ccextractor: 0.93 -> 0.94-unstable-2024-08-12

  lib,

  stdenv,

  fetchFromGitHub,

  writeTextFile,

  pkg-config,

  cmake,

  libiconv,

  ninja,

  cargo,

  rustc,

  corrosion,

  rustPlatform,

  gpac,

  protobufc,

  libpng,

  zlib,

  utf8proc,

  freetype,

  ffmpeg_7,

  libarchive,

  curl,

  libiconv,

  enableOcr ? true,

  makeWrapper,

  tesseract4,

  leptonica,

  ffmpeg_4,

  tesseract,

}:

stdenv.mkDerivation (finalAttrs: {

  pname = "ccextractor";

  version = "0.93";

  version = "0.94-unstable-2024-08-12";

  src = fetchFromGitHub {

    owner = "CCExtractor";

    repo = "ccextractor";

    rev = "v${finalAttrs.version}";

    hash = "sha256-usVAKBkdd8uz9cD5eLd0hnwGonOJLscRdc+iWDlNXVc=";

    rev = "92f2ce0fa026b01fb07db6751210e6bd8c8944d3";

    hash = "sha256-bp7T9uJK4bauR2Co4lKqqnM6oGa3WZ+1toEKmzOx4mI=";

  };

  postPatch =

    ''

      # https://github.com/CCExtractor/ccextractor/issues/1467

      sed -i '/allheaders.h/a#include <leptonica/pix_internal.h>' src/lib_ccx/ocr.c

    ''

    + lib.optionalString stdenv.isDarwin ''

      substituteInPlace src/CMakeLists.txt \

      --replace 'add_definitions(-DGPAC_CONFIG_LINUX)' 'add_definitions(-DGPAC_CONFIG_DARWIN)'

    '';

  patches = [

    ./remove-default-commit-hash.patch

    ./remove-vendored-libraries.patch

  ] ++ finalAttrs.cargoDeps.patches;

  cmakeDir = "../src";

  cargoRoot = "src/rust";

  cargoDeps = rustPlatform.fetchCargoTarball {

    inherit (finalAttrs) src;

    sourceRoot = "${finalAttrs.src.name}/${finalAttrs.cargoRoot}";

    patches = [ ./use-rsmpeg-0.15.patch ];

    patchFlags = [ "-p3" ];

    hash = "sha256-jh8hHKAad+tCJGwuGdoJp/TMm/IsMrZmz8aag9lj0BA=";

  };

  nativeBuildInputs = [

    pkg-config

    cmake

    makeWrapper

    ninja

    cargo

    rustc

    corrosion

    rustPlatform.cargoSetupHook

    rustPlatform.bindgenHook

  ];

  buildInputs =

    [ zlib ]

    ++ lib.optional (!stdenv.isLinux) libiconv

    [

      gpac

      protobufc

      libpng

      zlib

      utf8proc

      freetype

      ffmpeg_7

      libarchive

      curl

      libiconv

    ]

    ++ lib.optionals enableOcr [

      leptonica

      tesseract4

      ffmpeg_4

      tesseract

    ];

  cmakeFlags =

    [

      # file RPATH_CHANGE could not write new RPATH:

      "-DCMAKE_SKIP_BUILD_RPATH=ON"

      # The tests are all part of one `cargo test` invocation, so let’s

      # get the output from it.

      (lib.cmakeFeature "CMAKE_CTEST_ARGUMENTS" "--verbose")

      # TODO: This (and the corresponding patch) should probably be

      # removed for the next stable release.

      (lib.cmakeFeature "GIT_COMMIT_HASH" finalAttrs.src.rev)

    ]

    ++ lib.optionals enableOcr [

      "-DWITH_OCR=on"

      "-DWITH_HARDSUBX=on"

      (lib.cmakeBool "WITH_OCR" true)

      (lib.cmakeBool "WITH_HARDSUBX" true)

    ];

  postInstall = lib.optionalString enableOcr ''

    wrapProgram "$out/bin/ccextractor" \

      --set TESSDATA_PREFIX "${tesseract4}/share/"

  env = {

    FFMPEG_INCLUDE_DIR = "${lib.getDev ffmpeg_7}/include";

    # Upstream’s FFmpeg binding crate needs an explicit path to a shared

    # object to do dynamic linking. The key word is *an* explicit path;

    # they don’t support passing more than one. This linker script hack

    # pulls in all the FFmpeg libraries they bind to.

    #

    # See: <https://github.com/CCExtractor/rusty_ffmpeg/pull/69>

    FFMPEG_DLL_PATH =

      let

        ffmpegLibNames = [

          "avcodec"

          "avdevice"

          "avfilter"

          "avformat"

          "avutil"

