Merge pull request #4589 from bosu/k2pdfopt

New package: k2pdfopt - mobile PDF/DJVU optimizer

3 files changed, 217 insertions, 0 deletions

diff --git a/pkgs/applications/misc/k2pdfopt/default.nix b/pkgs/applications/misc/k2pdfopt/default.nix
new file mode 100644
index 000000000000..34bef86f4380
--- /dev/null
+++ b/pkgs/applications/misc/k2pdfopt/default.nix
@@ -0,0 +1,110 @@
+# 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-1.5-source.tar.gz;
+    sha256 = "0sl47zqf4c9fhs4h5zg046vixjmwgy4vhljhr5g4md733nash7z4";
+  };
+
+  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 {
+  name = "k2pdfopt";
+  src = fetchzip {
+    url = http://www.willus.com/k2pdfopt/src/k2pdfopt_v2.21_src.zip;
+    sha256 = "1vy0yw41z6p95gmivjk4r534zbg0kqap4lr9ps56kvjw51q8r54j";
+  };
+
+  buildInputs = [ libX11 libXext autoconf automake libtool leptonica libpng libtiff zlib
+                    openjpeg freetype jbig2dec djvulibre openssl ];
+  NIX_LDFLAGS = "-lX11 -lXext";
+
+  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-1.5-source/source/fitz/
+    cp $src/mupdf_mod/pdf-* mupdf-1.5-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 $plibs/lib
+
+    cd ../mupdf-1.5-source
+    make prefix=$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
+  '';
+
+  meta = with stdenv.lib; {
+    description = "K2pdfopt optimizes PDF/DJVU files for mobile e-readers (e.g. the Kindle) and smartphones.";
+    homepage = http://www.willus.com/k2pdfopt;
+    license = licenses.gpl3;
+    platforms = platforms.linux;
+    maintainers = [ maintainers.bosu ];
+  };
+}
+
diff --git a/pkgs/applications/misc/k2pdfopt/k2pdfopt.patch b/pkgs/applications/misc/k2pdfopt/k2pdfopt.patch
new file mode 100644
index 000000000000..00ac5770ea42
--- /dev/null
+++ b/pkgs/applications/misc/k2pdfopt/k2pdfopt.patch
@@ -0,0 +1,95 @@
+diff -aur k2pdfopt_v2.21/willuslib/array.c k2pdfopt_v2.21.new/willuslib/array.c
+--- k2pdfopt_v2.21/willuslib/array.c	2014-05-23 16:29:58.000000000 -0300
++++ k2pdfopt_v2.21.new/willuslib/array.c	2014-07-26 11:35:49.829825567 -0300
+@@ -1055,7 +1055,7 @@
+ void arrayf_sort(float *a,int n)
+ 
+     {
+-    sort(a,(long)n);
++    willus_sort(a,(long)n);
+     }
+ 
+ 
+diff -aur k2pdfopt_v2.21/willuslib/math.c k2pdfopt_v2.21.new/willuslib/math.c
+--- k2pdfopt_v2.21/willuslib/math.c	2013-08-15 21:33:50.000000000 -0300
++++ k2pdfopt_v2.21.new/willuslib/math.c	2014-07-26 11:36:02.853170659 -0300
+@@ -532,7 +532,7 @@
+ 
+ 
+ 
+-void sort(float *x,int n)
++void willus_sort(float *x,int n)
+ 
+     {
+     int top,n1;
+diff -aur k2pdfopt_v2.21/willuslib/ocrjocr.c k2pdfopt_v2.21.new/willuslib/ocrjocr.c
+--- k2pdfopt_v2.21/willuslib/ocrjocr.c	2012-11-12 13:09:42.000000000 -0300
++++ k2pdfopt_v2.21.new/willuslib/ocrjocr.c	2014-07-26 11:36:46.699837185 -0300
+@@ -29,6 +29,8 @@
+ #ifdef HAVE_GOCR_LIB
+ #include <gocr.h>
+ 
++job_t *JOB;
++
+ /*
