k2pdfopt: 2.32 -> 2.42

1 files changed, 85 insertions, 98 deletions

diff --git a/pkgs/applications/misc/k2pdfopt/default.nix b/pkgs/applications/misc/k2pdfopt/default.nix
index 7c0d615f3663..587e6a588a73 100644
--- a/pkgs/applications/misc/k2pdfopt/default.nix
+++ b/pkgs/applications/misc/k2pdfopt/default.nix
@@ -1,105 +1,92 @@
-# Build procedure lifted from https://aur.archlinux.org/packages/k2/k2pdfopt/PKGBUILD
-{ stdenv, fetchzip, fetchurl, writeScript, libX11, libXext, autoconf, automake, libtool
-      , leptonica, libpng, libtiff, zlib, openjpeg, freetype, jbig2dec, djvulibre
-      , openssl }:
-
-let
-  mupdf_src = fetchurl {
-    url = http://www.mupdf.com/downloads/archive/mupdf-1.6-source.tar.gz;
-    sha256 = "0qx51rj6alzcagcixm59rvdpm54w6syrwr4184v439jh14ryw4wq";
-  };
-
-  tess_src = fetchurl {
-    url = http://tesseract-ocr.googlecode.com/files/tesseract-ocr-3.02.02.tar.gz;
-    sha256 = "0g81m9y4iydp7kgr56mlkvjdwpp3mb01q385yhdnyvra7z5kkk96";
-  };
-
-  gocr_src = fetchurl {
-    url = http://www-e.uni-magdeburg.de/jschulen/ocr/gocr-0.49.tar.gz;
-    sha256 = "06hpzp7rkkwfr1fvmc8kcfz9v490i9yir7f7imh13gmka0fr6afc";
-  };
-
-in stdenv.mkDerivation rec {
+{ stdenv, fetchzip, fetchurl, fetchpatch, cmake, pkgconfig
+, zlib, libpng
+, enableGSL ? true, gsl
+, enableGhostScript ? true, ghostscript
+, enableMuPDF ? true, jbig2dec, openjpeg, freetype, harfbuzz, mupdf
+, enableJPEG2K ? true, jasper
+, enableDJVU ? true, djvulibre
+, enableGOCR ? false, gocr # Disabled by default due to crashes
+, enableTesseract ? true, leptonica, tesseract
+}:
+
+with stdenv.lib;
+
+stdenv.mkDerivation rec {
   name = "k2pdfopt-${version}";
-  version = "2.32";
+  version = "2.42";
+
   src = fetchzip {
     url = "http://www.willus.com/k2pdfopt/src/k2pdfopt_v${version}_src.zip";
-    sha256 = "1v3cj5bwpjvy7s66sfqcmkxs91f7nxaykjpdjm2wn87vn6q7n19m";
+    sha256 = "1zag4jmkr0qrcpqqb5davmvdrabhdyz87q4zz0xpfkl6xw2dn9bk";
   };
 
