Attachment #96960 for bug #135665

View | Details | Raw Unified | Return to bug 135665
Collapse All | Expand All




#

PORTNAME=	ocropus
PORTVERSION=	0.4
CATEGORIES=	graphics
MASTER_SITES=	${MASTER_SITE_GOOGLE_CODE}

MAINTAINER=	hiroto.kagotani@gmail.com
COMMENT=	The OCRopus(tm) open source document analysis and OCR system

BUILD_DEPENDS=	${LOCALBASE}/lib/libtesseract_full.a:${PORTSDIR}/graphics/tesseract
LIB_DEPENDS=	png.5:${PORTSDIR}/graphics/png \
		jpeg.9:${PORTSDIR}/graphics/jpeg \
		tiff.4:${PORTSDIR}/graphics/tiff \
		iulib.0:${PORTSDIR}/graphics/iulib \
		gsl.13:${PORTSDIR}/math/gsl

WRKSRC=		${WRKDIR}/ocropus-0.4/ocropus

OPTIONS=	SDL "Enable SDL for graphical debugging" off \
		LEPTONICA "Enable Leptonica image analysis" off

MAKE_JOBS_UNSAFE=	yes
USE_PYTHON_BUILD=	yes
USE_AUTOTOOLS=	aclocal:110 automake:110 autoconf:262
USE_GMAKE=	yes
CONFIGURE_ARGS=	--without-fst --without-tesseract
CONFIGURE_ENV=	CPPFLAGS=-I${LOCALBASE}/include LDFLAGS=-L${LOCALBASE}/lib

.include <bsd.port.pre.mk>


.if !defined(WITH_LEPTONICA)
CONFIGURE_ARGS+=--without-leptonica
PLIST_SUB+=	LEPTONICA="@comment "
.else
LIB_DEPENDS+=	lept.0:${PORTSDIR}/graphics/leptonlib
PLIST_SUB+=	LEPTONICA=""
.endif

pre-configure:
	@(cd ${CONFIGURE_WRKSRC} \
	  && ${SETENV} ${PYTHON_CMD} genAM.py >Makefile.am)

run-autotools: run-autotools-aclocal run-autotools-automake run-autotools-autoconf

run-autotools-autoconf:
	@(cd ${CONFIGURE_WRKSRC} && ${SETENV} ${AUTOTOOLS_ENV} ${AUTOCONF} \
		${AUTOCONF_ARGS})
	@${REINPLACE_CMD} -e 's|-lpthread|${PTHREAD_LIBS}|'\
		${WRKSRC}/configure

.include <bsd.port.post.mk>

Lines 1-3 Link Here

(-)/usr/ports/graphics/ocropus/distinfo (-3 / +3 lines)
1	MD5 (ocropus-0.3.1.tar.gz) = 2a1b66419ae69ef031d5e6269db15bb5	1	MD5 (ocropus-0.4.tar.gz) = d883ee9c9fd63bfdd42b25992f63c9ed
2	SHA256 (ocropus-0.3.1.tar.gz) = ee02d209a1c823090f0bceba7ec4a884029f66fc44147a2d34922f8148a699df	2	SHA256 (ocropus-0.4.tar.gz) = 2a564916d05e1badf1a527cc59015f3397e57f7ab9dc1269781f046f87d1b8ff
3	SIZE (ocropus-0.3.1.tar.gz) = 12061574	3	SIZE (ocropus-0.4.tar.gz) = 21311501

Lines 1-11 Link Here

(-)/usr/ports/graphics/ocropus/files/patch-Makefile.am (-11 lines)
1	--- ./Makefile.am.orig 2008-10-16 05:40:47.000000000 +0900
2	+++ ./Makefile.am 2009-05-26 21:25:34.000000000 +0900
3	@@ -110,7 +110,7 @@
4
5	# run check-style everytime and give a hint about make check
6	all:
7	- $(srcdir)/utilities/check-style -f $(srcdir)
8	+# $(srcdir)/utilities/check-style -f $(srcdir)
9	@echo
10	@echo "Use 'make check' to run tests!"
11	@echo




