Merge pull request #261113 from lilyinstarlight/fix/mupdf-hell

pkgs/applications/misc/llpp/default.nix

···
 
   patches = [
     (fetchpatch {
-      name = "system-makedeps.patch";
-      url = "https://aur.archlinux.org/cgit/aur.git/plain/system-makedeps.patch?h=llpp&id=0d2913056aaf3dbf7431e57b7b08b55568ba076c";
-      hash = "sha256-t9PLXsM8+exCeYqJBe0LSDK0D2rpktmozS8qNcEAcHo=";
+      name = "system-makedeps-and-ocaml5.patch";
+      url = "https://aur.archlinux.org/cgit/aur.git/plain/system-makedeps-and-ocaml5.patch?h=llpp&id=32955e115f914bb96348d288f9af9c6e3e80a02b";
+      hash = "sha256-3rcPsR+M8Jx7M8GHUIsw0WNBvp6aE7BcPr4yk2vT9Ik=";
     })
-    ./fix-mupdf.patch
   ];
 
   postPatch = ''

pkgs/applications/misc/llpp/fix-mupdf.patch

···
---- a/link.c
-+++ b/link.c
-@@ -1522,8 +1522,9 @@ static void *mainloop (void UNUSED_ATTR *unused)
-             if (pdf && nameddest && *nameddest) {
-                 fz_point xy;
-                 struct pagedim *pdim;
--                int pageno = pdf_lookup_anchor (state.ctx, pdf, nameddest,
-+                fz_location location = fz_resolve_link (state.ctx, state.doc, nameddest,
-                                                 &xy.x, &xy.y);
-+                int pageno = location.page;
-                 pdim = pdimofpageno (pageno);
-                 xy = fz_transform_point (xy, pdim->ctm);
-                 printd ("a %d %d %d", pageno, (int) xy.x, (int) xy.y);

pkgs/applications/misc/mupdf/0003-Fix-cpp-build.patch

···
+diff --git a/scripts/wrap/cpp.py b/scripts/wrap/cpp.py
+index 51ac5f1..b5c0b5a 100644
+--- a/scripts/wrap/cpp.py
++++ b/scripts/wrap/cpp.py
+@@ -4595,7 +4595,7 @@ def cpp_source(
+                     */
+                     typedef unsigned long size_t;
+                     '''))
+-            if state.state_.macos:
++            if state.state_.linux or state.state_.macos:
+                 f.write( textwrap.dedent('''
+                     /*
+                     Workaround on MacOS: we need to define fixed-size int types

pkgs/applications/misc/mupdf/default.nix

···
 { stdenv
 , lib
 , fetchurl
-, fetchpatch
 , fetchFromGitHub
 , copyDesktopItems
 , makeDesktopItem
···
 , enableGL ? true
 , freeglut
 , libGLU
+, enableOcr ? false
+, leptonica
+, tesseract
+, enableCxx ? false
+, python3
+, enablePython ? false
+, which
+, swig
 , xcbuild
 , gitUpdater
 
 # for passthru.tests
 , cups-filters
-, python3
 , zathura
+, mupdf
 }:
-let
 
-  # OpenJPEG version is hardcoded in package source
-  openJpegVersion = with stdenv;
-    lib.versions.majorMinor (lib.getVersion openjpeg);
+assert enablePython -> enableCxx;
+
+let
 
   freeglut-mupdf = freeglut.overrideAttrs (old: rec {
     pname = "freeglut-mupdf";
···
 
 in
 stdenv.mkDerivation rec {
-  version = "1.23.0";
+  version = "1.23.5";
   pname = "mupdf";
 
   src = fetchurl {
     url = "https://mupdf.com/downloads/archive/${pname}-${version}-source.tar.gz";
-    sha256 = "sha256-3kFAaS5pMULDEeAwrBVuOO4XXXq2wb4QxcmuljhGFk4=";
+    sha256 = "sha256-blZ5zfqu+cfoniljlSIM4sEz7T3K1RpHhmczbG6uxwY=";
   };
 
   patches = [ ./0001-Use-command-v-in-favor-of-which.patch
               ./0002-Add-Darwin-deps.patch
+              ./0003-Fix-cpp-build.patch
             ];
 
   postPatch = ''
-    sed -i "s/__OPENJPEG__VERSION__/${openJpegVersion}/" source/fitz/load-jpx.c
     substituteInPlace Makerules --replace "(shell pkg-config" "(shell $PKG_CONFIG"
+
+    patchShebangs scripts/mupdfwrap.py
+
+    # slip in makeFlags when building bindings
+    sed -i -e 's/^\( *make_args *=\)/\1 """ $(echo ''${makeFlagsArray[@]@Q})"""/' scripts/wrap/__main__.py
+
+    # fix libclang unnamed struct format
+    for wrapper in ./scripts/wrap/{cpp,state}.py; do
+      substituteInPlace "$wrapper" --replace 'struct (unnamed' '(unnamed struct'
+    done
   '';
 