-  buildInputs = [ libX11 libXext autoconf automake libtool leptonica libpng libtiff zlib
-                    openjpeg freetype jbig2dec djvulibre openssl ];
-  NIX_LDFLAGS = "-lX11 -lXext";
-
-  hardeningDisable = [ "format" ];
-
-  k2_pa = ./k2pdfopt.patch;
-  tess_pa = ./tesseract.patch;
-
-  builder = writeScript "builder.sh" ''
-    . ${stdenv}/setup
-    set -e
-
-    plibs=`pwd`/patched_libraries
-
-    tar zxf ${mupdf_src}
-    cp $src/mupdf_mod/font.c $src/mupdf_mod/string.c mupdf-1.6-source/source/fitz/
-    cp $src/mupdf_mod/pdf-* mupdf-1.6-source/source/pdf
-
-    tar zxf ${tess_src}
-    cp $src/tesseract_mod/dawg.cpp tesseract-ocr/dict
-    cp $src/tesseract_mod/tessdatamanager.cpp tesseract-ocr/ccutil
-    cp $src/tesseract_mod/tessedit.cpp tesseract-ocr/ccmain
-    cp $src/tesseract_mod/tesscapi.cpp tesseract-ocr/api
-    cp $src/include_mod/tesseract.h $src/include_mod/leptonica.h tesseract-ocr/api
-
-    cp -a $src k2pdfopt_v2.21
-    chmod -R +w k2pdfopt_v2.21
-
-    patch -p0 -i $tess_pa
-    patch -p0 -i $k2_pa
-
-    cd tesseract-ocr
-    ./autogen.sh
-    substituteInPlace "configure" \
-            --replace 'LIBLEPT_HEADERSDIR="/usr/local/include /usr/include"' \
-            'LIBLEPT_HEADERSDIR=${leptonica}/include'
-    ./configure --prefix=$plibs --disable-shared
-    make install
-
-    cd ..
-    tar zxf ${gocr_src}
-    cd gocr-0.49
-    ./configure
-    cp src/{gocr.h,pnm.h,unicode.h,list.h} $plibs/include
-    cp include/config.h $plibs/include
-    make libs
-    cp src/libPgm2asc.a $plibs/lib
-
-    cd ../mupdf-1.6-source
-    make prefix=$plibs install
-    install -Dm644 build/debug/libmujs.a $plibs/lib
-
-    cd ../k2pdfopt_v2.21/k2pdfoptlib
-    gcc -Ofast -Wall -c *.c -I ../include_mod/ -I $plibs/include \
-        -I . -I ../willuslib
-    ar rcs libk2pdfopt.a *.o
-
-    cd ../willuslib
-    gcc -Ofast -Wall -c *.c -I ../include_mod/ -I $plibs/include
-    ar rcs libwillus.a *.o
-
-    cd ..
-    gcc -Wall -Ofast -o k2pdfopt.o -c k2pdfopt.c -I k2pdfoptlib/ -I willuslib/ \
-            -I include_mod/ -I $plibs/include
-    g++ -Ofast k2pdfopt.o -o k2pdfopt -I willuslib/ -I k2pdfoptlib/ -I include_mod/ \
-            -I $plibs/include -L $plibs/lib/ \
-            -L willuslib/ -L k2pdfoptlib/ -lk2pdfopt -lwillus -ldjvulibre -lz -lmupdf  \
-            -ljbig2dec -ljpeg -lopenjp2 -lpng -lfreetype -lpthread -lmujs \
-            -lPgm2asc -llept -ltesseract -lcrypto
-
-    mkdir -p $out/bin
-    cp k2pdfopt $out/bin
+  patches = [ ./k2pdfopt.patch ];
+
+  nativeBuildInputs = [ cmake pkgconfig ];
+
+  buildInputs =
+  let
+    mupdf_modded = mupdf.overrideAttrs (attrs: {
+      name = "mupdf-1.10a";
+      src = fetchurl {
+        url = "http://mupdf.com/downloads/archive/mupdf-1.10a-source.tar.gz";
+        sha256 = "0dm8wcs8i29aibzkqkrn8kcnk4q0kd1v66pg48h5c3qqp4v1zk5a";
+      };
+      # Excluded the pdf-*.c files, since they mostly just broke the #includes
+      prePatch = ''
+        cp ${src}/mupdf_mod/{font,stext-device,string}.c source/fitz/
+        cp ${src}/mupdf_mod/font-win32.c source/pdf/
+      '';
+      # Patches from previous 1.10a version in nixpkgs
+      patches = [
+        # Compatibility with new openjpeg
+        (fetchpatch {
+         name = "mupdf-1.9a-openjpeg-2.1.1.patch";
+         url = "https://git.archlinux.org/svntogit/community.git/plain/mupdf/trunk/0001-mupdf-openjpeg.patch?id=5a28ad0a8999a9234aa7848096041992cc988099";
+         sha256 = "1i24qr4xagyapx4bijjfksj4g3bxz8vs5c2mn61nkm29c63knp75";
+        })
+
+        (fetchurl {
+         name = "CVE-2017-5896.patch";
+         url = "http://git.ghostscript.com/?p=mupdf.git;a=patch;h=2c4e5867ee699b1081527bc6c6ea0e99a35a5c27";
+         sha256 = "14k7x47ifx82sds1c06ibzbmcparfg80719jhgwjk6w1vkh4r693";
+        })
+      ];
+    });
+    leptonica_modded = leptonica.overrideAttrs (attrs: {
+      prePatch = ''
+        cp ${src}/leptonica_mod/* src/
+      '';
+    });
+    tesseract_modded = tesseract.overrideAttrs (attrs: {
+      prePatch = ''
+        cp ${src}/tesseract_mod/{ambigs.cpp,ccutil.h,ccutil.cpp} ccutil/
+        cp ${src}/tesseract_mod/dawg.cpp api/
+        cp ${src}/tesseract_mod/{imagedata.cpp,tessdatamanager.cpp} ccstruct/
+        cp ${src}/tesseract_mod/openclwrapper.h opencl/
+        cp ${src}/tesseract_mod/{tessedit.cpp,thresholder.cpp} ccmain/
+        cp ${src}/tesseract_mod/tess_lang_mod_edge.h cube/
+        cp ${src}/tesseract_mod/tesscapi.cpp api/
+        cp ${src}/include_mod/{tesseract.h,leptonica.h} api/
+      '';
+      patches = [ ./tesseract.patch ];
+    });
+  in
+    [ zlib libpng ] ++
+    optional enableGSL gsl ++
+    optional enableGhostScript ghostscript ++
+    optionals enableMuPDF [ jbig2dec openjpeg freetype harfbuzz mupdf_modded ] ++
+    optionals enableJPEG2K [ jasper ] ++
+    optional enableDJVU djvulibre ++
+    optional enableGOCR gocr ++
+    optionals enableTesseract [ leptonica_modded tesseract_modded ];
+
+  dontUseCmakeBuildDir = true;
+
+  cmakeFlags = [ "-DCMAKE_C_FLAGS=-I${src}/include_mod" ];
+
+  installPhase = ''
+    install -D -m 755 k2pdfopt $out/bin/k2pdfopt
   '';
 
   meta = with stdenv.lib; {
@@ -107,7 +94,7 @@ in stdenv.mkDerivation rec {
     homepage = http://www.willus.com/k2pdfopt;
     license = licenses.gpl3;
     platforms = platforms.linux;
-    maintainers = [ maintainers.bosu ];
+    maintainers = with maintainers; [ bosu danielfullmer ];
   };
 }