
-noinst_LTLIBRARIES =
+noinst_LTLIBRARIES = libimageio_util.la libimagedec.la libimageenc.la
-noinst_LTLIBRARIES += libimageio_util.la
-if WANT_DEMUX
-  noinst_LTLIBRARIES += libimagedec.la
-endif
-noinst_LTLIBRARIES += libimageenc.la
 
 noinst_HEADERS =
 noinst_HEADERS += ../src/webp/decode.h
 noinst_HEADERS += ../src/webp/types.h
 
-libimageio_util_la_SOURCES =
+libimageio_util_la_SOURCES = imageio_util.c imageio_util.h
-libimageio_util_la_SOURCES += imageio_util.c imageio_util.h
 
-libimagedec_la_SOURCES  =
+libimagedec_la_SOURCES  = image_dec.c image_dec.h
-libimagedec_la_SOURCES += image_dec.c image_dec.h
 libimagedec_la_SOURCES += jpegdec.c jpegdec.h
 libimagedec_la_SOURCES += metadata.c metadata.h
 libimagedec_la_SOURCES += pngdec.c pngdec.h
@@ -23,10 +16,8 @@ libimagedec_la_SOURCES += tiffdec.c tiffdec.h
 libimagedec_la_SOURCES += webpdec.c webpdec.h
 libimagedec_la_SOURCES += wicdec.c wicdec.h
 libimagedec_la_CPPFLAGS = $(JPEG_INCLUDES) $(PNG_INCLUDES) $(TIFF_INCLUDES)
-libimagedec_la_CPPFLAGS += $(AM_CPPFLAGS)
+libimagedec_la_CPPFLAGS += $(AM_CPPFLAGS) $(USE_EXPERIMENTAL_CODE)
-libimagedec_la_LIBADD = ../src/demux/libwebpdemux.la
 
-libimageenc_la_SOURCES  =
+libimageenc_la_SOURCES  = image_enc.c image_enc.h
-libimageenc_la_SOURCES += image_enc.c image_enc.h
 libimageenc_la_CPPFLAGS = $(JPEG_INCLUDES) $(PNG_INCLUDES) $(TIFF_INCLUDES)
-libimageenc_la_CPPFLAGS += $(AM_CPPFLAGS)
+libimageenc_la_CPPFLAGS += $(AM_CPPFLAGS) $(USE_EXPERIMENTAL_CODE)

imageio/image_enc.c

@@ -158,8 +158,14 @@ static void PNGAPI PNGErrorFunction(png_structp png, png_const_charp dummy) {
 }
 
 int WebPWritePNG(FILE* out_file, const WebPDecBuffer* const buffer) {
+  const uint32_t width = buffer->width;
+  const uint32_t height = buffer->height;
+  png_bytep row = buffer->u.RGBA.rgba;
+  const int stride = buffer->u.RGBA.stride;
+  const int has_alpha = WebPIsAlphaMode(buffer->colorspace);
   volatile png_structp png;
   volatile png_infop info;
+  png_uint_32 y;
 
   if (out_file == NULL || buffer == NULL) return 0;
 
@@ -178,23 +184,14 @@ int WebPWritePNG(FILE* out_file, const WebPDecBuffer* const buffer) {
     return 0;
   }
   png_init_io(png, out_file);
-  {
+  png_set_IHDR(png, info, width, height, 8,
-    const uint32_t width = buffer->width;
+               has_alpha ? PNG_COLOR_TYPE_RGBA : PNG_COLOR_TYPE_RGB,
-    const uint32_t height = buffer->height;
+               PNG_INTERLACE_NONE, PNG_COMPRESSION_TYPE_DEFAULT,
-    png_bytep row = buffer->u.RGBA.rgba;
+               PNG_FILTER_TYPE_DEFAULT);
-    const int stride = buffer->u.RGBA.stride;
+  png_write_info(png, info);
-    const int has_alpha = WebPIsAlphaMode(buffer->colorspace);
+  for (y = 0; y < height; ++y) {
-    uint32_t y;
+    png_write_rows(png, &row, 1);
-
+    row += stride;
-    png_set_IHDR(png, info, width, height, 8,
-                 has_alpha ? PNG_COLOR_TYPE_RGBA : PNG_COLOR_TYPE_RGB,
-                 PNG_INTERLACE_NONE, PNG_COMPRESSION_TYPE_DEFAULT,
-                 PNG_FILTER_TYPE_DEFAULT);
-    png_write_info(png, info);
-    for (y = 0; y < height; ++y) {
-      png_write_rows(png, &row, 1);
-      row += stride;
-    }
   }
   png_write_end(png, info);
   png_destroy_write_struct((png_structpp)&png, (png_infopp)&info);
@@ -545,24 +542,22 @@ int WebPWriteYUV(FILE* fout, const WebPDecBuffer* const buffer) {
 // Generic top-level call
 
 int WebPSaveImage(const WebPDecBuffer* const buffer,
-                  WebPOutputFileFormat format,
+                  WebPOutputFileFormat format, const char* const out_file) {
-                  const char* const out_file_name) {
   FILE* fout = NULL;
   int needs_open_file = 1;
-  const int use_stdout = (out_file_name != NULL) && !strcmp(out_file_name, "-");
+  const int use_stdout = (out_file != NULL) && !strcmp(out_file, "-");
   int ok = 1;
 
-  if (buffer == NULL || out_file_name == NULL) return 0;
+  if (buffer == NULL || out_file == NULL) return 0;
 
 #ifdef HAVE_WINCODEC_H
   needs_open_file = (format != PNG);
 #endif
 
   if (needs_open_file) {
-    fout = use_stdout ? ImgIoUtilSetBinaryMode(stdout)
+    fout = use_stdout ? ImgIoUtilSetBinaryMode(stdout) : fopen(out_file, "wb");
-                      : fopen(out_file_name, "wb");
     if (fout == NULL) {
-      fprintf(stderr, "Error opening output file %s\n", out_file_name);
+      fprintf(stderr, "Error opening output file %s\n", out_file);
       return 0;
     }
   }
@@ -571,7 +566,7 @@ int WebPSaveImage(const WebPDecBuffer* const buffer,
       format == RGBA || format == BGRA || format == ARGB ||
       format == rgbA || format == bgrA || format == Argb) {
 #ifdef HAVE_WINCODEC_H
-    ok &= WebPWritePNG(out_file_name, use_stdout, buffer);
+    ok &= WebPWritePNG(out_file, use_stdout, buffer);
 #else
     ok &= WebPWritePNG(fout, buffer);
 #endif

imageio/imageio_util.c

@@ -47,8 +47,7 @@ int ImgIoUtilReadFromStdin(const uint8_t** data, size_t* data_size) {
   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;
-    // we allocate one extra byte for the \0 terminator
+    void* const new_data = realloc(input, max_size + extra_size);
-    void* const new_data = realloc(input, max_size + extra_size + 1);
     if (new_data == NULL) goto Error;
     input = (uint8_t*)new_data;
     max_size += extra_size;
@@ -56,7 +55,6 @@ int ImgIoUtilReadFromStdin(const uint8_t** data, size_t* data_size) {
     if (size < max_size) break;
   }
   if (ferror(stdin)) goto Error;
-  if (input != NULL) input[size] = '\0';  // convenient 0-terminator
   *data = input;
   *data_size = size;
   return 1;
@@ -70,7 +68,7 @@ int ImgIoUtilReadFromStdin(const uint8_t** data, size_t* data_size) {
 int ImgIoUtilReadFile(const char* const file_name,
                       const uint8_t** data, size_t* data_size) {
   int ok;
-  uint8_t* file_data;
+  void* file_data;
   size_t file_size;
   FILE* in;
   const int from_stdin = (file_name == NULL) || !strcmp(file_name, "-");
@@ -89,14 +87,8 @@ int ImgIoUtilReadFile(const char* const file_name,
   fseek(in, 0, SEEK_END);
   file_size = ftell(in);
   fseek(in, 0, SEEK_SET);
-  // we allocate one extra byte for the \0 terminator
+  file_data = malloc(file_size);
-  file_data = (uint8_t*)malloc(file_size + 1);
+  if (file_data == NULL) return 0;
-  if (file_data == NULL) {
-    fclose(in);
-    fprintf(stderr, "memory allocation failure when reading file %s\n",
-            file_name);
-    return 0;
-  }
   ok = (fread(file_data, file_size, 1, in) == 1);
   fclose(in);
 
@@ -106,14 +98,11 @@ int ImgIoUtilReadFile(const char* const file_name,
     free(file_data);
     return 0;
   }
-  file_data[file_size] = '\0';  // convenient 0-terminator
+  *data = (uint8_t*)file_data;
-  *data = file_data;
   *data_size = file_size;
   return 1;
 }
 
-// -----------------------------------------------------------------------------
-
 int ImgIoUtilWriteFile(const char* const file_name,
                        const uint8_t* data, size_t data_size) {
   int ok;
@@ -148,11 +137,7 @@ void ImgIoUtilCopyPlane(const uint8_t* src, int src_stride,
 
 int ImgIoUtilCheckSizeArgumentsOverflow(uint64_t nmemb, size_t size) {
   const uint64_t total_size = nmemb * size;
-  int ok = (total_size == (size_t)total_size);
+  return (total_size == (size_t)total_size);
-#if defined(WEBP_MAX_IMAGE_SIZE)
-  ok = ok && (total_size <= (uint64_t)WEBP_MAX_IMAGE_SIZE);
-#endif
-  return ok;
 }
 
 // -----------------------------------------------------------------------------

imageio/imageio_util.h

@@ -30,9 +30,6 @@ FILE* ImgIoUtilSetBinaryMode(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().
-// Note: for convenience, the data will be null-terminated with an extra byte
-// (not accounted for in *data_size), in case the file is text and intended
-// to be used as a C-string.
 // If 'file_name' is NULL or equal to "-", input is read from stdin by calling
 // the function ImgIoUtilReadFromStdin().
 int ImgIoUtilReadFile(const char* const file_name,

imageio/jpegdec.c

@@ -304,18 +304,18 @@ int ReadJPEG(const uint8_t* const data, size_t data_size,
 
   if (stride != (int)stride ||
       !ImgIoUtilCheckSizeArgumentsOverflow(stride, height)) {
-    goto Error;
+    goto End;
   }
 
   rgb = (uint8_t*)malloc((size_t)stride * height);
   if (rgb == NULL) {
-    goto Error;
+    goto End;
   }
   buffer[0] = (JSAMPLE*)rgb;
 
   while (dinfo.output_scanline < dinfo.output_height) {
     if (jpeg_read_scanlines((j_decompress_ptr)&dinfo, buffer, 1) != 1) {
-      goto Error;
+      goto End;
     }
     buffer[0] += stride;
   }

imageio/pngdec.c

@@ -185,6 +185,7 @@ static int ExtractMetadataFromPNG(png_structp png,
       }
     }
   }
+
   return 1;
 }
 
@@ -264,16 +265,6 @@ int ReadPNG(const uint8_t* const data, size_t data_size,
     has_alpha = !!(color_type & PNG_COLOR_MASK_ALPHA);
   }
 
-  // Apply gamma correction if needed.
-  {
-    double image_gamma = 1 / 2.2, screen_gamma = 2.2;
-    int srgb_intent;
-    if (png_get_sRGB(png, info, &srgb_intent) ||
-        png_get_gAMA(png, info, &image_gamma)) {
-      png_set_gamma(png, screen_gamma, image_gamma);
-    }
-  }
-
   if (!keep_alpha) {
     png_set_strip_alpha(png);
     has_alpha = 0;

imageio/pnmdec.c

@@ -117,13 +117,8 @@ static size_t ReadPAMFields(PNMInfo* const info, size_t off) {
     }
   }
   if (!(info->seen_flags & TUPLE_FLAG)) {
-    if (info->depth > 0 && info->depth <= 4 && info->depth != 2) {
+    info->seen_flags |= TUPLE_FLAG;
-      info->seen_flags |= TUPLE_FLAG;
+    info->bytes_per_px = info->depth * (info->max_value > 255 ? 2 : 1);
-      info->bytes_per_px = info->depth * (info->max_value > 255 ? 2 : 1);
-    } else {
-      fprintf(stderr, "PAM: invalid bitdepth (%d).\n", info->depth);
-      return 0;
-    }
   }
   if (info->seen_flags != ALL_NEEDED_FLAGS) {
     fprintf(stderr, "PAM: incomplete header.\n");
@@ -165,7 +160,7 @@ static size_t ReadHeader(PNMInfo* const info) {
   // perform some basic numerical validation
   if (info->width <= 0 || info->height <= 0 ||
       info->type <= 0 || info->type >= 9 ||
-      info->depth <= 0 || info->depth == 2 || info->depth > 4 ||
+      info->depth <= 0 || info->depth > 4 ||
       info->bytes_per_px < info->depth ||
       info->max_value <= 0 || info->max_value >= 65536) {
     return 0;

imageio/webpdec.c

@@ -9,18 +9,12 @@
 //
 // WebP decode.
 
-#ifdef HAVE_CONFIG_H
-#include "webp/config.h"
-#endif
-
 #include "./webpdec.h"
 
-#include <assert.h>
 #include <stdio.h>
 #include <stdlib.h>
 
 #include "webp/decode.h"
-#include "webp/demux.h"
 #include "webp/encode.h"
 #include "./imageio_util.h"
 #include "./metadata.h"
@@ -97,47 +91,25 @@ VP8StatusCode DecodeWebPIncremental(
       fprintf(stderr, "Failed during WebPINewDecoder().\n");
       return VP8_STATUS_OUT_OF_MEMORY;
     } else {
+#ifdef WEBP_EXPERIMENTAL_FEATURES
+      size_t size = 0;
+      const size_t incr = 2 + (data_size / 20);
+      while (size < data_size) {
+        size_t next_size = size + (rand() % incr);
+        if (next_size > data_size) next_size = data_size;
+        status = WebPIUpdate(idec, data, next_size);
+        if (status != VP8_STATUS_OK && status != VP8_STATUS_SUSPENDED) break;
+        size = next_size;
+      }
+#else
       status = WebPIUpdate(idec, data, data_size);
+#endif
       WebPIDelete(idec);
     }
   }
   return status;
 }
 
-// -----------------------------------------------------------------------------
-// Metadata
-
-static int ExtractMetadata(const uint8_t* const data, size_t data_size,
-                           Metadata* const metadata) {
-  WebPData webp_data = { data, data_size };
-  WebPDemuxer* const demux = WebPDemux(&webp_data);
-  WebPChunkIterator chunk_iter;
-  uint32_t flags;
-
-  if (demux == NULL) return 0;
-  assert(metadata != NULL);
-
-  flags = WebPDemuxGetI(demux, WEBP_FF_FORMAT_FLAGS);
-
-  if ((flags & ICCP_FLAG) && WebPDemuxGetChunk(demux, "ICCP", 1, &chunk_iter)) {
-    MetadataCopy((const char*)chunk_iter.chunk.bytes, chunk_iter.chunk.size,
-                 &metadata->iccp);
-    WebPDemuxReleaseChunkIterator(&chunk_iter);
-  }
-  if ((flags & EXIF_FLAG) && WebPDemuxGetChunk(demux, "EXIF", 1, &chunk_iter)) {
-    MetadataCopy((const char*)chunk_iter.chunk.bytes, chunk_iter.chunk.size,
-                 &metadata->exif);
-    WebPDemuxReleaseChunkIterator(&chunk_iter);
-  }
-  if ((flags & XMP_FLAG) && WebPDemuxGetChunk(demux, "XMP ", 1, &chunk_iter)) {
-    MetadataCopy((const char*)chunk_iter.chunk.bytes, chunk_iter.chunk.size,
-                 &metadata->xmp);
-    WebPDemuxReleaseChunkIterator(&chunk_iter);
-  }
-  WebPDemuxDelete(demux);
-  return 1;
-}
-
 // -----------------------------------------------------------------------------
 
 int ReadWebP(const uint8_t* const data, size_t data_size,
@@ -151,6 +123,11 @@ int ReadWebP(const uint8_t* const data, size_t data_size,
 
   if (data == NULL || data_size == 0 || pic == NULL) return 0;
 
+  // TODO(jzern): add Exif/XMP/ICC extraction.
+  if (metadata != NULL) {
+    fprintf(stderr, "Warning: metadata extraction from WebP is unsupported.\n");
+  }
+
   if (!WebPInitDecoderConfig(&config)) {
     fprintf(stderr, "Library version mismatch!\n");
     return 0;
@@ -164,32 +141,17 @@ int ReadWebP(const uint8_t* const data, size_t data_size,
 
   do {
     const int has_alpha = keep_alpha && bitstream->has_alpha;
-    uint64_t stride;
     pic->width = bitstream->width;
     pic->height = bitstream->height;
-    if (pic->use_argb) {
+    if (!pic->use_argb) pic->colorspace = has_alpha ? WEBP_YUV420A
-      stride = (uint64_t)bitstream->width * 4;
+                                                    : WEBP_YUV420;
-    } else {
-      stride = (uint64_t)bitstream->width * (has_alpha ? 5 : 3) / 2;
-      pic->colorspace = has_alpha ? WEBP_YUV420A : WEBP_YUV420;
-    }
-
-    if (!ImgIoUtilCheckSizeArgumentsOverflow(stride, bitstream->height)) {
-      status = VP8_STATUS_OUT_OF_MEMORY;
-      break;
-    }
-
     ok = WebPPictureAlloc(pic);
     if (!ok) {
       status = VP8_STATUS_OUT_OF_MEMORY;
       break;
     }
     if (pic->use_argb) {
-#ifdef WORDS_BIGENDIAN
-      output_buffer->colorspace = MODE_ARGB;
-#else
       output_buffer->colorspace = MODE_BGRA;
-#endif
       output_buffer->u.RGBA.rgba = (uint8_t*)pic->argb;
       output_buffer->u.RGBA.stride = pic->argb_stride * sizeof(uint32_t);
       output_buffer->u.RGBA.size = output_buffer->u.RGBA.stride * pic->height;
@@ -212,6 +174,7 @@ int ReadWebP(const uint8_t* const data, size_t data_size,
 
     status = DecodeWebP(data, data_size, &config);
     ok = (status == VP8_STATUS_OK);
+    if (!ok) WebPPictureFree(pic);
     if (ok && !keep_alpha && pic->use_argb) {
       // Need to wipe out the alpha value, as requested.
       int x, y;
@@ -225,18 +188,9 @@ int ReadWebP(const uint8_t* const data, size_t data_size,
 
   if (status != VP8_STATUS_OK) {
     PrintWebPError("input data", status);
-    ok = 0;
   }
 
   WebPFreeDecBuffer(output_buffer);
-
-  if (ok && metadata != NULL) {
-    ok = ExtractMetadata(data, data_size, metadata);
-    if (!ok) {
-      PrintWebPError("metadata", VP8_STATUS_BITSTREAM_ERROR);
-    }
-  }
-  if (!ok) WebPPictureFree(pic);
   return ok;
 }
 

makefile.unix

@@ -25,7 +25,6 @@ ifeq ($(strip $(shell uname)), Darwin)
   # Failure observed with: gcc 4.2.1 and 4.0.1.
   EXTRA_FLAGS += -fno-common
   EXTRA_FLAGS += -DHAVE_GLUT_GLUT_H
-  EXTRA_FLAGS += -Wno-deprecated-declarations
   EXTRA_FLAGS += -I/opt/local/include
   EXTRA_LIBS  += -L/opt/local/lib
   GL_LIBS = -framework GLUT -framework OpenGL
@@ -35,16 +34,6 @@ else
   GL_LIBS = -lglut -lGL
 endif
 
-# SDL flags: use sdl-config if it exists
-SDL_CONFIG = $(shell sdl-config --version 2> /dev/null)
-ifneq ($(SDL_CONFIG),)
-  SDL_LIBS = $(shell sdl-config --libs)
-  SDL_FLAGS = $(shell sdl-config --cflags)
-else
-  # use best-guess
-  SDL_LIBS = -lSDL
-  SDL_FLAGS =
-endif
 
 # To install libraries on Mac OS X:
 # 1. Install MacPorts (http://www.macports.org/install.php)
@@ -64,8 +53,11 @@ endif
 # 'make -f makefile.unix EXTRA_FLAGS=-m32' to that effect.
 # EXTRA_FLAGS += -m32
 
+# Extra flags to enable experimental features and code
+# EXTRA_FLAGS += -DWEBP_EXPERIMENTAL_FEATURES
+
 # Extra flags to enable byte swap for 16 bit colorspaces.
-# EXTRA_FLAGS += -DWEBP_SWAP_16BIT_CSP=1
+# EXTRA_FLAGS += -DWEBP_SWAP_16BIT_CSP
 
 # Extra flags to enable multi-threading
 EXTRA_FLAGS += -DWEBP_USE_THREAD
@@ -111,13 +103,8 @@ endif
 
 AR = ar
 ARFLAGS = r
-CPPFLAGS = -I. -Isrc/ -Wall
+CPPFLAGS = -Isrc/ -Wall
-ifeq ($(DEBUG), 1)
+CFLAGS = -O3 -DNDEBUG $(EXTRA_FLAGS)
-  CFLAGS = -g
-else
-  CFLAGS = -O3 -DNDEBUG
-endif
-CFLAGS += $(EXTRA_FLAGS)
 CC = gcc
 INSTALL = install
 GROFF = /usr/bin/groff
@@ -179,15 +166,16 @@ DSP_DEC_OBJS = \
     src/dsp/upsampling_msa.o \
     src/dsp/upsampling_neon.o \
     src/dsp/upsampling_sse2.o \
-    src/dsp/upsampling_sse41.o \
     src/dsp/yuv.o \
     src/dsp/yuv_mips32.o \
     src/dsp/yuv_mips_dsp_r2.o \
     src/dsp/yuv_neon.o \
     src/dsp/yuv_sse2.o \
-    src/dsp/yuv_sse41.o \
 
 DSP_ENC_OBJS = \
+    src/dsp/argb.o \
+    src/dsp/argb_mips_dsp_r2.o \
+    src/dsp/argb_sse2.o \
     src/dsp/cost.o \
     src/dsp/cost_mips32.o \
     src/dsp/cost_mips_dsp_r2.o \
@@ -217,6 +205,7 @@ ENC_OBJS = \
     src/enc/backward_references_enc.o \
     src/enc/config_enc.o \
     src/enc/cost_enc.o \
+    src/enc/delta_palettization_enc.o \
     src/enc/filter_enc.o \
     src/enc/frame_enc.o \
     src/enc/histogram_enc.o \
@@ -314,6 +303,7 @@ HDRS = \
     src/dsp/yuv.h \
     src/enc/backward_references_enc.h \
     src/enc/cost_enc.h \
+    src/enc/delta_palettization_enc.h \
     src/enc/histogram_enc.h \
     src/enc/vp8i_enc.h \
     src/enc/vp8li_enc.h \
@@ -345,8 +335,7 @@ OUT_LIBS += src/libwebp.a
 EXTRA_LIB = extras/libwebpextras.a
 OUT_EXAMPLES = examples/cwebp examples/dwebp
 EXTRA_EXAMPLES = examples/gif2webp examples/vwebp examples/webpmux \
-                 examples/anim_diff examples/anim_dump \
+                 examples/anim_diff examples/img2webp examples/webpinfo
-                 examples/img2webp examples/webpinfo
 OTHER_EXAMPLES = extras/get_disto extras/webp_quality extras/vwebp_sdl
 
 OUTPUT = $(OUT_LIBS) $(OUT_EXAMPLES)
@@ -374,7 +363,7 @@ src/utils/bit_reader_utils.o: src/utils/endian_inl_utils.h
 src/utils/bit_writer_utils.o: src/utils/endian_inl_utils.h
 
 %.o: %.c $(HDRS)
-	$(CC) $(CPPFLAGS) $(CFLAGS) -c $< -o $@
+	$(CC) $(CFLAGS) $(CPPFLAGS) -c $< -o $@
 
 examples/libanim_util.a: $(ANIM_UTIL_OBJS)
 examples/libexample_util.a: $(EX_UTIL_OBJS)
@@ -392,7 +381,6 @@ src/demux/libwebpdemux.a: $(LIBWEBPDEMUX_OBJS)
 	$(AR) $(ARFLAGS) $@ $^
 
 examples/anim_diff: examples/anim_diff.o $(ANIM_UTIL_OBJS) $(GIFDEC_OBJS)
-examples/anim_dump: examples/anim_dump.o $(ANIM_UTIL_OBJS)
 examples/cwebp: examples/cwebp.o
 examples/dwebp: examples/dwebp.o
 examples/gif2webp: examples/gif2webp.o $(GIFDEC_OBJS)
@@ -406,22 +394,13 @@ examples/anim_diff: src/demux/libwebpdemux.a examples/libexample_util.a
 examples/anim_diff: imageio/libimageio_util.a src/libwebp.a
 examples/anim_diff: EXTRA_LIBS += $(GIF_LIBS)
 examples/anim_diff: EXTRA_FLAGS += -DWEBP_HAVE_GIF
-examples/anim_dump: examples/libanim_util.a
-examples/anim_dump: src/demux/libwebpdemux.a
-examples/anim_dump: examples/libexample_util.a
-examples/anim_dump: imageio/libimageio_util.a
-examples/anim_dump: imageio/libimageenc.a
-examples/anim_dump: src/libwebp.a
-examples/anim_dump: EXTRA_LIBS += $(GIF_LIBS) $(DWEBP_LIBS)
 examples/cwebp: examples/libexample_util.a
 examples/cwebp: imageio/libimagedec.a
-examples/cwebp: src/demux/libwebpdemux.a
 examples/cwebp: imageio/libimageio_util.a
 examples/cwebp: src/libwebp.a
 examples/cwebp: EXTRA_LIBS += $(CWEBP_LIBS)
 examples/dwebp: examples/libexample_util.a
 examples/dwebp: imageio/libimagedec.a
-examples/dwebp: src/demux/libwebpdemux.a
 examples/dwebp: imageio/libimageenc.a
 examples/dwebp: imageio/libimageio_util.a
 examples/dwebp: src/libwebp.a
@@ -438,17 +417,13 @@ examples/webpmux: examples/libexample_util.a imageio/libimageio_util.a
 examples/webpmux: src/mux/libwebpmux.a src/libwebpdecoder.a
 examples/img2webp: examples/libexample_util.a imageio/libimageio_util.a
 examples/img2webp: imageio/libimagedec.a
-examples/img2webp: src/demux/libwebpdemux.a
 examples/img2webp: src/mux/libwebpmux.a src/libwebp.a
 examples/img2webp: EXTRA_LIBS += $(CWEBP_LIBS)
 examples/webpinfo: examples/libexample_util.a imageio/libimageio_util.a
 examples/webpinfo: src/libwebpdecoder.a
 
 extras/get_disto: extras/get_disto.o
-extras/get_disto: imageio/libimagedec.a
+extras/get_disto: imageio/libimagedec.a imageio/libimageio_util.a src/libwebp.a
-extras/get_disto: src/demux/libwebpdemux.a
-extras/get_disto: imageio/libimageio_util.a
-extras/get_disto: src/libwebp.a
 extras/get_disto: EXTRA_LIBS += $(CWEBP_LIBS)
 
 extras/webp_quality: extras/webp_quality.o
@@ -459,8 +434,8 @@ extras/vwebp_sdl: extras/vwebp_sdl.o
 extras/vwebp_sdl: extras/webp_to_sdl.o
 extras/vwebp_sdl: imageio/libimageio_util.a
 extras/vwebp_sdl: src/libwebp.a
-extras/vwebp_sdl: EXTRA_FLAGS += -DWEBP_HAVE_SDL $(SDL_FLAGS)
+extras/vwebp_sdl: EXTRA_FLAGS += -DWEBP_HAVE_SDL
-extras/vwebp_sdl: EXTRA_LIBS += $(SDL_LIBS)
+extras/vwebp_sdl: EXTRA_LIBS += -lSDL
 
 $(OUT_EXAMPLES) $(EXTRA_EXAMPLES) $(OTHER_EXAMPLES):
 	$(CC) -o $@ $^ $(LDFLAGS)
@@ -479,7 +454,7 @@ dist: all
 	for m in man/[cdv]webp.1 man/gif2webp.1 man/webpmux.1 \
                  man/img2webp.1 man/webpinfo.1; do \
 	  basenam=$$(basename $$m .1); \
-	  $(GROFF) -t -e -man -T ascii $$m \
+	  $(GROFF) -t -e -man -T utf8 $$m \
 	    | $(COL) -bx >$(DESTDIR)/doc/$${basenam}.txt; \
 	  $(GROFF) -t -e -man -T html $$m \
 	    | $(COL) -bx >$(DESTDIR)/doc/$${basenam}.html; \

man/gif2webp.1

@@ -1,5 +1,5 @@
 .\"                                      Hey, EMACS: -*- nroff -*-
-.TH GIF2WEBP 1 "January 25, 2018"
+.TH GIF2WEBP 1 "January 25, 2017"
 .SH NAME
 gif2webp \- Convert a GIF image to WebP
 .SH SYNOPSIS
@@ -20,12 +20,6 @@ Specify the name of the output WebP file. If omitted, \fBgif2webp\fP will
 perform conversion but only report statistics.
 Using "\-" as output name will direct output to 'stdout'.
 .TP
-.BI \-\- " string
-Explicitly specify the input file. This option is useful if the input
-file starts with an '\-' for instance. This option must appear \fBlast\fP.
-Any other options afterward will be ignored. If the input file is "\-",
-the data will be read from \fIstdin\fP instead of a file.
-.TP
 .B \-h, \-help
 Usage information.
 .TP
@@ -116,10 +110,6 @@ the range of 20 to 50.
 .B \-mt
 Use multi-threading for encoding, if possible.
 .TP
-.B \-loop_compatibility
-If enabled, handle the loop information in a compatible fashion for Chrome
-version prior to M62 (inclusive) and Firefox.
-.TP
 .B \-v
 Print extra information.
 .TP
@@ -143,8 +133,6 @@ gif2webp \-lossy \-m 3 picture.gif \-o picture_lossy.webp
 gif2webp \-lossy \-f 50 picture.gif \-o picture.webp
 .br
 gif2webp \-q 70 \-o picture.webp \-\- \-\-\-picture.gif
-.br
-cat picture.gif | gif2webp \-o \- \-\- \- > output.webp
 
 .SH AUTHORS
 \fBgif2webp\fP is a part of libwebp and was written by the WebP team.

man/img2webp.1

@@ -1,13 +1,11 @@
 .\"                                      Hey, EMACS: -*- nroff -*-
-.TH IMG2WEBP 1 "April 3, 2018"
+.TH IMG2WEBP 1 "January 23, 2017"
 .SH NAME
 img2webp \- create animated WebP file from a sequence of input images.
 .SH SYNOPSIS
 .B img2webp
 [file_level_options] [files] [per_frame_options...]
 .br
-.B img2webp argument_file_name
-.br
 .SH DESCRIPTION
 This manual page documents the
 .B img2webp
@@ -15,9 +13,6 @@ command.
 .PP
 \fBimg2webp\fP compresses a sequence of images using the animated WebP format.
 Input images can either be PNG, JPEG, TIFF or WebP.
-If a single file name (not starting with the character '\-') is supplied as
-the argument, the command line argument are actually tokenized from this file.
-This allows for easy scripting or using large number of arguments.
 .SH FILE-LEVEL OPTIONS
 The file-level options are applied at the beginning of the compression process,
 before the input frames are read.
@@ -45,17 +40,14 @@ lossy or lossless compression for each frame heuristically. This global
 option disables the local option \fB-lossy\fP and \fB-lossless\fP .
 .TP
 .BI \-loop " int
-Specifies the number of times the animation should loop. Using '0'
+Specifies the number of times the animation should loop. Using '0' means
-means 'loop indefinitely'.
+'loop indefinitely'.
 .TP
 .BI \-v
 Be more verbose.
 .TP
 .B \-h, \-help
 A short usage summary.
-.TP
-.B \-version
-Print the version numbers of the relevant libraries used.
 