-  # Use shared libraries to decrease size
-  buildFlags = [ "shared" ];
-
   makeFlags = [
     "prefix=$(out)"
+    "shared=yes"
     "USE_SYSTEM_LIBS=yes"
     "PKG_CONFIG=${buildPackages.pkg-config}/bin/${buildPackages.pkg-config.targetPrefix}pkg-config"
   ] ++ lib.optionals (!enableX11) [ "HAVE_X11=no" ]
-    ++ lib.optionals (!enableGL) [ "HAVE_GLUT=no" ];
+    ++ lib.optionals (!enableGL) [ "HAVE_GLUT=no" ]
+    ++ lib.optionals (enableOcr) [ "USE_TESSERACT=yes" ];
 
   nativeBuildInputs = [ pkg-config ]
     ++ lib.optional (enableGL || enableX11) copyDesktopItems
+    ++ lib.optionals (enableCxx || enablePython) [ python3 python3.pkgs.setuptools python3.pkgs.libclang ]
+    ++ lib.optionals (enablePython) [ which swig ]
     ++ lib.optional stdenv.isDarwin desktopToDarwinBundle;
 
   buildInputs = [ freetype harfbuzz openjpeg jbig2dec libjpeg gumbo ]
···
     ++ lib.optionals enableX11 [ libX11 libXext libXi libXrandr ]
     ++ lib.optionals enableCurl [ curl openssl ]
     ++ lib.optionals enableGL (
-    if stdenv.isDarwin then
-      with darwin.apple_sdk.frameworks; [ GLUT OpenGL ]
-    else
-      [ freeglut-mupdf libGLU ]
-  )
+      if stdenv.isDarwin then
+        with darwin.apple_sdk.frameworks; [ GLUT OpenGL ]
+      else
+        [ freeglut-mupdf libGLU ]
+    )
+    ++ lib.optionals enableOcr [ leptonica tesseract ]
   ;
   outputs = [ "bin" "dev" "out" "man" "doc" ];
 
···
     rm -rf thirdparty/{curl,freetype,glfw,harfbuzz,jbig2dec,libjpeg,openjpeg,zlib}
   '';
 
+  postBuild = lib.optionalString (enableCxx || enablePython) ''
+    for dir in build/*; do
+      ./scripts/mupdfwrap.py -d "$dir" -b ${lib.optionalString (enableCxx) "01"}${lib.optionalString (enablePython) "23"}
+    done
+  '';
+
   desktopItems = [
     (makeDesktopItem {
       name = pname;
···
     Name: mupdf
     Description: Library for rendering PDF documents
     Version: ${version}
-    Libs: -L$out/lib -lmupdf -lmupdf-third
+    Libs: -L$out/lib -lmupdf
     Cflags: -I$dev/include
     EOF
 
···
     ln -s "$bin/bin/mupdf-gl" "$bin/bin/mupdf"
   '' else lib.optionalString (enableX11) ''
     ln -s "$bin/bin/mupdf-x11" "$bin/bin/mupdf"
-  '');
+  '') + (lib.optionalString (enableCxx) ''
+    cp platform/c++/include/mupdf/*.h $out/include/mupdf
+    cp build/*/libmupdfcpp.so $out/lib
+  '') + (lib.optionalString (enablePython) (''
+    mkdir -p $out/${python3.sitePackages}/mupdf
+    cp build/*/_mupdf.so $out/${python3.sitePackages}
+    cp build/*/mupdf.py $out/${python3.sitePackages}/mupdf/__init__.py
+  '' + lib.optionalString (stdenv.isDarwin) ''
+    install_name_tool -add_rpath $out/lib $out/${python3.sitePackages}/_mupdf.so
+  ''));
 
   enableParallelBuilding = true;
 
