version up: Tesseract 4.0.0-beta.3 & Leptonica 1.76.0

Add osd trainned data

component/leptonica/buildout.cfg

@@ -11,14 +11,12 @@ extends =
 
 [leptonica]
 recipe = slapos.recipe.cmmi
-url = http://leptonica.googlecode.com/files/leptonica-1.68.tar.gz
-md5sum = 5cd7092f9ff2ca7e3f3e73bfcd556403
+url = http://www.leptonica.com/source/leptonica-1.76.0.tar.gz
+md5sum = a263a5e4f7e8f8a661fb121a265d2d20
 configure-options =
   --disable-static
-patch-options = -p1
-patches =
-  ${:_profile_base_location_}/leptonica-1.69-zlib-include.patch#cff3dc942075190939b407c38e0d3201
+
 environment =
   CPPFLAGS=-I${zlib:location}/include -I${libjpeg:location}/include -I${libpng:location}/include -I${libtiff:location}/include -I${webp:location}/include -I${giflib:location}/include
   LDFLAGS=-L${zlib:location}/lib -Wl,-rpath=${zlib:location}/lib -L${libjpeg:location}/lib -Wl,-rpath=${libjpeg:location}/lib -L${libpng:location}/lib -Wl,-rpath=${libpng:location}/lib -L${libtiff:location}/lib -Wl,-rpath=${libtiff:location}/lib -L${webp:location}/lib -Wl,-rpath=${webp:location}/lib -L${giflib:location}/lib -Wl,-rpath=${giflib:location}/lib
-  PATH=${patch:location}/bin:%(PATH)s
+  PATH=${patch:location}/bin:%(PATH)s
\ No newline at end of file

component/tesseract/buildout.cfg

 [buildout]
+
 extends =
+  ../pkgconfig/buildout.cfg
   ../autoconf/buildout.cfg
   ../automake/buildout.cfg
   ../jbigkit/buildout.cfg
@@ -8,45 +10,56 @@ extends =
   ../libtool/buildout.cfg
   ../patch/buildout.cfg
   ../webp/buildout.cfg
-
+  
 parts =
   tesseract
+  tesseract-traineddata
+  tesseract-eng-traineddata
+  tesseract-osd-traineddata
 
 [tesseract]
 recipe = slapos.recipe.cmmi
-url = http://tesseract-ocr.googlecode.com/files/tesseract-3.01.tar.gz
-md5sum = 1ba496e51a42358fb9d3ffe781b2d20a
-patch-options =
-  -p1
-patches =
-  ${:_profile_base_location_}/tesseract-3.00-gcc-4.7-build.patch#ca80db3ec489c547b03f3ee48879c1b1
-  ${:_profile_base_location_}/tesseract-3.01-remove-bom.patch#2e691858cb492b7c17d23bf0912b3d24
-  ${:_profile_base_location_}/tesseract-3.01-gcc6-ftbs.patch#f7a6140c0fe390b96fe753a70e9d59fd
+url = https://github.com/tesseract-ocr/tesseract/archive/6b250b58121a9858d3e3019a78a6f7d421bd0fc7.tar.gz
+md5sum = fdc38148ad8eb1bd0485a217503dd6d5
+pkg_config_depends = ${leptonica:location}/lib/pkgconfig:${fontconfig:location}/lib/pkgconfig:${fontconfig:pkg_config_depends}:${lcms2:location}/lib/pkgconfig:${xz-utils:location}/lib/pkgconfig
 pre-configure =
-  libtoolize -f -c
-  aclocal -I ${libtool:location}/share/aclocal -I config
-  autoheader -f
-  automake -c -a -f
+  aclocal -I${pkgconfig:location}/share/aclocal -I${libtool:location}/share/aclocal
+  ./autogen.sh
+  autoreconf -ivf
   autoconf -Wno-portability
+
 configure-options =
   --disable-static
-  --datarootdir=${tesseract-eng-traineddata:location}
+  --datarootdir=${tesseract-traineddata:location}
 # XXX: tesseract seems not easily configurable at runtime about where to find
 # its trained data, so we set its datarootdir above to a controlled location
 