 .SH PER-FRAME OPTIONS
 The per-frame options are applied for the images following as arguments in the

man/webpinfo.1

@@ -1,5 +1,5 @@
 .\"                                      Hey, EMACS: -*- nroff -*-
-.TH WEBPINFO 1 "November 24, 2017"
+.TH WEBPINFO 1 "May 08, 2017"
 .SH NAME
 webpinfo \- print out the chunk level structure of WebP files
 along with basic integrity checks.
@@ -22,19 +22,16 @@ WebP format.
 
 .SH OPTIONS
 .TP
-.B \-version
+.B -quiet
-Print the version number (as major.minor.revision) and exit.
-.TP
-.B \-quiet
 Do not show chunk parsing information.
 .TP
-.B \-diag
+.B -diag
 Show parsing error diagnosis.
 .TP
-.B \-summary
+.B -summary
 Show chunk stats summary.
 .TP
-.BI \-bitstream_info
+.BI -bitstream_info
 Parse bitstream header.
 .TP
 .B \-h, \-help

man/webpmux.1

@@ -1,5 +1,5 @@
 .\"                                      Hey, EMACS: -*- nroff -*-
-.TH WEBPMUX 1 "December 1, 2017"
+.TH WEBPMUX 1 "November 10, 2016"
 .SH NAME
 webpmux \- create animated WebP files from non\-animated WebP images, extract
 frames from animated WebP images, and manage XMP/EXIF metadata and ICC profile.
@@ -48,8 +48,6 @@ frames from animated WebP images, and manage XMP/EXIF metadata and ICC profile.
 .B webpmux [\-h|\-help]
 .br
 .B webpmux \-version
-.br
-.B webpmux argument_file_name
 .SH DESCRIPTION
 This manual page documents the
 .B webpmux
@@ -57,9 +55,6 @@ command.
 .PP
 \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.
-If a single file name (not starting with the character '\-') is supplied as
-the argument, the command line argument are actually tokenized from this file.
-This allows for easy scripting or using large number of arguments.
 .SH OPTIONS
 .SS GET_OPTIONS (\-get):
 .TP

src/Makefile.am

@@ -22,7 +22,6 @@ commondir = $(includedir)/webp
 libwebp_la_SOURCES =
 libwebpinclude_HEADERS =
 libwebpinclude_HEADERS += webp/encode.h
-
 noinst_HEADERS =
 noinst_HEADERS += webp/format_constants.h
 
@@ -36,7 +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 7:2:0
+libwebp_la_LDFLAGS = -no-undefined -version-info 7:0:0
 libwebpincludedir = $(includedir)/webp
 pkgconfig_DATA = libwebp.pc
 
@@ -48,7 +47,7 @@ if BUILD_LIBWEBPDECODER
   libwebpdecoder_la_LIBADD += dsp/libwebpdspdecode.la
   libwebpdecoder_la_LIBADD += utils/libwebputilsdecode.la
 
-  libwebpdecoder_la_LDFLAGS = -no-undefined -version-info 3:2:0
+  libwebpdecoder_la_LDFLAGS = -no-undefined -version-info 3:0:0
   pkgconfig_DATA += libwebpdecoder.pc
 endif
 

src/dec/Makefile.am

@@ -1,4 +1,3 @@
-AM_CPPFLAGS += -I$(top_builddir) -I$(top_srcdir)
 noinst_LTLIBRARIES = libwebpdecode.la
 
 libwebpdecode_la_SOURCES =
@@ -25,5 +24,5 @@ libwebpdecodeinclude_HEADERS += ../webp/types.h
 noinst_HEADERS =
 noinst_HEADERS += ../webp/format_constants.h
 
-libwebpdecode_la_CPPFLAGS = $(AM_CPPFLAGS)
+libwebpdecode_la_CPPFLAGS = $(AM_CPPFLAGS) $(USE_EXPERIMENTAL_CODE)
 libwebpdecodeincludedir = $(includedir)/webp

src/dec/alpha_dec.c

@@ -12,13 +12,13 @@
 // Author: Skal (pascal.massimino@gmail.com)
 
 #include <stdlib.h>
-#include "src/dec/alphai_dec.h"
+#include "./alphai_dec.h"
-#include "src/dec/vp8i_dec.h"
+#include "./vp8i_dec.h"
-#include "src/dec/vp8li_dec.h"
+#include "./vp8li_dec.h"
-#include "src/dsp/dsp.h"
+#include "../dsp/dsp.h"
-#include "src/utils/quant_levels_dec_utils.h"
+#include "../utils/quant_levels_dec_utils.h"
-#include "src/utils/utils.h"
+#include "../utils/utils.h"
-#include "src/webp/format_constants.h"
+#include "../webp/format_constants.h"
 
 //------------------------------------------------------------------------------
 // ALPHDecoder object.

src/dec/alphai_dec.h

@@ -11,11 +11,11 @@
 //
 // Author: Urvang (urvang@google.com)
 
-#ifndef WEBP_DEC_ALPHAI_DEC_H_
+#ifndef WEBP_DEC_ALPHAI_H_
-#define WEBP_DEC_ALPHAI_DEC_H_
+#define WEBP_DEC_ALPHAI_H_
 
-#include "src/dec/webpi_dec.h"
+#include "./webpi_dec.h"
-#include "src/utils/filters_utils.h"
+#include "../utils/filters_utils.h"
 
 #ifdef __cplusplus
 extern "C" {
@@ -51,4 +51,4 @@ void WebPDeallocateAlphaMemory(VP8Decoder* const dec);
 }    // extern "C"
 #endif
 
-#endif  /* WEBP_DEC_ALPHAI_DEC_H_ */
+#endif  /* WEBP_DEC_ALPHAI_H_ */

src/dec/buffer_dec.c

@@ -13,15 +13,15 @@
 
 #include <stdlib.h>
 
-#include "src/dec/vp8i_dec.h"
+#include "./vp8i_dec.h"
-#include "src/dec/webpi_dec.h"
+#include "./webpi_dec.h"
-#include "src/utils/utils.h"
+#include "../utils/utils.h"
 
 //------------------------------------------------------------------------------
 // WebPDecBuffer
 
 // Number of bytes per pixel for the different color-spaces.
-static const uint8_t kModeBpp[MODE_LAST] = {
+static const int kModeBpp[MODE_LAST] = {
   3, 4, 3, 4, 4, 2, 2,
   4, 4, 4, 2,    // pre-multiplied modes
   1, 1 };
@@ -36,7 +36,7 @@ static int IsValidColorspace(int webp_csp_mode) {
 // strictly speaking, the very last (or first, if flipped) row
 // doesn't require padding.
 #define MIN_BUFFER_SIZE(WIDTH, HEIGHT, STRIDE)       \
-    ((uint64_t)(STRIDE) * ((HEIGHT) - 1) + (WIDTH))
+    (uint64_t)(STRIDE) * ((HEIGHT) - 1) + (WIDTH)
 
 static VP8StatusCode CheckDecBuffer(const WebPDecBuffer* const buffer) {
   int ok = 1;
@@ -98,14 +98,9 @@ static VP8StatusCode AllocateBuffer(WebPDecBuffer* const buffer) {
     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.
-    int stride;
+    const int stride = w * kModeBpp[mode];
-    uint64_t size;
+    const uint64_t size = (uint64_t)stride * h;
 
-    if ((uint64_t)w * kModeBpp[mode] >= (1ull << 32)) {
-      return VP8_STATUS_INVALID_PARAM;
-    }
-    stride = w * kModeBpp[mode];
-    size = (uint64_t)stride * h;
     if (!WebPIsRGBMode(mode)) {
       uv_stride = (w + 1) / 2;
       uv_size = (uint64_t)uv_stride * ((h + 1) / 2);
@@ -174,11 +169,11 @@ VP8StatusCode WebPFlipBuffer(WebPDecBuffer* const buffer) {
   return VP8_STATUS_OK;
 }
 
-VP8StatusCode WebPAllocateDecBuffer(int width, int height,
+VP8StatusCode WebPAllocateDecBuffer(int w, int h,
                                     const WebPDecoderOptions* const options,
-                                    WebPDecBuffer* const buffer) {
+                                    WebPDecBuffer* const out) {
   VP8StatusCode status;
-  if (buffer == NULL || width <= 0 || height <= 0) {
+  if (out == NULL || w <= 0 || h <= 0) {
     return VP8_STATUS_INVALID_PARAM;
   }
   if (options != NULL) {    // First, apply options if there is any.
@@ -187,39 +182,33 @@ VP8StatusCode WebPAllocateDecBuffer(int width, int height,
       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 ||
+      if (x < 0 || y < 0 || cw <= 0 || ch <= 0 || x + cw > w || y + ch > h) {
-          x + cw > width || y + ch > height) {
         return VP8_STATUS_INVALID_PARAM;   // out of frame boundary.
       }
-      width = cw;
+      w = cw;
-      height = ch;
+      h = ch;
     }
-
     if (options->use_scaling) {
-#if !defined(WEBP_REDUCE_SIZE)
       int scaled_width = options->scaled_width;
       int scaled_height = options->scaled_height;
       if (!WebPRescalerGetScaledDimensions(
-              width, height, &scaled_width, &scaled_height)) {
+              w, h, &scaled_width, &scaled_height)) {
         return VP8_STATUS_INVALID_PARAM;
       }
-      width = scaled_width;
+      w = scaled_width;
-      height = scaled_height;
+      h = scaled_height;
-#else
-      return VP8_STATUS_INVALID_PARAM;   // rescaling not supported
-#endif
     }
   }
-  buffer->width = width;
+  out->width = w;
-  buffer->height = height;
+  out->height = h;
 
   // Then, allocate buffer for real.
-  status = AllocateBuffer(buffer);
+  status = AllocateBuffer(out);
   if (status != VP8_STATUS_OK) return status;
 
   // Use the stride trick if vertical flip is needed.
   if (options != NULL && options->flip) {
-    status = WebPFlipBuffer(buffer);
+    status = WebPFlipBuffer(out);
   }
   return status;
 }

src/dec/common_dec.h

@@ -11,8 +11,8 @@
 //
 // Author: Skal (pascal.massimino@gmail.com)
 
-#ifndef WEBP_DEC_COMMON_DEC_H_
+#ifndef WEBP_DEC_COMMON_H_
-#define WEBP_DEC_COMMON_DEC_H_
+#define WEBP_DEC_COMMON_H_
 
 // intra prediction modes
 enum { B_DC_PRED = 0,   // 4x4 modes
@@ -51,4 +51,4 @@ enum { MB_FEATURE_TREE_PROBS = 3,
        NUM_PROBAS = 11
      };
 
-#endif    // WEBP_DEC_COMMON_DEC_H_
+#endif    // WEBP_DEC_COMMON_H_

src/dec/frame_dec.c

@@ -12,13 +12,13 @@
 // Author: Skal (pascal.massimino@gmail.com)
 
 #include <stdlib.h>
-#include "src/dec/vp8i_dec.h"
+#include "./vp8i_dec.h"
-#include "src/utils/utils.h"
+#include "../utils/utils.h"
 
 //------------------------------------------------------------------------------
 // Main reconstruction function.
 
-static const uint16_t kScan[16] = {
+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,
@@ -320,7 +320,7 @@ static void PrecomputeFilterStrengths(VP8Decoder* const dec) {
 #define MIN_DITHER_AMP 4
 
 #define DITHER_AMP_TAB_SIZE 12
-static const uint8_t kQuantToDitherAmp[DITHER_AMP_TAB_SIZE] = {
+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
 };
@@ -400,9 +400,7 @@ static void DitherRow(VP8Decoder* const dec) {
 #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(void* arg1, void* arg2) {
+static int FinishRow(VP8Decoder* const dec, VP8Io* const io) {
-  VP8Decoder* const dec = (VP8Decoder*)arg1;
-  VP8Io* const io = (VP8Io*)arg2;
   int ok = 1;
   const VP8ThreadContext* const ctx = &dec->thread_ctx_;
   const int cache_id = ctx->id_;
@@ -450,9 +448,10 @@ static int FinishRow(void* arg1, void* arg2) {
     if (y_end > io->crop_bottom) {
       y_end = io->crop_bottom;    // make sure we don't overflow on last row.
     }
-    // If dec->alpha_data_ is not NULL, we have some alpha plane present.
     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, io, y_start, y_end - y_start);
       if (io->a == NULL) {
         return VP8SetError(dec, VP8_STATUS_BITSTREAM_ERROR,
@@ -559,6 +558,7 @@ VP8StatusCode VP8EnterCritical(VP8Decoder* const dec, VP8Io* const io) {
   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.
   //
@@ -569,6 +569,8 @@ VP8StatusCode VP8EnterCritical(VP8Decoder* const dec, VP8Io* const io) {
   // 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) {
@@ -649,7 +651,7 @@ static int InitThreadContext(VP8Decoder* const dec) {
     }
     worker->data1 = dec;
     worker->data2 = (void*)&dec->thread_ctx_.io_;
-    worker->hook = FinishRow;
+    worker->hook = (WebPWorkerHook)FinishRow;
     dec->num_caches_ =
       (dec->filter_type_ > 0) ? MT_CACHE_LINES : MT_CACHE_LINES - 1;
   } else {
@@ -726,7 +728,7 @@ static int AllocateMemory(VP8Decoder* const dec) {
   }
 
   mem = (uint8_t*)dec->mem_;
-  dec->intra_t_ = mem;
+  dec->intra_t_ = (uint8_t*)mem;
   mem += intra_pred_mode_size;
 
   dec->yuv_t_ = (VP8TopSamples*)mem;
@@ -748,7 +750,7 @@ static int AllocateMemory(VP8Decoder* const dec) {
 
   mem = (uint8_t*)WEBP_ALIGN(mem);
   assert((yuv_size & WEBP_ALIGN_CST) == 0);
-  dec->yuv_b_ = mem;
+  dec->yuv_b_ = (uint8_t*)mem;
   mem += yuv_size;
 
   dec->mb_data_ = (VP8MBData*)mem;
@@ -764,7 +766,7 @@ static int AllocateMemory(VP8Decoder* const dec) {
     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_ = mem + extra_y;
+    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_
@@ -774,7 +776,7 @@ static int AllocateMemory(VP8Decoder* const dec) {
   mem += cache_size;
 
   // alpha plane
-  dec->alpha_plane_ = alpha_size ? mem : NULL;
+  dec->alpha_plane_ = alpha_size ? (uint8_t*)mem : NULL;
   mem += alpha_size;
   assert(mem <= (uint8_t*)dec->mem_ + dec->mem_size_);
 

src/dec/idec_dec.c

@@ -15,10 +15,10 @@
 #include <string.h>
 #include <stdlib.h>
 
-#include "src/dec/alphai_dec.h"
+#include "./alphai_dec.h"
-#include "src/dec/webpi_dec.h"
+#include "./webpi_dec.h"
-#include "src/dec/vp8i_dec.h"
+#include "./vp8i_dec.h"
-#include "src/utils/utils.h"
+#include "../utils/utils.h"
 
 // In append mode, buffer allocations increase as multiples of this value.
 // Needs to be a power of 2.
@@ -673,12 +673,12 @@ void WebPIDelete(WebPIDecoder* idec) {
 //------------------------------------------------------------------------------
 // Wrapper toward WebPINewDecoder
 
-WebPIDecoder* WebPINewRGB(WEBP_CSP_MODE csp, uint8_t* output_buffer,
+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) ? 1 : 0;
   WebPIDecoder* idec;
 
-  if (csp >= MODE_YUV) return NULL;
+  if (mode >= MODE_YUV) return NULL;
   if (is_external_memory == 0) {    // Overwrite parameters to sane values.
     output_buffer_size = 0;
     output_stride = 0;
@@ -689,7 +689,7 @@ WebPIDecoder* WebPINewRGB(WEBP_CSP_MODE csp, uint8_t* output_buffer,
   }
   idec = WebPINewDecoder(NULL);
   if (idec == NULL) return NULL;
-  idec->output_.colorspace = csp;
+  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;

src/dec/io_dec.c

@@ -13,11 +13,11 @@
 
 #include <assert.h>
 #include <stdlib.h>
-#include "src/dec/vp8i_dec.h"
+#include "../dec/vp8i_dec.h"
-#include "src/dec/webpi_dec.h"
+#include "./webpi_dec.h"
-#include "src/dsp/dsp.h"
+#include "../dsp/dsp.h"
-#include "src/dsp/yuv.h"
+#include "../dsp/yuv.h"
-#include "src/utils/utils.h"
+#include "../utils/utils.h"
 
 //------------------------------------------------------------------------------
 // Main YUV<->RGB conversion functions
@@ -212,7 +212,7 @@ static int EmitAlphaRGBA4444(const VP8Io* const io, WebPDecParams* const p,
     int num_rows;
     const int start_y = GetAlphaSourceRow(io, &alpha, &num_rows);
     uint8_t* const base_rgba = buf->rgba + start_y * buf->stride;
-#if (WEBP_SWAP_16BIT_CSP == 1)
+#ifdef WEBP_SWAP_16BIT_CSP
     uint8_t* alpha_dst = base_rgba;
 #else
     uint8_t* alpha_dst = base_rgba + 1;
@@ -241,7 +241,6 @@ static int EmitAlphaRGBA4444(const VP8Io* const io, WebPDecParams* const p,
 //------------------------------------------------------------------------------
 // YUV rescaling (no final RGB conversion needed)
 
-#if !defined(WEBP_REDUCE_SIZE)
 static int Rescale(const uint8_t* src, int src_stride,
                    int new_lines, WebPRescaler* const wrk) {
   int num_lines_out = 0;
@@ -432,7 +431,7 @@ static int ExportAlphaRGBA4444(WebPDecParams* const p, int y_pos,
                                int max_lines_out) {
   const WebPRGBABuffer* const buf = &p->output->u.RGBA;
   uint8_t* const base_rgba = buf->rgba + y_pos * buf->stride;
-#if (WEBP_SWAP_16BIT_CSP == 1)
+#ifdef WEBP_SWAP_16BIT_CSP
   uint8_t* alpha_dst = base_rgba;
 #else
   uint8_t* alpha_dst = base_rgba + 1;
@@ -542,8 +541,6 @@ static int InitRGBRescaler(const VP8Io* const io, WebPDecParams* const p) {
   return 1;
 }
 
-#endif  // WEBP_REDUCE_SIZE
-
 //------------------------------------------------------------------------------
 // Default custom functions
 
@@ -564,14 +561,10 @@ static int CustomSetup(VP8Io* io) {
     WebPInitUpsamplers();
   }
   if (io->use_scaling) {
-#if !defined(WEBP_REDUCE_SIZE)
     const int ok = is_rgb ? InitRGBRescaler(io, p) : InitYUVRescaler(io, p);
     if (!ok) {
       return 0;    // memory error
     }
-#else
-    return 0;   // rescaling support not compiled
-#endif
   } else {
     if (is_rgb) {
       WebPInitSamplers();
@@ -605,6 +598,9 @@ static int CustomSetup(VP8Io* io) {
     }
   }
 
+  if (is_rgb) {
+    VP8YUVInit();
+  }
   return 1;
 }
 

src/dec/quant_dec.c

@@ -11,7 +11,7 @@
 //
 // Author: Skal (pascal.massimino@gmail.com)
 
-#include "src/dec/vp8i_dec.h"
+#include "./vp8i_dec.h"
 
 static WEBP_INLINE int clip(int v, int M) {
   return v < 0 ? 0 : v > M ? M : v;

src/dec/tree_dec.c

@@ -11,19 +11,15 @@
 //
 // Author: Skal (pascal.massimino@gmail.com)
 
-#include "src/dec/vp8i_dec.h"
+#include "./vp8i_dec.h"
-#include "src/utils/bit_reader_inl_utils.h"
+#include "../utils/bit_reader_inl_utils.h"
 
-#if !defined(USE_GENERIC_TREE)
 #if !defined(__arm__) && !defined(_M_ARM) && !defined(__aarch64__)
 // using a table is ~1-2% slower on ARM. Prefer the coded-tree approach then.
-#define USE_GENERIC_TREE 1   // ALTERNATE_CODE
+#define USE_GENERIC_TREE
-#else
-#define USE_GENERIC_TREE 0
 #endif
-#endif  // USE_GENERIC_TREE
 
-#if (USE_GENERIC_TREE == 1)
+#ifdef USE_GENERIC_TREE
 static const int8_t kYModesIntra4[18] = {
   -B_DC_PRED, 1,
     -B_TM_PRED, 2,
@@ -321,7 +317,7 @@ static void ParseIntraMode(VP8BitReader* const br,
       int x;
       for (x = 0; x < 4; ++x) {
         const uint8_t* const prob = kBModesProba[top[x]][ymode];
-#if (USE_GENERIC_TREE == 1)
+#ifdef USE_GENERIC_TREE
         // Generic tree-parsing
         int i = kYModesIntra4[VP8GetBit(br, prob[0])];
         while (i > 0) {
@@ -339,7 +335,7 @@ static void ParseIntraMode(VP8BitReader* const br,
                         (!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
+#endif    // USE_GENERIC_TREE
         top[x] = ymode;
       }
       memcpy(modes, top, 4 * sizeof(*top));
@@ -502,7 +498,7 @@ static const uint8_t
 
 // Paragraph 9.9
 
-static const uint8_t kBands[16 + 1] = {
+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
 };

src/dec/vp8_dec.c

@@ -13,12 +13,12 @@
 
 #include <stdlib.h>
 
-#include "src/dec/alphai_dec.h"
+#include "./alphai_dec.h"
-#include "src/dec/vp8i_dec.h"
+#include "./vp8i_dec.h"
-#include "src/dec/vp8li_dec.h"
+#include "./vp8li_dec.h"
-#include "src/dec/webpi_dec.h"
+#include "./webpi_dec.h"
-#include "src/utils/bit_reader_inl_utils.h"
+#include "../utils/bit_reader_inl_utils.h"
-#include "src/utils/utils.h"
+#include "../utils/utils.h"
 
 //------------------------------------------------------------------------------
 
@@ -491,7 +491,7 @@ static int GetCoeffsAlt(VP8BitReader* const br,
   return 16;
 }
 
-static WEBP_TSAN_IGNORE_FUNCTION void InitGetCoeffs(void) {
+WEBP_TSAN_IGNORE_FUNCTION static void InitGetCoeffs(void) {
   if (GetCoeffs == NULL) {
     if (VP8GetCPUInfo != NULL && VP8GetCPUInfo(kSlowSSSE3)) {
       GetCoeffs = GetCoeffsAlt;

src/dec/vp8_dec.h

@@ -11,10 +11,10 @@
 //
 // Author: Skal (pascal.massimino@gmail.com)
 
-#ifndef WEBP_DEC_VP8_DEC_H_
+#ifndef WEBP_WEBP_DECODE_VP8_H_
-#define WEBP_DEC_VP8_DEC_H_
+#define WEBP_WEBP_DECODE_VP8_H_
 
-#include "src/webp/decode.h"
+#include "../webp/decode.h"
 
 #ifdef __cplusplus
 extern "C" {
@@ -157,24 +157,24 @@ 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);
+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(
+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);
+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(
+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);
 
@@ -182,4 +182,4 @@ WEBP_EXTERN int VP8LGetInfo(
 }    // extern "C"
 #endif
 
-#endif  /* WEBP_DEC_VP8_DEC_H_ */
+#endif  /* WEBP_WEBP_DECODE_VP8_H_ */

src/dec/vp8i_dec.h

@@ -11,16 +11,16 @@
 //
 // Author: Skal (pascal.massimino@gmail.com)
 
-#ifndef WEBP_DEC_VP8I_DEC_H_
+#ifndef WEBP_DEC_VP8I_H_
-#define WEBP_DEC_VP8I_DEC_H_
+#define WEBP_DEC_VP8I_H_
 
 #include <string.h>     // for memcpy()
-#include "src/dec/common_dec.h"
+#include "./common_dec.h"
-#include "src/dec/vp8li_dec.h"
+#include "./vp8li_dec.h"
-#include "src/utils/bit_reader_utils.h"
+#include "../utils/bit_reader_utils.h"
-#include "src/utils/random_utils.h"
+#include "../utils/random_utils.h"
-#include "src/utils/thread_utils.h"
+#include "../utils/thread_utils.h"
-#include "src/dsp/dsp.h"
+#include "../dsp/dsp.h"
 
 #ifdef __cplusplus
 extern "C" {
@@ -30,8 +30,8 @@ extern "C" {
 // Various defines and enums
 
 // version numbers
-#define DEC_MAJ_VERSION 1
+#define DEC_MAJ_VERSION 0
-#define DEC_MIN_VERSION 0
+#define DEC_MIN_VERSION 6
 #define DEC_REV_VERSION 0
 
 // YUV-cache parameters. Cache is 32-bytes wide (= one cacheline).
@@ -57,6 +57,7 @@ extern "C" {
 //  '|' = left sample,   '-' = top sample,    '+' = top-left sample
 //  't' = extra top-right sample for 4x4 modes
 #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)
@@ -316,4 +317,4 @@ const uint8_t* VP8DecompressAlphaRows(VP8Decoder* const dec,
 }    // extern "C"
 #endif
 
-#endif  /* WEBP_DEC_VP8I_DEC_H_ */
+#endif  /* WEBP_DEC_VP8I_H_ */

src/dec/vp8l_dec.c

@@ -14,22 +14,22 @@
 
 #include <stdlib.h>
 
-#include "src/dec/alphai_dec.h"
+#include "./alphai_dec.h"
-#include "src/dec/vp8li_dec.h"
+#include "./vp8li_dec.h"
-#include "src/dsp/dsp.h"
+#include "../dsp/dsp.h"
-#include "src/dsp/lossless.h"
+#include "../dsp/lossless.h"
-#include "src/dsp/lossless_common.h"
+#include "../dsp/lossless_common.h"
-#include "src/dsp/yuv.h"
+#include "../dsp/yuv.h"
-#include "src/utils/endian_inl_utils.h"
+#include "../utils/endian_inl_utils.h"
-#include "src/utils/huffman_utils.h"
+#include "../utils/huffman_utils.h"
-#include "src/utils/utils.h"
+#include "../utils/utils.h"
 
 #define NUM_ARGB_CACHE_ROWS          16
 
 static const int kCodeLengthLiterals = 16;
 static const int kCodeLengthRepeatCode = 16;
-static const uint8_t kCodeLengthExtraBits[3] = { 2, 3, 7 };
+static const int kCodeLengthExtraBits[3] = { 2, 3, 7 };
-static const uint8_t kCodeLengthRepeatOffsets[3] = { 3, 3, 11 };
+static const int kCodeLengthRepeatOffsets[3] = { 3, 3, 11 };
 
 // -----------------------------------------------------------------------------
 //  Five Huffman codes are used at each meta code:
@@ -86,7 +86,7 @@ static const uint8_t kCodeToPlane[CODE_TO_PLANE_CODES] = {
 // All values computed for 8-bit first level lookup with Mark Adler's tool:
 // http://www.hdfgroup.org/ftp/lib-external/zlib/zlib-1.2.5/examples/enough.c
 #define FIXED_TABLE_SIZE (630 * 3 + 410)
-static const uint16_t kTableSize[12] = {
+static const int kTableSize[12] = {
   FIXED_TABLE_SIZE + 654,
   FIXED_TABLE_SIZE + 656,
   FIXED_TABLE_SIZE + 658,
@@ -485,7 +485,6 @@ static int ReadHuffmanCodes(VP8LDecoder* const dec, int xsize, int ysize,
 //------------------------------------------------------------------------------
 // Scaling.
 