--- ./configure.ac.orig	2009-06-01 05:18:41.000000000 +0900
+++ ./configure.ac	2009-06-17 19:47:20.000000000 +0900
@@ -114,11 +114,19 @@
     AC_MSG_ERROR([no TIFFOpen; please install libtiff4-dev or equivalent]))
 
 AC_LANG_CPLUSPLUS

 
 # --- iulib (required) ---
 # NB: we can only use functions with C linkage here
 AC_CHECK_LIB(iulib, exit,,
-    AC_MSG_ERROR([no iulib; please install iulib first (see INSTALL)]))
+    AC_CHECK_LIB(avcodec, avcodec_open, ,
+        AC_MSG_ERROR([no iulib; please install iulib first (see INSTALL)]))
+    AC_CHECK_LIB(avformat, url_fopen, ,
+        AC_MSG_ERROR([no iulib; please install iulib first (see INSTALL)]))
+    AC_CHECK_LIB(iulib, sleep, ,
+        AC_MSG_ERROR([no iulib; please install iulib first (see INSTALL)]))
+)
 
 # --- libpthread (needed by tesseract) ---
 AC_CHECK_LIB(pthread, pthread_create,,)
@@ -176,14 +184,14 @@
     LDFLAGS="$LDFLAGS -L$leptheaders/../../lib"
     AC_CHECK_LIB(lept,pixCreate,,AC_MSG_ERROR([leptonica not found! Choose --without-leptonica if you don't want to use it.]))
 fi
-AM_CONDITIONAL([use_leptonica], [test x$use_leptonica == xyes])
+AM_CONDITIONAL([use_leptonica], [test x$use_leptonica = xyes])
 
 # --- GSL (optional for glinerec) ---
-AC_SUBST(use_gsl, "yes")
+AC_SUBST(use_gsl, "no")
 AC_CHECK_LIB(gslcblas, abort,,AC_SUBST(use_gsl, "no"))
 AC_CHECK_LIB(gsl, gsl_error,,AC_SUBST(use_gsl, "no"))
 AC_CHECK_LIB(blas, exit,,AC_SUBST(use_gsl, "no"))
-AM_CONDITIONAL([use_gsl], [test x$use_gsl == xyes])
+AM_CONDITIONAL([use_gsl], [test x$use_gsl = xyes])
 
 
 # --- SDL (optional for graphical debugging) ---




--- ./genAM.py.orig	2009-06-01 05:18:41.000000000 +0900
+++ ./genAM.py	2009-06-17 19:44:58.000000000 +0900
@@ -62,7 +62,7 @@
 ocropusincludedir=$(includedir)/ocropus
 
 AM_CPPFLAGS = -I$(srcdir)/include -I$(srcdir)/ocr-utils \
--I@iulibheaders@ -I@tessheaders@
+-I@iulibheaders@
 
 AM_LDFLAGS =
 
@@ -180,7 +180,7 @@
 print """
 # run check-style everytime and give a hint about make check
 all:
-	$(srcdir)/utilities/check-style -f $(srcdir)
+#	$(srcdir)/utilities/check-style -f $(srcdir)
 	@echo
 	@echo "Use 'make check' to run tests!"
 	@echo

Line 0 Link Here

(-)/usr/ports/graphics/ocropus/files/patch-ocr-utils__components.cc (+10 lines)
	1	--- ./ocr-utils/components.cc.orig 2009-06-01 05:18:41.000000000 +0900
	2	+++ ./ocr-utils/components.cc 2009-06-17 19:44:58.000000000 +0900
	3	@@ -26,6 +26,7 @@
	4	#include "colib/colib.h"
	5	#include "iulib/iulib.h"
	6	#include "components.h"
	7	+extern char **environ;
	8
	9	using namespace colib;
	10