···
     tests = {
       inherit cups-filters zathura;
       inherit (python3.pkgs) pikepdf pymupdf;
+      mupdf-all = mupdf.override { enableCurl = true; enableGL = true; enableOcr = true; enableCxx = true; enablePython = true; };
     };
 
     updateScript = gitUpdater {
···
     description = "Lightweight PDF, XPS, and E-book viewer and toolkit written in portable C";
     changelog = "https://git.ghostscript.com/?p=mupdf.git;a=blob_plain;f=CHANGES;hb=${version}";
     license = licenses.agpl3Plus;
-    maintainers = with maintainers; [ vrthra fpletz ];
+    maintainers = with maintainers; [ vrthra fpletz lilyinstarlight ];
     platforms = platforms.unix;
     mainProgram = "mupdf";
   };

pkgs/applications/misc/sioyek/default.nix

···
 
   postPatch = ''
     substituteInPlace pdf_viewer_build_config.pro \
-      --replace "-lmupdf-threads" "-lgumbo -lharfbuzz -lfreetype -ljbig2dec -ljpeg -lopenjp2"
+      --replace "-lmupdf-threads" "-lgumbo -lharfbuzz -lfreetype -ljbig2dec -ljpeg -lopenjp2" \
+      --replace "-lmupdf-third" ""
     substituteInPlace pdf_viewer/main.cpp \
       --replace "/usr/share/sioyek" "$out/share" \
       --replace "/etc/sioyek" "$out/etc"

pkgs/applications/misc/zathura/pdf-mupdf/default.nix

···
 
   PKG_CONFIG_ZATHURA_PLUGINDIR= "lib/zathura";
 
+  postPatch = ''
+    sed -i -e '/^mupdfthird =/d' -e 's/, mupdfthird//g' meson.build
+  '';
+
   passthru.updateScript = gitUpdater {
     url = "https://git.pwmt.org/pwmt/zathura-pdf-mupdf.git";
   };

pkgs/applications/office/beamerpresenter/default.nix

···
     "-DUSE_POPPLER=${if usePoppler then "ON" else "OFF"}"
     "-DUSE_MUPDF=${if useMupdf then "ON" else "OFF"}"
     "-DUSE_QTPDF=OFF"
-    "-DLINK_MUPDF_THIRD=ON"
+    "-DLINK_MUPDF_THIRD=OFF"
     "-DUSE_EXTERNAL_RENDERER=${if useExternalRenderer then "ON" else "OFF"}"
     "-DLINK_MUJS=OFF"
     "-DLINK_GUMBO=ON"

pkgs/desktops/arcan/arcan/default.nix

···
 { lib
 , stdenv
 , fetchFromGitHub
-, fetchgit
 , SDL2
 , cmake
 , espeak
···
 , libXfixes
 , libdrm
 , libffi
+, libjpeg
 , libusb1
 , libuvc
 , libvlc
 , libvncserver
 , libxcb
 , libxkbcommon
-, lua5_1
-, luajit
 , makeWrapper
 , mesa
 , mupdf
-, ninja
 , openal
 , openjpeg
 , pcre2
···
     libXfixes
     libdrm
     libffi
+    libjpeg
     libusb1
     libuvc
     libvlc
···
     libxcb
     libxkbcommon
     mesa
-    mupdf.dev
+    mupdf
     openal
-    openjpeg.dev
+    openjpeg
     pcre2
     sqlite
     tesseract

pkgs/development/python-modules/libclang/default.nix

···
+{ lib
+, buildPythonPackage
+, llvmPackages
+, setuptools
+, writeText
+}:
+
+let
+  libclang = llvmPackages.libclang;
+
+  pyproject_toml = writeText "pyproject.toml" ''
+    [build-system]
+    requires = ["setuptools>=42", "wheel"]
+    build-backend = "setuptools.build_meta"
+  '';
+
+  setup_cfg = writeText "setup.cfg" ''
+    [metadata]
+    name = clang
+    version = ${libclang.version}
+
+    [options]
+    packages = clang
+  '';
+in buildPythonPackage {
+  pname = "libclang";
+  format = "pyproject";
+
+  inherit (libclang) version src;
+
+  buildInputs = [ setuptools ];
+
+  postUnpack = ''
+    # set source root to python bindings
+    if [ -e "$sourceRoot/clang/bindings/python" ]; then
+      # LLVM 13+ puts clang sources in subdirectory instead of plain tarball
+      sourceRoot="$sourceRoot/clang/bindings/python"
+    else
+      sourceRoot="$sourceRoot/bindings/python"
+    fi
+  '';
+
+  postPatch = ''
+    # link in our own build info to build as a python package
+    ln -s ${pyproject_toml} ./pyproject.toml
+    ln -s ${setup_cfg} ./setup.cfg
+
+    # set passed libclang for runtime
+    echo 'Config.set_library_path("${lib.getLib libclang}/lib")' >>./clang/cindex.py
+  '';
+
+  meta = libclang.meta // {
+    description = "Python bindings for the C language family frontend for LLVM";
+    maintainers = with lib.maintainers; [ lilyinstarlight ];
+  };
+}

pkgs/development/python-modules/pymupdf/default.nix

···
 , stdenv
 , buildPythonPackage
 , pythonOlder
-, fetchPypi
+, fetchFromGitHub
+, pytestCheckHook
+, python
 , swig
-, xcbuild
 , mupdf
 , freetype
 , harfbuzz
···
 , libjpeg_turbo
 , gumbo
 , memstreamHook
+, fonttools
 }:
 