 # tesseract has a non-standard way of testing for leptonica, hence the
 # LIBLEPT_HEADERSDIR entry below:
 environment =
-  PATH=${autoconf:location}/bin:${automake:location}/bin:${libtool:location}/bin:${m4:location}/bin:${patch:location}/bin:%(PATH)s
+  PATH=${pkgconfig:location}/bin:${autoconf:location}/bin:${automake:location}/bin:${libtool:location}/bin:${m4:location}/bin:${patch:location}/bin:%(PATH)s
+  PKG_CONFIG_PATH=${:pkg_config_depends}
   ACLOCAL_ARGS=-I${libtool:location}/share/aclocal
+  ACLOCAL_PATH=${pkgconfig:location}/share/aclocal:${libtool:location}/share/aclocal
   LIBLEPT_HEADERSDIR=${leptonica:location}/include
   CPPFLAGS=-I${leptonica:location}/include
   LDFLAGS =-L${leptonica:location}/lib -Wl,-rpath=${leptonica:location}/lib -L${jbigkit:location}/lib -Wl,-rpath=${jbigkit:location}/lib -L${zlib:location}/lib -Wl,-rpath=${zlib:location}/lib
+
 make-options =
   LIBTOOL=${libtool:location}/bin/libtool
 
+[tesseract-traineddata]
+location = ${buildout:parts-directory}/${:_buildout_section_name_}
+
 [tesseract-eng-traineddata]
-recipe = slapos.recipe.build:download-unpacked
-strip-top-level-dir = true
-url = http://tesseract-ocr.googlecode.com/files/tesseract-ocr-3.01.eng.tar.gz
-md5sum = 89c139a73e0e7b1225809fc7b226b6c9
+recipe = slapos.recipe.build:download
+destination = ${tesseract-traineddata:location}/tessdata/eng.traineddata
+url = https://github.com/tesseract-ocr/tessdata/raw/590567f20dc044f6948a8e2c61afc714c360ad0e/eng.traineddata
+md5sum = 57e0df3d84fed9fbf8c7a8e589f8f012
+
+[tesseract-osd-traineddata]
+recipe = slapos.recipe.build:download
+destination = ${tesseract-traineddata:location}/tessdata/osd.traineddata
+url = https://github.com/tesseract-ocr/tessdata/raw/590567f20dc044f6948a8e2c61afc714c360ad0e/osd.traineddata
+md5sum = 7611737524efd1ce2dde67eff629bbcf
+

component/tesseract/tesseract-3.00-gcc-4.7-build.patch

---- tesseract-3.00/viewer/svutil.cpp.old	2012-04-11 09:34:26.168608940 +0200
-+++ tesseract-3.00/viewer/svutil.cpp	2012-04-11 09:34:46.108565692 +0200
-@@ -21,6 +21,7 @@
- // thread/process creation & synchronization and network connection.
- 
- #include <stdio.h>
-+#include <unistd.h>
- #ifdef WIN32
- #include <windows.h>
- #include <winsock.h>