Lines 1-10 Link Here

(-)/usr/ports/graphics/ocropus/files/patch-ocr-utils__narray-io.h (-3 / +3 lines)
1	--- ./ocr-utils/narray-io.h.orig 2008-10-16 05:40:46.000000000 +0900	1	--- ./ocr-utils/narray-io.h.orig 2009-06-01 05:18:41.000000000 +0900
2	+++ ./ocr-utils/narray-io.h 2009-05-26 21:22:11.000000000 +0900	2	+++ ./ocr-utils/narray-io.h 2009-06-17 19:44:58.000000000 +0900
3	@@ -31,6 +31,7 @@	3	@@ -31,6 +31,7 @@
4		4
5	#include <stdio.h>	5	#include <stdio.h>
6	#include <stdlib.h>	6	#include <stdlib.h>
7	+#include <stdint.h>	7	+#include <stdint.h>
8	#include "colib.h"	8	#include "ocropus.h"
9		9
10	namespace ocropus {	10	namespace ocropus {




--- ./ocroscript/scripts/rec-tess-complete.lua.orig	2008-10-16 05:40:35.000000000 +0900
+++ ./ocroscript/scripts/rec-tess-complete.lua	2009-05-26 21:22:11.000000000 +0900
@@ -20,11 +20,20 @@
 -- Reviewer: 
 -- Primary Repository: 
 -- Web Sites: www.iupr.org, www.dfki.de, www.ocropus.org
+--
+-- Patch applied:
+-- http://code.google.com/p/ocropus/issues/detail?id=137
 
 
 require 'lib.util'
 require 'lib.headings'
 require 'lib.paragraphs'
+require 'lib.path'
+require 'lib.hocr'
+import_all(ocr)
+import_all(graphics)
+import_all(iulib)
+import_all(nustring)
 
 remove_hyphens = true
 
@@ -74,7 +83,7 @@
 -- RecognizedPage is a transport object of tesseract_recognize_blockwise().
 -- This function will convert it to a DOM.
 function convert_RecognizedPage_to_DOM(p, image_path, keep_char_boxes)
-    page_DOM = get_page_DOM(p, image_path)
+    page_DOM = hocr.get_page_DOM(p, image_path)
     for i = 0, p:linesCount() - 1 do
         local bbox = p:bbox(i)
         local text = nustring()
@@ -85,13 +94,12 @@
             bboxes = narray_to_table(r)
         end
         p:text(text, i)
-        line_DOM = get_line_DOM(bbox, text, bboxes, p)
+        line_DOM = hocr.get_line_DOM(bbox, text, bboxes, p)
         table.insert(page_DOM, line_DOM)
     end
     return page_DOM
 end
 
-
 function get_images_DOM(tiseg_image, html_path, images_dir, page_image)
     os.execute('mkdir -p "'..images_dir..'"')
     local rects = rectarray()
@@ -102,12 +110,11 @@
     local dom = {{tag = 'hr', size = '0'}}
     for i = 0, rects:length() - 1 do
         local src = images_dir .. ('/%04d.png'):format(i + 1)
-        local img_path = util.combine_paths(html_path, src)
         img = bytearray()
         r = rects:at(i)
         extract_subimage(img, page_image, r.x0, r.y0, r.x1, r.y1)
-        write_image_gray(img_path, img)
-        local props = {bbox = bbox_to_string(page_image, r)}
+        iulib.write_image_gray(src, img)
+        local props = {bbox = hocr.bbox_to_string(page_image, r)}
         local link = {tag = 'a', href=src}
         local width = r.x1 - r.x0
         local height = r.y1 -r.y0
@@ -119,7 +126,7 @@
            height = "200px"
         end
         local tag = {tag = 'img', src = src, width=width, height=height,
-                     class = 'ocr_image', title = hocr_properties_attribute(props)}
+                     class = 'ocr_image', title = hocr.properties_attribute(props)}
         table.insert(link, tag)
         table.insert(dom, link)
         table.insert(dom, '\n')
@@ -146,8 +153,8 @@
     get_nontext_mask(nontext_mask,tiseg_image)
     remove_masked_region(text_image,nontext_mask,clean_image)
     segmenter:segment(page_segmentation,text_image)
-    local p = RecognizedPage()
-    tesseract_recognize_blockwise(p, page_image, page_segmentation)
+    local p = tesseract.RecognizedPage()
+    tesseract.recognize_blockwise(p, page_image, page_segmentation)
     page_DOM = convert_RecognizedPage_to_DOM(p, pages:getFileName(), 
                                              option("charboxes"))
     page_DOM = detect_headings(page_DOM, page_image)
@@ -157,10 +164,10 @@
     table.insert(body_DOM, page_DOM)
 end
 --end
-doc_DOM = get_html_tag()
-table.insert(doc_DOM, get_head_tag())
+doc_DOM = hocr.get_html_tag()
+table.insert(doc_DOM, hocr.get_head_tag())
 table.insert(doc_DOM, '\n')
 table.insert(doc_DOM, body_DOM)
 file = io.open(output_file, 'w')
-dump_DOM(file, doc_DOM, html_preamble)
+hocr.dump(file, doc_DOM, html_preamble)
 file:close()