-#if !defined(WEBP_REDUCE_SIZE)
 static int AllocateAndInitRescaler(VP8LDecoder* const dec, VP8Io* const io) {
   const int num_channels = 4;
   const int in_width = io->mb_w;
@@ -517,13 +516,10 @@ static int AllocateAndInitRescaler(VP8LDecoder* const dec, VP8Io* const io) {
                    out_width, out_height, 0, num_channels, work);
   return 1;
 }
-#endif   // WEBP_REDUCE_SIZE
 
 //------------------------------------------------------------------------------
 // Export to ARGB
 
-#if !defined(WEBP_REDUCE_SIZE)
-
 // We have special "export" function since we need to convert from BGRA
 static int Export(WebPRescaler* const rescaler, WEBP_CSP_MODE colorspace,
                   int rgba_stride, uint8_t* const rgba) {
@@ -565,8 +561,6 @@ static int EmitRescaledRowsRGBA(const VP8LDecoder* const dec,
   return num_lines_out;
 }
 
-#endif   // WEBP_REDUCE_SIZE
-
 // Emit rows without any scaling.
 static int EmitRows(WEBP_CSP_MODE colorspace,
                     const uint8_t* row_in, int in_stride,
@@ -752,12 +746,9 @@ static void ProcessRows(VP8LDecoder* const dec, int row) {
       if (WebPIsRGBMode(output->colorspace)) {  // convert to RGBA
         const WebPRGBABuffer* const buf = &output->u.RGBA;
         uint8_t* const rgba = buf->rgba + dec->last_out_row_ * buf->stride;
-        const int num_rows_out =
+        const int num_rows_out = io->use_scaling ?
-#if !defined(WEBP_REDUCE_SIZE)
-         io->use_scaling ?
             EmitRescaledRowsRGBA(dec, rows_data, in_stride, io->mb_h,
                                  rgba, buf->stride) :
-#endif  // WEBP_REDUCE_SIZE
             EmitRows(output->colorspace, rows_data, in_stride,
                      io->mb_w, io->mb_h, rgba, buf->stride);
         // Update 'last_out_row_'.
@@ -1641,19 +1632,12 @@ int VP8LDecodeImage(VP8LDecoder* const dec) {
 
     if (!AllocateInternalBuffers32b(dec, io->width)) goto Err;
 
-#if !defined(WEBP_REDUCE_SIZE)
     if (io->use_scaling && !AllocateAndInitRescaler(dec, io)) goto Err;
-#else
+
-    if (io->use_scaling) {
-      dec->status_ = VP8_STATUS_INVALID_PARAM;
-      goto Err;
-    }
-#endif
     if (io->use_scaling || WebPIsPremultipliedMode(dec->output_->colorspace)) {
       // need the alpha-multiply functions for premultiplied output or rescaling
       WebPInitAlphaProcessing();
     }
-
     if (!WebPIsRGBMode(dec->output_->colorspace)) {
       WebPInitConvertARGBToYUV();
       if (dec->output_->u.YUVA.a != NULL) WebPInitAlphaProcessing();

src/dec/vp8li_dec.h

@@ -12,14 +12,14 @@
 // Author: Skal (pascal.massimino@gmail.com)
 //         Vikas Arora(vikaas.arora@gmail.com)
 
-#ifndef WEBP_DEC_VP8LI_DEC_H_
+#ifndef WEBP_DEC_VP8LI_H_
-#define WEBP_DEC_VP8LI_DEC_H_
+#define WEBP_DEC_VP8LI_H_
 
 #include <string.h>     // for memcpy()
-#include "src/dec/webpi_dec.h"
+#include "./webpi_dec.h"
-#include "src/utils/bit_reader_utils.h"
+#include "../utils/bit_reader_utils.h"
-#include "src/utils/color_cache_utils.h"
+#include "../utils/color_cache_utils.h"
-#include "src/utils/huffman_utils.h"
+#include "../utils/huffman_utils.h"
 
 #ifdef __cplusplus
 extern "C" {
@@ -132,4 +132,4 @@ void VP8LDelete(VP8LDecoder* const dec);
 }    // extern "C"
 #endif
 
-#endif  /* WEBP_DEC_VP8LI_DEC_H_ */
+#endif  /* WEBP_DEC_VP8LI_H_ */

src/dec/webp_dec.c

@@ -13,11 +13,11 @@
 
 #include <stdlib.h>
 
-#include "src/dec/vp8i_dec.h"
+#include "./vp8i_dec.h"
-#include "src/dec/vp8li_dec.h"
+#include "./vp8li_dec.h"
-#include "src/dec/webpi_dec.h"
+#include "./webpi_dec.h"
-#include "src/utils/utils.h"
+#include "../utils/utils.h"
-#include "src/webp/mux_types.h"  // ALPHA_FLAG
+#include "../webp/mux_types.h"  // ALPHA_FLAG
 
 //------------------------------------------------------------------------------
 // RIFF layout is:
@@ -421,9 +421,7 @@ VP8StatusCode WebPParseHeaders(WebPHeaderStructure* const headers) {
                                 NULL, NULL, NULL, &has_animation,
                                 NULL, headers);
   if (status == VP8_STATUS_OK || status == VP8_STATUS_NOT_ENOUGH_DATA) {
-    // The WebPDemux API + libwebp can be used to decode individual
+    // TODO(jzern): full support of animation frames will require API additions.
-    // uncomposited frames or the WebPAnimDecoder can be used to fully
-    // reconstruct them (see webp/demux.h).
     if (has_animation) {
       status = VP8_STATUS_UNSUPPORTED_FEATURE;
     }

src/dec/webpi_dec.h

@@ -11,15 +11,15 @@
 //
 // Author: somnath@google.com (Somnath Banerjee)
 
-#ifndef WEBP_DEC_WEBPI_DEC_H_
+#ifndef WEBP_DEC_WEBPI_H_
-#define WEBP_DEC_WEBPI_DEC_H_
+#define WEBP_DEC_WEBPI_H_
 
 #ifdef __cplusplus
 extern "C" {
 #endif
 
-#include "src/utils/rescaler_utils.h"
+#include "../utils/rescaler_utils.h"
-#include "src/dec/vp8_dec.h"
+#include "./vp8_dec.h"
 
 //------------------------------------------------------------------------------
 // WebPDecParams: Decoding output parameters. Transient internal object.
@@ -130,4 +130,4 @@ int WebPAvoidSlowMemory(const WebPDecBuffer* const output,
 }    // extern "C"
 #endif
 
-#endif  /* WEBP_DEC_WEBPI_DEC_H_ */
+#endif  /* WEBP_DEC_WEBPI_H_ */

src/demux/Makefile.am

@@ -1,18 +1,14 @@
-AM_CPPFLAGS += -I$(top_builddir) -I$(top_srcdir)
 lib_LTLIBRARIES = libwebpdemux.la
 
 libwebpdemux_la_SOURCES =
 libwebpdemux_la_SOURCES += anim_decode.c demux.c
 
 libwebpdemuxinclude_HEADERS =
-libwebpdemuxinclude_HEADERS += ../webp/decode.h
 libwebpdemuxinclude_HEADERS += ../webp/demux.h
 libwebpdemuxinclude_HEADERS += ../webp/mux_types.h
 libwebpdemuxinclude_HEADERS += ../webp/types.h
-noinst_HEADERS =
-noinst_HEADERS += ../webp/format_constants.h
 
 libwebpdemux_la_LIBADD = ../libwebp.la
-libwebpdemux_la_LDFLAGS = -no-undefined -version-info 2:4:0
+libwebpdemux_la_LDFLAGS = -no-undefined -version-info 2:2:0
 libwebpdemuxincludedir = $(includedir)/webp
 pkgconfig_DATA = libwebpdemux.pc

src/demux/anim_decode.c

@@ -11,15 +11,15 @@
 //
 
 #ifdef HAVE_CONFIG_H
-#include "src/webp/config.h"
+#include "../webp/config.h"
 #endif
 
 #include <assert.h>
 #include <string.h>
 
-#include "src/utils/utils.h"
+#include "../utils/utils.h"
-#include "src/webp/decode.h"
+#include "../webp/decode.h"
-#include "src/webp/demux.h"
+#include "../webp/demux.h"
 
 #define NUM_CHANNELS 4
 

src/demux/demux.c

@@ -11,21 +11,21 @@
 //
 
 #ifdef HAVE_CONFIG_H
-#include "src/webp/config.h"
+#include "../webp/config.h"
 #endif
 
 #include <assert.h>
 #include <stdlib.h>
 #include <string.h>
 
-#include "src/utils/utils.h"
+#include "../utils/utils.h"
-#include "src/webp/decode.h"     // WebPGetFeatures
+#include "../webp/decode.h"     // WebPGetFeatures
-#include "src/webp/demux.h"
+#include "../webp/demux.h"
-#include "src/webp/format_constants.h"
+#include "../webp/format_constants.h"
 
-#define DMUX_MAJ_VERSION 1
+#define DMUX_MAJ_VERSION 0
-#define DMUX_MIN_VERSION 0
+#define DMUX_MIN_VERSION 3
-#define DMUX_REV_VERSION 0
+#define DMUX_REV_VERSION 2
 
 typedef struct {
   size_t start_;        // start location of the data
@@ -205,14 +205,12 @@ static void SetFrameInfo(size_t start_offset, size_t size,
   frame->complete_ = complete;
 }
 
-// Store image bearing chunks to 'frame'. 'min_size' is an optional size
+// Store image bearing chunks to 'frame'.
-// requirement, it may be zero.
 static ParseStatus StoreFrame(int frame_num, uint32_t min_size,
                               MemBuffer* const mem, Frame* const frame) {
   int alpha_chunks = 0;
   int image_chunks = 0;
-  int done = (MemDataSize(mem) < CHUNK_HEADER_SIZE ||
+  int done = (MemDataSize(mem) < min_size);
-              MemDataSize(mem) < min_size);
   ParseStatus status = PARSE_OK;
 
   if (done) return PARSE_NEED_MORE_DATA;
@@ -403,9 +401,9 @@ static ParseStatus ParseSingleImage(WebPDemuxer* const dmux) {
   frame = (Frame*)WebPSafeCalloc(1ULL, sizeof(*frame));
   if (frame == NULL) return PARSE_ERROR;
 
-  // For the single image case we allow parsing of a partial frame, so no
+  // For the single image case we allow parsing of a partial frame, but we need
-  // minimum size is imposed here.
+  // at least CHUNK_HEADER_SIZE for parsing.
-  status = StoreFrame(1, 0, &dmux->mem_, frame);
+  status = StoreFrame(1, CHUNK_HEADER_SIZE, &dmux->mem_, frame);
   if (status != PARSE_ERROR) {
     const int has_alpha = !!(dmux->feature_flags_ & ALPHA_FLAG);
     // Clear any alpha when the alpha flag is missing.

src/demux/libwebpdemux.rc

@@ -6,8 +6,8 @@
 LANGUAGE LANG_ENGLISH, SUBLANG_ENGLISH_US
 
 VS_VERSION_INFO VERSIONINFO
- FILEVERSION 1,0,0,0
+ FILEVERSION 0,3,0,2
- PRODUCTVERSION 1,0,0,0
+ PRODUCTVERSION 0,3,0,2
  FILEFLAGSMASK 0x3fL
 #ifdef _DEBUG
  FILEFLAGS 0x1L
@@ -24,12 +24,12 @@ BEGIN
         BEGIN
             VALUE "CompanyName", "Google, Inc."
             VALUE "FileDescription", "libwebpdemux DLL"
-            VALUE "FileVersion", "1.0.0"
+            VALUE "FileVersion", "0.3.2"
             VALUE "InternalName", "libwebpdemux.dll"
-            VALUE "LegalCopyright", "Copyright (C) 2018"
+            VALUE "LegalCopyright", "Copyright (C) 2017"
             VALUE "OriginalFilename", "libwebpdemux.dll"
             VALUE "ProductName", "WebP Image Demuxer"
-            VALUE "ProductVersion", "1.0.0"
+            VALUE "ProductVersion", "0.3.2"
         END
     END
     BLOCK "VarFileInfo"

src/dsp/Makefile.am

@@ -1,15 +1,9 @@
-AM_CPPFLAGS += -I$(top_builddir) -I$(top_srcdir)
+noinst_LTLIBRARIES = libwebpdsp.la libwebpdsp_avx2.la
-noinst_LTLIBRARIES =
+noinst_LTLIBRARIES += libwebpdsp_sse2.la libwebpdspdecode_sse2.la
-noinst_LTLIBRARIES += libwebpdsp.la
+noinst_LTLIBRARIES += libwebpdsp_sse41.la libwebpdspdecode_sse41.la
-noinst_LTLIBRARIES += libwebpdsp_avx2.la
+noinst_LTLIBRARIES += libwebpdsp_neon.la libwebpdspdecode_neon.la
-noinst_LTLIBRARIES += libwebpdsp_sse2.la
+noinst_LTLIBRARIES += libwebpdsp_msa.la libwebpdspdecode_msa.la
-noinst_LTLIBRARIES += libwebpdspdecode_sse2.la
+noinst_LTLIBRARIES += libwebpdspdecode_wasm.la
-noinst_LTLIBRARIES += libwebpdsp_sse41.la
-noinst_LTLIBRARIES += libwebpdspdecode_sse41.la
-noinst_LTLIBRARIES += libwebpdsp_neon.la
-noinst_LTLIBRARIES += libwebpdspdecode_neon.la
-noinst_LTLIBRARIES += libwebpdsp_msa.la
-noinst_LTLIBRARIES += libwebpdspdecode_msa.la
 
 if BUILD_LIBWEBPDECODER
   noinst_LTLIBRARIES += libwebpdspdecode.la
@@ -46,6 +40,8 @@ COMMON_SOURCES += yuv_mips32.c
 COMMON_SOURCES += yuv_mips_dsp_r2.c
 
 ENC_SOURCES =
+ENC_SOURCES += argb.c
+ENC_SOURCES += argb_mips_dsp_r2.c
 ENC_SOURCES += cost.c
 ENC_SOURCES += cost_mips32.c
 ENC_SOURCES += cost_mips_dsp_r2.c
@@ -65,8 +61,6 @@ libwebpdsp_avx2_la_CFLAGS = $(AM_CFLAGS) $(AVX2_FLAGS)
 libwebpdspdecode_sse41_la_SOURCES =
 libwebpdspdecode_sse41_la_SOURCES += alpha_processing_sse41.c
 libwebpdspdecode_sse41_la_SOURCES += dec_sse41.c
-libwebpdspdecode_sse41_la_SOURCES += upsampling_sse41.c
-libwebpdspdecode_sse41_la_SOURCES += yuv_sse41.c
 libwebpdspdecode_sse41_la_CPPFLAGS = $(libwebpdsp_la_CPPFLAGS)
 libwebpdspdecode_sse41_la_CFLAGS = $(AM_CFLAGS) $(SSE41_FLAGS)
 
@@ -103,7 +97,12 @@ libwebpdspdecode_msa_la_SOURCES += upsampling_msa.c
 libwebpdspdecode_msa_la_CPPFLAGS = $(libwebpdsp_msa_la_CPPFLAGS)
 libwebpdspdecode_msa_la_CFLAGS = $(libwebpdsp_msa_la_CFLAGS)
 
+# WASM is not fully integrated into configure; the addition here keeps source
+# extraction by cmake simple.
+libwebpdspdecode_wasm_la_SOURCES = dec_wasm.c
+
 libwebpdsp_sse2_la_SOURCES =
+libwebpdsp_sse2_la_SOURCES += argb_sse2.c
 libwebpdsp_sse2_la_SOURCES += cost_sse2.c
 libwebpdsp_sse2_la_SOURCES += enc_sse2.c
 libwebpdsp_sse2_la_SOURCES += lossless_enc_sse2.c
@@ -141,11 +140,10 @@ noinst_HEADERS += ../webp/decode.h
 
 libwebpdsp_la_CPPFLAGS =
 libwebpdsp_la_CPPFLAGS += $(AM_CPPFLAGS)
-libwebpdsp_la_CPPFLAGS += $(USE_SWAP_16BIT_CSP)
+libwebpdsp_la_CPPFLAGS += $(USE_EXPERIMENTAL_CODE) $(USE_SWAP_16BIT_CSP)
 libwebpdsp_la_LDFLAGS = -lm
 libwebpdsp_la_LIBADD =
-libwebpdsp_la_LIBADD += libwebpdsp_avx2.la
+libwebpdsp_la_LIBADD += libwebpdsp_avx2.la libwebpdsp_sse2.la
-libwebpdsp_la_LIBADD += libwebpdsp_sse2.la
 libwebpdsp_la_LIBADD += libwebpdsp_sse41.la
 libwebpdsp_la_LIBADD += libwebpdsp_neon.la
 libwebpdsp_la_LIBADD += libwebpdsp_msa.la

src/dsp/alpha_processing.c

@@ -12,13 +12,10 @@
 // Author: Skal (pascal.massimino@gmail.com)
 
 #include <assert.h>
-#include "src/dsp/dsp.h"
+#include "./dsp.h"
 
 // Tables can be faster on some platform but incur some extra binary size (~2k).
-#if !defined(USE_TABLES_FOR_ALPHA_MULT)
+// #define USE_TABLES_FOR_ALPHA_MULT
-#define USE_TABLES_FOR_ALPHA_MULT 0   // ALTERNATE_CODE
-#endif
-
 
 // -----------------------------------------------------------------------------
 
@@ -32,7 +29,7 @@ static uint32_t Mult(uint8_t x, uint32_t mult) {
   return v;
 }
 
-#if (USE_TABLES_FOR_ALPHA_MULT == 1)
+#ifdef USE_TABLES_FOR_ALPHA_MULT
 
 static const uint32_t kMultTables[2][256] = {
   {    // (255u << MFIX) / alpha
@@ -135,9 +132,9 @@ 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
+#endif    // USE_TABLES_FOR_ALPHA_MULT
 
-void WebPMultARGBRow_C(uint32_t* const ptr, int width, int inverse) {
+void WebPMultARGBRowC(uint32_t* const ptr, int width, int inverse) {
   int x;
   for (x = 0; x < width; ++x) {
     const uint32_t argb = ptr[x];
@@ -157,8 +154,8 @@ void WebPMultARGBRow_C(uint32_t* const ptr, int width, int inverse) {
   }
 }
 
-void WebPMultRow_C(uint8_t* const ptr, const uint8_t* const alpha,
+void WebPMultRowC(uint8_t* const ptr, const uint8_t* const alpha,
-                   int width, int inverse) {
+                  int width, int inverse) {
   int x;
   for (x = 0; x < width; ++x) {
     const uint32_t a = alpha[x];
@@ -220,9 +217,8 @@ void WebPMultRows(uint8_t* ptr, int stride,
 #define PREMULTIPLY(x, m) (((x) * (m) + (1U << 23)) >> 24)
 #endif
 
-#if !WEBP_NEON_OMIT_C_CODE
+static void ApplyAlphaMultiply(uint8_t* rgba, int alpha_first,
-static void ApplyAlphaMultiply_C(uint8_t* rgba, int alpha_first,
+                               int w, int h, int stride) {
-                                 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);
@@ -239,7 +235,6 @@ static void ApplyAlphaMultiply_C(uint8_t* rgba, int alpha_first,
     rgba += stride;
   }
 }
-#endif  // !WEBP_NEON_OMIT_C_CODE
 #undef MULTIPLIER
 #undef PREMULTIPLY
 
@@ -259,9 +254,9 @@ static WEBP_INLINE uint8_t multiply(uint8_t x, uint32_t m) {
   return (x * m) >> 16;
 }
 
-static WEBP_INLINE void ApplyAlphaMultiply4444_C(uint8_t* rgba4444,
+static WEBP_INLINE void ApplyAlphaMultiply4444(uint8_t* rgba4444,
-                                                 int w, int h, int stride,
+                                               int w, int h, int stride,
-                                                 int rg_byte_pos /* 0 or 1 */) {
+                                               int rg_byte_pos /* 0 or 1 */) {
   while (h-- > 0) {
     int i;
     for (i = 0; i < w; ++i) {
@@ -280,16 +275,15 @@ static WEBP_INLINE void ApplyAlphaMultiply4444_C(uint8_t* rgba4444,
 }
 #undef MULTIPLIER
 
-static void ApplyAlphaMultiply_16b_C(uint8_t* rgba4444,
+static void ApplyAlphaMultiply_16b(uint8_t* rgba4444,
-                                     int w, int h, int stride) {
+                                   int w, int h, int stride) {
-#if (WEBP_SWAP_16BIT_CSP == 1)
+#ifdef WEBP_SWAP_16BIT_CSP
-  ApplyAlphaMultiply4444_C(rgba4444, w, h, stride, 1);
+  ApplyAlphaMultiply4444(rgba4444, w, h, stride, 1);
 #else
-  ApplyAlphaMultiply4444_C(rgba4444, w, h, stride, 0);
+  ApplyAlphaMultiply4444(rgba4444, w, h, stride, 0);
 #endif
 }
 
-#if !WEBP_NEON_OMIT_C_CODE
 static int DispatchAlpha_C(const uint8_t* alpha, int alpha_stride,
                            int width, int height,
                            uint8_t* dst, int dst_stride) {
@@ -344,46 +338,6 @@ static void ExtractGreen_C(const uint32_t* argb, uint8_t* alpha, int size) {
   int i;
   for (i = 0; i < size; ++i) alpha[i] = argb[i] >> 8;
 }
-#endif  // !WEBP_NEON_OMIT_C_CODE
-
-//------------------------------------------------------------------------------
-
-static int HasAlpha8b_C(const uint8_t* src, int length) {
-  while (length-- > 0) if (*src++ != 0xff) return 1;
-  return 0;
-}
-
-static int HasAlpha32b_C(const uint8_t* src, int length) {
-  int x;
-  for (x = 0; length-- > 0; x += 4) if (src[x] != 0xff) return 1;
-  return 0;
-}
-
-//------------------------------------------------------------------------------
-// Simple channel manipulations.
-
-static WEBP_INLINE uint32_t MakeARGB32(int a, int r, int g, int b) {
-  return (((uint32_t)a << 24) | (r << 16) | (g << 8) | b);
-}
-
-#ifdef WORDS_BIGENDIAN
-static void PackARGB_C(const uint8_t* a, const uint8_t* r, const uint8_t* g,
-                       const uint8_t* b, int len, uint32_t* out) {
-  int i;
-  for (i = 0; i < len; ++i) {
-    out[i] = MakeARGB32(a[4 * i], r[4 * i], g[4 * i], b[4 * i]);
-  }
-}
-#endif
-
-static void PackRGB_C(const uint8_t* r, const uint8_t* g, const uint8_t* b,
-                      int len, int step, uint32_t* out) {
-  int i, offset = 0;
-  for (i = 0; i < len; ++i) {
-    out[i] = MakeARGB32(0xff, r[offset], g[offset], b[offset]);
-    offset += step;
-  }
-}
 
 void (*WebPApplyAlphaMultiply)(uint8_t*, int, int, int, int);
 void (*WebPApplyAlphaMultiply4444)(uint8_t*, int, int, int);
@@ -391,15 +345,6 @@ int (*WebPDispatchAlpha)(const uint8_t*, int, int, int, uint8_t*, int);
 void (*WebPDispatchAlphaToGreen)(const uint8_t*, int, int, int, uint32_t*, int);
 int (*WebPExtractAlpha)(const uint8_t*, int, int, int, uint8_t*, int);
 void (*WebPExtractGreen)(const uint32_t* argb, uint8_t* alpha, int size);
-#ifdef WORDS_BIGENDIAN
-void (*WebPPackARGB)(const uint8_t* a, const uint8_t* r, const uint8_t* g,
-                     const uint8_t* b, int, uint32_t*);
-#endif
-void (*WebPPackRGB)(const uint8_t* r, const uint8_t* g, const uint8_t* b,
-                    int len, int step, uint32_t* out);
-
-int (*WebPHasAlpha8b)(const uint8_t* src, int length);
-int (*WebPHasAlpha32b)(const uint8_t* src, int length);
 
 //------------------------------------------------------------------------------
 // Init function
@@ -409,25 +354,21 @@ extern void WebPInitAlphaProcessingSSE2(void);
 extern void WebPInitAlphaProcessingSSE41(void);
 extern void WebPInitAlphaProcessingNEON(void);
 
-WEBP_DSP_INIT_FUNC(WebPInitAlphaProcessing) {
+static volatile VP8CPUInfo alpha_processing_last_cpuinfo_used =
-  WebPMultARGBRow = WebPMultARGBRow_C;
+    (VP8CPUInfo)&alpha_processing_last_cpuinfo_used;
-  WebPMultRow = WebPMultRow_C;
+
-  WebPApplyAlphaMultiply4444 = ApplyAlphaMultiply_16b_C;
+WEBP_TSAN_IGNORE_FUNCTION void WebPInitAlphaProcessing(void) {
+  if (alpha_processing_last_cpuinfo_used == VP8GetCPUInfo) return;
+
+  WebPMultARGBRow = WebPMultARGBRowC;
+  WebPMultRow = WebPMultRowC;
+  WebPApplyAlphaMultiply = ApplyAlphaMultiply;
+  WebPApplyAlphaMultiply4444 = ApplyAlphaMultiply_16b;
 
-#ifdef WORDS_BIGENDIAN
-  WebPPackARGB = PackARGB_C;
-#endif
-  WebPPackRGB = PackRGB_C;
-#if !WEBP_NEON_OMIT_C_CODE
-  WebPApplyAlphaMultiply = ApplyAlphaMultiply_C;
   WebPDispatchAlpha = DispatchAlpha_C;
   WebPDispatchAlphaToGreen = DispatchAlphaToGreen_C;
   WebPExtractAlpha = ExtractAlpha_C;
   WebPExtractGreen = ExtractGreen_C;
-#endif
-
-  WebPHasAlpha8b = HasAlpha8b_C;
-  WebPHasAlpha32b = HasAlpha32b_C;
 
   // If defined, use CPUInfo() to overwrite some pointers with faster versions.
   if (VP8GetCPUInfo != NULL) {
@@ -441,32 +382,16 @@ WEBP_DSP_INIT_FUNC(WebPInitAlphaProcessing) {
 #endif
     }
 #endif
+#if defined(WEBP_USE_NEON)
+    if (VP8GetCPUInfo(kNEON)) {
+      WebPInitAlphaProcessingNEON();
+    }
+#endif
 #if defined(WEBP_USE_MIPS_DSP_R2)
     if (VP8GetCPUInfo(kMIPSdspR2)) {
       WebPInitAlphaProcessingMIPSdspR2();
     }
 #endif
   }
-
+  alpha_processing_last_cpuinfo_used = VP8GetCPUInfo;
-#if defined(WEBP_USE_NEON)
-  if (WEBP_NEON_OMIT_C_CODE ||
-      (VP8GetCPUInfo != NULL && VP8GetCPUInfo(kNEON))) {
-    WebPInitAlphaProcessingNEON();
-  }
-#endif
-
-  assert(WebPMultARGBRow != NULL);
-  assert(WebPMultRow != NULL);
-  assert(WebPApplyAlphaMultiply != NULL);
-  assert(WebPApplyAlphaMultiply4444 != NULL);
-  assert(WebPDispatchAlpha != NULL);
-  assert(WebPDispatchAlphaToGreen != NULL);
-  assert(WebPExtractAlpha != NULL);
-  assert(WebPExtractGreen != NULL);
-#ifdef WORDS_BIGENDIAN
-  assert(WebPPackARGB != NULL);
-#endif
-  assert(WebPPackRGB != NULL);
-  assert(WebPHasAlpha8b != NULL);
-  assert(WebPHasAlpha32b != NULL);
 }

src/dsp/alpha_processing_mips_dsp_r2.c

@@ -12,13 +12,13 @@
 // Author(s): Branimir Vasic (branimir.vasic@imgtec.com)
 //            Djordje Pesut  (djordje.pesut@imgtec.com)
 
-#include "src/dsp/dsp.h"
+#include "./dsp.h"
 
 #if defined(WEBP_USE_MIPS_DSP_R2)
 
-static int DispatchAlpha_MIPSdspR2(const uint8_t* alpha, int alpha_stride,
+static int DispatchAlpha(const uint8_t* alpha, int alpha_stride,
-                                   int width, int height,
+                         int width, int height,
-                                   uint8_t* dst, int dst_stride) {
+                         uint8_t* dst, int dst_stride) {
   uint32_t alpha_mask = 0xffffffff;
   int i, j, temp0;
 
@@ -79,8 +79,7 @@ static int DispatchAlpha_MIPSdspR2(const uint8_t* alpha, int alpha_stride,
   return (alpha_mask != 0xff);
 }
 
-static void MultARGBRow_MIPSdspR2(uint32_t* const ptr, int width,
+static void MultARGBRow(uint32_t* const ptr, int width, int inverse) {
-                                  int inverse) {
   int x;
   const uint32_t c_00ffffff = 0x00ffffffu;
   const uint32_t c_ff000000 = 0xff000000u;
@@ -125,100 +124,14 @@ static void MultARGBRow_MIPSdspR2(uint32_t* const ptr, int width,
   }
 }
 
-#ifdef WORDS_BIGENDIAN
-static void PackARGB_MIPSdspR2(const uint8_t* a, const uint8_t* r,
-                               const uint8_t* g, const uint8_t* b, int len,
-                               uint32_t* out) {
-  int temp0, temp1, temp2, temp3, offset;
-  const int rest = len & 1;
-  const uint32_t* const loop_end = out + len - rest;
-  const int step = 4;
-  __asm__ volatile (
-    "xor          %[offset],   %[offset], %[offset]    \n\t"
-    "beq          %[loop_end], %[out],    0f           \n\t"
-  "2:                                                  \n\t"
-    "lbux         %[temp0],    %[offset](%[a])         \n\t"
-    "lbux         %[temp1],    %[offset](%[r])         \n\t"
-    "lbux         %[temp2],    %[offset](%[g])         \n\t"
-    "lbux         %[temp3],    %[offset](%[b])         \n\t"
-    "ins          %[temp1],    %[temp0],  16,     16   \n\t"
-    "ins          %[temp3],    %[temp2],  16,     16   \n\t"
-    "addiu        %[out],      %[out],    4            \n\t"
-    "precr.qb.ph  %[temp0],    %[temp1],  %[temp3]     \n\t"
-    "sw           %[temp0],    -4(%[out])              \n\t"
-    "addu         %[offset],   %[offset], %[step]      \n\t"
-    "bne          %[loop_end], %[out],    2b           \n\t"
-  "0:                                                  \n\t"
-    "beq          %[rest],     $zero,     1f           \n\t"
-    "lbux         %[temp0],    %[offset](%[a])         \n\t"
-    "lbux         %[temp1],    %[offset](%[r])         \n\t"
-    "lbux         %[temp2],    %[offset](%[g])         \n\t"
-    "lbux         %[temp3],    %[offset](%[b])         \n\t"
-    "ins          %[temp1],    %[temp0],  16,     16   \n\t"
-    "ins          %[temp3],    %[temp2],  16,     16   \n\t"
-    "precr.qb.ph  %[temp0],    %[temp1],  %[temp3]     \n\t"
-    "sw           %[temp0],    0(%[out])               \n\t"
-  "1:                                                  \n\t"
-    : [temp0]"=&r"(temp0), [temp1]"=&r"(temp1), [temp2]"=&r"(temp2),
-      [temp3]"=&r"(temp3), [offset]"=&r"(offset), [out]"+&r"(out)
-    : [a]"r"(a), [r]"r"(r), [g]"r"(g), [b]"r"(b), [step]"r"(step),
-      [loop_end]"r"(loop_end), [rest]"r"(rest)
-    : "memory"
-  );
-}
-#endif  // WORDS_BIGENDIAN
-
-static void PackRGB_MIPSdspR2(const uint8_t* r, const uint8_t* g,
-                              const uint8_t* b, int len, int step,
-                              uint32_t* out) {
-  int temp0, temp1, temp2, offset;
-  const int rest = len & 1;
-  const int a = 0xff;
-  const uint32_t* const loop_end = out + len - rest;
-  __asm__ volatile (
-    "xor          %[offset],   %[offset], %[offset]    \n\t"
-    "beq          %[loop_end], %[out],    0f           \n\t"
-  "2:                                                  \n\t"
-    "lbux         %[temp0],    %[offset](%[r])         \n\t"
-    "lbux         %[temp1],    %[offset](%[g])         \n\t"
-    "lbux         %[temp2],    %[offset](%[b])         \n\t"
-    "ins          %[temp0],    %[a],      16,     16   \n\t"
-    "ins          %[temp2],    %[temp1],  16,     16   \n\t"
-    "addiu        %[out],      %[out],    4            \n\t"
-    "precr.qb.ph  %[temp0],    %[temp0],  %[temp2]     \n\t"
-    "sw           %[temp0],    -4(%[out])              \n\t"
-    "addu         %[offset],   %[offset], %[step]      \n\t"
-    "bne          %[loop_end], %[out],    2b           \n\t"
-  "0:                                                  \n\t"
-    "beq          %[rest],     $zero,     1f           \n\t"
-    "lbux         %[temp0],    %[offset](%[r])         \n\t"
-    "lbux         %[temp1],    %[offset](%[g])         \n\t"
-    "lbux         %[temp2],    %[offset](%[b])         \n\t"
-    "ins          %[temp0],    %[a],      16,     16   \n\t"
-    "ins          %[temp2],    %[temp1],  16,     16   \n\t"
-    "precr.qb.ph  %[temp0],    %[temp0],  %[temp2]     \n\t"
-    "sw           %[temp0],    0(%[out])               \n\t"
-  "1:                                                  \n\t"
-    : [temp0]"=&r"(temp0), [temp1]"=&r"(temp1), [temp2]"=&r"(temp2),
-      [offset]"=&r"(offset), [out]"+&r"(out)
-    : [a]"r"(a), [r]"r"(r), [g]"r"(g), [b]"r"(b), [step]"r"(step),
-      [loop_end]"r"(loop_end), [rest]"r"(rest)
-    : "memory"
-  );
-}
-
 //------------------------------------------------------------------------------
 // Entry point
 
 extern void WebPInitAlphaProcessingMIPSdspR2(void);
 