          "swresample"

          "swscale"

        ];

        ffmpegLibDir = "${lib.getLib ffmpeg_7}/lib";

        ffmpegLibExt = stdenv.hostPlatform.extensions.library;

        ffmpegLibPath = ffmpegLibName: "${ffmpegLibDir}/lib${ffmpegLibName}.${ffmpegLibExt}";

        ffmpegLinkerScript = writeTextFile {

          name = "ccextractor-ffmpeg-linker-script";

          destination = "/lib/ffmpeg.ld";

          text = "INPUT(${lib.concatMapStringsSep " " ffmpegLibPath ffmpegLibNames})";

        };

      in

      "${ffmpegLinkerScript}/lib/ffmpeg.ld";

  };

  doCheck = true;

  postPatch = lib.optionalString enableOcr ''

    substituteInPlace src/lib_ccx/ocr.c \

      --replace-fail 'getenv("TESSDATA_PREFIX")' '"${tesseract}/share"'

  '';

  meta = with lib; {

    homepage = "https://www.ccextractor.org";

  meta = {

    homepage = "https://www.ccextractor.org/";

    changelog = "${finalAttrs.src.meta.homepage}/blob/${finalAttrs.src.rev}/docs/CHANGES.TXT";

    description = "Tool that produces subtitles from closed caption data in videos";

    longDescription = ''

      A tool that analyzes video files and produces independent subtitle files from

      It works on Linux, Windows, and OSX.

    '';

    platforms = lib.platforms.unix;

    # undefined reference to `png_do_expand_palette_rgba8_neon'

    # undefined reference to `png_riffle_palette_neon'

    # undefined reference to `png_do_expand_palette_rgb8_neon'

    # undefined reference to `png_init_filter_functions_neon'

    # during Linking C executable ccextractor

    broken = stdenv.isAarch64;

    sourceProvenance = [ lib.sourceTypes.fromSource ];

    license = lib.licenses.gpl2Only;

    maintainers = [ lib.maintainers.emily ];

    mainProgram = "ccextractor";

diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt

index d7fdda02e3...2738cab631 100644

--- a/src/CMakeLists.txt

+++ b/src/CMakeLists.txt

@@ -24,9 +24,6 @@

             OUTPUT_VARIABLE GIT_COMMIT_HASH

             OUTPUT_STRIP_TRAILING_WHITESPACE

     )

-ELSE(EXISTS "${BASE_PROJ_DIR}/.git")

-    set(GIT_BRANCH "Unknown")

-    set(GIT_COMMIT_HASH "Unknown")

 ENDIF(EXISTS "${BASE_PROJ_DIR}/.git")

 #Get the date

diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt

index 2738cab631...5bb2b7d17a 100644

--- a/src/CMakeLists.txt

+++ b/src/CMakeLists.txt

@@ -48,93 +48,20 @@

 include_directories(${PROJECT_SOURCE_DIR})

 include_directories(${PROJECT_SOURCE_DIR}/lib_ccx)

 include_directories(${PROJECT_SOURCE_DIR}/lib_ccx/zvbi)

-include_directories(${PROJECT_SOURCE_DIR}/thirdparty)

-include_directories(${PROJECT_SOURCE_DIR}/thirdparty/protobuf-c)

 include_directories(${PROJECT_SOURCE_DIR}/thirdparty/lib_hash)

-include_directories(${PROJECT_SOURCE_DIR}/thirdparty/libpng)

-# Check if the operating system is macOS (Darwin)

-if (${CMAKE_SYSTEM_NAME} MATCHES "Darwin")

-  if(${CMAKE_HOST_SYSTEM_PROCESSOR} MATCHES "arm64")

-    # ARM Macs

-    include_directories("/opt/homebrew/include")

-    include_directories(${PROJECT_SOURCE_DIR}/thirdparty/libpng/arm)

-    aux_source_directory(${PROJECT_SOURCE_DIR}/thirdparty/libpng/arm SOURCEFILE)

-  else()

-    include_directories("/usr/local/include")

-  endif()

-endif()

-

-include_directories(${PROJECT_SOURCE_DIR}/thirdparty/zlib)

-include_directories(${PROJECT_SOURCE_DIR}/thirdparty/freetype/include)

 aux_source_directory(${PROJECT_SOURCE_DIR}/thirdparty/lib_hash/ SOURCEFILE)

-aux_source_directory(${PROJECT_SOURCE_DIR}/thirdparty/libpng/ SOURCEFILE)

-aux_source_directory(${PROJECT_SOURCE_DIR}/thirdparty/protobuf-c/ SOURCEFILE)