bin/ocroscript
lib/libocropus.a
bin/ocr-distance
bin/ocropus
include/ocropus/ocr-openfst.h
include/ocropus/glclass.h
include/ocropus/defs.h
include/ocropus/voronoi-ocropus.h
include/ocropus/const.h
include/ocropus/beam-search.h
include/ocropus/langmod-shortest-path.h
include/ocropus/lattice.h
include/ocropus/ocr-binarize-sauvola.h
include/ocropus/ocr-binarize-otsu.h
include/ocropus/grouping.h
include/ocropus/make-garbage.h
include/ocropus/charlib.h
include/ocropus/feature-extractor.h
include/ocropus/confusion-matrix.h
include/ocropus/bpnet.h
include/ocropus/additions.h
include/ocropus/feature-stream.h
include/ocropus/mnist.h
include/ocropus/classmap.h
include/ocropus/classify-chars.h
include/ocropus/bpnetline.h
include/ocropus/ocr-deskew-rast.h
include/ocropus/ocr-noisefilter.h
include/ocropus/ocr-doc-clean.h
include/ocropus/ocr-doc-clean-concomp.h
include/ocropus/ocr-pageframe-rast.h
include/ocropus/ocr-whitespace-cover.h
include/ocropus/ocr-char-stats.h
include/ocropus/ocr-layout-rast.h
include/ocropus/line-info.h
include/ocropus/glcuts.h
include/ocropus/tesseract.h
include/ocropus/ocr-pfst.h
include/ocropus/glfmaps.h
include/ocropus/gldataset.h
include/ocropus/glutils.h
include/ocropus/grouper.h
include/ocropus/ocropus.h
include/ocropus/gsl.h
include/ocropus/ocrcomponents.h
include/ocropus/kmeans.h
include/ocropus/glinerec.h
include/ocropus/ocr-layout.h
include/ocropus/narray-io.h
include/ocropus/queue.h
include/ocropus/grid.h
include/ocropus/sysutil.h
include/ocropus/logger.h
include/ocropus/grouper.h
include/ocropus/segmentation.h
include/ocropus/pages.h
include/ocropus/linesegs.h
include/ocropus/components.h
include/ocropus/stringutil.h
include/ocropus/init-ocropus.h
include/ocropus/editdist.h
include/ocropus/ocrinterfaces.h
include/ocropus/arraypaint.h
include/ocropus/narray-binio.h
include/ocropus/sysutil.h
include/ocropus/segmentation.h
include/ocropus/resource-path.h
include/ocropus/xml-entities.h
include/ocropus/pagesegs.h
include/ocropus/ocr-utils.h