 WEBP_TSAN_IGNORE_FUNCTION void WebPInitAlphaProcessingMIPSdspR2(void) {
-  WebPDispatchAlpha = DispatchAlpha_MIPSdspR2;
+  WebPDispatchAlpha = DispatchAlpha;
-  WebPMultARGBRow = MultARGBRow_MIPSdspR2;
+  WebPMultARGBRow = MultARGBRow;
-#ifdef WORDS_BIGENDIAN
-  WebPPackARGB = PackARGB_MIPSdspR2;
-#endif
-  WebPPackRGB = PackRGB_MIPSdspR2;
 }
 
 #else  // !WEBP_USE_MIPS_DSP_R2

src/dsp/alpha_processing_neon.c

@@ -11,11 +11,11 @@
 //
 // Author: Skal (pascal.massimino@gmail.com)
 
-#include "src/dsp/dsp.h"
+#include "./dsp.h"
 
 #if defined(WEBP_USE_NEON)
 
-#include "src/dsp/neon.h"
+#include "./neon.h"
 
 //------------------------------------------------------------------------------
 

src/dsp/alpha_processing_sse2.c

@@ -11,16 +11,16 @@
 //
 // Author: Skal (pascal.massimino@gmail.com)
 
-#include "src/dsp/dsp.h"
+#include "./dsp.h"
 
 #if defined(WEBP_USE_SSE2)
 #include <emmintrin.h>
 
 //------------------------------------------------------------------------------
 
-static int DispatchAlpha_SSE2(const uint8_t* alpha, int alpha_stride,
+static int DispatchAlpha(const uint8_t* alpha, int alpha_stride,
-                              int width, int height,
+                         int width, int height,
-                              uint8_t* dst, int dst_stride) {
+                         uint8_t* dst, int dst_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;
@@ -72,9 +72,9 @@ static int DispatchAlpha_SSE2(const uint8_t* alpha, int alpha_stride,
   return (alpha_and != 0xff);
 }
 
-static void DispatchAlphaToGreen_SSE2(const uint8_t* alpha, int alpha_stride,
+static void DispatchAlphaToGreen(const uint8_t* alpha, int alpha_stride,
-                                      int width, int height,
+                                 int width, int height,
-                                      uint32_t* dst, int dst_stride) {
+                                 uint32_t* dst, int dst_stride) {
   int i, j;
   const __m128i zero = _mm_setzero_si128();
   const int limit = width & ~15;
@@ -98,9 +98,9 @@ static void DispatchAlphaToGreen_SSE2(const uint8_t* alpha, int alpha_stride,
   }
 }
 
-static int ExtractAlpha_SSE2(const uint8_t* argb, int argb_stride,
+static int ExtractAlpha(const uint8_t* argb, int argb_stride,
-                             int width, int height,
+                        int width, int height,
-                             uint8_t* alpha, int alpha_stride) {
+                        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;
@@ -210,61 +210,6 @@ static void ApplyAlphaMultiply_SSE2(uint8_t* rgba, int alpha_first,
 #undef MULTIPLIER
 #undef PREMULTIPLY
 
-//------------------------------------------------------------------------------
-// Alpha detection
-
-static int HasAlpha8b_SSE2(const uint8_t* src, int length) {
-  const __m128i all_0xff = _mm_set1_epi8(0xff);
-  int i = 0;
-  for (; i + 16 <= length; i += 16) {
-    const __m128i v = _mm_loadu_si128((const __m128i*)(src + i));
-    const __m128i bits = _mm_cmpeq_epi8(v, all_0xff);
-    const int mask = _mm_movemask_epi8(bits);
-    if (mask != 0xffff) return 1;
-  }
-  for (; i < length; ++i) if (src[i] != 0xff) return 1;
-  return 0;
-}
-
-static int HasAlpha32b_SSE2(const uint8_t* src, int length) {
-  const __m128i alpha_mask = _mm_set1_epi32(0xff);
-  const __m128i all_0xff = _mm_set1_epi8(0xff);
-  int i = 0;
-  // We don't know if we can access the last 3 bytes after the last alpha
-  // value 'src[4 * length - 4]' (because we don't know if alpha is the first
-  // or the last byte of the quadruplet). Hence the '-3' protection below.
-  length = length * 4 - 3;   // size in bytes
-  for (; i + 64 <= length; i += 64) {
-    const __m128i a0 = _mm_loadu_si128((const __m128i*)(src + i +  0));
-    const __m128i a1 = _mm_loadu_si128((const __m128i*)(src + i + 16));
-    const __m128i a2 = _mm_loadu_si128((const __m128i*)(src + i + 32));
-    const __m128i a3 = _mm_loadu_si128((const __m128i*)(src + i + 48));
-    const __m128i b0 = _mm_and_si128(a0, alpha_mask);
-    const __m128i b1 = _mm_and_si128(a1, alpha_mask);
-    const __m128i b2 = _mm_and_si128(a2, alpha_mask);
-    const __m128i b3 = _mm_and_si128(a3, alpha_mask);
-    const __m128i c0 = _mm_packs_epi32(b0, b1);
-    const __m128i c1 = _mm_packs_epi32(b2, b3);
-    const __m128i d  = _mm_packus_epi16(c0, c1);
-    const __m128i bits = _mm_cmpeq_epi8(d, all_0xff);
-    const int mask = _mm_movemask_epi8(bits);
-    if (mask != 0xffff) return 1;
-  }
-  for (; i + 32 <= length; i += 32) {
-    const __m128i a0 = _mm_loadu_si128((const __m128i*)(src + i +  0));
-    const __m128i a1 = _mm_loadu_si128((const __m128i*)(src + i + 16));
-    const __m128i b0 = _mm_and_si128(a0, alpha_mask);
-    const __m128i b1 = _mm_and_si128(a1, alpha_mask);
-    const __m128i c  = _mm_packs_epi32(b0, b1);
-    const __m128i d  = _mm_packus_epi16(c, c);
-    const __m128i bits = _mm_cmpeq_epi8(d, all_0xff);
-    const int mask = _mm_movemask_epi8(bits);
-    if (mask != 0xffff) return 1;
-  }
-  for (; i <= length; i += 4) if (src[i] != 0xff) return 1;
-  return 0;
-}
-
 // -----------------------------------------------------------------------------
 // Apply alpha value to rows
 
@@ -293,7 +238,7 @@ static void MultARGBRow_SSE2(uint32_t* const ptr, int width, int inverse) {
     }
   }
   width -= x;
-  if (width > 0) WebPMultARGBRow_C(ptr + x, width, inverse);
+  if (width > 0) WebPMultARGBRowC(ptr + x, width, inverse);
 }
 
 static void MultRow_SSE2(uint8_t* const ptr, const uint8_t* const alpha,
@@ -316,7 +261,7 @@ static void MultRow_SSE2(uint8_t* const ptr, const uint8_t* const alpha,
     }
   }
   width -= x;
-  if (width > 0) WebPMultRow_C(ptr + x, alpha + x, width, inverse);
+  if (width > 0) WebPMultRowC(ptr + x, alpha + x, width, inverse);
 }
 
 //------------------------------------------------------------------------------
@@ -328,12 +273,9 @@ WEBP_TSAN_IGNORE_FUNCTION void WebPInitAlphaProcessingSSE2(void) {
   WebPMultARGBRow = MultARGBRow_SSE2;
   WebPMultRow = MultRow_SSE2;
   WebPApplyAlphaMultiply = ApplyAlphaMultiply_SSE2;
-  WebPDispatchAlpha = DispatchAlpha_SSE2;
+  WebPDispatchAlpha = DispatchAlpha;
-  WebPDispatchAlphaToGreen = DispatchAlphaToGreen_SSE2;
+  WebPDispatchAlphaToGreen = DispatchAlphaToGreen;
-  WebPExtractAlpha = ExtractAlpha_SSE2;
+  WebPExtractAlpha = ExtractAlpha;
-
-  WebPHasAlpha8b = HasAlpha8b_SSE2;
-  WebPHasAlpha32b = HasAlpha32b_SSE2;
 }
 
 #else  // !WEBP_USE_SSE2

src/dsp/alpha_processing_sse41.c

@@ -11,7 +11,7 @@
 //
 // Author: Skal (pascal.massimino@gmail.com)
 
-#include "src/dsp/dsp.h"
+#include "./dsp.h"
 
 #if defined(WEBP_USE_SSE41)
 
@@ -19,9 +19,9 @@
 
 //------------------------------------------------------------------------------
 
-static int ExtractAlpha_SSE41(const uint8_t* argb, int argb_stride,
+static int ExtractAlpha(const uint8_t* argb, int argb_stride,
-                              int width, int height,
+                        int width, int height,
-                              uint8_t* alpha, int alpha_stride) {
+                        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;
@@ -82,7 +82,7 @@ static int ExtractAlpha_SSE41(const uint8_t* argb, int argb_stride,
 extern void WebPInitAlphaProcessingSSE41(void);
 
 WEBP_TSAN_IGNORE_FUNCTION void WebPInitAlphaProcessingSSE41(void) {
-  WebPExtractAlpha = ExtractAlpha_SSE41;
+  WebPExtractAlpha = ExtractAlpha;
 }
 
 #else  // !WEBP_USE_SSE41

src/dsp/argb.c

@@ -0,0 +1,68 @@
+// 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.
+// -----------------------------------------------------------------------------
+//
+//   ARGB making functions.
+//
+// Author: Djordje Pesut (djordje.pesut@imgtec.com)
+
+#include "./dsp.h"
+
+static WEBP_INLINE uint32_t MakeARGB32(int a, int r, int g, int b) {
+  return (((uint32_t)a << 24) | (r << 16) | (g << 8) | b);
+}
+
+static void PackARGB(const uint8_t* a, const uint8_t* r, const uint8_t* g,
+                     const uint8_t* b, int len, uint32_t* out) {
+  int i;
+  for (i = 0; i < len; ++i) {
+    out[i] = MakeARGB32(a[4 * i], r[4 * i], g[4 * i], b[4 * i]);
+  }
+}
+
+static void PackRGB(const uint8_t* r, const uint8_t* g, const uint8_t* b,
+                    int len, int step, uint32_t* out) {
+  int i, offset = 0;
+  for (i = 0; i < len; ++i) {
+    out[i] = MakeARGB32(0xff, r[offset], g[offset], b[offset]);
+    offset += step;
+  }
+}
+
+void (*VP8PackARGB)(const uint8_t*, const uint8_t*, const uint8_t*,
+                    const uint8_t*, int, uint32_t*);
+void (*VP8PackRGB)(const uint8_t*, const uint8_t*, const uint8_t*,
+                   int, int, uint32_t*);
+
+extern void VP8EncDspARGBInitMIPSdspR2(void);
+extern void VP8EncDspARGBInitSSE2(void);
+
+static volatile VP8CPUInfo argb_last_cpuinfo_used =
+    (VP8CPUInfo)&argb_last_cpuinfo_used;
+
+WEBP_TSAN_IGNORE_FUNCTION void VP8EncDspARGBInit(void) {
+  if (argb_last_cpuinfo_used == VP8GetCPUInfo) return;
+
+  VP8PackARGB = PackARGB;
+  VP8PackRGB = PackRGB;
+
+  // If defined, use CPUInfo() to overwrite some pointers with faster versions.
+  if (VP8GetCPUInfo != NULL) {
+#if defined(WEBP_USE_SSE2)
+    if (VP8GetCPUInfo(kSSE2)) {
+      VP8EncDspARGBInitSSE2();
+    }
+#endif
+#if defined(WEBP_USE_MIPS_DSP_R2)
+    if (VP8GetCPUInfo(kMIPSdspR2)) {
+      VP8EncDspARGBInitMIPSdspR2();
+    }
+#endif
+  }
+  argb_last_cpuinfo_used = VP8GetCPUInfo;
+}

src/dsp/argb_mips_dsp_r2.c

@@ -0,0 +1,110 @@
+// 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.
+// -----------------------------------------------------------------------------
+//
+//   ARGB making functions (mips version).
+//
+// Author: Djordje Pesut (djordje.pesut@imgtec.com)
+
+#include "./dsp.h"
+
+#if defined(WEBP_USE_MIPS_DSP_R2)
+
+static void PackARGB(const uint8_t* a, const uint8_t* r, const uint8_t* g,
+                     const uint8_t* b, int len, uint32_t* out) {
+  int temp0, temp1, temp2, temp3, offset;
+  const int rest = len & 1;
+  const uint32_t* const loop_end = out + len - rest;
+  const int step = 4;
+  __asm__ volatile (
+    "xor          %[offset],   %[offset], %[offset]    \n\t"
+    "beq          %[loop_end], %[out],    0f           \n\t"
+  "2:                                                  \n\t"
+    "lbux         %[temp0],    %[offset](%[a])         \n\t"
+    "lbux         %[temp1],    %[offset](%[r])         \n\t"
+    "lbux         %[temp2],    %[offset](%[g])         \n\t"
+    "lbux         %[temp3],    %[offset](%[b])         \n\t"
+    "ins          %[temp1],    %[temp0],  16,     16   \n\t"
+    "ins          %[temp3],    %[temp2],  16,     16   \n\t"
+    "addiu        %[out],      %[out],    4            \n\t"
+    "precr.qb.ph  %[temp0],    %[temp1],  %[temp3]     \n\t"
+    "sw           %[temp0],    -4(%[out])              \n\t"
+    "addu         %[offset],   %[offset], %[step]      \n\t"
+    "bne          %[loop_end], %[out],    2b           \n\t"
+  "0:                                                  \n\t"
+    "beq          %[rest],     $zero,     1f           \n\t"
+    "lbux         %[temp0],    %[offset](%[a])         \n\t"
+    "lbux         %[temp1],    %[offset](%[r])         \n\t"
+    "lbux         %[temp2],    %[offset](%[g])         \n\t"
+    "lbux         %[temp3],    %[offset](%[b])         \n\t"
+    "ins          %[temp1],    %[temp0],  16,     16   \n\t"
+    "ins          %[temp3],    %[temp2],  16,     16   \n\t"
+    "precr.qb.ph  %[temp0],    %[temp1],  %[temp3]     \n\t"
+    "sw           %[temp0],    0(%[out])               \n\t"
+  "1:                                                  \n\t"
+    : [temp0]"=&r"(temp0), [temp1]"=&r"(temp1), [temp2]"=&r"(temp2),
+      [temp3]"=&r"(temp3), [offset]"=&r"(offset), [out]"+&r"(out)
+    : [a]"r"(a), [r]"r"(r), [g]"r"(g), [b]"r"(b), [step]"r"(step),
+      [loop_end]"r"(loop_end), [rest]"r"(rest)
+    : "memory"
+  );
+}
+
+static void PackRGB(const uint8_t* r, const uint8_t* g, const uint8_t* b,
+                    int len, int step, uint32_t* out) {
+  int temp0, temp1, temp2, offset;
+  const int rest = len & 1;
+  const int a = 0xff;
+  const uint32_t* const loop_end = out + len - rest;
+  __asm__ volatile (
+    "xor          %[offset],   %[offset], %[offset]    \n\t"
+    "beq          %[loop_end], %[out],    0f           \n\t"
+  "2:                                                  \n\t"
+    "lbux         %[temp0],    %[offset](%[r])         \n\t"
+    "lbux         %[temp1],    %[offset](%[g])         \n\t"
+    "lbux         %[temp2],    %[offset](%[b])         \n\t"
+    "ins          %[temp0],    %[a],      16,     16   \n\t"
+    "ins          %[temp2],    %[temp1],  16,     16   \n\t"
+    "addiu        %[out],      %[out],    4            \n\t"
+    "precr.qb.ph  %[temp0],    %[temp0],  %[temp2]     \n\t"
+    "sw           %[temp0],    -4(%[out])              \n\t"
+    "addu         %[offset],   %[offset], %[step]      \n\t"
+    "bne          %[loop_end], %[out],    2b           \n\t"
+  "0:                                                  \n\t"
+    "beq          %[rest],     $zero,     1f           \n\t"
+    "lbux         %[temp0],    %[offset](%[r])         \n\t"
+    "lbux         %[temp1],    %[offset](%[g])         \n\t"
+    "lbux         %[temp2],    %[offset](%[b])         \n\t"
+    "ins          %[temp0],    %[a],      16,     16   \n\t"
+    "ins          %[temp2],    %[temp1],  16,     16   \n\t"
+    "precr.qb.ph  %[temp0],    %[temp0],  %[temp2]     \n\t"
+    "sw           %[temp0],    0(%[out])               \n\t"
+  "1:                                                  \n\t"
+    : [temp0]"=&r"(temp0), [temp1]"=&r"(temp1), [temp2]"=&r"(temp2),
+      [offset]"=&r"(offset), [out]"+&r"(out)
+    : [a]"r"(a), [r]"r"(r), [g]"r"(g), [b]"r"(b), [step]"r"(step),
+      [loop_end]"r"(loop_end), [rest]"r"(rest)
+    : "memory"
+  );
+}
+
+//------------------------------------------------------------------------------
+// Entry point
+
+extern void VP8EncDspARGBInitMIPSdspR2(void);
+
+WEBP_TSAN_IGNORE_FUNCTION void VP8EncDspARGBInitMIPSdspR2(void) {
+  VP8PackARGB = PackARGB;
+  VP8PackRGB = PackRGB;
+}
+
+#else  // !WEBP_USE_MIPS_DSP_R2
+
+WEBP_DSP_INIT_STUB(VP8EncDspARGBInitMIPSdspR2)
+
+#endif  // WEBP_USE_MIPS_DSP_R2

src/dsp/argb_sse2.c

@@ -0,0 +1,53 @@
+// 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.
+// -----------------------------------------------------------------------------
+//
+//   ARGB making functions (SSE2 version).
+//
+// Author: Skal (pascal.massimino@gmail.com)
+
+#include "./dsp.h"
+#include "./lossless.h"
+
+#if defined(WEBP_USE_SSE2)
+
+#include <assert.h>
+#include <emmintrin.h>
+#include <string.h>
+
+static void PackARGB(const uint8_t* a, const uint8_t* r, const uint8_t* g,
+                     const uint8_t* b, int len, uint32_t* out) {
+  (void)a;
+  if (g == r + 1) {  // RGBA input order. Need to swap R and B.
+    assert(b == r + 2);
+    assert(a == r + 3);
+    VP8LConvertBGRAToRGBA((const uint32_t*)r, len, (uint8_t*)out);
+  } else {
+    assert(g == b + 1);
+    assert(r == b + 2);
+    assert(a == b + 3);
+    memcpy(out, b, len * 4);
+  }
+}
+
+//------------------------------------------------------------------------------
+// Entry point
+
+extern void VP8EncDspARGBInitSSE2(void);
+extern void VP8LDspInitSSE2(void);
+
+WEBP_TSAN_IGNORE_FUNCTION void VP8EncDspARGBInitSSE2(void) {
+  VP8LDspInitSSE2();
+  VP8PackARGB = PackARGB;
+}
+
+#else  // !WEBP_USE_SSE2
+
+WEBP_DSP_INIT_STUB(VP8EncDspARGBInitSSE2)
+
+#endif  // WEBP_USE_SSE2

src/dsp/common_sse2.h

@@ -128,9 +128,9 @@ static WEBP_INLINE void VP8Transpose_2_4x4_16b(
 // Pack the planar buffers
 // rrrr... rrrr... gggg... gggg... bbbb... bbbb....
 // triplet by triplet in the output buffer rgb as rgbrgbrgbrgb ...
-static WEBP_INLINE void VP8PlanarTo24b_SSE2(
+static WEBP_INLINE void VP8PlanarTo24b(__m128i* const in0, __m128i* const in1,
-    __m128i* const in0, __m128i* const in1, __m128i* const in2,
+                                       __m128i* const in2, __m128i* const in3,
-    __m128i* const in3, __m128i* const in4, __m128i* const in5) {
+                                       __m128i* const in4, __m128i* const in5) {
   // The input is 6 registers of sixteen 8b but for the sake of explanation,
   // let's take 6 registers of four 8b values.
   // To pack, we will keep taking one every two 8b integer and move it
@@ -159,10 +159,10 @@ static WEBP_INLINE void VP8PlanarTo24b_SSE2(
 
 // Convert four packed four-channel buffers like argbargbargbargb... into the
 // split channels aaaaa ... rrrr ... gggg .... bbbbb ......
-static WEBP_INLINE void VP8L32bToPlanar_SSE2(__m128i* const in0,
+static WEBP_INLINE void VP8L32bToPlanar(__m128i* const in0,
-                                             __m128i* const in1,
+                                        __m128i* const in1,
-                                             __m128i* const in2,
+                                        __m128i* const in2,
-                                             __m128i* const in3) {
+                                        __m128i* const in3) {
   // Column-wise transpose.
   const __m128i A0 = _mm_unpacklo_epi8(*in0, *in1);
   const __m128i A1 = _mm_unpackhi_epi8(*in0, *in1);

src/dsp/common_sse41.h

@@ -1,132 +0,0 @@
-// Copyright 2016 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.
-// -----------------------------------------------------------------------------
-//
-// SSE4 code common to several files.
-//
-// Author: Vincent Rabaud (vrabaud@google.com)
-
-#ifndef WEBP_DSP_COMMON_SSE41_H_
-#define WEBP_DSP_COMMON_SSE41_H_
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-#if defined(WEBP_USE_SSE41)
-#include <smmintrin.h>
-
-//------------------------------------------------------------------------------
-// Channel mixing.
-// Shuffles the input buffer as A0 0 0 A1 0 0 A2 ...
-#define WEBP_SSE41_SHUFF(OUT, IN0, IN1)    \
-  OUT##0 = _mm_shuffle_epi8(*IN0, shuff0); \
-  OUT##1 = _mm_shuffle_epi8(*IN0, shuff1); \
-  OUT##2 = _mm_shuffle_epi8(*IN0, shuff2); \
-  OUT##3 = _mm_shuffle_epi8(*IN1, shuff0); \
-  OUT##4 = _mm_shuffle_epi8(*IN1, shuff1); \
-  OUT##5 = _mm_shuffle_epi8(*IN1, shuff2);
-
-// Pack the planar buffers
-// rrrr... rrrr... gggg... gggg... bbbb... bbbb....
-// triplet by triplet in the output buffer rgb as rgbrgbrgbrgb ...
-static WEBP_INLINE void VP8PlanarTo24b_SSE41(
-    __m128i* const in0, __m128i* const in1, __m128i* const in2,
-    __m128i* const in3, __m128i* const in4, __m128i* const in5) {
-  __m128i R0, R1, R2, R3, R4, R5;
-  __m128i G0, G1, G2, G3, G4, G5;
-  __m128i B0, B1, B2, B3, B4, B5;
-
-  // Process R.
-  {
-    const __m128i shuff0 = _mm_set_epi8(
-        5, -1, -1, 4, -1, -1, 3, -1, -1, 2, -1, -1, 1, -1, -1, 0);
-    const __m128i shuff1 = _mm_set_epi8(
-        -1, 10, -1, -1, 9, -1, -1, 8, -1, -1, 7, -1, -1, 6, -1, -1);
-    const __m128i shuff2 = _mm_set_epi8(
-     -1, -1, 15, -1, -1, 14, -1, -1, 13, -1, -1, 12, -1, -1, 11, -1);
-    WEBP_SSE41_SHUFF(R, in0, in1)
-  }
-
-  // Process G.
-  {
-    // Same as before, just shifted to the left by one and including the right
-    // padding.
-    const __m128i shuff0 = _mm_set_epi8(
-        -1, -1, 4, -1, -1, 3, -1, -1, 2, -1, -1, 1, -1, -1, 0, -1);
-    const __m128i shuff1 = _mm_set_epi8(
-        10, -1, -1, 9, -1, -1, 8, -1, -1, 7, -1, -1, 6, -1, -1, 5);
-    const __m128i shuff2 = _mm_set_epi8(
-     -1, 15, -1, -1, 14, -1, -1, 13, -1, -1, 12, -1, -1, 11, -1, -1);
-    WEBP_SSE41_SHUFF(G, in2, in3)
-  }
-
-  // Process B.
-  {
-    const __m128i shuff0 = _mm_set_epi8(
-        -1, 4, -1, -1, 3, -1, -1, 2, -1, -1, 1, -1, -1, 0, -1, -1);
-    const __m128i shuff1 = _mm_set_epi8(
-        -1, -1, 9, -1, -1, 8, -1, -1, 7, -1, -1, 6, -1, -1, 5, -1);
-    const __m128i shuff2 = _mm_set_epi8(
-      15, -1, -1, 14, -1, -1, 13, -1, -1, 12, -1, -1, 11, -1, -1, 10);
-    WEBP_SSE41_SHUFF(B, in4, in5)
-  }
-
-  // OR the different channels.
-  {
-    const __m128i RG0 = _mm_or_si128(R0, G0);
-    const __m128i RG1 = _mm_or_si128(R1, G1);
-    const __m128i RG2 = _mm_or_si128(R2, G2);
-    const __m128i RG3 = _mm_or_si128(R3, G3);
-    const __m128i RG4 = _mm_or_si128(R4, G4);
-    const __m128i RG5 = _mm_or_si128(R5, G5);
-    *in0 = _mm_or_si128(RG0, B0);
-    *in1 = _mm_or_si128(RG1, B1);
-    *in2 = _mm_or_si128(RG2, B2);
-    *in3 = _mm_or_si128(RG3, B3);
-    *in4 = _mm_or_si128(RG4, B4);
-    *in5 = _mm_or_si128(RG5, B5);
-  }
-}
-
-#undef WEBP_SSE41_SHUFF
-
-// Convert four packed four-channel buffers like argbargbargbargb... into the
-// split channels aaaaa ... rrrr ... gggg .... bbbbb ......
-static WEBP_INLINE void VP8L32bToPlanar_SSE41(__m128i* const in0,
-                                              __m128i* const in1,
-                                              __m128i* const in2,
-                                              __m128i* const in3) {
-  // aaaarrrrggggbbbb
-  const __m128i shuff0 =
-      _mm_set_epi8(15, 11, 7, 3, 14, 10, 6, 2, 13, 9, 5, 1, 12, 8, 4, 0);
-  const __m128i A0 = _mm_shuffle_epi8(*in0, shuff0);
-  const __m128i A1 = _mm_shuffle_epi8(*in1, shuff0);
-  const __m128i A2 = _mm_shuffle_epi8(*in2, shuff0);
-  const __m128i A3 = _mm_shuffle_epi8(*in3, shuff0);
-  // A0A1R0R1
-  // G0G1B0B1
-  // A2A3R2R3
-  // G0G1B0B1
-  const __m128i B0 = _mm_unpacklo_epi32(A0, A1);
-  const __m128i B1 = _mm_unpackhi_epi32(A0, A1);
-  const __m128i B2 = _mm_unpacklo_epi32(A2, A3);
-  const __m128i B3 = _mm_unpackhi_epi32(A2, A3);
-  *in3 = _mm_unpacklo_epi64(B0, B2);
-  *in2 = _mm_unpackhi_epi64(B0, B2);
-  *in1 = _mm_unpacklo_epi64(B1, B3);
-  *in0 = _mm_unpackhi_epi64(B1, B3);
-}
-
-#endif  // WEBP_USE_SSE41
-
-#ifdef __cplusplus
-}  // extern "C"
-#endif
-
-#endif  // WEBP_DSP_COMMON_SSE41_H_

src/dsp/cost.c

@@ -9,8 +9,8 @@
 //
 // Author: Skal (pascal.massimino@gmail.com)
 
-#include "src/dsp/dsp.h"
+#include "./dsp.h"
-#include "src/enc/cost_enc.h"
+#include "../enc/cost_enc.h"
 
 //------------------------------------------------------------------------------
 // Boolean-cost cost table
@@ -319,7 +319,7 @@ const uint8_t VP8EncBands[16 + 1] = {
 //------------------------------------------------------------------------------
 // Mode costs
 
-static int GetResidualCost_C(int ctx0, const VP8Residual* const res) {
+static int GetResidualCost(int ctx0, const VP8Residual* const res) {
   int n = res->first;
   // should be prob[VP8EncBands[n]], but it's equivalent for n=0 or 1
   const int p0 = res->prob[n][ctx0][0];
@@ -354,8 +354,8 @@ static int GetResidualCost_C(int ctx0, const VP8Residual* const res) {
   return cost;
 }
 
-static void SetResidualCoeffs_C(const int16_t* const coeffs,
+static void SetResidualCoeffs(const int16_t* const coeffs,
-                                VP8Residual* const res) {
+                              VP8Residual* const res) {
   int n;
   res->last = -1;
   assert(res->first == 0 || coeffs[0] == 0);
@@ -378,9 +378,14 @@ extern void VP8EncDspCostInitMIPS32(void);
 extern void VP8EncDspCostInitMIPSdspR2(void);
 extern void VP8EncDspCostInitSSE2(void);
 