+ ** bmp8 must be grayscale
+ ** (x1,y1) and (x2,y2) from top left of bitmap
+@@ -66,6 +68,7 @@
+     h=y2-y1+1;
+     dh=h+bw*2;
+     job=&_job;
++    JOB=job;
+     job_init(job);
+     job_init_image(job);
+     // willus_mem_alloc_warn((void **)&job->src.p.p,w*h,funcname,10);
+diff -aur k2pdfopt_v2.21/willuslib/string.c k2pdfopt_v2.21.new/willuslib/string.c
+--- k2pdfopt_v2.21/willuslib/string.c	2014-02-03 00:37:44.000000000 -0300
++++ k2pdfopt_v2.21.new/willuslib/string.c	2014-07-26 11:37:01.766506277 -0300
+@@ -81,7 +81,7 @@
+ ** Returns NULL if EOF, otherwise returns pointer to the string.
+ **
+ */
+-char *get_line(char *buf,int max,FILE *f)
++char *willus_get_line(char *buf,int max,FILE *f)
+ 
+     {
+     int     i;
+diff -aur k2pdfopt_v2.21/willuslib/willus.h k2pdfopt_v2.21.new/willuslib/willus.h
+--- k2pdfopt_v2.21/willuslib/willus.h	2014-07-25 15:03:51.000000000 -0300
++++ k2pdfopt_v2.21.new/willuslib/willus.h	2014-07-26 11:37:56.316506038 -0300
+@@ -214,9 +214,6 @@
+ ** CMAKE handles the defines, not this source
+ ** (Mod from Dirk Thierbach, 31-Dec-2013)
+ */
+-#ifdef USE_CMAKE
+-#include "config.h"
+-#else /* USE_CMAKE */
+ 
+ #ifndef HAVE_Z_LIB
+ #define HAVE_Z_LIB
+@@ -268,7 +265,6 @@
+ #undef HAVE_GSL_LIB
+ #endif
+ 
+-#endif /* USE_CMAKE */
+ /*
+ ** Consistency check
+ */
+@@ -533,7 +529,7 @@
+                       int *n,FILE *err);
+ int    readxyz_ex    (char *filename,double **x,double **y,double **z,
+                       int *n,FILE *err,int ignore_after_semicolon);
+-void   sort          (float *x,int n);
++void   willus_sort          (float *x,int n);
+ void   sortd         (double *x,int n);
+ void   sorti         (int *x,int n);
+ void   sortxy        (float *x,float *y,int n);
+@@ -602,7 +598,7 @@
+ /* string.c */
+ void   clean_line    (char *buf);
+ void   clean_line_end(char *buf);
+-char  *get_line      (char *buf,int max,FILE *f);
++char  *willus_get_line      (char *buf,int max,FILE *f);
+ char  *get_line_cf   (char *buf,int max,FILE *f);
+ int   mem_get_line_cf(char *buf,int maxlen,char *cptr,long *cindex,long csize);
+ int    in_string     (char *buffer,char *pattern);
diff --git a/pkgs/applications/misc/k2pdfopt/tesseract.patch b/pkgs/applications/misc/k2pdfopt/tesseract.patch
new file mode 100644
index 000000000000..5cb6e0fe3176
--- /dev/null
+++ b/pkgs/applications/misc/k2pdfopt/tesseract.patch
@@ -0,0 +1,12 @@
+diff -aur tesseract-ocr/api/Makefile.am tesseract-ocr.new/api/Makefile.am
+--- tesseract-ocr/api/Makefile.am	2012-10-09 14:18:39.000000000 -0300
++++ tesseract-ocr.new/api/Makefile.am	2014-03-20 18:43:13.926030341 -0300
+@@ -36,7 +36,7 @@
+ if VISIBILITY
+ libtesseract_api_la_CPPFLAGS += -DTESS_EXPORTS
+ endif
+-libtesseract_api_la_SOURCES = baseapi.cpp capi.cpp
++libtesseract_api_la_SOURCES = baseapi.cpp capi.cpp tesscapi.cpp
+ 
+ lib_LTLIBRARIES += libtesseract.la
+ libtesseract_la_LDFLAGS =