component/tesseract/tesseract-3.01-gcc6-ftbs.patch

-commit 58e79a222e12280984ed19ab4d3bcac654e121fa
-Author: Arnaud Fontaine <arnaud.fontaine@nexedi.com>
-Date:   Fri Nov 4 19:10:13 2016 +0900
-
-    Fix compilation errors.
-
-diff --git a/cube/char_set.cpp b/cube/char_set.cpp
-index 3cf4798..6b29883 100644
---- a/cube/char_set.cpp
-+++ b/cube/char_set.cpp
-@@ -65,13 +65,13 @@ CharSet *CharSet::Create(TessdataManager *tessdata_manager,
-       !tessdata_manager->SeekToStart(TESSDATA_UNICHARSET)) {
-     fprintf(stderr, "Cube ERROR (CharSet::Create): could not find "
-             "either cube or tesseract unicharset\n");
--    return false;
-+    return NULL;
-   }
-   FILE *charset_fp = tessdata_manager->GetDataFilePtr();
-   if (!charset_fp) {
-     fprintf(stderr, "Cube ERROR (CharSet::Create): could not load "
-             "a unicharset\n");
--    return false;
-+    return NULL;
-   }
- 
-   // If we found a cube unicharset separate from tesseract's, load it and
-@@ -84,7 +84,7 @@ CharSet *CharSet::Create(TessdataManager *tessdata_manager,
-     loaded = char_set->LoadSupportedCharList(charset_fp, NULL);
-   if (!loaded) {
-     delete char_set;
--    return false;
-+    return NULL;
-   }
- 
-   char_set->init_ = true;
-diff --git a/cube/cube_line_segmenter.cpp b/cube/cube_line_segmenter.cpp
-index deee573..3f0b762 100644
---- a/cube/cube_line_segmenter.cpp
-+++ b/cube/cube_line_segmenter.cpp
-@@ -124,7 +124,7 @@ Pixa *CubeLineSegmenter::CrackLine(Pix *cracked_line_pix,
- 
-   if (line_con_comps == NULL) {
-     delete []lines_pixa;
--    return false;
-+    return NULL;
-   }
- 
-   // assign each conn comp to the a line based on its centroid
-@@ -142,7 +142,7 @@ Pixa *CubeLineSegmenter::CrackLine(Pix *cracked_line_pix,
-         delete []lines_pixa;
-         boxaDestroy(&line_con_comps);
-         pixaDestroy(&line_con_comps_pix);
--        return false;
-+        return NULL;
-       }
-     }
- 
-@@ -413,14 +413,14 @@ Pix *CubeLineSegmenter::Pixa2Pix(Pixa *pixa, Box **dest_box,
- 
-   (*dest_box) = boxCreate(min_x, min_y, max_x - min_x, max_y - min_y);
-   if ((*dest_box) == NULL) {
--    return false;
-+    return NULL;
-   }
- 
-   // create the union pix
-   Pix *union_pix = pixCreate((*dest_box)->w, (*dest_box)->h, img_->d);
-   if (union_pix == NULL) {
-     boxDestroy(dest_box);
--    return false;
-+    return NULL;
-   }
- 
-   // create a pix corresponding to the union of all pixs
-diff --git a/cube/cube_object.cpp b/cube/cube_object.cpp
-index 48bce64..b9a7113 100644
---- a/cube/cube_object.cpp
-+++ b/cube/cube_object.cpp
-@@ -165,7 +165,7 @@ WordAltList *CubeObject::Recognize(LangModel *lang_mod, bool word_mode) {
-       if (deslanted_beam_obj_ == NULL) {
-         fprintf(stderr, "Cube ERROR (CubeObject::Recognize): could not "
-                 "construct deslanted BeamSearch\n");
--        return false;
-+        return NULL;
-       }
-     }
- 
-diff --git a/cube/word_list_lang_model.cpp b/cube/word_list_lang_model.cpp
-index 18f85c1..0f7f562 100644
---- a/cube/word_list_lang_model.cpp
-+++ b/cube/word_list_lang_model.cpp
-@@ -74,7 +74,7 @@ LangModEdge **WordListLangModel::GetEdges(CharAltList *alt_list,
-   // initialize if necessary
-   if (init_ == false) {
-     if (Init() == false) {
--      return false;
-+      return NULL;
-     }
-   }
- 

component/tesseract/tesseract-3.01-remove-bom.patch

-The patch below removes a utf-8 BOM mark.
-
-Avoid touching it as the BOM is invisible, and copy/pasting might not work.
-
-It is needed because old compilers treat the BOM as garbage instead of
-whitespace.
-
---- tesseract-3.01/ccutil/strngs.h.orig	2012-05-24 15:13:22.743808379 +0200
-+++ tesseract-3.01/ccutil/strngs.h	2012-05-24 15:16:54.468858282 +0200
-@@ -1,4 +1,4 @@
--/**********************************************************************
-+/**********************************************************************
-  * File:        strngs.h  (Formerly strings.h)
-  * Description: STRING class definition.
-  * Author:					Ray Smith