-WEBP_DSP_INIT_FUNC(VP8EncDspCostInit) {
+static volatile VP8CPUInfo cost_last_cpuinfo_used =
-  VP8GetResidualCost = GetResidualCost_C;
+    (VP8CPUInfo)&cost_last_cpuinfo_used;
-  VP8SetResidualCoeffs = SetResidualCoeffs_C;
+
+WEBP_TSAN_IGNORE_FUNCTION void VP8EncDspCostInit(void) {
+  if (cost_last_cpuinfo_used == VP8GetCPUInfo) return;
+
+  VP8GetResidualCost = GetResidualCost;
+  VP8SetResidualCoeffs = SetResidualCoeffs;
 
   // If defined, use CPUInfo() to overwrite some pointers with faster versions.
   if (VP8GetCPUInfo != NULL) {
@@ -400,6 +405,8 @@ WEBP_DSP_INIT_FUNC(VP8EncDspCostInit) {
     }
 #endif
   }
+
+  cost_last_cpuinfo_used = VP8GetCPUInfo;
 }
 
 //------------------------------------------------------------------------------

src/dsp/cost_mips32.c

@@ -9,13 +9,13 @@
 //
 // Author: Djordje Pesut (djordje.pesut@imgtec.com)
 
-#include "src/dsp/dsp.h"
+#include "./dsp.h"
 
 #if defined(WEBP_USE_MIPS32)
 
-#include "src/enc/cost_enc.h"
+#include "../enc/cost_enc.h"
 
-static int GetResidualCost_MIPS32(int ctx0, const VP8Residual* const res) {
+static int GetResidualCost(int ctx0, const VP8Residual* const res) {
   int temp0, temp1;
   int v_reg, ctx_reg;
   int n = res->first;
@@ -96,8 +96,8 @@ static int GetResidualCost_MIPS32(int ctx0, const VP8Residual* const res) {
   return cost;
 }
 
-static void SetResidualCoeffs_MIPS32(const int16_t* const coeffs,
+static void SetResidualCoeffs(const int16_t* const coeffs,
-                                     VP8Residual* const res) {
+                              VP8Residual* const res) {
   const int16_t* p_coeffs = (int16_t*)coeffs;
   int temp0, temp1, temp2, n, n1;
   assert(res->first == 0 || coeffs[0] == 0);
@@ -143,8 +143,8 @@ static void SetResidualCoeffs_MIPS32(const int16_t* const coeffs,
 extern void VP8EncDspCostInitMIPS32(void);
 
 WEBP_TSAN_IGNORE_FUNCTION void VP8EncDspCostInitMIPS32(void) {
-  VP8GetResidualCost = GetResidualCost_MIPS32;
+  VP8GetResidualCost = GetResidualCost;
-  VP8SetResidualCoeffs = SetResidualCoeffs_MIPS32;
+  VP8SetResidualCoeffs = SetResidualCoeffs;
 }
 
 #else  // !WEBP_USE_MIPS32

src/dsp/cost_mips_dsp_r2.c

@@ -9,13 +9,13 @@
 //
 // Author: Djordje Pesut (djordje.pesut@imgtec.com)
 
-#include "src/dsp/dsp.h"
+#include "./dsp.h"
 
 #if defined(WEBP_USE_MIPS_DSP_R2)
 
-#include "src/enc/cost_enc.h"
+#include "../enc/cost_enc.h"
 
-static int GetResidualCost_MIPSdspR2(int ctx0, const VP8Residual* const res) {
+static int GetResidualCost(int ctx0, const VP8Residual* const res) {
   int temp0, temp1;
   int v_reg, ctx_reg;
   int n = res->first;
@@ -97,7 +97,7 @@ static int GetResidualCost_MIPSdspR2(int ctx0, const VP8Residual* const res) {
 extern void VP8EncDspCostInitMIPSdspR2(void);
 
 WEBP_TSAN_IGNORE_FUNCTION void VP8EncDspCostInitMIPSdspR2(void) {
-  VP8GetResidualCost = GetResidualCost_MIPSdspR2;
+  VP8GetResidualCost = GetResidualCost;
 }
 
 #else  // !WEBP_USE_MIPS_DSP_R2

src/dsp/cost_sse2.c

@@ -11,19 +11,19 @@
 //
 // Author: Skal (pascal.massimino@gmail.com)
 
-#include "src/dsp/dsp.h"
+#include "./dsp.h"
 
 #if defined(WEBP_USE_SSE2)
 #include <emmintrin.h>
 
-#include "src/enc/cost_enc.h"
+#include "../enc/cost_enc.h"
-#include "src/enc/vp8i_enc.h"
+#include "../enc/vp8i_enc.h"
-#include "src/utils/utils.h"
+#include "../utils/utils.h"
 
 //------------------------------------------------------------------------------
 
-static void SetResidualCoeffs_SSE2(const int16_t* const coeffs,
+static void SetResidualCoeffsSSE2(const int16_t* const coeffs,
-                                   VP8Residual* const res) {
+                                  VP8Residual* const res) {
   const __m128i c0 = _mm_loadu_si128((const __m128i*)(coeffs + 0));
   const __m128i c1 = _mm_loadu_si128((const __m128i*)(coeffs + 8));
   // Use SSE2 to compare 16 values with a single instruction.
@@ -42,7 +42,7 @@ static void SetResidualCoeffs_SSE2(const int16_t* const coeffs,
   res->coeffs = coeffs;
 }
 
-static int GetResidualCost_SSE2(int ctx0, const VP8Residual* const res) {
+static int GetResidualCostSSE2(int ctx0, const VP8Residual* const res) {
   uint8_t levels[16], ctxs[16];
   uint16_t abs_levels[16];
   int n = res->first;
@@ -108,8 +108,8 @@ static int GetResidualCost_SSE2(int ctx0, const VP8Residual* const res) {
 extern void VP8EncDspCostInitSSE2(void);
 
 WEBP_TSAN_IGNORE_FUNCTION void VP8EncDspCostInitSSE2(void) {
-  VP8SetResidualCoeffs = SetResidualCoeffs_SSE2;
+  VP8SetResidualCoeffs = SetResidualCoeffsSSE2;
-  VP8GetResidualCost = GetResidualCost_SSE2;
+  VP8GetResidualCost = GetResidualCostSSE2;
 }
 
 #else  // !WEBP_USE_SSE2

src/dsp/cpu.c

@@ -11,7 +11,7 @@
 //
 // Author: Christian Duvivier (cduvivier@google.com)
 
-#include "src/dsp/dsp.h"
+#include "./dsp.h"
 
 #if defined(WEBP_HAVE_NEON_RTCD)
 #include <stdio.h>
@@ -23,11 +23,13 @@
 #endif
 
 //------------------------------------------------------------------------------
-// SSE2 detection.
+// x86/x86-64 micro-arch detection.
 //
 
+// skip x86 specific code for WASM builds
+#if defined(WEBP_USE_WASM)
 // apple/darwin gcc-4.0.1 defines __PIC__, but not __pic__ with -fPIC.
-#if (defined(__pic__) || defined(__PIC__)) && defined(__i386__)
+#elif (defined(__pic__) || defined(__PIC__)) && defined(__i386__)
 static WEBP_INLINE void GetCPUInfo(int cpu_info[4], int info_type) {
   __asm__ volatile (
     "mov %%ebx, %%edi\n"
@@ -63,8 +65,10 @@ static WEBP_INLINE void GetCPUInfo(int cpu_info[4], int info_type) {
 #define GetCPUInfo __cpuid
 #endif
 
+// skip xgetbv definition for WASM builds
+#if defined(WEBP_USE_WASM)
 // NaCl has no support for xgetbv or the raw opcode.
-#if !defined(__native_client__) && (defined(__i386__) || defined(__x86_64__))
+#elif !defined(__native_client__) && (defined(__i386__) || defined(__x86_64__))
 static WEBP_INLINE uint64_t xgetbv(void) {
   const uint32_t ecx = 0;
   uint32_t eax, edx;
@@ -94,7 +98,19 @@ static WEBP_INLINE uint64_t xgetbv(void) {
 #define xgetbv() 0U  // no AVX for older x64 or unrecognized toolchains.
 #endif
 
-#if defined(__i386__) || defined(__x86_64__) || defined(WEBP_MSC_SSE2)
+//------------------------------------------------------------------------------
+// Platform specific VP8CPUInfo functions.
+//
+
+// WASM needs to precede platform specific architecture checks as the defines
+// will still be present when building this target.
+#if defined(WEBP_USE_WASM)
+static int wasmCPUInfo(CPUFeature feature) {
+  if (feature != kWASM) return 0;
+  return 1;
+}
+VP8CPUInfo VP8GetCPUInfo = wasmCPUInfo;
+#elif defined(__i386__) || defined(__x86_64__) || defined(WEBP_MSC_SSE2)
 
 // helper function for run-time detection of slow SSSE3 platforms
 static int CheckSlowModel(int info) {
@@ -143,7 +159,7 @@ static int x86CPUInfo(CPUFeature feature) {
     return !!(cpu_info[2] & (1 << 0));
   }
   if (feature == kSlowSSSE3) {
-    if (is_intel && (cpu_info[2] & (1 << 9))) {   // SSSE3?
+    if (is_intel && (cpu_info[2] & (1 << 0))) {   // SSSE3?
       return CheckSlowModel(cpu_info[0]);
     }
     return 0;

src/dsp/dec.c

@@ -11,11 +11,9 @@
 //
 // Author: Skal (pascal.massimino@gmail.com)
 
-#include <assert.h>
+#include "./dsp.h"
-
+#include "../dec/vp8i_dec.h"
-#include "src/dsp/dsp.h"
+#include "../utils/utils.h"
-#include "src/dec/vp8i_dec.h"
-#include "src/utils/utils.h"
 
 //------------------------------------------------------------------------------
 
@@ -27,7 +25,7 @@ static WEBP_INLINE uint8_t clip_8b(int v) {
 // Transforms (Paragraph 14.4)
 
 #define STORE(x, y, v) \
-  dst[(x) + (y) * BPS] = clip_8b(dst[(x) + (y) * BPS] + ((v) >> 3))
+  dst[x + y * BPS] = clip_8b(dst[x + y * BPS] + ((v) >> 3))
 
 #define STORE2(y, dc, d, c) do {    \
   const int DC = (dc);              \
@@ -40,8 +38,7 @@ static WEBP_INLINE uint8_t clip_8b(int v) {
 #define MUL1(a) ((((a) * 20091) >> 16) + (a))
 #define MUL2(a) (((a) * 35468) >> 16)
 
-#if !WEBP_NEON_OMIT_C_CODE
+static void TransformOne(const int16_t* in, uint8_t* dst) {
-static void TransformOne_C(const int16_t* in, uint8_t* dst) {
   int C[4 * 4], *tmp;
   int i;
   tmp = C;
@@ -81,7 +78,7 @@ static void TransformOne_C(const int16_t* in, uint8_t* dst) {
 }
 
 // Simplified transform when only in[0], in[1] and in[4] are non-zero
-static void TransformAC3_C(const int16_t* in, uint8_t* dst) {
+static void TransformAC3(const int16_t* in, uint8_t* dst) {
   const int a = in[0] + 4;
   const int c4 = MUL2(in[4]);
   const int d4 = MUL1(in[4]);
@@ -96,21 +93,19 @@ static void TransformAC3_C(const int16_t* in, uint8_t* dst) {
 #undef MUL2
 #undef STORE2
 
-static void TransformTwo_C(const int16_t* in, uint8_t* dst, int do_two) {
+static void TransformTwo(const int16_t* in, uint8_t* dst, int do_two) {
-  TransformOne_C(in, dst);
+  TransformOne(in, dst);
   if (do_two) {
-    TransformOne_C(in + 16, dst + 4);
+    TransformOne(in + 16, dst + 4);
   }
 }
-#endif  // !WEBP_NEON_OMIT_C_CODE
 
-static void TransformUV_C(const int16_t* in, uint8_t* dst) {
+static void TransformUV(const int16_t* in, uint8_t* dst) {
   VP8Transform(in + 0 * 16, dst, 1);
   VP8Transform(in + 2 * 16, dst + 4 * BPS, 1);
 }
 
-#if !WEBP_NEON_OMIT_C_CODE
+static void TransformDC(const int16_t* in, uint8_t* dst) {
-static void TransformDC_C(const int16_t* in, uint8_t* dst) {
   const int DC = in[0] + 4;
   int i, j;
   for (j = 0; j < 4; ++j) {
@@ -119,9 +114,8 @@ static void TransformDC_C(const int16_t* in, uint8_t* dst) {
     }
   }
 }
-#endif  // !WEBP_NEON_OMIT_C_CODE
 
-static void TransformDCUV_C(const int16_t* in, uint8_t* dst) {
+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);
@@ -133,8 +127,7 @@ static void TransformDCUV_C(const int16_t* in, uint8_t* dst) {
 //------------------------------------------------------------------------------
 // Paragraph 14.3
 
-#if !WEBP_NEON_OMIT_C_CODE
+static void TransformWHT(const int16_t* in, int16_t* out) {
-static void TransformWHT_C(const int16_t* in, int16_t* out) {
   int tmp[16];
   int i;
   for (i = 0; i < 4; ++i) {
@@ -160,7 +153,6 @@ static void TransformWHT_C(const int16_t* in, int16_t* out) {
     out += 64;
   }
 }
-#endif  // !WEBP_NEON_OMIT_C_CODE
 
 void (*VP8TransformWHT)(const int16_t* in, int16_t* out);
 
@@ -169,7 +161,6 @@ void (*VP8TransformWHT)(const int16_t* in, int16_t* out);
 
 #define DST(x, y) dst[(x) + (y) * BPS]
 
-#if !WEBP_NEON_OMIT_C_CODE
 static WEBP_INLINE void TrueMotion(uint8_t* dst, int size) {
   const uint8_t* top = dst - BPS;
   const uint8_t* const clip0 = VP8kclip1 - top[-1];
@@ -183,21 +174,21 @@ static WEBP_INLINE void TrueMotion(uint8_t* dst, int size) {
     dst += BPS;
   }
 }
-static void TM4_C(uint8_t* dst)   { TrueMotion(dst, 4); }
+static void TM4(uint8_t* dst)   { TrueMotion(dst, 4); }
-static void TM8uv_C(uint8_t* dst) { TrueMotion(dst, 8); }
+static void TM8uv(uint8_t* dst) { TrueMotion(dst, 8); }
-static void TM16_C(uint8_t* dst)  { TrueMotion(dst, 16); }
+static void TM16(uint8_t* dst)  { TrueMotion(dst, 16); }
 
 //------------------------------------------------------------------------------
 // 16x16
 
-static void VE16_C(uint8_t* dst) {     // vertical
+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_C(uint8_t* dst) {     // horizontal
+static void HE16(uint8_t* dst) {     // horizontal
   int j;
   for (j = 16; j > 0; --j) {
     memset(dst, dst[-1], 16);
@@ -212,7 +203,7 @@ static WEBP_INLINE void Put16(int v, uint8_t* dst) {
   }
 }
 
-static void DC16_C(uint8_t* dst) {    // DC
+static void DC16(uint8_t* dst) {    // DC
   int DC = 16;
   int j;
   for (j = 0; j < 16; ++j) {
@@ -221,7 +212,7 @@ static void DC16_C(uint8_t* dst) {    // DC
   Put16(DC >> 5, dst);
 }
 
-static void DC16NoTop_C(uint8_t* dst) {   // DC with top samples not available
+static void DC16NoTop(uint8_t* dst) {   // DC with top samples not available
   int DC = 8;
   int j;
   for (j = 0; j < 16; ++j) {
@@ -230,7 +221,7 @@ static void DC16NoTop_C(uint8_t* dst) {   // DC with top samples not available
   Put16(DC >> 4, dst);
 }
 
-static void DC16NoLeft_C(uint8_t* dst) {  // DC with left samples not available
+static void DC16NoLeft(uint8_t* dst) {  // DC with left samples not available
   int DC = 8;
   int i;
   for (i = 0; i < 16; ++i) {
@@ -239,10 +230,9 @@ static void DC16NoLeft_C(uint8_t* dst) {  // DC with left samples not available
   Put16(DC >> 4, dst);
 }
 
-static void DC16NoTopLeft_C(uint8_t* dst) {  // DC with no top and left samples
+static void DC16NoTopLeft(uint8_t* dst) {  // DC with no top and left samples
   Put16(0x80, dst);
 }
-#endif  // !WEBP_NEON_OMIT_C_CODE
 
 VP8PredFunc VP8PredLuma16[NUM_B_DC_MODES];
 
@@ -252,8 +242,7 @@ VP8PredFunc VP8PredLuma16[NUM_B_DC_MODES];
 #define AVG3(a, b, c) ((uint8_t)(((a) + 2 * (b) + (c) + 2) >> 2))
 #define AVG2(a, b) (((a) + (b) + 1) >> 1)
 
-#if !WEBP_NEON_OMIT_C_CODE
+static void VE4(uint8_t* dst) {    // vertical
-static void VE4_C(uint8_t* dst) {    // vertical
   const uint8_t* top = dst - BPS;
   const uint8_t vals[4] = {
     AVG3(top[-1], top[0], top[1]),
@@ -266,9 +255,8 @@ static void VE4_C(uint8_t* dst) {    // vertical
     memcpy(dst + i * BPS, vals, sizeof(vals));
   }
 }
-#endif  // !WEBP_NEON_OMIT_C_CODE
 
-static void HE4_C(uint8_t* dst) {    // horizontal
+static void HE4(uint8_t* dst) {    // horizontal
   const int A = dst[-1 - BPS];
   const int B = dst[-1];
   const int C = dst[-1 + BPS];
@@ -280,8 +268,7 @@ static void HE4_C(uint8_t* dst) {    // horizontal
   WebPUint32ToMem(dst + 3 * BPS, 0x01010101U * AVG3(D, E, E));
 }
 
-#if !WEBP_NEON_OMIT_C_CODE
+static void DC4(uint8_t* dst) {   // DC
-static void DC4_C(uint8_t* dst) {   // DC
   uint32_t dc = 4;
   int i;
   for (i = 0; i < 4; ++i) dc += dst[i - BPS] + dst[-1 + i * BPS];
@@ -289,7 +276,7 @@ static void DC4_C(uint8_t* dst) {   // DC
   for (i = 0; i < 4; ++i) memset(dst + i * BPS, dc, 4);
 }
 
-static void RD4_C(uint8_t* dst) {   // Down-right
+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];
@@ -308,7 +295,7 @@ static void RD4_C(uint8_t* dst) {   // Down-right
                                       DST(3, 0) = AVG3(D, C, B);
 }
 
-static void LD4_C(uint8_t* dst) {   // Down-Left
+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];
@@ -325,9 +312,8 @@ static void LD4_C(uint8_t* dst) {   // Down-Left
                           DST(3, 2) = DST(2, 3) = AVG3(F, G, H);
                                       DST(3, 3) = AVG3(G, H, H);
 }
-#endif  // !WEBP_NEON_OMIT_C_CODE
 
-static void VR4_C(uint8_t* dst) {   // Vertical-Right
+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];
@@ -349,7 +335,7 @@ static void VR4_C(uint8_t* dst) {   // Vertical-Right
   DST(3, 1) =             AVG3(B, C, D);
 }
 
-static void VL4_C(uint8_t* dst) {   // Vertical-Left
+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];
@@ -371,7 +357,7 @@ static void VL4_C(uint8_t* dst) {   // Vertical-Left
               DST(3, 3) = AVG3(F, G, H);
 }
 
-static void HU4_C(uint8_t* dst) {   // Horizontal-Up
+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];
@@ -386,7 +372,7 @@ static void HU4_C(uint8_t* dst) {   // Horizontal-Up
     DST(0, 3) = DST(1, 3) = DST(2, 3) = DST(3, 3) = L;
 }
 
-static void HD4_C(uint8_t* dst) {  // Horizontal-Down
+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];
@@ -418,15 +404,14 @@ VP8PredFunc VP8PredLuma4[NUM_BMODES];
 //------------------------------------------------------------------------------
 // Chroma
 
-#if !WEBP_NEON_OMIT_C_CODE
+static void VE8uv(uint8_t* dst) {    // vertical
-static void VE8uv_C(uint8_t* dst) {    // vertical
   int j;
   for (j = 0; j < 8; ++j) {
     memcpy(dst + j * BPS, dst - BPS, 8);
   }
 }
 
-static void HE8uv_C(uint8_t* dst) {    // horizontal
+static void HE8uv(uint8_t* dst) {    // horizontal
   int j;
   for (j = 0; j < 8; ++j) {
     memset(dst, dst[-1], 8);
@@ -442,7 +427,7 @@ static WEBP_INLINE void Put8x8uv(uint8_t value, uint8_t* dst) {
   }
 }
 
-static void DC8uv_C(uint8_t* dst) {     // DC
+static void DC8uv(uint8_t* dst) {     // DC
   int dc0 = 8;
   int i;
   for (i = 0; i < 8; ++i) {
@@ -451,7 +436,7 @@ static void DC8uv_C(uint8_t* dst) {     // DC
   Put8x8uv(dc0 >> 4, dst);
 }
 
-static void DC8uvNoLeft_C(uint8_t* dst) {   // DC with no left samples
+static void DC8uvNoLeft(uint8_t* dst) {   // DC with no left samples
   int dc0 = 4;
   int i;
   for (i = 0; i < 8; ++i) {
@@ -460,7 +445,7 @@ static void DC8uvNoLeft_C(uint8_t* dst) {   // DC with no left samples
   Put8x8uv(dc0 >> 3, dst);
 }
 
-static void DC8uvNoTop_C(uint8_t* dst) {  // DC with no top samples
+static void DC8uvNoTop(uint8_t* dst) {  // DC with no top samples
   int dc0 = 4;
   int i;
   for (i = 0; i < 8; ++i) {
@@ -469,19 +454,17 @@ static void DC8uvNoTop_C(uint8_t* dst) {  // DC with no top samples
   Put8x8uv(dc0 >> 3, dst);
 }
 
-static void DC8uvNoTopLeft_C(uint8_t* dst) {    // DC with nothing
+static void DC8uvNoTopLeft(uint8_t* dst) {    // DC with nothing
   Put8x8uv(0x80, dst);
 }
-#endif  // !WEBP_NEON_OMIT_C_CODE
 
 VP8PredFunc VP8PredChroma8[NUM_B_DC_MODES];
 
 //------------------------------------------------------------------------------
 // Edge filtering functions
 
-#if !WEBP_NEON_OMIT_C_CODE || WEBP_NEON_WORK_AROUND_GCC
 // 4 pixels in, 2 pixels out
-static WEBP_INLINE void DoFilter2_C(uint8_t* p, int step) {
+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]
@@ -491,7 +474,7 @@ static WEBP_INLINE void DoFilter2_C(uint8_t* p, int step) {
 }
 
 // 4 pixels in, 4 pixels out
-static WEBP_INLINE void DoFilter4_C(uint8_t* p, int step) {
+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];
@@ -504,7 +487,7 @@ static WEBP_INLINE void DoFilter4_C(uint8_t* p, int step) {
 }
 
 // 6 pixels in, 6 pixels out
-static WEBP_INLINE void DoFilter6_C(uint8_t* p, int step) {
+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]];
@@ -520,22 +503,18 @@ static WEBP_INLINE void DoFilter6_C(uint8_t* p, int step) {
   p[ 2*step] = VP8kclip1[q2 - a3];
 }
 
-static WEBP_INLINE int Hev(const uint8_t* p, int step, int thresh) {
+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);
 }
-#endif  // !WEBP_NEON_OMIT_C_CODE || WEBP_NEON_WORK_AROUND_GCC
 
-#if !WEBP_NEON_OMIT_C_CODE
+static WEBP_INLINE int needs_filter(const uint8_t* p, int step, int t) {
-static WEBP_INLINE int NeedsFilter_C(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);
 }
-#endif  // !WEBP_NEON_OMIT_C_CODE
 
-#if !WEBP_NEON_OMIT_C_CODE || WEBP_NEON_WORK_AROUND_GCC
+static WEBP_INLINE int needs_filter2(const uint8_t* p,
-static WEBP_INLINE int NeedsFilter2_C(const uint8_t* p,
+                                     int step, int t, int it) {
-                                      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];
@@ -544,159 +523,140 @@ static WEBP_INLINE int NeedsFilter2_C(const uint8_t* p,
          VP8kabs0[p1 - p0] <= it && VP8kabs0[q3 - q2] <= it &&
          VP8kabs0[q2 - q1] <= it && VP8kabs0[q1 - q0] <= it;
 }
-#endif  // !WEBP_NEON_OMIT_C_CODE || WEBP_NEON_WORK_AROUND_GCC
 
 //------------------------------------------------------------------------------
 // Simple In-loop filtering (Paragraph 15.2)
 
-#if !WEBP_NEON_OMIT_C_CODE
+static void SimpleVFilter16(uint8_t* p, int stride, int thresh) {
-static void SimpleVFilter16_C(uint8_t* p, int stride, int thresh) {
   int i;
   const int thresh2 = 2 * thresh + 1;
   for (i = 0; i < 16; ++i) {
-    if (NeedsFilter_C(p + i, stride, thresh2)) {
+    if (needs_filter(p + i, stride, thresh2)) {
-      DoFilter2_C(p + i, stride);
+      do_filter2(p + i, stride);
     }
   }
 }
 
-static void SimpleHFilter16_C(uint8_t* p, int stride, int thresh) {
+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 (NeedsFilter_C(p + i * stride, 1, thresh2)) {
+    if (needs_filter(p + i * stride, 1, thresh2)) {
-      DoFilter2_C(p + i * stride, 1);
+      do_filter2(p + i * stride, 1);
     }
   }
 }
 
-static void SimpleVFilter16i_C(uint8_t* p, int stride, int thresh) {
+static void SimpleVFilter16i(uint8_t* p, int stride, int thresh) {
   int k;
   for (k = 3; k > 0; --k) {
     p += 4 * stride;
-    SimpleVFilter16_C(p, stride, thresh);
+    SimpleVFilter16(p, stride, thresh);
   }
 }
 
-static void SimpleHFilter16i_C(uint8_t* p, int stride, int thresh) {
+static void SimpleHFilter16i(uint8_t* p, int stride, int thresh) {
   int k;
   for (k = 3; k > 0; --k) {
     p += 4;
-    SimpleHFilter16_C(p, stride, thresh);
+    SimpleHFilter16(p, stride, thresh);
   }
 }
-#endif  // !WEBP_NEON_OMIT_C_CODE
 
 //------------------------------------------------------------------------------
 // Complex In-loop filtering (Paragraph 15.3)
 
-#if !WEBP_NEON_OMIT_C_CODE || WEBP_NEON_WORK_AROUND_GCC
+static WEBP_INLINE void FilterLoop26(uint8_t* p,
-static WEBP_INLINE void FilterLoop26_C(uint8_t* p,
+                                     int hstride, int vstride, int size,
-                                       int hstride, int vstride, int size,
+                                     int thresh, int ithresh, int hev_thresh) {
-                                       int thresh, int ithresh,
-                                       int hev_thresh) {
   const int thresh2 = 2 * thresh + 1;
   while (size-- > 0) {
-    if (NeedsFilter2_C(p, hstride, thresh2, ithresh)) {
+    if (needs_filter2(p, hstride, thresh2, ithresh)) {
-      if (Hev(p, hstride, hev_thresh)) {
+      if (hev(p, hstride, hev_thresh)) {
-        DoFilter2_C(p, hstride);
+        do_filter2(p, hstride);
       } else {
-        DoFilter6_C(p, hstride);
+        do_filter6(p, hstride);
       }
     }
     p += vstride;
   }
 }
 
-static WEBP_INLINE void FilterLoop24_C(uint8_t* p,
+static WEBP_INLINE void FilterLoop24(uint8_t* p,
-                                       int hstride, int vstride, int size,
+                                     int hstride, int vstride, int size,
-                                       int thresh, int ithresh,
+                                     int thresh, int ithresh, int hev_thresh) {
-                                       int hev_thresh) {
   const int thresh2 = 2 * thresh + 1;
   while (size-- > 0) {
-    if (NeedsFilter2_C(p, hstride, thresh2, ithresh)) {
+    if (needs_filter2(p, hstride, thresh2, ithresh)) {
-      if (Hev(p, hstride, hev_thresh)) {
+      if (hev(p, hstride, hev_thresh)) {
-        DoFilter2_C(p, hstride);
+        do_filter2(p, hstride);
       } else {
-        DoFilter4_C(p, hstride);
+        do_filter4(p, hstride);
       }
     }
     p += vstride;
   }
 }
-#endif  // !WEBP_NEON_OMIT_C_CODE || WEBP_NEON_WORK_AROUND_GCC
 
-#if !WEBP_NEON_OMIT_C_CODE
 // on macroblock edges
-static void VFilter16_C(uint8_t* p, int stride,
+static void VFilter16(uint8_t* p, int stride,
-                        int thresh, int ithresh, int hev_thresh) {
+                      int thresh, int ithresh, int hev_thresh) {
-  FilterLoop26_C(p, stride, 1, 16, thresh, ithresh, hev_thresh);
+  FilterLoop26(p, stride, 1, 16, thresh, ithresh, hev_thresh);
 }
 
-static void HFilter16_C(uint8_t* p, int stride,
+static void HFilter16(uint8_t* p, int stride,
-                        int thresh, int ithresh, int hev_thresh) {
+                      int thresh, int ithresh, int hev_thresh) {
-  FilterLoop26_C(p, 1, stride, 16, thresh, ithresh, hev_thresh);
+  FilterLoop26(p, 1, stride, 16, thresh, ithresh, hev_thresh);
 }
 
 // on three inner edges
-static void VFilter16i_C(uint8_t* p, int stride,
+static void VFilter16i(uint8_t* p, int stride,
-                         int thresh, int ithresh, int hev_thresh) {
+                       int thresh, int ithresh, int hev_thresh) {
   int k;
   for (k = 3; k > 0; --k) {
     p += 4 * stride;
-    FilterLoop24_C(p, stride, 1, 16, thresh, ithresh, hev_thresh);
+    FilterLoop24(p, stride, 1, 16, thresh, ithresh, hev_thresh);
   }
 }
-#endif  // !WEBP_NEON_OMIT_C_CODE
 