include/ocropus/didegrade.h
include/ocropus/docproc.h
include/ocropus/enumerator.h
%%DATADIR%%/models/default.fst.gz
%%DATADIR%%/models/default.model
%%DATADIR%%/models/latin3-full.model
%%DATADIR%%/models/ocr-dict-case.fst.gz
%%DATADIR%%/models/neural-net-file.nn
%%DATADIR%%/words/en-us
%%DATADIR%%/scripts/lib/align.lua
%%DATADIR%%/scripts/lib/datasets.lua
%%DATADIR%%/scripts/lib/editdist.lua
%%DATADIR%%/scripts/lib/getopt.lua
%%DATADIR%%/scripts/lib/headings.lua
%%DATADIR%%/scripts/lib/hocr.lua
%%DATADIR%%/scripts/lib/paragraphs.lua
%%DATADIR%%/scripts/lib/path.lua
%%DATADIR%%/scripts/lib/util.lua
%%DATADIR%%/scripts/lib/xml.lua
%%DATADIR%%/scripts/align-lines-wordwise.lua
%%DATADIR%%/scripts/align-transcription.lua
%%DATADIR%%/scripts/align.lua
%%DATADIR%%/scripts/build-ngram-model.lua
%%DATADIR%%/scripts/check-train-valid-bpnet-feature.lua
%%DATADIR%%/scripts/degrade.lua
%%DATADIR%%/scripts/deskew.lua
%%DATADIR%%/scripts/editdist.lua
%%DATADIR%%/scripts/erode3.lua
%%DATADIR%%/scripts/eval-bpnet-on-words.lua
%%DATADIR%%/scripts/eval-editdist-layout.lua
%%DATADIR%%/scripts/eval-on-word-list.lua
%%DATADIR%%/scripts/hocr-to-text.lua
%%DATADIR%%/scripts/line-clean.lua
%%DATADIR%%/scripts/matra-clipping.lua
%%DATADIR%%/scripts/rec-bpnet-isolated.lua
%%DATADIR%%/scripts/rec-bpnet.lua
%%DATADIR%%/scripts/rec-guided.lua
%%DATADIR%%/scripts/rec-line.lua
%%DATADIR%%/scripts/rec-ltess.lua
%%DATADIR%%/scripts/rec-minimal.lua
%%DATADIR%%/scripts/rec-tess-complete.lua
%%DATADIR%%/scripts/recognize.lua
%%DATADIR%%/scripts/reflow.lua
%%DATADIR%%/scripts/sauvola.lua
%%DATADIR%%/scripts/segment-line.lua
%%DATADIR%%/scripts/show.lua
%%DATADIR%%/scripts/showseg.lua
%%DATADIR%%/scripts/strict.lua
%%DATADIR%%/scripts/text-to-hocr.lua
%%DATADIR%%/scripts/train-bpnet-isolated.lua
%%DATADIR%%/scripts/train-bpnet-lines.lua
@dirrm include/ocropus
@dirrm %%DATADIR%%/models
@dirrm %%DATADIR%%/scripts/lib
@dirrmtry %%DATADIR%%/scripts
@dirrm %%DATADIR%%/words
@dirrmtry %%DATADIR%%