-aux_source_directory(${PROJECT_SOURCE_DIR}/thirdparty/zlib/ SOURCEFILE)

 aux_source_directory(${PROJECT_SOURCE_DIR}/lib_ccx/zvbi/ SOURCEFILE)

-set(UTF8PROC_SOURCE ${PROJECT_SOURCE_DIR}/thirdparty/utf8proc/utf8proc.c)

+set(UTF8PROC_SOURCE)

-set(FREETYPE_SOURCE

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/autofit/autofit.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/base/ftbase.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/base/ftbbox.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/base/ftbdf.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/base/ftbitmap.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/base/ftcid.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/base/ftfntfmt.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/base/ftfstype.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/base/ftgasp.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/base/ftglyph.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/base/ftgxval.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/base/ftinit.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/base/ftlcdfil.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/base/ftmm.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/base/ftotval.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/base/ftpatent.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/base/ftpfr.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/base/ftstroke.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/base/ftsynth.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/base/ftsystem.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/base/fttype1.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/base/ftwinfnt.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/bdf/bdf.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/bzip2/ftbzip2.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/cache/ftcache.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/cff/cff.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/cid/type1cid.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/gzip/ftgzip.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/lzw/ftlzw.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/pcf/pcf.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/pfr/pfr.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/psaux/psaux.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/pshinter/pshinter.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/psnames/psnames.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/raster/raster.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/sfnt/sfnt.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/smooth/smooth.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/truetype/truetype.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/type1/type1.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/type42/type42.c

-        ${PROJECT_SOURCE_DIR}/thirdparty/freetype/winfonts/winfnt.c

-        )

+set(FREETYPE_SOURCE)

 #Windows specific libraries and linker flags

 if(WIN32)

     include_directories ("${PROJECT_SOURCE_DIR}/thirdparty/win_spec_incld/")

     include_directories ("${PROJECT_SOURCE_DIR}/thirdparty/win_iconv/")

     aux_source_directory ("${PROJECT_SOURCE_DIR}/thirdparty/win_iconv/" SOURCEFILE)

     set (EXTRA_LIBS ${EXTRA_LIBS} ws2_32 winmm Bcrypt)

-else (WIN32)

-    # Adding some platform specific library path

-    if(UNIX AND NOT APPLE)

-        link_directories (/usr/local/lib)

-    endif()

-

-    if(APPLE)

-        # Homebrew library paths

-        link_directories(/usr/local/lib)

-        link_directories(/opt/homebrew/lib)

-    endif()

 endif(WIN32)

 if(MSVC)

@@ -212,9 +139,6 @@

   pkg_check_modules (NANOMSG REQUIRED libnanomsg)

   set (EXTRA_LIBS ${EXTRA_LIBS} ${NANOMSG_STATIC_LIBRARIES})

-  include_directories ("${PROJECT_SOURCE_DIR}/thirdparty/protobuf-c/")

-  aux_source_directory ("${PROJECT_SOURCE_DIR}/thirdparty/protobuf-c/" SOURCEFILE)

-

   set (CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -DENABLE_SHARING")

 endif (PKG_CONFIG_FOUND AND WITH_SHARING)

diff --git a/src/lib_ccx/CMakeLists.txt b/src/lib_ccx/CMakeLists.txt

index 4f329bcaab...a334d20c4d 100644

--- a/src/lib_ccx/CMakeLists.txt

+++ b/src/lib_ccx/CMakeLists.txt

@@ -13,9 +13,39 @@

 find_package(PkgConfig)

 pkg_check_modules (GPAC REQUIRED gpac)

+set (REQUIRES_PRIVATE "libpng libprotobuf-c")

+

+if (WITH_FFMPEG)

+  set (REQUIRES_PRIVATE "${REQUIRES_PRIVATE} libavutil")

+endif (WITH_FFMPEG)

+

+if (WITH_HARDSUBX)

+  set (REQUIRES_PRIVATE "${REQUIRES_PRIVATE} libavcodec libavformat libswscale tesseract lept")

+endif (WITH_HARDSUBX)

+

 set (EXTRA_INCLUDES ${EXTRA_INCLUDES} ${GPAC_INCLUDE_DIRS})

 set (EXTRA_LIBS ${EXTRA_LIBS} ${GPAC_LIBRARIES})

+pkg_check_modules (PROTOBUFC REQUIRED libprotobuf-c)

+set (EXTRA_LIBS ${EXTRA_LIBS} ${PROTOBUFC_LIBRARIES})

+set (EXTRA_INCLUDES ${EXTRA_INCLUDES} ${PROTOBUFC_INCLUDE_DIRS}/protobuf-c)

+

+pkg_check_modules (LIBPNG REQUIRED libpng)

+set (EXTRA_LIBS ${EXTRA_LIBS} ${LIBPNG_LIBRARIES})