-#if !WEBP_NEON_OMIT_C_CODE || WEBP_NEON_WORK_AROUND_GCC
+static void HFilter16i(uint8_t* p, int stride,
-static void HFilter16i_C(uint8_t* p, int stride,
+                       int thresh, int ithresh, int hev_thresh) {
-                         int thresh, int ithresh, int hev_thresh) {
   int k;
   for (k = 3; k > 0; --k) {
     p += 4;
-    FilterLoop24_C(p, 1, stride, 16, thresh, ithresh, hev_thresh);
+    FilterLoop24(p, 1, stride, 16, thresh, ithresh, hev_thresh);
   }
 }
-#endif  // !WEBP_NEON_OMIT_C_CODE || WEBP_NEON_WORK_AROUND_GCC
 
-#if !WEBP_NEON_OMIT_C_CODE
 // 8-pixels wide variant, for chroma filtering
-static void VFilter8_C(uint8_t* u, uint8_t* v, int stride,
+static void VFilter8(uint8_t* u, uint8_t* v, int stride,
-                       int thresh, int ithresh, int hev_thresh) {
+                     int thresh, int ithresh, int hev_thresh) {
-  FilterLoop26_C(u, stride, 1, 8, thresh, ithresh, hev_thresh);
+  FilterLoop26(u, stride, 1, 8, thresh, ithresh, hev_thresh);
-  FilterLoop26_C(v, stride, 1, 8, thresh, ithresh, hev_thresh);
+  FilterLoop26(v, stride, 1, 8, thresh, ithresh, hev_thresh);
 }
-#endif  // !WEBP_NEON_OMIT_C_CODE
 
-#if !WEBP_NEON_OMIT_C_CODE || WEBP_NEON_WORK_AROUND_GCC
+static void HFilter8(uint8_t* u, uint8_t* v, int stride,
-static void HFilter8_C(uint8_t* u, uint8_t* v, int stride,
+                     int thresh, int ithresh, int hev_thresh) {
-                       int thresh, int ithresh, int hev_thresh) {
+  FilterLoop26(u, 1, stride, 8, thresh, ithresh, hev_thresh);
-  FilterLoop26_C(u, 1, stride, 8, thresh, ithresh, hev_thresh);
+  FilterLoop26(v, 1, stride, 8, thresh, ithresh, hev_thresh);
-  FilterLoop26_C(v, 1, stride, 8, thresh, ithresh, hev_thresh);
 }
-#endif  // !WEBP_NEON_OMIT_C_CODE || WEBP_NEON_WORK_AROUND_GCC
 
-#if !WEBP_NEON_OMIT_C_CODE
+static void VFilter8i(uint8_t* u, uint8_t* v, int stride,
-static void VFilter8i_C(uint8_t* u, uint8_t* v, int stride,
+                      int thresh, int ithresh, int hev_thresh) {
-                        int thresh, int ithresh, int hev_thresh) {
+  FilterLoop24(u + 4 * stride, stride, 1, 8, thresh, ithresh, hev_thresh);
-  FilterLoop24_C(u + 4 * stride, stride, 1, 8, thresh, ithresh, hev_thresh);
+  FilterLoop24(v + 4 * stride, stride, 1, 8, thresh, ithresh, hev_thresh);
-  FilterLoop24_C(v + 4 * stride, stride, 1, 8, thresh, ithresh, hev_thresh);
 }
-#endif  // !WEBP_NEON_OMIT_C_CODE
 
-#if !WEBP_NEON_OMIT_C_CODE || WEBP_NEON_WORK_AROUND_GCC
+static void HFilter8i(uint8_t* u, uint8_t* v, int stride,
-static void HFilter8i_C(uint8_t* u, uint8_t* v, int stride,
+                      int thresh, int ithresh, int hev_thresh) {
-                        int thresh, int ithresh, int hev_thresh) {
+  FilterLoop24(u + 4, 1, stride, 8, thresh, ithresh, hev_thresh);
-  FilterLoop24_C(u + 4, 1, stride, 8, thresh, ithresh, hev_thresh);
+  FilterLoop24(v + 4, 1, stride, 8, thresh, ithresh, hev_thresh);
-  FilterLoop24_C(v + 4, 1, stride, 8, thresh, ithresh, hev_thresh);
 }
-#endif  // !WEBP_NEON_OMIT_C_CODE || WEBP_NEON_WORK_AROUND_GCC
 
 //------------------------------------------------------------------------------
 
-static void DitherCombine8x8_C(const uint8_t* dither, uint8_t* dst,
+static void DitherCombine8x8(const uint8_t* dither, uint8_t* dst,
-                               int dst_stride) {
+                             int dst_stride) {
   int i, j;
   for (j = 0; j < 8; ++j) {
     for (i = 0; i < 8; ++i) {
@@ -740,70 +700,64 @@ extern void VP8DspInitNEON(void);
 extern void VP8DspInitMIPS32(void);
 extern void VP8DspInitMIPSdspR2(void);
 extern void VP8DspInitMSA(void);
+extern void VP8DspInitWASM(void);
+
+static volatile VP8CPUInfo dec_last_cpuinfo_used =
+    (VP8CPUInfo)&dec_last_cpuinfo_used;
+
+WEBP_TSAN_IGNORE_FUNCTION void VP8DspInit(void) {
+  if (dec_last_cpuinfo_used == VP8GetCPUInfo) return;
 
-WEBP_DSP_INIT_FUNC(VP8DspInit) {
   VP8InitClipTables();
 
-#if !WEBP_NEON_OMIT_C_CODE
+  VP8TransformWHT = TransformWHT;
-  VP8TransformWHT = TransformWHT_C;
+  VP8Transform = TransformTwo;
-  VP8Transform = TransformTwo_C;
+  VP8TransformUV = TransformUV;
-  VP8TransformDC = TransformDC_C;
+  VP8TransformDC = TransformDC;
-  VP8TransformAC3 = TransformAC3_C;
+  VP8TransformDCUV = TransformDCUV;
-#endif
+  VP8TransformAC3 = TransformAC3;
-  VP8TransformUV = TransformUV_C;
-  VP8TransformDCUV = TransformDCUV_C;
 
-#if !WEBP_NEON_OMIT_C_CODE
+  VP8VFilter16 = VFilter16;
-  VP8VFilter16 = VFilter16_C;
+  VP8HFilter16 = HFilter16;
-  VP8VFilter16i = VFilter16i_C;
+  VP8VFilter8 = VFilter8;
-  VP8HFilter16 = HFilter16_C;
+  VP8HFilter8 = HFilter8;
-  VP8VFilter8 = VFilter8_C;
+  VP8VFilter16i = VFilter16i;
-  VP8VFilter8i = VFilter8i_C;
+  VP8HFilter16i = HFilter16i;
-  VP8SimpleVFilter16 = SimpleVFilter16_C;
+  VP8VFilter8i = VFilter8i;
-  VP8SimpleHFilter16 = SimpleHFilter16_C;
+  VP8HFilter8i = HFilter8i;
-  VP8SimpleVFilter16i = SimpleVFilter16i_C;
+  VP8SimpleVFilter16 = SimpleVFilter16;
-  VP8SimpleHFilter16i = SimpleHFilter16i_C;
+  VP8SimpleHFilter16 = SimpleHFilter16;
-#endif
+  VP8SimpleVFilter16i = SimpleVFilter16i;
+  VP8SimpleHFilter16i = SimpleHFilter16i;
 
-#if !WEBP_NEON_OMIT_C_CODE || WEBP_NEON_WORK_AROUND_GCC
+  VP8PredLuma4[0] = DC4;
-  VP8HFilter16i = HFilter16i_C;
+  VP8PredLuma4[1] = TM4;
-  VP8HFilter8 = HFilter8_C;
+  VP8PredLuma4[2] = VE4;
-  VP8HFilter8i = HFilter8i_C;
+  VP8PredLuma4[3] = HE4;
-#endif
+  VP8PredLuma4[4] = RD4;
+  VP8PredLuma4[5] = VR4;
+  VP8PredLuma4[6] = LD4;
+  VP8PredLuma4[7] = VL4;
+  VP8PredLuma4[8] = HD4;
+  VP8PredLuma4[9] = HU4;
 
-#if !WEBP_NEON_OMIT_C_CODE
+  VP8PredLuma16[0] = DC16;
-  VP8PredLuma4[0] = DC4_C;
+  VP8PredLuma16[1] = TM16;
-  VP8PredLuma4[1] = TM4_C;
+  VP8PredLuma16[2] = VE16;
-  VP8PredLuma4[2] = VE4_C;
+  VP8PredLuma16[3] = HE16;
-  VP8PredLuma4[4] = RD4_C;
+  VP8PredLuma16[4] = DC16NoTop;
-  VP8PredLuma4[6] = LD4_C;
+  VP8PredLuma16[5] = DC16NoLeft;
-#endif
+  VP8PredLuma16[6] = DC16NoTopLeft;
 
-  VP8PredLuma4[3] = HE4_C;
+  VP8PredChroma8[0] = DC8uv;
-  VP8PredLuma4[5] = VR4_C;
+  VP8PredChroma8[1] = TM8uv;
-  VP8PredLuma4[7] = VL4_C;
+  VP8PredChroma8[2] = VE8uv;
-  VP8PredLuma4[8] = HD4_C;
+  VP8PredChroma8[3] = HE8uv;
-  VP8PredLuma4[9] = HU4_C;
+  VP8PredChroma8[4] = DC8uvNoTop;
+  VP8PredChroma8[5] = DC8uvNoLeft;
+  VP8PredChroma8[6] = DC8uvNoTopLeft;
 
-#if !WEBP_NEON_OMIT_C_CODE
+  VP8DitherCombine8x8 = DitherCombine8x8;
-  VP8PredLuma16[0] = DC16_C;
-  VP8PredLuma16[1] = TM16_C;
-  VP8PredLuma16[2] = VE16_C;
-  VP8PredLuma16[3] = HE16_C;
-  VP8PredLuma16[4] = DC16NoTop_C;
-  VP8PredLuma16[5] = DC16NoLeft_C;
-  VP8PredLuma16[6] = DC16NoTopLeft_C;
-
-  VP8PredChroma8[0] = DC8uv_C;
-  VP8PredChroma8[1] = TM8uv_C;
-  VP8PredChroma8[2] = VE8uv_C;
-  VP8PredChroma8[3] = HE8uv_C;
-  VP8PredChroma8[4] = DC8uvNoTop_C;
-  VP8PredChroma8[5] = DC8uvNoLeft_C;
-  VP8PredChroma8[6] = DC8uvNoTopLeft_C;
-#endif
-
-  VP8DitherCombine8x8 = DitherCombine8x8_C;
 
   // If defined, use CPUInfo() to overwrite some pointers with faster versions.
   if (VP8GetCPUInfo != NULL) {
@@ -817,6 +771,11 @@ WEBP_DSP_INIT_FUNC(VP8DspInit) {
 #endif
     }
 #endif
+#if defined(WEBP_USE_NEON)
+    if (VP8GetCPUInfo(kNEON)) {
+      VP8DspInitNEON();
+    }
+#endif
 #if defined(WEBP_USE_MIPS32)
     if (VP8GetCPUInfo(kMIPS32)) {
       VP8DspInitMIPS32();
@@ -832,56 +791,11 @@ WEBP_DSP_INIT_FUNC(VP8DspInit) {
       VP8DspInitMSA();
     }
 #endif
-  }
+#if defined(WEBP_USE_WASM)
-
+    if (VP8GetCPUInfo(kWASM)) {
-#if defined(WEBP_USE_NEON)
+      VP8DspInitWASM();
-  if (WEBP_NEON_OMIT_C_CODE ||
+    }
-      (VP8GetCPUInfo != NULL && VP8GetCPUInfo(kNEON))) {
-    VP8DspInitNEON();
-  }
 #endif
-
+  }
-  assert(VP8TransformWHT != NULL);
+  dec_last_cpuinfo_used = VP8GetCPUInfo;
-  assert(VP8Transform != NULL);
-  assert(VP8TransformDC != NULL);
-  assert(VP8TransformAC3 != NULL);
-  assert(VP8TransformUV != NULL);
-  assert(VP8TransformDCUV != NULL);
-  assert(VP8VFilter16 != NULL);
-  assert(VP8HFilter16 != NULL);
-  assert(VP8VFilter8 != NULL);
-  assert(VP8HFilter8 != NULL);
-  assert(VP8VFilter16i != NULL);
-  assert(VP8HFilter16i != NULL);
-  assert(VP8VFilter8i != NULL);
-  assert(VP8HFilter8i != NULL);
-  assert(VP8SimpleVFilter16 != NULL);
-  assert(VP8SimpleHFilter16 != NULL);
-  assert(VP8SimpleVFilter16i != NULL);
-  assert(VP8SimpleHFilter16i != NULL);
-  assert(VP8PredLuma4[0] != NULL);
-  assert(VP8PredLuma4[1] != NULL);
-  assert(VP8PredLuma4[2] != NULL);
-  assert(VP8PredLuma4[3] != NULL);
-  assert(VP8PredLuma4[4] != NULL);
-  assert(VP8PredLuma4[5] != NULL);
-  assert(VP8PredLuma4[6] != NULL);
-  assert(VP8PredLuma4[7] != NULL);
-  assert(VP8PredLuma4[8] != NULL);
-  assert(VP8PredLuma4[9] != NULL);
-  assert(VP8PredLuma16[0] != NULL);
-  assert(VP8PredLuma16[1] != NULL);
-  assert(VP8PredLuma16[2] != NULL);
-  assert(VP8PredLuma16[3] != NULL);
-  assert(VP8PredLuma16[4] != NULL);
-  assert(VP8PredLuma16[5] != NULL);
-  assert(VP8PredLuma16[6] != NULL);
-  assert(VP8PredChroma8[0] != NULL);
-  assert(VP8PredChroma8[1] != NULL);
-  assert(VP8PredChroma8[2] != NULL);
-  assert(VP8PredChroma8[3] != NULL);
-  assert(VP8PredChroma8[4] != NULL);
-  assert(VP8PredChroma8[5] != NULL);
-  assert(VP8PredChroma8[6] != NULL);
-  assert(VP8DitherCombine8x8 != NULL);
 }

src/dsp/dec_clip_tables.c

@@ -11,14 +11,11 @@
 //
 // Author: Skal (pascal.massimino@gmail.com)
 
-#include "src/dsp/dsp.h"
+#include "./dsp.h"
 
-// define to 0 to have run-time table initialization
+#define USE_STATIC_TABLES     // undefine to have run-time table initialization
-#if !defined(USE_STATIC_TABLES)
-#define USE_STATIC_TABLES 1   // ALTERNATE_CODE
-#endif
 
-#if (USE_STATIC_TABLES == 1)
+#ifdef USE_STATIC_TABLES
 
 static const uint8_t abs0[255 + 255 + 1] = {
   0xff, 0xfe, 0xfd, 0xfc, 0xfb, 0xfa, 0xf9, 0xf8, 0xf7, 0xf6, 0xf5, 0xf4,
@@ -340,7 +337,7 @@ static uint8_t clip1[255 + 511 + 1];
 // and make sure it's set to true _last_ (so as to be thread-safe)
 static volatile int tables_ok = 0;
 
-#endif    // USE_STATIC_TABLES
+#endif
 
 const int8_t* const VP8ksclip1 = (const int8_t*)&sclip1[1020];
 const int8_t* const VP8ksclip2 = (const int8_t*)&sclip2[112];
@@ -348,7 +345,7 @@ const uint8_t* const VP8kclip1 = &clip1[255];
 const uint8_t* const VP8kabs0 = &abs0[255];
 
 WEBP_TSAN_IGNORE_FUNCTION void VP8InitClipTables(void) {
-#if (USE_STATIC_TABLES == 0)
+#if !defined(USE_STATIC_TABLES)
   int i;
   if (!tables_ok) {
     for (i = -255; i <= 255; ++i) {

src/dsp/dec_mips32.c

@@ -12,11 +12,11 @@
 // Author(s):  Djordje Pesut    (djordje.pesut@imgtec.com)
 //             Jovan Zelincevic (jovan.zelincevic@imgtec.com)
 
-#include "src/dsp/dsp.h"
+#include "./dsp.h"
 
 #if defined(WEBP_USE_MIPS32)
 
-#include "src/dsp/mips_macro.h"
+#include "./mips_macro.h"
 
 static const int kC1 = 20091 + (1 << 16);
 static const int kC2 = 35468;

src/dsp/dec_mips_dsp_r2.c

@@ -12,11 +12,11 @@
 // Author(s):  Djordje Pesut    (djordje.pesut@imgtec.com)
 //             Jovan Zelincevic (jovan.zelincevic@imgtec.com)
 
-#include "src/dsp/dsp.h"
+#include "./dsp.h"
 
 #if defined(WEBP_USE_MIPS_DSP_R2)
 
-#include "src/dsp/mips_macro.h"
+#include "./mips_macro.h"
 
 static const int kC1 = 20091 + (1 << 16);
 static const int kC2 = 35468;

src/dsp/dec_msa.c

@@ -12,11 +12,11 @@
 // Author(s):  Prashant Patil   (prashant.patil@imgtec.com)
 
 
-#include "src/dsp/dsp.h"
+#include "./dsp.h"
 
 #if defined(WEBP_USE_MSA)
 
-#include "src/dsp/msa_macro.h"
+#include "./msa_macro.h"
 
 //------------------------------------------------------------------------------
 // Transforms

src/dsp/dec_sse2.c

@@ -12,25 +12,23 @@
 // Author: somnath@google.com (Somnath Banerjee)
 //         cduvivier@google.com (Christian Duvivier)
 
-#include "src/dsp/dsp.h"
+#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:
-#if !defined(USE_TRANSFORM_AC3)
+// #define USE_TRANSFORM_AC3
-#define USE_TRANSFORM_AC3 0   // ALTERNATE_CODE
-#endif
 
 #include <emmintrin.h>
-#include "src/dsp/common_sse2.h"
+#include "./common_sse2.h"
-#include "src/dec/vp8i_dec.h"
+#include "../dec/vp8i_dec.h"
-#include "src/utils/utils.h"
+#include "../utils/utils.h"
 
 //------------------------------------------------------------------------------
 // Transforms (Paragraph 14.4)
 
-static void Transform_SSE2(const int16_t* in, uint8_t* dst, int do_two) {
+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
@@ -195,7 +193,7 @@ static void Transform_SSE2(const int16_t* in, uint8_t* dst, int do_two) {
   }
 }
 
-#if (USE_TRANSFORM_AC3 == 1)
+#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);
@@ -250,7 +248,7 @@ static void TransformAC3(const int16_t* in, uint8_t* dst) {
     _mm_subs_epu8((p), (q)))
 
 // Shift each byte of "x" by 3 bits while preserving by the sign bit.
-static WEBP_INLINE void SignedShift8b_SSE2(__m128i* const x) {
+static WEBP_INLINE void SignedShift8b(__m128i* const x) {
   const __m128i zero = _mm_setzero_si128();
   const __m128i lo_0 = _mm_unpacklo_epi8(zero, *x);
   const __m128i hi_0 = _mm_unpackhi_epi8(zero, *x);
@@ -260,8 +258,8 @@ static WEBP_INLINE void SignedShift8b_SSE2(__m128i* const x) {
 }
 
 #define FLIP_SIGN_BIT2(a, b) {                                                 \
-  (a) = _mm_xor_si128(a, sign_bit);                                            \
+  a = _mm_xor_si128(a, sign_bit);                                              \
-  (b) = _mm_xor_si128(b, sign_bit);                                            \
+  b = _mm_xor_si128(b, sign_bit);                                              \
 }
 
 #define FLIP_SIGN_BIT4(a, b, c, d) {                                           \
@@ -270,11 +268,11 @@ static WEBP_INLINE void SignedShift8b_SSE2(__m128i* const x) {
 }
 
 // input/output is uint8_t
-static WEBP_INLINE void GetNotHEV_SSE2(const __m128i* const p1,
+static WEBP_INLINE void GetNotHEV(const __m128i* const p1,
-                                       const __m128i* const p0,
+                                  const __m128i* const p0,
-                                       const __m128i* const q0,
+                                  const __m128i* const q0,
-                                       const __m128i* const q1,
+                                  const __m128i* const q1,
-                                       int hev_thresh, __m128i* const not_hev) {
+                                  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);
@@ -287,11 +285,11 @@ static WEBP_INLINE void GetNotHEV_SSE2(const __m128i* const p1,
 }
 
 // input pixels are int8_t
-static WEBP_INLINE void GetBaseDelta_SSE2(const __m128i* const p1,
+static WEBP_INLINE void GetBaseDelta(const __m128i* const p1,
-                                          const __m128i* const p0,
+                                     const __m128i* const p0,
-                                          const __m128i* const q0,
+                                     const __m128i* const q0,
-                                          const __m128i* const q1,
+                                     const __m128i* const q1,
-                                          __m128i* const delta) {
+                                     __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
@@ -302,16 +300,15 @@ static WEBP_INLINE void GetBaseDelta_SSE2(const __m128i* const p1,
 }
 
 // input and output are int8_t
-static WEBP_INLINE void DoSimpleFilter_SSE2(__m128i* const p0,
+static WEBP_INLINE void DoSimpleFilter(__m128i* const p0, __m128i* const q0,
-                                            __m128i* const q0,
+                                       const __m128i* const fl) {
-                                            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_SSE2(&v4);             // v4 >> 3
+  SignedShift8b(&v4);                  // v4 >> 3
-  SignedShift8b_SSE2(&v3);             // v3 >> 3
+  SignedShift8b(&v3);                  // v3 >> 3
   *q0 = _mm_subs_epi8(*q0, v4);        // q0 -= v4
   *p0 = _mm_adds_epi8(*p0, v3);        // p0 += v3
 }
@@ -320,9 +317,9 @@ static WEBP_INLINE void DoSimpleFilter_SSE2(__m128i* const p0,
 // 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_SSE2(__m128i* const pi, __m128i* const qi,
+static WEBP_INLINE void Update2Pixels(__m128i* const pi, __m128i* const qi,
-                                           const __m128i* const a0_lo,
+                                      const __m128i* const a0_lo,
-                                           const __m128i* const a0_hi) {
+                                      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);
@@ -333,11 +330,11 @@ static WEBP_INLINE void Update2Pixels_SSE2(__m128i* const pi, __m128i* const qi,
 }
 
 // input pixels are uint8_t
-static WEBP_INLINE void NeedsFilter_SSE2(const __m128i* const p1,
+static WEBP_INLINE void NeedsFilter(const __m128i* const p1,
-                                         const __m128i* const p0,
+                                    const __m128i* const p0,
-                                         const __m128i* const q0,
+                                    const __m128i* const q0,
-                                         const __m128i* const q1,
+                                    const __m128i* const q1,
-                                         int thresh, __m128i* const mask) {
+                                    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);
@@ -356,29 +353,28 @@ static WEBP_INLINE void NeedsFilter_SSE2(const __m128i* const p1,
 // Edge filtering functions
 
 // Applies filter on 2 pixels (p0 and q0)
-static WEBP_INLINE void DoFilter2_SSE2(__m128i* const p1, __m128i* const p0,
+static WEBP_INLINE void DoFilter2(__m128i* const p1, __m128i* const p0,
-                                       __m128i* const q0, __m128i* const q1,
+                                  __m128i* const q0, __m128i* const q1,
-                                       int thresh) {
+                                  int thresh) {
   __m128i a, mask;
   const __m128i sign_bit = _mm_set1_epi8(0x80);
-  // convert p1/q1 to int8_t (for GetBaseDelta_SSE2)
+  // 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_SSE2(p1, p0, q0, q1, thresh, &mask);
+  NeedsFilter(p1, p0, q0, q1, thresh, &mask);
 
   FLIP_SIGN_BIT2(*p0, *q0);
-  GetBaseDelta_SSE2(&p1s, p0, q0, &q1s, &a);
+  GetBaseDelta(&p1s, p0, q0, &q1s, &a);
   a = _mm_and_si128(a, mask);     // mask filter values we don't care about
-  DoSimpleFilter_SSE2(p0, q0, &a);
+  DoSimpleFilter(p0, q0, &a);
   FLIP_SIGN_BIT2(*p0, *q0);
 }
 
 // Applies filter on 4 pixels (p1, p0, q0 and q1)
-static WEBP_INLINE void DoFilter4_SSE2(__m128i* const p1, __m128i* const p0,
+static WEBP_INLINE void DoFilter4(__m128i* const p1, __m128i* const p0,
-                                       __m128i* const q0, __m128i* const q1,
+                                  __m128i* const q0, __m128i* const q1,
-                                       const __m128i* const mask,
+                                  const __m128i* const mask, int hev_thresh) {
-                                       int hev_thresh) {
   const __m128i zero = _mm_setzero_si128();
   const __m128i sign_bit = _mm_set1_epi8(0x80);
   const __m128i k64 = _mm_set1_epi8(64);
@@ -388,7 +384,7 @@ static WEBP_INLINE void DoFilter4_SSE2(__m128i* const p1, __m128i* const p0,
   __m128i t1, t2, t3;
 
   // compute hev mask
-  GetNotHEV_SSE2(p1, p0, q0, q1, hev_thresh, &not_hev);
+  GetNotHEV(p1, p0, q0, q1, hev_thresh, &not_hev);
 
   // convert to signed values
   FLIP_SIGN_BIT4(*p1, *p0, *q0, *q1);
@@ -403,8 +399,8 @@ static WEBP_INLINE void DoFilter4_SSE2(__m128i* const p1, __m128i* const p0,
 
   t2 = _mm_adds_epi8(t1, k3);        // 3 * (q0 - p0) + hev(p1 - q1) + 3
   t3 = _mm_adds_epi8(t1, k4);        // 3 * (q0 - p0) + hev(p1 - q1) + 4
-  SignedShift8b_SSE2(&t2);           // (3 * (q0 - p0) + hev(p1 - q1) + 3) >> 3
+  SignedShift8b(&t2);                // (3 * (q0 - p0) + hev(p1 - q1) + 3) >> 3
-  SignedShift8b_SSE2(&t3);           // (3 * (q0 - p0) + hev(p1 - q1) + 4) >> 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);
@@ -421,26 +417,25 @@ static WEBP_INLINE void DoFilter4_SSE2(__m128i* const p1, __m128i* const p0,
 }
 
 // Applies filter on 6 pixels (p2, p1, p0, q0, q1 and q2)
-static WEBP_INLINE void DoFilter6_SSE2(__m128i* const p2, __m128i* const p1,
+static WEBP_INLINE void DoFilter6(__m128i* const p2, __m128i* const p1,
-                                       __m128i* const p0, __m128i* const q0,
+                                  __m128i* const p0, __m128i* const q0,
-                                       __m128i* const q1, __m128i* const q2,
+                                  __m128i* const q1, __m128i* const q2,
-                                       const __m128i* const mask,
+                                  const __m128i* const mask, int hev_thresh) {
-                                       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_SSE2(p1, p0, q0, q1, hev_thresh, &not_hev);
+  GetNotHEV(p1, p0, q0, q1, hev_thresh, &not_hev);
 
   FLIP_SIGN_BIT4(*p1, *p0, *q0, *q1);
   FLIP_SIGN_BIT2(*p2, *q2);
-  GetBaseDelta_SSE2(p1, p0, q0, q1, &a);
+  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_SSE2(p0, q0, &f);
+    DoSimpleFilter(p0, q0, &f);
   }
 
   { // do strong filter on pixels with not hev
@@ -465,15 +460,15 @@ static WEBP_INLINE void DoFilter6_SSE2(__m128i* const p2, __m128i* const p1,
     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_SSE2(p2, q2, &a2_lo, &a2_hi);
+    Update2Pixels(p2, q2, &a2_lo, &a2_hi);
-    Update2Pixels_SSE2(p1, q1, &a1_lo, &a1_hi);
+    Update2Pixels(p1, q1, &a1_lo, &a1_hi);
-    Update2Pixels_SSE2(p0, q0, &a0_lo, &a0_hi);
+    Update2Pixels(p0, q0, &a0_lo, &a0_hi);
   }
 }
 
 // reads 8 rows across a vertical edge.
-static WEBP_INLINE void Load8x4_SSE2(const uint8_t* const b, int stride,
+static WEBP_INLINE void Load8x4(const uint8_t* const b, int stride,
-                                     __m128i* const p, __m128i* const q) {
+                                __m128i* const p, __m128i* const q) {
   // A0 = 63 62 61 60 23 22 21 20 43 42 41 40 03 02 01 00
   // A1 = 73 72 71 70 33 32 31 30 53 52 51 50 13 12 11 10
   const __m128i A0 = _mm_set_epi32(
@@ -499,11 +494,11 @@ static WEBP_INLINE void Load8x4_SSE2(const uint8_t* const b, int stride,
   *q = _mm_unpackhi_epi32(C0, C1);
 }
 
-static WEBP_INLINE void Load16x4_SSE2(const uint8_t* const r0,
+static WEBP_INLINE void Load16x4(const uint8_t* const r0,
-                                      const uint8_t* const r8,
+                                 const uint8_t* const r8,
-                                      int stride,
+                                 int stride,
-                                      __m128i* const p1, __m128i* const p0,
+                                 __m128i* const p1, __m128i* const p0,
-                                      __m128i* const q0, __m128i* const q1) {
+                                 __m128i* const q0, __m128i* const q1) {
   // Assume the pixels around the edge (|) are numbered as follows
   //                00 01 | 02 03
   //                10 11 | 12 13
@@ -519,8 +514,8 @@ static WEBP_INLINE void Load16x4_SSE2(const uint8_t* const r0,
   // 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_SSE2(r0, stride, p1, q0);
+  Load8x4(r0, stride, p1, q0);
-  Load8x4_SSE2(r8, stride, p0, q1);
+  Load8x4(r8, stride, p0, q1);
 
   {
     // p1 = f0 e0 d0 c0 b0 a0 90 80 70 60 50 40 30 20 10 00
@@ -536,8 +531,7 @@ static WEBP_INLINE void Load16x4_SSE2(const uint8_t* const r0,
   }
 }
 
-static WEBP_INLINE void Store4x4_SSE2(__m128i* const x,
+static WEBP_INLINE void Store4x4(__m128i* const x, uint8_t* dst, int stride) {
-                                      uint8_t* dst, int stride) {
   int i;
   for (i = 0; i < 4; ++i, dst += stride) {
     WebPUint32ToMem(dst, _mm_cvtsi128_si32(*x));
@@ -546,12 +540,12 @@ static WEBP_INLINE void Store4x4_SSE2(__m128i* const x,
 }
 
 // Transpose back and store
-static WEBP_INLINE void Store16x4_SSE2(const __m128i* const p1,
+static WEBP_INLINE void Store16x4(const __m128i* const p1,
-                                       const __m128i* const p0,
+                                  const __m128i* const p0,
-                                       const __m128i* const q0,
+                                  const __m128i* const q0,
-                                       const __m128i* const q1,
+                                  const __m128i* const q1,
-                                       uint8_t* r0, uint8_t* r8,
+                                  uint8_t* r0, uint8_t* r8,
-                                       int stride) {
+                                  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
@@ -578,55 +572,55 @@ static WEBP_INLINE void Store16x4_SSE2(const __m128i* const p1,
   p1_s = _mm_unpacklo_epi16(t1, q1_s);
   q1_s = _mm_unpackhi_epi16(t1, q1_s);
 