-buildPythonPackage rec {
+let
+  # PyMuPDF needs the C++ bindings generated
+  mupdf-cxx = mupdf.override { enableOcr = true; enableCxx = true; enablePython = true; };
+in buildPythonPackage rec {
   pname = "pymupdf";
-  version = "1.23.3";
+  version = "1.23.6";
   format = "setuptools";
 
   disabled = pythonOlder "3.7";
 
-  src = fetchPypi {
-    pname = "PyMuPDF";
-    inherit version;
-    hash = "sha256-AhR4rmx26IWSQdu5cGEskIColX2L1pe7oLRTHcHPT4c=";
+  src = fetchFromGitHub {
+    owner = "pymupdf";
+    repo = "PyMuPDF";
+    rev = version;
+    hash = "sha256-60KT5+EGP+s7HD4UIeaf9x2QVNU9IUbC5WKEJbrIBCI=";
   };
 
-  postPatch = ''
-    substituteInPlace setup.py \
-        --replace '/usr/include/mupdf' ${mupdf.dev}/include/mupdf
-  '';
   nativeBuildInputs = [
+    pytestCheckHook
     swig
-  ] ++ lib.optionals stdenv.isDarwin [
-    xcbuild
   ];
 
   buildInputs = [
-    mupdf
     freetype
     harfbuzz
     openjpeg
···
     memstreamHook
   ];
 
-  doCheck = false;
+  propagatedBuildInputs = [
+    mupdf-cxx
+  ];
+
+  env = {
+    # force using system MuPDF (must be defined in environment and empty)
+    PYMUPDF_SETUP_MUPDF_BUILD = "";
+    # provide MuPDF paths
+    PYMUPDF_MUPDF_LIB = "${lib.getLib mupdf-cxx}/lib";
+    PYMUPDF_MUPDF_INCLUDE = "${lib.getDev mupdf-cxx}/include";
+  };
+
+  # TODO: manually add mupdf rpath until upstream fixes it
+  postInstall = lib.optionalString stdenv.isDarwin ''
+    for lib in */*.so $out/${python.sitePackages}/*/*.so; do
+      install_name_tool -add_rpath ${lib.getLib mupdf-cxx}/lib "$lib"
+    done
+  '';
+
+  checkInputs = [
+    fonttools
+  ];
+
+  disabledTests = [
+    # fails for indeterminate reasons
+    "test_color_count"
+  ] ++ lib.optionals stdenv.isDarwin [
+    # darwin does not support OCR right now
+    "test_tesseract"
+  ];
 
   pythonImportsCheck = [
     "fitz"
+    "fitz_new"
   ];
 
   meta = with lib; {

pkgs/development/tools/hotdoc/default.nix

···
 , setuptools
 , toposort
 , wheezy-template
-, libclang
+, llvmPackages
 , gst_all_1
 }:
 
···
   postPatch = ''
     substituteInPlace hotdoc/extensions/c/c_extension.py \
       --replace "shutil.which('llvm-config')" 'True' \
-      --replace "subprocess.check_output(['llvm-config', '--version']).strip().decode()" '"${libclang.version}"' \
-      --replace "subprocess.check_output(['llvm-config', '--prefix']).strip().decode()" '"${libclang.lib}"' \
-      --replace "subprocess.check_output(['llvm-config', '--libdir']).strip().decode()" '"${libclang.lib}/lib"'
+      --replace "subprocess.check_output(['llvm-config', '--version']).strip().decode()" '"${llvmPackages.libclang.version}"' \
+      --replace "subprocess.check_output(['llvm-config', '--prefix']).strip().decode()" '"${llvmPackages.libclang.lib}"' \
+      --replace "subprocess.check_output(['llvm-config', '--libdir']).strip().decode()" '"${llvmPackages.libclang.lib}/lib"'
   '';
 
   # Make pytest run from a temp dir to have it pick up installed package for cmark

pkgs/top-level/python-packages.nix

···
     inherit (pkgs) libasyncns;
   };
 
+  libclang = callPackage ../development/python-modules/libclang { };
+
   libcloud = callPackage ../development/python-modules/libcloud { };
 
   libcst = callPackage ../development/python-modules/libcst { };