+set (EXTRA_INCLUDES ${EXTRA_INCLUDES} ${LIBPNG_INCLUDE_DIRS})

+

+pkg_check_modules (ZLIB REQUIRED zlib)

+set (EXTRA_LIBS ${EXTRA_LIBS} ${ZLIB_LIBRARIES})

+set (EXTRA_INCLUDES ${EXTRA_INCLUDES} ${ZLIB_INCLUDE_DIRS})

+

+pkg_check_modules (UTF8PROC REQUIRED libutf8proc)

+set (EXTRA_LIBS ${EXTRA_LIBS} ${UTF8PROC_LIBRARIES})

+set (EXTRA_INCLUDES ${EXTRA_INCLUDES} ${UTF8PROC_INCLUDE_DIRS})

+

+pkg_check_modules (FREETYPE REQUIRED freetype2)

+set (EXTRA_LIBS ${EXTRA_LIBS} ${FREETYPE_LIBRARIES})

+set (EXTRA_INCLUDES ${EXTRA_INCLUDES} ${FREETYPE_INCLUDE_DIRS})

+

 if (WITH_FFMPEG)

     find_package(PkgConfig)

@@ -94,7 +124,7 @@

     set (CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -DDISABLE_RUST")

 endif (WITHOUT_RUST)

-file (GLOB HeaderFiles *.h)

+file (GLOB_RECURSE HeaderFiles *.h)

 file (WRITE ccx.pc "prefix=${CMAKE_INSTALL_PREFIX}\n"

                   "includedir=\${prefix}/include\n"

                   "libdir=\${prefix}/lib\n\n"

@@ -102,8 +132,8 @@

                   "Description: Closed Caption Extraction library\n"

                   "Version: 0.75\n"

                   "Cflags: -I\${includedir}/\n"

-                  "Libs: -L\${libdir} -lccx -lpng\n"

-                  "Libs.private: -lpng\n"

+                  "Libs: -L\${libdir} -lccx\n"

+                  "Requires.private: ${REQUIRES_PRIVATE}\n"

                   )

 install (TARGETS ccx DESTINATION lib)

diff --git a/src/lib_ccx/params.c b/src/lib_ccx/params.c

index eb1562e50c...984070a285 100644

--- a/src/lib_ccx/params.c

+++ b/src/lib_ccx/params.c

@@ -14,7 +14,7 @@

 #include "../lib_hash/sha2.h"

 #include <string.h>

 #include <stdio.h>

-#include <utf8proc/utf8proc.h>

+#include <utf8proc.h>

 #ifdef ENABLE_OCR

 #include <tesseract/capi.h>

diff --git a/src/rust/Cargo.lock b/src/rust/Cargo.lock

index 5c49573775..3e855aa637 100644

--- a/src/rust/Cargo.lock

+++ b/src/rust/Cargo.lock

@@ -665,11 +665,10 @@

 [[package]]

 name = "rsmpeg"

-version = "0.14.2+ffmpeg.6.1"

+version = "0.15.1+ffmpeg.7.0"

 source = "registry+https://github.com/rust-lang/crates.io-index"

-checksum = "927012cd6ae43519f519741f4a69602ce3a47cf84750784da124dffd03527cc0"

+checksum = "d3ffbead667d06e0c77c4363f83d49a3481cc3838bc9a61882aa07b01e3f63e1"

 dependencies = [

- "libc",

  "paste",

  "rusty_ffmpeg",

  "thiserror",

@@ -711,9 +710,9 @@

 [[package]]

 name = "rusty_ffmpeg"

-version = "0.13.3+ffmpeg.6.1"

+version = "0.14.1+ffmpeg.7.0"

 source = "registry+https://github.com/rust-lang/crates.io-index"

-checksum = "716adffa5f909c8533611b1dab9ab5666bece35687845865b75ed6a990fc239c"

+checksum = "40f4db8e3e23d4a3044d53a41aba5324eae70d3e7fe82375ce833521533bc315"

 dependencies = [

  "bindgen 0.69.4",

  "camino",

diff --git a/src/rust/Cargo.toml b/src/rust/Cargo.toml

index 4c1e73dcf0..68502915dc 100644

--- a/src/rust/Cargo.toml

+++ b/src/rust/Cargo.toml

@@ -15,7 +15,7 @@

 env_logger = "0.8.4"

 iconv = "0.1.1"

 palette = "0.6.0"

-rsmpeg = { version = "0.14.1", optional = true, features = [

+rsmpeg = { version = "0.15.1", optional = true, features = [

     "link_system_ffmpeg",

 ] }

 tesseract-sys = { version = "0.5.14", optional = true, default-features = false }