-  Store4x4_SSE2(&p0_s, r0, stride);
+  Store4x4(&p0_s, r0, stride);
   r0 += 4 * stride;
-  Store4x4_SSE2(&q0_s, r0, stride);
+  Store4x4(&q0_s, r0, stride);
 
-  Store4x4_SSE2(&p1_s, r8, stride);
+  Store4x4(&p1_s, r8, stride);
   r8 += 4 * stride;
-  Store4x4_SSE2(&q1_s, r8, stride);
+  Store4x4(&q1_s, r8, stride);
 }
 
 //------------------------------------------------------------------------------
 // Simple In-loop filtering (Paragraph 15.2)
 
-static void SimpleVFilter16_SSE2(uint8_t* p, int stride, int thresh) {
+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_SSE2(&p1, &p0, &q0, &q1, thresh);
+  DoFilter2(&p1, &p0, &q0, &q1, thresh);
 
   // Store
   _mm_storeu_si128((__m128i*)&p[-stride], p0);
   _mm_storeu_si128((__m128i*)&p[0], q0);
 }
 
-static void SimpleHFilter16_SSE2(uint8_t* p, int stride, int thresh) {
+static void SimpleHFilter16(uint8_t* p, int stride, int thresh) {
   __m128i p1, p0, q0, q1;
 
   p -= 2;  // beginning of p1
 
-  Load16x4_SSE2(p, p + 8 * stride, stride, &p1, &p0, &q0, &q1);
+  Load16x4(p, p + 8 * stride, stride, &p1, &p0, &q0, &q1);
-  DoFilter2_SSE2(&p1, &p0, &q0, &q1, thresh);
+  DoFilter2(&p1, &p0, &q0, &q1, thresh);
-  Store16x4_SSE2(&p1, &p0, &q0, &q1, p, p + 8 * stride, stride);
+  Store16x4(&p1, &p0, &q0, &q1, p, p + 8 * stride, stride);
 }
 
-static void SimpleVFilter16i_SSE2(uint8_t* p, int stride, int thresh) {
+static void SimpleVFilter16i(uint8_t* p, int stride, int thresh) {
   int k;
   for (k = 3; k > 0; --k) {
     p += 4 * stride;
-    SimpleVFilter16_SSE2(p, stride, thresh);
+    SimpleVFilter16(p, stride, thresh);
   }
 }
 
-static void SimpleHFilter16i_SSE2(uint8_t* p, int stride, int thresh) {
+static void SimpleHFilter16i(uint8_t* p, int stride, int thresh) {
   int k;
   for (k = 3; k > 0; --k) {
     p += 4;
-    SimpleHFilter16_SSE2(p, stride, thresh);
+    SimpleHFilter16(p, stride, thresh);
   }
 }
 
@@ -634,60 +628,60 @@ static void SimpleHFilter16i_SSE2(uint8_t* p, int stride, int thresh) {
 // Complex In-loop filtering (Paragraph 15.3)
 
 #define MAX_DIFF1(p3, p2, p1, p0, m) do {                                      \
-  (m) = MM_ABS(p1, p0);                                                        \
+  m = MM_ABS(p1, p0);                                                          \
-  (m) = _mm_max_epu8(m, MM_ABS(p3, p2));                                       \
+  m = _mm_max_epu8(m, MM_ABS(p3, p2));                                         \
-  (m) = _mm_max_epu8(m, MM_ABS(p2, p1));                                       \
+  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(p1, p0));                                         \
-  (m) = _mm_max_epu8(m, MM_ABS(p3, p2));                                       \
+  m = _mm_max_epu8(m, MM_ABS(p3, p2));                                         \
-  (m) = _mm_max_epu8(m, MM_ABS(p2, p1));                                       \
+  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)]);                        \
+  e1 = _mm_loadu_si128((__m128i*)&(p)[0 * stride]);                            \
-  (e2) = _mm_loadu_si128((__m128i*)&(p)[1 * (stride)]);                        \
+  e2 = _mm_loadu_si128((__m128i*)&(p)[1 * stride]);                            \
-  (e3) = _mm_loadu_si128((__m128i*)&(p)[2 * (stride)]);                        \
+  e3 = _mm_loadu_si128((__m128i*)&(p)[2 * stride]);                            \
-  (e4) = _mm_loadu_si128((__m128i*)&(p)[3 * (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);                                              \
+  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(e1, u, v, 0 * stride);                                         \
-  LOADUV_H_EDGE(e2, u, v, 1 * (stride));                                       \
+  LOADUV_H_EDGE(e2, u, v, 1 * stride);                                         \
-  LOADUV_H_EDGE(e3, u, v, 2 * (stride));                                       \
+  LOADUV_H_EDGE(e3, u, v, 2 * stride);                                         \
-  LOADUV_H_EDGE(e4, u, v, 3 * (stride));                                       \
+  LOADUV_H_EDGE(e4, u, v, 3 * stride);                                         \
 }
 
 #define STOREUV(p, u, v, stride) {                                             \
-  _mm_storel_epi64((__m128i*)&(u)[(stride)], p);                               \
+  _mm_storel_epi64((__m128i*)&u[(stride)], p);                                 \
-  (p) = _mm_srli_si128(p, 8);                                                  \
+  p = _mm_srli_si128(p, 8);                                                    \
-  _mm_storel_epi64((__m128i*)&(v)[(stride)], p);                               \
+  _mm_storel_epi64((__m128i*)&v[(stride)], p);                                 \
 }
 
-static WEBP_INLINE void ComplexMask_SSE2(const __m128i* const p1,
+static WEBP_INLINE void ComplexMask(const __m128i* const p1,
-                                         const __m128i* const p0,
+                                    const __m128i* const p0,
-                                         const __m128i* const q0,
+                                    const __m128i* const q0,
-                                         const __m128i* const q1,
+                                    const __m128i* const q1,
-                                         int thresh, int ithresh,
+                                    int thresh, int ithresh,
-                                         __m128i* const mask) {
+                                    __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_SSE2(p1, p0, q0, q1, thresh, &filter_mask);
+  NeedsFilter(p1, p0, q0, q1, thresh, &filter_mask);
   *mask = _mm_and_si128(thresh_mask, filter_mask);
 }
 
 // on macroblock edges
-static void VFilter16_SSE2(uint8_t* p, int stride,
+static void VFilter16(uint8_t* p, int stride,
-                           int thresh, int ithresh, int hev_thresh) {
+                      int thresh, int ithresh, int hev_thresh) {
   __m128i t1;
   __m128i mask;
   __m128i p2, p1, p0, q0, q1, q2;
@@ -700,8 +694,8 @@ static void VFilter16_SSE2(uint8_t* p, int stride,
   LOAD_H_EDGES4(p, stride, q0, q1, q2, t1);
   MAX_DIFF2(t1, q2, q1, q0, mask);
 
-  ComplexMask_SSE2(&p1, &p0, &q0, &q1, thresh, ithresh, &mask);
+  ComplexMask(&p1, &p0, &q0, &q1, thresh, ithresh, &mask);
-  DoFilter6_SSE2(&p2, &p1, &p0, &q0, &q1, &q2, &mask, hev_thresh);
+  DoFilter6(&p2, &p1, &p0, &q0, &q1, &q2, &mask, hev_thresh);
 
   // Store
   _mm_storeu_si128((__m128i*)&p[-3 * stride], p2);
@@ -712,28 +706,28 @@ static void VFilter16_SSE2(uint8_t* p, int stride,
   _mm_storeu_si128((__m128i*)&p[+2 * stride], q2);
 }
 
-static void HFilter16_SSE2(uint8_t* p, int stride,
+static void HFilter16(uint8_t* p, int stride,
-                           int thresh, int ithresh, int hev_thresh) {
+                      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_SSE2(b, b + 8 * stride, stride, &p3, &p2, &p1, &p0);
+  Load16x4(b, b + 8 * stride, stride, &p3, &p2, &p1, &p0);  // p3, p2, p1, p0
   MAX_DIFF1(p3, p2, p1, p0, mask);
 
-  Load16x4_SSE2(p, p + 8 * stride, stride, &q0, &q1, &q2, &q3);
+  Load16x4(p, p + 8 * stride, stride, &q0, &q1, &q2, &q3);  // q0, q1, q2, q3
   MAX_DIFF2(q3, q2, q1, q0, mask);
 
-  ComplexMask_SSE2(&p1, &p0, &q0, &q1, thresh, ithresh, &mask);
+  ComplexMask(&p1, &p0, &q0, &q1, thresh, ithresh, &mask);
-  DoFilter6_SSE2(&p2, &p1, &p0, &q0, &q1, &q2, &mask, hev_thresh);
+  DoFilter6(&p2, &p1, &p0, &q0, &q1, &q2, &mask, hev_thresh);
 
-  Store16x4_SSE2(&p3, &p2, &p1, &p0, b, b + 8 * stride, stride);
+  Store16x4(&p3, &p2, &p1, &p0, b, b + 8 * stride, stride);
-  Store16x4_SSE2(&q0, &q1, &q2, &q3, p, p + 8 * stride, stride);
+  Store16x4(&q0, &q1, &q2, &q3, p, p + 8 * stride, stride);
 }
 
 // on three inner edges
-static void VFilter16i_SSE2(uint8_t* p, int stride,
+static void VFilter16i(uint8_t* p, int stride,
-                            int thresh, int ithresh, int hev_thresh) {
+                       int thresh, int ithresh, int hev_thresh) {
   int k;
   __m128i p3, p2, p1, p0;   // loop invariants
 
@@ -750,8 +744,8 @@ static void VFilter16i_SSE2(uint8_t* p, int stride,
 
     // 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_SSE2(&p1, &p0, &p3, &p2, thresh, ithresh, &mask);
+    ComplexMask(&p1, &p0, &p3, &p2, thresh, ithresh, &mask);
-    DoFilter4_SSE2(&p1, &p0, &p3, &p2, &mask, hev_thresh);
+    DoFilter4(&p1, &p0, &p3, &p2, &mask, hev_thresh);
 
     // Store
     _mm_storeu_si128((__m128i*)&b[0 * stride], p1);
@@ -765,12 +759,12 @@ static void VFilter16i_SSE2(uint8_t* p, int stride,
   }
 }
 
-static void HFilter16i_SSE2(uint8_t* p, int stride,
+static void HFilter16i(uint8_t* p, int stride,
-                            int thresh, int ithresh, int hev_thresh) {
+                       int thresh, int ithresh, int hev_thresh) {
   int k;
   __m128i p3, p2, p1, p0;   // loop invariants
 
-  Load16x4_SSE2(p, p + 8 * stride, stride, &p3, &p2, &p1, &p0);  // prologue
+  Load16x4(p, p + 8 * stride, stride, &p3, &p2, &p1, &p0);  // prologue
 
   for (k = 3; k > 0; --k) {
     __m128i mask, tmp1, tmp2;
@@ -779,13 +773,13 @@ static void HFilter16i_SSE2(uint8_t* p, int stride,
     p += 4;  // beginning of q0 (and next span)
 
     MAX_DIFF1(p3, p2, p1, p0, mask);   // compute partial mask
-    Load16x4_SSE2(p, p + 8 * stride, stride, &p3, &p2, &tmp1, &tmp2);
+    Load16x4(p, p + 8 * stride, stride, &p3, &p2, &tmp1, &tmp2);
     MAX_DIFF2(p3, p2, tmp1, tmp2, mask);
 
-    ComplexMask_SSE2(&p1, &p0, &p3, &p2, thresh, ithresh, &mask);
+    ComplexMask(&p1, &p0, &p3, &p2, thresh, ithresh, &mask);
-    DoFilter4_SSE2(&p1, &p0, &p3, &p2, &mask, hev_thresh);
+    DoFilter4(&p1, &p0, &p3, &p2, &mask, hev_thresh);
 
-    Store16x4_SSE2(&p1, &p0, &p3, &p2, b, b + 8 * stride, stride);
+    Store16x4(&p1, &p0, &p3, &p2, b, b + 8 * stride, stride);
 
     // rotate samples
     p1 = tmp1;
@@ -794,8 +788,8 @@ static void HFilter16i_SSE2(uint8_t* p, int stride,
 }
 
 // 8-pixels wide variant, for chroma filtering
-static void VFilter8_SSE2(uint8_t* u, uint8_t* v, int stride,
+static void VFilter8(uint8_t* u, uint8_t* v, int stride,
-                          int thresh, int ithresh, int hev_thresh) {
+                     int thresh, int ithresh, int hev_thresh) {
   __m128i mask;
   __m128i t1, p2, p1, p0, q0, q1, q2;
 
@@ -807,8 +801,8 @@ static void VFilter8_SSE2(uint8_t* u, uint8_t* v, int stride,
   LOADUV_H_EDGES4(u, v, stride, q0, q1, q2, t1);
   MAX_DIFF2(t1, q2, q1, q0, mask);
 
-  ComplexMask_SSE2(&p1, &p0, &q0, &q1, thresh, ithresh, &mask);
+  ComplexMask(&p1, &p0, &q0, &q1, thresh, ithresh, &mask);
-  DoFilter6_SSE2(&p2, &p1, &p0, &q0, &q1, &q2, &mask, hev_thresh);
+  DoFilter6(&p2, &p1, &p0, &q0, &q1, &q2, &mask, hev_thresh);
 
   // Store
   STOREUV(p2, u, v, -3 * stride);
@@ -819,28 +813,28 @@ static void VFilter8_SSE2(uint8_t* u, uint8_t* v, int stride,
   STOREUV(q2, u, v, 2 * stride);
 }
 
-static void HFilter8_SSE2(uint8_t* u, uint8_t* v, int stride,
+static void HFilter8(uint8_t* u, uint8_t* v, int stride,
-                          int thresh, int ithresh, int hev_thresh) {
+                     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_SSE2(tu, tv, stride, &p3, &p2, &p1, &p0);
+  Load16x4(tu, tv, stride, &p3, &p2, &p1, &p0);  // p3, p2, p1, p0
   MAX_DIFF1(p3, p2, p1, p0, mask);
 
-  Load16x4_SSE2(u, v, stride, &q0, &q1, &q2, &q3);
+  Load16x4(u, v, stride, &q0, &q1, &q2, &q3);    // q0, q1, q2, q3
   MAX_DIFF2(q3, q2, q1, q0, mask);
 
-  ComplexMask_SSE2(&p1, &p0, &q0, &q1, thresh, ithresh, &mask);
+  ComplexMask(&p1, &p0, &q0, &q1, thresh, ithresh, &mask);
-  DoFilter6_SSE2(&p2, &p1, &p0, &q0, &q1, &q2, &mask, hev_thresh);
+  DoFilter6(&p2, &p1, &p0, &q0, &q1, &q2, &mask, hev_thresh);
 
-  Store16x4_SSE2(&p3, &p2, &p1, &p0, tu, tv, stride);
+  Store16x4(&p3, &p2, &p1, &p0, tu, tv, stride);
-  Store16x4_SSE2(&q0, &q1, &q2, &q3, u, v, stride);
+  Store16x4(&q0, &q1, &q2, &q3, u, v, stride);
 }
 
-static void VFilter8i_SSE2(uint8_t* u, uint8_t* v, int stride,
+static void VFilter8i(uint8_t* u, uint8_t* v, int stride,
-                           int thresh, int ithresh, int hev_thresh) {
+                      int thresh, int ithresh, int hev_thresh) {
   __m128i mask;
   __m128i t1, t2, p1, p0, q0, q1;
 
@@ -855,8 +849,8 @@ static void VFilter8i_SSE2(uint8_t* u, uint8_t* v, int stride,
   LOADUV_H_EDGES4(u, v, stride, q0, q1, t1, t2);
   MAX_DIFF2(t2, t1, q1, q0, mask);
 
-  ComplexMask_SSE2(&p1, &p0, &q0, &q1, thresh, ithresh, &mask);
+  ComplexMask(&p1, &p0, &q0, &q1, thresh, ithresh, &mask);
-  DoFilter4_SSE2(&p1, &p0, &q0, &q1, &mask, hev_thresh);
+  DoFilter4(&p1, &p0, &q0, &q1, &mask, hev_thresh);
 
   // Store
   STOREUV(p1, u, v, -2 * stride);
@@ -865,24 +859,24 @@ static void VFilter8i_SSE2(uint8_t* u, uint8_t* v, int stride,
   STOREUV(q1, u, v, 1 * stride);
 }
 
-static void HFilter8i_SSE2(uint8_t* u, uint8_t* v, int stride,
+static void HFilter8i(uint8_t* u, uint8_t* v, int stride,
-                           int thresh, int ithresh, int hev_thresh) {
+                      int thresh, int ithresh, int hev_thresh) {
   __m128i mask;
   __m128i t1, t2, p1, p0, q0, q1;
-  Load16x4_SSE2(u, v, stride, &t2, &t1, &p1, &p0);   // p3, p2, p1, p0
+  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_SSE2(u, v, stride, &q0, &q1, &t1, &t2);  // q0, q1, q2, q3
+  Load16x4(u, v, stride, &q0, &q1, &t1, &t2);  // q0, q1, q2, q3
   MAX_DIFF2(t2, t1, q1, q0, mask);
 
-  ComplexMask_SSE2(&p1, &p0, &q0, &q1, thresh, ithresh, &mask);
+  ComplexMask(&p1, &p0, &q0, &q1, thresh, ithresh, &mask);
-  DoFilter4_SSE2(&p1, &p0, &q0, &q1, &mask, hev_thresh);
+  DoFilter4(&p1, &p0, &q0, &q1, &mask, hev_thresh);
 
   u -= 2;  // beginning of p1
   v -= 2;
-  Store16x4_SSE2(&p1, &p0, &q0, &q1, u, v, stride);
+  Store16x4(&p1, &p0, &q0, &q1, u, v, stride);
 }
 
 //------------------------------------------------------------------------------
@@ -899,7 +893,7 @@ static void HFilter8i_SSE2(uint8_t* u, uint8_t* v, int stride,
 //   where: AC = (a + b + 1) >> 1,   BC = (b + c + 1) >> 1
 //   and ab = a ^ b, bc = b ^ c, lsb = (AC^BC)&1
 
-static void VE4_SSE2(uint8_t* dst) {    // vertical
+static void VE4(uint8_t* dst) {    // vertical
   const __m128i one = _mm_set1_epi8(1);
   const __m128i ABCDEFGH = _mm_loadl_epi64((__m128i*)(dst - BPS - 1));
   const __m128i BCDEFGH0 = _mm_srli_si128(ABCDEFGH, 1);
@@ -915,7 +909,7 @@ static void VE4_SSE2(uint8_t* dst) {    // vertical
   }
 }
 
-static void LD4_SSE2(uint8_t* dst) {   // Down-Left
+static void LD4(uint8_t* dst) {   // Down-Left
   const __m128i one = _mm_set1_epi8(1);
   const __m128i ABCDEFGH = _mm_loadl_epi64((__m128i*)(dst - BPS));
   const __m128i BCDEFGH0 = _mm_srli_si128(ABCDEFGH, 1);
@@ -931,7 +925,7 @@ static void LD4_SSE2(uint8_t* dst) {   // Down-Left
   WebPUint32ToMem(dst + 3 * BPS, _mm_cvtsi128_si32(_mm_srli_si128(abcdefg, 3)));
 }
 
-static void VR4_SSE2(uint8_t* dst) {   // Vertical-Right
+static void VR4(uint8_t* dst) {   // Vertical-Right
   const __m128i one = _mm_set1_epi8(1);
   const int I = dst[-1 + 0 * BPS];
   const int J = dst[-1 + 1 * BPS];
@@ -956,7 +950,7 @@ static void VR4_SSE2(uint8_t* dst) {   // Vertical-Right
   DST(0, 3) = AVG3(K, J, I);
 }
 
-static void VL4_SSE2(uint8_t* dst) {   // Vertical-Left
+static void VL4(uint8_t* dst) {   // Vertical-Left
   const __m128i one = _mm_set1_epi8(1);
   const __m128i ABCDEFGH = _mm_loadl_epi64((__m128i*)(dst - BPS));
   const __m128i BCDEFGH_ = _mm_srli_si128(ABCDEFGH, 1);
@@ -981,7 +975,7 @@ static void VL4_SSE2(uint8_t* dst) {   // Vertical-Left
   DST(3, 3) = (extra_out >> 8) & 0xff;
 }
 
-static void RD4_SSE2(uint8_t* dst) {   // Down-right
+static void RD4(uint8_t* dst) {   // Down-right
   const __m128i one = _mm_set1_epi8(1);
   const __m128i XABCD = _mm_loadl_epi64((__m128i*)(dst - BPS - 1));
   const __m128i ____XABCD = _mm_slli_si128(XABCD, 4);
@@ -1010,7 +1004,7 @@ static void RD4_SSE2(uint8_t* dst) {   // Down-right
 //------------------------------------------------------------------------------
 // Luma 16x16
 
-static WEBP_INLINE void TrueMotion_SSE2(uint8_t* dst, int size) {
+static WEBP_INLINE void TrueMotion(uint8_t* dst, int size) {
   const uint8_t* top = dst - BPS;
   const __m128i zero = _mm_setzero_si128();
   int y;
@@ -1047,11 +1041,11 @@ static WEBP_INLINE void TrueMotion_SSE2(uint8_t* dst, int size) {
   }
 }
 
-static void TM4_SSE2(uint8_t* dst)   { TrueMotion_SSE2(dst, 4); }
+static void TM4(uint8_t* dst)   { TrueMotion(dst, 4); }
-static void TM8uv_SSE2(uint8_t* dst) { TrueMotion_SSE2(dst, 8); }
+static void TM8uv(uint8_t* dst) { TrueMotion(dst, 8); }
-static void TM16_SSE2(uint8_t* dst)  { TrueMotion_SSE2(dst, 16); }
+static void TM16(uint8_t* dst)  { TrueMotion(dst, 16); }
 
-static void VE16_SSE2(uint8_t* dst) {
+static void VE16(uint8_t* dst) {
   const __m128i top = _mm_loadu_si128((const __m128i*)(dst - BPS));
   int j;
   for (j = 0; j < 16; ++j) {
@@ -1059,7 +1053,7 @@ static void VE16_SSE2(uint8_t* dst) {
   }
 }
 
-static void HE16_SSE2(uint8_t* dst) {     // horizontal
+static void HE16(uint8_t* dst) {     // horizontal
   int j;
   for (j = 16; j > 0; --j) {
     const __m128i values = _mm_set1_epi8(dst[-1]);
@@ -1068,7 +1062,7 @@ static void HE16_SSE2(uint8_t* dst) {     // horizontal
   }
 }
 
-static WEBP_INLINE void Put16_SSE2(uint8_t v, uint8_t* dst) {
+static WEBP_INLINE void Put16(uint8_t v, uint8_t* dst) {
   int j;
   const __m128i values = _mm_set1_epi8(v);
   for (j = 0; j < 16; ++j) {
@@ -1076,7 +1070,7 @@ static WEBP_INLINE void Put16_SSE2(uint8_t v, uint8_t* dst) {
   }
 }
 
-static void DC16_SSE2(uint8_t* dst) {  // DC
+static void DC16(uint8_t* dst) {    // DC
   const __m128i zero = _mm_setzero_si128();
   const __m128i top = _mm_loadu_si128((const __m128i*)(dst - BPS));
   const __m128i sad8x2 = _mm_sad_epu8(top, zero);
@@ -1089,37 +1083,37 @@ static void DC16_SSE2(uint8_t* dst) {  // DC
   }
   {
     const int DC = _mm_cvtsi128_si32(sum) + left + 16;
-    Put16_SSE2(DC >> 5, dst);
+    Put16(DC >> 5, dst);
   }
 }
 
-static void DC16NoTop_SSE2(uint8_t* dst) {  // DC with top samples unavailable
+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_SSE2(DC >> 4, dst);
+  Put16(DC >> 4, dst);
 }
 
-static void DC16NoLeft_SSE2(uint8_t* dst) {  // DC with left samples unavailable
+static void DC16NoLeft(uint8_t* dst) {  // DC with left samples not available
   const __m128i zero = _mm_setzero_si128();
   const __m128i top = _mm_loadu_si128((const __m128i*)(dst - BPS));
   const __m128i sad8x2 = _mm_sad_epu8(top, zero);
   // sum the two sads: sad8x2[0:1] + sad8x2[8:9]
   const __m128i sum = _mm_add_epi16(sad8x2, _mm_shuffle_epi32(sad8x2, 2));
   const int DC = _mm_cvtsi128_si32(sum) + 8;
-  Put16_SSE2(DC >> 4, dst);
+  Put16(DC >> 4, dst);
 }
 
-static void DC16NoTopLeft_SSE2(uint8_t* dst) {  // DC with no top & left samples
+static void DC16NoTopLeft(uint8_t* dst) {  // DC with no top and left samples
-  Put16_SSE2(0x80, dst);
+  Put16(0x80, dst);
 }
 
 //------------------------------------------------------------------------------
 // Chroma
 
-static void VE8uv_SSE2(uint8_t* dst) {    // vertical
+static void VE8uv(uint8_t* dst) {    // vertical
   int j;
   const __m128i top = _mm_loadl_epi64((const __m128i*)(dst - BPS));
   for (j = 0; j < 8; ++j) {
@@ -1127,8 +1121,17 @@ static void VE8uv_SSE2(uint8_t* dst) {    // vertical
   }
 }
 
+static void HE8uv(uint8_t* dst) {    // horizontal
+  int j;
+  for (j = 0; j < 8; ++j) {
+    const __m128i values = _mm_set1_epi8(dst[-1]);
+    _mm_storel_epi64((__m128i*)dst, values);
+    dst += BPS;
+  }
+}
+
 // helper for chroma-DC predictions
-static WEBP_INLINE void Put8x8uv_SSE2(uint8_t v, uint8_t* dst) {
+static WEBP_INLINE void Put8x8uv(uint8_t v, uint8_t* dst) {
   int j;
   const __m128i values = _mm_set1_epi8(v);
   for (j = 0; j < 8; ++j) {
@@ -1136,7 +1139,7 @@ static WEBP_INLINE void Put8x8uv_SSE2(uint8_t v, uint8_t* dst) {
   }
 }
 
-static void DC8uv_SSE2(uint8_t* dst) {     // DC
+static void DC8uv(uint8_t* dst) {     // DC
   const __m128i zero = _mm_setzero_si128();
   const __m128i top = _mm_loadl_epi64((const __m128i*)(dst - BPS));
   const __m128i sum = _mm_sad_epu8(top, zero);
@@ -1147,29 +1150,29 @@ static void DC8uv_SSE2(uint8_t* dst) {     // DC
   }
   {
     const int DC = _mm_cvtsi128_si32(sum) + left + 8;
-    Put8x8uv_SSE2(DC >> 4, dst);
+    Put8x8uv(DC >> 4, dst);
   }
 }
 
-static void DC8uvNoLeft_SSE2(uint8_t* dst) {   // DC with no left samples
+static void DC8uvNoLeft(uint8_t* dst) {   // DC with no left samples
   const __m128i zero = _mm_setzero_si128();
   const __m128i top = _mm_loadl_epi64((const __m128i*)(dst - BPS));
   const __m128i sum = _mm_sad_epu8(top, zero);
   const int DC = _mm_cvtsi128_si32(sum) + 4;
-  Put8x8uv_SSE2(DC >> 3, dst);
+  Put8x8uv(DC >> 3, dst);
 }
 
-static void DC8uvNoTop_SSE2(uint8_t* dst) {  // DC with no top samples
+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_SSE2(dc0 >> 3, dst);
+  Put8x8uv(dc0 >> 3, dst);
 }
 
-static void DC8uvNoTopLeft_SSE2(uint8_t* dst) {    // DC with nothing
+static void DC8uvNoTopLeft(uint8_t* dst) {    // DC with nothing
-  Put8x8uv_SSE2(0x80, dst);
+  Put8x8uv(0x80, dst);
 }
 
 //------------------------------------------------------------------------------
@@ -1178,46 +1181,47 @@ static void DC8uvNoTopLeft_SSE2(uint8_t* dst) {    // DC with nothing
 extern void VP8DspInitSSE2(void);
 
 WEBP_TSAN_IGNORE_FUNCTION void VP8DspInitSSE2(void) {
-  VP8Transform = Transform_SSE2;
+  VP8Transform = Transform;
-#if (USE_TRANSFORM_AC3 == 1)
+#if defined(USE_TRANSFORM_AC3)
-  VP8TransformAC3 = TransformAC3_SSE2;
+  VP8TransformAC3 = TransformAC3;
 #endif
 
-  VP8VFilter16 = VFilter16_SSE2;
+  VP8VFilter16 = VFilter16;
-  VP8HFilter16 = HFilter16_SSE2;
+  VP8HFilter16 = HFilter16;
-  VP8VFilter8 = VFilter8_SSE2;
+  VP8VFilter8 = VFilter8;
-  VP8HFilter8 = HFilter8_SSE2;
+  VP8HFilter8 = HFilter8;
-  VP8VFilter16i = VFilter16i_SSE2;
+  VP8VFilter16i = VFilter16i;
-  VP8HFilter16i = HFilter16i_SSE2;
+  VP8HFilter16i = HFilter16i;
-  VP8VFilter8i = VFilter8i_SSE2;
+  VP8VFilter8i = VFilter8i;
-  VP8HFilter8i = HFilter8i_SSE2;
+  VP8HFilter8i = HFilter8i;
 
-  VP8SimpleVFilter16 = SimpleVFilter16_SSE2;
+  VP8SimpleVFilter16 = SimpleVFilter16;
-  VP8SimpleHFilter16 = SimpleHFilter16_SSE2;
+  VP8SimpleHFilter16 = SimpleHFilter16;
-  VP8SimpleVFilter16i = SimpleVFilter16i_SSE2;
+  VP8SimpleVFilter16i = SimpleVFilter16i;
-  VP8SimpleHFilter16i = SimpleHFilter16i_SSE2;
+  VP8SimpleHFilter16i = SimpleHFilter16i;
 
-  VP8PredLuma4[1] = TM4_SSE2;
+  VP8PredLuma4[1] = TM4;
-  VP8PredLuma4[2] = VE4_SSE2;
+  VP8PredLuma4[2] = VE4;
-  VP8PredLuma4[4] = RD4_SSE2;
+  VP8PredLuma4[4] = RD4;
-  VP8PredLuma4[5] = VR4_SSE2;
+  VP8PredLuma4[5] = VR4;
-  VP8PredLuma4[6] = LD4_SSE2;
+  VP8PredLuma4[6] = LD4;
-  VP8PredLuma4[7] = VL4_SSE2;
+  VP8PredLuma4[7] = VL4;
 
-  VP8PredLuma16[0] = DC16_SSE2;
+  VP8PredLuma16[0] = DC16;
-  VP8PredLuma16[1] = TM16_SSE2;
+  VP8PredLuma16[1] = TM16;
-  VP8PredLuma16[2] = VE16_SSE2;
+  VP8PredLuma16[2] = VE16;
-  VP8PredLuma16[3] = HE16_SSE2;
+  VP8PredLuma16[3] = HE16;
-  VP8PredLuma16[4] = DC16NoTop_SSE2;
+  VP8PredLuma16[4] = DC16NoTop;
-  VP8PredLuma16[5] = DC16NoLeft_SSE2;
+  VP8PredLuma16[5] = DC16NoLeft;
-  VP8PredLuma16[6] = DC16NoTopLeft_SSE2;
+  VP8PredLuma16[6] = DC16NoTopLeft;
 
-  VP8PredChroma8[0] = DC8uv_SSE2;
+  VP8PredChroma8[0] = DC8uv;
-  VP8PredChroma8[1] = TM8uv_SSE2;
+  VP8PredChroma8[1] = TM8uv;
-  VP8PredChroma8[2] = VE8uv_SSE2;
+  VP8PredChroma8[2] = VE8uv;
-  VP8PredChroma8[4] = DC8uvNoTop_SSE2;
+  VP8PredChroma8[3] = HE8uv;
-  VP8PredChroma8[5] = DC8uvNoLeft_SSE2;
+  VP8PredChroma8[4] = DC8uvNoTop;
-  VP8PredChroma8[6] = DC8uvNoTopLeft_SSE2;
+  VP8PredChroma8[5] = DC8uvNoLeft;
+  VP8PredChroma8[6] = DC8uvNoTopLeft;
 }
 
 #else  // !WEBP_USE_SSE2

src/dsp/dec_sse41.c

@@ -11,15 +11,15 @@
 //
 // Author: Skal (pascal.massimino@gmail.com)
 
-#include "src/dsp/dsp.h"
+#include "./dsp.h"
 
 #if defined(WEBP_USE_SSE41)
 
 #include <smmintrin.h>
-#include "src/dec/vp8i_dec.h"
+#include "../dec/vp8i_dec.h"
-#include "src/utils/utils.h"
+#include "../utils/utils.h"
 
-static void HE16_SSE41(uint8_t* dst) {     // horizontal
+static void HE16(uint8_t* dst) {     // horizontal
   int j;
   const __m128i kShuffle3 = _mm_set1_epi8(3);
   for (j = 16; j > 0; --j) {
@@ -36,7 +36,7 @@ static void HE16_SSE41(uint8_t* dst) {     // horizontal
 extern void VP8DspInitSSE41(void);
 
 WEBP_TSAN_IGNORE_FUNCTION void VP8DspInitSSE41(void) {
-  VP8PredLuma16[3] = HE16_SSE41;
+  VP8PredLuma16[3] = HE16;
 }
 
 #else  // !WEBP_USE_SSE41

src/dsp/dsp.h

@@ -15,10 +15,10 @@
 #define WEBP_DSP_DSP_H_
 
 #ifdef HAVE_CONFIG_H
-#include "src/webp/config.h"
+#include "../webp/config.h"
 #endif
 
-#include "src/webp/types.h"
+#include "../webp/types.h"
 
 #ifdef __cplusplus
 extern "C" {
@@ -51,8 +51,9 @@ extern "C" {
 # define __has_builtin(x) 0
 #endif
 
-// for now, none of the optimizations below are available in emscripten
+// For now, none of the optimizations below are available in emscripten.
-#if !defined(EMSCRIPTEN)
+// WebAssembly overrides native optimizations.
+#if !(defined(EMSCRIPTEN) || defined(WEBP_USE_WASM))
 
 #if defined(_MSC_VER) && _MSC_VER > 1310 && \
     (defined(_M_X64) || defined(_M_IX86))
@@ -104,7 +105,7 @@ extern "C" {
 #define WEBP_USE_MIPS32
 #if (__mips_isa_rev >= 2)
 #define WEBP_USE_MIPS32_R2
-#if defined(__mips_dspr2) || (defined(__mips_dsp_rev) && __mips_dsp_rev >= 2)
+#if defined(__mips_dspr2) || (__mips_dsp_rev >= 2)
 #define WEBP_USE_MIPS_DSP_R2
 #endif
 #endif
@@ -116,22 +117,6 @@ extern "C" {
 
 #endif  /* EMSCRIPTEN */
 
-#ifndef WEBP_DSP_OMIT_C_CODE
-#define WEBP_DSP_OMIT_C_CODE 1
-#endif
-
-#if (defined(__aarch64__) || defined(__ARM_NEON__)) && WEBP_DSP_OMIT_C_CODE
-#define WEBP_NEON_OMIT_C_CODE 1
-#else
-#define WEBP_NEON_OMIT_C_CODE 0
-#endif
-
-#if !(LOCAL_CLANG_PREREQ(3,8) || LOCAL_GCC_PREREQ(4,8) || defined(__aarch64__))
-#define WEBP_NEON_WORK_AROUND_GCC 1
-#else
-#define WEBP_NEON_WORK_AROUND_GCC 0
-#endif
-
 // This macro prevents thread_sanitizer from reporting known concurrent writes.
 #define WEBP_TSAN_IGNORE_FUNCTION
 #if defined(__has_feature)
@@ -141,42 +126,6 @@ extern "C" {
 #endif
 #endif
 
-#if defined(WEBP_USE_THREAD) && !defined(_WIN32)
-#include <pthread.h>  // NOLINT
-
-#define WEBP_DSP_INIT(func) do {                                    \
-  static volatile VP8CPUInfo func ## _last_cpuinfo_used =           \
-      (VP8CPUInfo)&func ## _last_cpuinfo_used;                      \
-  static pthread_mutex_t func ## _lock = PTHREAD_MUTEX_INITIALIZER; \
-  if (pthread_mutex_lock(&func ## _lock)) break;                    \
-  if (func ## _last_cpuinfo_used != VP8GetCPUInfo) func();          \
-  func ## _last_cpuinfo_used = VP8GetCPUInfo;                       \
-  (void)pthread_mutex_unlock(&func ## _lock);                       \
-} while (0)
-#else  // !(defined(WEBP_USE_THREAD) && !defined(_WIN32))
-#define WEBP_DSP_INIT(func) do {                                    \
-  static volatile VP8CPUInfo func ## _last_cpuinfo_used =           \
-      (VP8CPUInfo)&func ## _last_cpuinfo_used;                      \
-  if (func ## _last_cpuinfo_used == VP8GetCPUInfo) break;           \
-  func();                                                           \
-  func ## _last_cpuinfo_used = VP8GetCPUInfo;                       \
-} while (0)
-#endif  // defined(WEBP_USE_THREAD) && !defined(_WIN32)
-
-// Defines an Init + helper function that control multiple initialization of
-// function pointers / tables.
-/* Usage:
-   WEBP_DSP_INIT_FUNC(InitFunc) {
-     ...function body
-   }
-*/
-#define WEBP_DSP_INIT_FUNC(name)                             \
-  static WEBP_TSAN_IGNORE_FUNCTION void name ## _body(void); \
-  WEBP_TSAN_IGNORE_FUNCTION void name(void) {                \
-    WEBP_DSP_INIT(name ## _body);                            \
-  }                                                          \
-  static WEBP_TSAN_IGNORE_FUNCTION void name ## _body(void)
-
 #define WEBP_UBSAN_IGNORE_UNDEF
 #define WEBP_UBSAN_IGNORE_UNSIGNED_OVERFLOW
 #if defined(__clang__) && defined(__has_attribute)
@@ -197,18 +146,6 @@ extern "C" {
 #endif
 #endif
 
-// Regularize the definition of WEBP_SWAP_16BIT_CSP (backward compatibility)
-#if !defined(WEBP_SWAP_16BIT_CSP)
-#define WEBP_SWAP_16BIT_CSP 0
-#endif
-
-// some endian fix (e.g.: mips-gcc doesn't define __BIG_ENDIAN__)
-#if !defined(WORDS_BIGENDIAN) && \
-    (defined(__BIG_ENDIAN__) || defined(_M_PPC) || \
-     (defined(__BYTE_ORDER__) && (__BYTE_ORDER__ == __ORDER_BIG_ENDIAN__)))
-#define WORDS_BIGENDIAN
-#endif
-
 typedef enum {
   kSSE2,
   kSSE3,
@@ -219,11 +156,12 @@ typedef enum {
   kNEON,
   kMIPS32,
   kMIPSdspR2,
-  kMSA
+  kMSA,
+  kWASM
 } CPUFeature;
 // returns true if the CPU supports the feature.
 typedef int (*VP8CPUInfo)(CPUFeature feature);
-WEBP_EXTERN VP8CPUInfo VP8GetCPUInfo;
+WEBP_EXTERN(VP8CPUInfo) VP8GetCPUInfo;
 
 //------------------------------------------------------------------------------
 // Init stub generator
@@ -232,7 +170,7 @@ WEBP_EXTERN VP8CPUInfo VP8GetCPUInfo;
 // avoiding a compiler warning.
 #define WEBP_DSP_INIT_STUB(func) \
   extern void func(void); \
-  void func(void) {}
+  WEBP_TSAN_IGNORE_FUNCTION void func(void) {}
 
 //------------------------------------------------------------------------------
 // Encoding
@@ -351,7 +289,6 @@ typedef double (*VP8SSIMGetClippedFunc)(const uint8_t* src1, int stride1,
                                         int xo, int yo,  // center position
                                         int W, int H);   // plane dimension
 
-#if !defined(WEBP_REDUCE_SIZE)
 // This version is called with the guarantee that you can load 8 bytes and
 // 8 rows at offset src1 and src2
 typedef double (*VP8SSIMGetFunc)(const uint8_t* src1, int stride1,
@@ -359,13 +296,10 @@ typedef double (*VP8SSIMGetFunc)(const uint8_t* src1, int stride1,
 
 extern VP8SSIMGetFunc VP8SSIMGet;         // unclipped / unchecked
 extern VP8SSIMGetClippedFunc VP8SSIMGetClipped;   // with clipping
-#endif
 
-#if !defined(WEBP_DISABLE_STATS)
 typedef uint32_t (*VP8AccumulateSSEFunc)(const uint8_t* src1,
                                          const uint8_t* src2, int len);
 extern VP8AccumulateSSEFunc VP8AccumulateSSE;
-#endif
 
 // must be called before using any of the above directly
 void VP8SSIMDspInit(void);
@@ -546,12 +480,12 @@ extern WebPRescalerExportRowFunc WebPRescalerExportRowExpand;
 extern WebPRescalerExportRowFunc WebPRescalerExportRowShrink;
 
 // Plain-C implementation, as fall-back.
-extern void WebPRescalerImportRowExpand_C(struct WebPRescaler* const wrk,
+extern void WebPRescalerImportRowExpandC(struct WebPRescaler* const wrk,
-                                          const uint8_t* src);
+                                         const uint8_t* src);
-extern void WebPRescalerImportRowShrink_C(struct WebPRescaler* const wrk,
+extern void WebPRescalerImportRowShrinkC(struct WebPRescaler* const wrk,
-                                          const uint8_t* src);
+                                         const uint8_t* src);
-extern void WebPRescalerExportRowExpand_C(struct WebPRescaler* const wrk);
+extern void WebPRescalerExportRowExpandC(struct WebPRescaler* const wrk);
-extern void WebPRescalerExportRowShrink_C(struct WebPRescaler* const wrk);
+extern void WebPRescalerExportRowShrinkC(struct WebPRescaler* const wrk);
 
 // Main entry calls:
 extern void WebPRescalerImportRow(struct WebPRescaler* const wrk,
@@ -617,29 +551,25 @@ void WebPMultRows(uint8_t* ptr, int stride,
                   int width, int num_rows, int inverse);
 
 // Plain-C versions, used as fallback by some implementations.
-void WebPMultRow_C(uint8_t* const ptr, const uint8_t* const alpha,
+void WebPMultRowC(uint8_t* const ptr, const uint8_t* const alpha,
-                   int width, int inverse);
+                  int width, int inverse);
-void WebPMultARGBRow_C(uint32_t* const ptr, int width, int inverse);
+void WebPMultARGBRowC(uint32_t* const ptr, int width, int inverse);
-
-#ifdef WORDS_BIGENDIAN
-// ARGB packing function: a/r/g/b input is rgba or bgra order.
-extern void (*WebPPackARGB)(const uint8_t* a, const uint8_t* r,
-                            const uint8_t* g, const uint8_t* b, int len,
-                            uint32_t* out);
-#endif
-
-// RGB packing function. 'step' can be 3 or 4. r/g/b input is rgb or bgr order.
-extern void (*WebPPackRGB)(const uint8_t* r, const uint8_t* g, const uint8_t* b,
-                           int len, int step, uint32_t* out);
-
-// This function returns true if src[i] contains a value different from 0xff.
-extern int (*WebPHasAlpha8b)(const uint8_t* src, int length);
-// This function returns true if src[4*i] contains a value different from 0xff.
-extern int (*WebPHasAlpha32b)(const uint8_t* src, int length);
 
 // To be called first before using the above.
 void WebPInitAlphaProcessing(void);
 
+// ARGB packing function: a/r/g/b input is rgba or bgra order.
+extern void (*VP8PackARGB)(const uint8_t* a, const uint8_t* r,
+                           const uint8_t* g, const uint8_t* b, int len,
+                           uint32_t* out);
+
+// RGB packing function. 'step' can be 3 or 4. r/g/b input is rgb or bgr order.
+extern void (*VP8PackRGB)(const uint8_t* r, const uint8_t* g, const uint8_t* b,
+                          int len, int step, uint32_t* out);
+
+// To be called first before using the above.
+void VP8EncDspARGBInit(void);
+
 //------------------------------------------------------------------------------
 // Filter functions
 

src/dsp/enc.c

@@ -14,18 +14,16 @@
 #include <assert.h>
 #include <stdlib.h>  // for abs()
 
-#include "src/dsp/dsp.h"
+#include "./dsp.h"
-#include "src/enc/vp8i_enc.h"
+#include "../enc/vp8i_enc.h"
 
 static WEBP_INLINE uint8_t clip_8b(int v) {
   return (!(v & ~0xff)) ? v : (v < 0) ? 0 : 255;
 }
 
-#if !WEBP_NEON_OMIT_C_CODE
 static WEBP_INLINE int clip_max(int v, int max) {
   return (v > max) ? max : v;
 }
-#endif  // !WEBP_NEON_OMIT_C_CODE
 
 //------------------------------------------------------------------------------
 // Compute susceptibility based on DCT-coeff histograms:
@@ -58,10 +56,9 @@ void VP8SetHistogramData(const int distribution[MAX_COEFF_THRESH + 1],
   histo->last_non_zero = last_non_zero;
 }
 
-#if !WEBP_NEON_OMIT_C_CODE
+static void CollectHistogram(const uint8_t* ref, const uint8_t* pred,
-static void CollectHistogram_C(const uint8_t* ref, const uint8_t* pred,
+                             int start_block, int end_block,
-                               int start_block, int end_block,
+                             VP8Histogram* const histo) {
-                               VP8Histogram* const histo) {
   int j;
   int distribution[MAX_COEFF_THRESH + 1] = { 0 };
   for (j = start_block; j < end_block; ++j) {
@@ -79,7 +76,6 @@ static void CollectHistogram_C(const uint8_t* ref, const uint8_t* pred,
   }
   VP8SetHistogramData(distribution, histo);
 }
-#endif  // !WEBP_NEON_OMIT_C_CODE
 
 //------------------------------------------------------------------------------
 // run-time tables (~4k)
@@ -104,8 +100,6 @@ static WEBP_TSAN_IGNORE_FUNCTION void InitTables(void) {
 //------------------------------------------------------------------------------
 // Transforms (Paragraph 14.4)
 
-#if !WEBP_NEON_OMIT_C_CODE
-
 #define STORE(x, y, v) \
   dst[(x) + (y) * BPS] = clip_8b(ref[(x) + (y) * BPS] + ((v) >> 3))
 
@@ -146,15 +140,15 @@ static WEBP_INLINE void ITransformOne(const uint8_t* ref, const int16_t* in,
   }
 }
 
-static void ITransform_C(const uint8_t* ref, const int16_t* in, uint8_t* dst,
+static void ITransform(const uint8_t* ref, const int16_t* in, uint8_t* dst,
-                         int do_two) {
+                       int do_two) {
   ITransformOne(ref, in, dst);
   if (do_two) {
     ITransformOne(ref + 4, in + 16, dst + 4);
   }
 }
 
-static void FTransform_C(const uint8_t* src, const uint8_t* ref, int16_t* out) {
+static void FTransform(const uint8_t* src, const uint8_t* ref, int16_t* out) {
   int i;
   int tmp[16];
   for (i = 0; i < 4; ++i, src += BPS, ref += BPS) {
@@ -182,16 +176,13 @@ static void FTransform_C(const uint8_t* src, const uint8_t* ref, int16_t* out) {
     out[12+ i] = ((a3 * 2217 - a2 * 5352 + 51000) >> 16);
   }
 }
-#endif  // !WEBP_NEON_OMIT_C_CODE
 
-static void FTransform2_C(const uint8_t* src, const uint8_t* ref,
+static void FTransform2(const uint8_t* src, const uint8_t* ref, int16_t* out) {
-                          int16_t* out) {
   VP8FTransform(src, ref, out);
   VP8FTransform(src + 4, ref + 4, out + 16);
 }
 
-#if !WEBP_NEON_OMIT_C_CODE
+static void FTransformWHT(const int16_t* in, int16_t* out) {
-static void FTransformWHT_C(const int16_t* in, int16_t* out) {
   // input is 12b signed
   int32_t tmp[16];
   int i;
@@ -220,7 +211,6 @@ static void FTransformWHT_C(const int16_t* in, int16_t* out) {
     out[12 + i] = b3 >> 1;
   }
 }
-#endif  // !WEBP_NEON_OMIT_C_CODE
 
 #undef MUL
 #undef STORE
@@ -313,8 +303,8 @@ static WEBP_INLINE void DCMode(uint8_t* dst, const uint8_t* left,
 //------------------------------------------------------------------------------
 // Chroma 8x8 prediction (paragraph 12.2)
 
-static void IntraChromaPreds_C(uint8_t* dst, const uint8_t* left,
+static void IntraChromaPreds(uint8_t* dst, const uint8_t* left,
-                               const uint8_t* top) {
+                             const uint8_t* top) {
   // U block
   DCMode(C8DC8 + dst, left, top, 8, 8, 4);
   VerticalPred(C8VE8 + dst, top, 8);
@@ -333,8 +323,8 @@ static void IntraChromaPreds_C(uint8_t* dst, const uint8_t* left,
 //------------------------------------------------------------------------------
 // luma 16x16 prediction (paragraph 12.3)
 
-static void Intra16Preds_C(uint8_t* dst,
+static void Intra16Preds(uint8_t* dst,
-                           const uint8_t* left, const uint8_t* top) {
+                         const uint8_t* left, const uint8_t* top) {
   DCMode(I16DC16 + dst, left, top, 16, 16, 5);
   VerticalPred(I16VE16 + dst, top, 16);
   HorizontalPred(I16HE16 + dst, left, 16);
@@ -517,7 +507,7 @@ static void TM4(uint8_t* dst, const uint8_t* top) {
 
 // Left samples are top[-5 .. -2], top_left is top[-1], top are
 // located at top[0..3], and top right is top[4..7]
-static void Intra4Preds_C(uint8_t* dst, const uint8_t* top) {
+static void Intra4Preds(uint8_t* dst, const uint8_t* top) {
   DC4(I4DC4 + dst, top);
   TM4(I4TM4 + dst, top);
   VE4(I4VE4 + dst, top);
@@ -533,7 +523,6 @@ static void Intra4Preds_C(uint8_t* dst, const uint8_t* top) {
 //------------------------------------------------------------------------------
 // Metric
 
-#if !WEBP_NEON_OMIT_C_CODE
 static WEBP_INLINE int GetSSE(const uint8_t* a, const uint8_t* b,
                               int w, int h) {
   int count = 0;
@@ -549,21 +538,20 @@ static WEBP_INLINE int GetSSE(const uint8_t* a, const uint8_t* b,
   return count;
 }
 
-static int SSE16x16_C(const uint8_t* a, const uint8_t* b) {
+static int SSE16x16(const uint8_t* a, const uint8_t* b) {
   return GetSSE(a, b, 16, 16);
 }
-static int SSE16x8_C(const uint8_t* a, const uint8_t* b) {
+static int SSE16x8(const uint8_t* a, const uint8_t* b) {
   return GetSSE(a, b, 16, 8);
 }
-static int SSE8x8_C(const uint8_t* a, const uint8_t* b) {
+static int SSE8x8(const uint8_t* a, const uint8_t* b) {
   return GetSSE(a, b, 8, 8);
 }
-static int SSE4x4_C(const uint8_t* a, const uint8_t* b) {
+static int SSE4x4(const uint8_t* a, const uint8_t* b) {
   return GetSSE(a, b, 4, 4);
 }
-#endif  // !WEBP_NEON_OMIT_C_CODE
 
-static void Mean16x4_C(const uint8_t* ref, uint32_t dc[4]) {
+static void Mean16x4(const uint8_t* ref, uint32_t dc[4]) {
   int k, x, y;
   for (k = 0; k < 4; ++k) {
     uint32_t avg = 0;
@@ -583,7 +571,6 @@ static void Mean16x4_C(const uint8_t* ref, uint32_t dc[4]) {
 // We try to match the spectral content (weighted) between source and
 // reconstructed samples.
 
-#if !WEBP_NEON_OMIT_C_CODE
 // Hadamard transform
 // Returns the weighted sum of the absolute value of transformed coefficients.
 // w[] contains a row-major 4 by 4 symmetric matrix.
@@ -621,25 +608,24 @@ static int TTransform(const uint8_t* in, const uint16_t* w) {
   return sum;
 }
 
-static int Disto4x4_C(const uint8_t* const a, const uint8_t* const b,
+static int Disto4x4(const uint8_t* const a, const uint8_t* const b,
-                      const uint16_t* const w) {
+                    const uint16_t* const w) {
   const int sum1 = TTransform(a, w);
   const int sum2 = TTransform(b, w);
   return abs(sum2 - sum1) >> 5;
 }
 
-static int Disto16x16_C(const uint8_t* const a, const uint8_t* const b,
+static int Disto16x16(const uint8_t* const a, const uint8_t* const b,
-                        const uint16_t* const w) {
+                      const uint16_t* const w) {
   int D = 0;
   int x, y;
   for (y = 0; y < 16 * BPS; y += 4 * BPS) {
     for (x = 0; x < 16; x += 4) {
-      D += Disto4x4_C(a + x + y, b + x + y, w);
+      D += Disto4x4(a + x + y, b + x + y, w);
     }
   }
   return D;
 }
-#endif  // !WEBP_NEON_OMIT_C_CODE
 
 //------------------------------------------------------------------------------
 // Quantization
@@ -650,8 +636,8 @@ static const uint8_t kZigzag[16] = {
 };
 
 // Simple quantization
-static int QuantizeBlock_C(int16_t in[16], int16_t out[16],
+static int QuantizeBlock(int16_t in[16], int16_t out[16],
-                           const VP8Matrix* const mtx) {
+                         const VP8Matrix* const mtx) {
   int last = -1;
   int n;
   for (n = 0; n < 16; ++n) {
@@ -676,15 +662,13 @@ static int QuantizeBlock_C(int16_t in[16], int16_t out[16],
   return (last >= 0);
 }
 
-#if !WEBP_NEON_OMIT_C_CODE || WEBP_NEON_WORK_AROUND_GCC
+static int Quantize2Blocks(int16_t in[32], int16_t out[32],
-static int Quantize2Blocks_C(int16_t in[32], int16_t out[32],
+                           const VP8Matrix* const mtx) {
-                             const VP8Matrix* const mtx) {
   int nz;
   nz  = VP8EncQuantizeBlock(in + 0 * 16, out + 0 * 16, mtx) << 0;
   nz |= VP8EncQuantizeBlock(in + 1 * 16, out + 1 * 16, mtx) << 1;
   return nz;
 }
-#endif  // !WEBP_NEON_OMIT_C_CODE || WEBP_NEON_WORK_AROUND_GCC
 
 //------------------------------------------------------------------------------
 // Block copy
@@ -698,11 +682,11 @@ static WEBP_INLINE void Copy(const uint8_t* src, uint8_t* dst, int w, int h) {
   }
 }
 
-static void Copy4x4_C(const uint8_t* src, uint8_t* dst) {
+static void Copy4x4(const uint8_t* src, uint8_t* dst) {
   Copy(src, dst, 4, 4);
 }
 
-static void Copy16x8_C(const uint8_t* src, uint8_t* dst) {
+static void Copy16x8(const uint8_t* src, uint8_t* dst) {
   Copy(src, dst, 16, 8);
 }
 
@@ -740,37 +724,36 @@ extern void VP8EncDspInitMIPS32(void);
 extern void VP8EncDspInitMIPSdspR2(void);
 extern void VP8EncDspInitMSA(void);
 
-WEBP_DSP_INIT_FUNC(VP8EncDspInit) {
+static volatile VP8CPUInfo enc_last_cpuinfo_used =
+    (VP8CPUInfo)&enc_last_cpuinfo_used;
+
+WEBP_TSAN_IGNORE_FUNCTION void VP8EncDspInit(void) {
+  if (enc_last_cpuinfo_used == VP8GetCPUInfo) return;
+
   VP8DspInit();  // common inverse transforms
   InitTables();
 
   // default C implementations
-#if !WEBP_NEON_OMIT_C_CODE
+  VP8CollectHistogram = CollectHistogram;
-  VP8ITransform = ITransform_C;
+  VP8ITransform = ITransform;
-  VP8FTransform = FTransform_C;
+  VP8FTransform = FTransform;
-  VP8FTransformWHT = FTransformWHT_C;
+  VP8FTransform2 = FTransform2;
-  VP8TDisto4x4 = Disto4x4_C;
+  VP8FTransformWHT = FTransformWHT;
-  VP8TDisto16x16 = Disto16x16_C;
+  VP8EncPredLuma4 = Intra4Preds;
-  VP8CollectHistogram = CollectHistogram_C;
+  VP8EncPredLuma16 = Intra16Preds;
-  VP8SSE16x16 = SSE16x16_C;
+  VP8EncPredChroma8 = IntraChromaPreds;
-  VP8SSE16x8 = SSE16x8_C;
+  VP8SSE16x16 = SSE16x16;
-  VP8SSE8x8 = SSE8x8_C;
+  VP8SSE8x8 = SSE8x8;
-  VP8SSE4x4 = SSE4x4_C;
+  VP8SSE16x8 = SSE16x8;
-#endif
+  VP8SSE4x4 = SSE4x4;
-
+  VP8TDisto4x4 = Disto4x4;
-#if !WEBP_NEON_OMIT_C_CODE || WEBP_NEON_WORK_AROUND_GCC
+  VP8TDisto16x16 = Disto16x16;
-  VP8EncQuantizeBlock = QuantizeBlock_C;
+  VP8Mean16x4 = Mean16x4;
-  VP8EncQuantize2Blocks = Quantize2Blocks_C;
+  VP8EncQuantizeBlock = QuantizeBlock;
-#endif
+  VP8EncQuantize2Blocks = Quantize2Blocks;
-
+  VP8EncQuantizeBlockWHT = QuantizeBlock;
-  VP8FTransform2 = FTransform2_C;
+  VP8Copy4x4 = Copy4x4;
-  VP8EncPredLuma4 = Intra4Preds_C;
+  VP8Copy16x8 = Copy16x8;
-  VP8EncPredLuma16 = Intra16Preds_C;
-  VP8EncPredChroma8 = IntraChromaPreds_C;
-  VP8Mean16x4 = Mean16x4_C;
-  VP8EncQuantizeBlockWHT = QuantizeBlock_C;
-  VP8Copy4x4 = Copy4x4_C;
-  VP8Copy16x8 = Copy16x8_C;
 
   // If defined, use CPUInfo() to overwrite some pointers with faster versions.
   if (VP8GetCPUInfo != NULL) {
@@ -789,6 +772,11 @@ WEBP_DSP_INIT_FUNC(VP8EncDspInit) {
       VP8EncDspInitAVX2();
     }
 #endif
+#if defined(WEBP_USE_NEON)
+    if (VP8GetCPUInfo(kNEON)) {
+      VP8EncDspInitNEON();
+    }
+#endif
 #if defined(WEBP_USE_MIPS32)
     if (VP8GetCPUInfo(kMIPS32)) {
       VP8EncDspInitMIPS32();
@@ -805,32 +793,5 @@ WEBP_DSP_INIT_FUNC(VP8EncDspInit) {
     }
 #endif
   }
-
+  enc_last_cpuinfo_used = VP8GetCPUInfo;
-#if defined(WEBP_USE_NEON)
-  if (WEBP_NEON_OMIT_C_CODE ||
-      (VP8GetCPUInfo != NULL && VP8GetCPUInfo(kNEON))) {
-    VP8EncDspInitNEON();
-  }
-#endif
-
-  assert(VP8ITransform != NULL);
-  assert(VP8FTransform != NULL);
-  assert(VP8FTransformWHT != NULL);
-  assert(VP8TDisto4x4 != NULL);
-  assert(VP8TDisto16x16 != NULL);
-  assert(VP8CollectHistogram != NULL);
-  assert(VP8SSE16x16 != NULL);
-  assert(VP8SSE16x8 != NULL);
-  assert(VP8SSE8x8 != NULL);
-  assert(VP8SSE4x4 != NULL);
-  assert(VP8EncQuantizeBlock != NULL);
-  assert(VP8EncQuantize2Blocks != NULL);
-  assert(VP8FTransform2 != NULL);
-  assert(VP8EncPredLuma4 != NULL);
-  assert(VP8EncPredLuma16 != NULL);
-  assert(VP8EncPredChroma8 != NULL);
-  assert(VP8Mean16x4 != NULL);
-  assert(VP8EncQuantizeBlockWHT != NULL);
-  assert(VP8Copy4x4 != NULL);
-  assert(VP8Copy16x8 != NULL);
 }