Return to bug 135665

Lines 6-33 Link Here

(-)/usr/ports/graphics/ocropus/Makefile (-7 / +16 lines)
6	#	6	#
7		7
8	PORTNAME= ocropus	8	PORTNAME= ocropus
9	PORTVERSION= 0.3.1	9	PORTVERSION= 0.4
10	CATEGORIES= graphics	10	CATEGORIES= graphics
11	MASTER_SITES= ${MASTER_SITE_GOOGLE_CODE}	11	MASTER_SITES= ${MASTER_SITE_GOOGLE_CODE}
12		12
13	MAINTAINER= hiroto.kagotani@gmail.com	13	MAINTAINER= hiroto.kagotani@gmail.com
14	COMMENT= The OCRopus(tm) open source document analysis and OCR system	14	COMMENT= The OCRopus(tm) open source document analysis and OCR system
15		15
16	BUILD_DEPENDS= ${LOCALBASE}/lib/libtesseract_full.a:${PORTSDIR}/graphics/tesseract
17	LIB_DEPENDS= png.5:${PORTSDIR}/graphics/png \	16	LIB_DEPENDS= png.5:${PORTSDIR}/graphics/png \
18	jpeg.9:${PORTSDIR}/graphics/jpeg \	17	jpeg.9:${PORTSDIR}/graphics/jpeg \
19	tiff.4:${PORTSDIR}/graphics/tiff \	18	tiff.4:${PORTSDIR}/graphics/tiff \
20	iulib.0:${PORTSDIR}/graphics/iulib	19	iulib.0:${PORTSDIR}/graphics/iulib \
		20	gsl.13:${PORTSDIR}/math/gsl
21		21
22	WRKSRC= ${WRKDIR}/ocropus-0.3	22	WRKSRC= ${WRKDIR}/ocropus-0.4/ocropus
23		23
24	OPTIONS= SDL "Enable SDL for graphical debugging" off \	24	OPTIONS= SDL "Enable SDL for graphical debugging" off \
25	LEPTONICA "Enable Leptonica image analysis" off	25	LEPTONICA "Enable Leptonica image analysis" off
26		26
27	MAKE_JOBS_UNSAFE= yes	27	MAKE_JOBS_UNSAFE= yes
		28	USE_PYTHON_BUILD= yes
28	USE_AUTOTOOLS= aclocal:110 automake:110 autoconf:262	29	USE_AUTOTOOLS= aclocal:110 automake:110 autoconf:262
29	USE_GMAKE= yes	30	USE_GMAKE= yes
30	CONFIGURE_ARGS= --without-fst	31	CONFIGURE_ARGS= --without-fst --without-tesseract
31	CONFIGURE_ENV= CPPFLAGS=-I${LOCALBASE}/include LDFLAGS=-L${LOCALBASE}/lib	32	CONFIGURE_ENV= CPPFLAGS=-I${LOCALBASE}/include LDFLAGS=-L${LOCALBASE}/lib
32		33
33	.include <bsd.port.pre.mk>	34	.include <bsd.port.pre.mk>
Lines 40-51 Link Here
40		41
41	.if !defined(WITH_LEPTONICA)	42	.if !defined(WITH_LEPTONICA)
42	CONFIGURE_ARGS+=--without-leptonica	43	CONFIGURE_ARGS+=--without-leptonica
43	PLIST_SUB+= LEPTONICA="@comment "
44	.else	44	.else
45	LIB_DEPENDS+= lept.0:${PORTSDIR}/graphics/leptonlib	45	LIB_DEPENDS+= lept.0:${PORTSDIR}/graphics/leptonlib
46	PLIST_SUB+= LEPTONICA=""
47	.endif	46	.endif
48		47
		48	pre-configure:
		49	@(cd ${CONFIGURE_WRKSRC} \
		50	&& ${SETENV} ${PYTHON_CMD} genAM.py >Makefile.am)
		51
49	run-autotools: run-autotools-aclocal run-autotools-automake run-autotools-autoconf	52	run-autotools: run-autotools-aclocal run-autotools-automake run-autotools-autoconf
50		53
		54	run-autotools-autoconf:
		55	@(cd ${CONFIGURE_WRKSRC} && ${SETENV} ${AUTOTOOLS_ENV} ${AUTOCONF} \
		56	${AUTOCONF_ARGS})
		57	@${REINPLACE_CMD} -e 's\|-lpthread\|${PTHREAD_LIBS}\|'\
		58	${WRKSRC}/configure
		59
51	.include <bsd.port.post.mk>	60	.include <bsd.port.post.mk>

Lines 1-6 Link Here

(-)/usr/ports/graphics/ocropus/files/patch-configure.ac (-5 / +26 lines)
1	--- ./configure.ac.orig 2008-10-16 05:40:35.000000000 +0900	1	--- ./configure.ac.orig 2009-06-01 05:18:41.000000000 +0900
2	+++ ./configure.ac 2009-05-26 21:22:11.000000000 +0900	2	+++ ./configure.ac 2009-06-17 19:47:20.000000000 +0900
3	@@ -116,6 +116,8 @@	3	@@ -114,11 +114,19 @@
4	AC_MSG_ERROR([no TIFFOpen; please install libtiff4-dev or equivalent]))	4	AC_MSG_ERROR([no TIFFOpen; please install libtiff4-dev or equivalent]))
5		5
6	AC_LANG_CPLUSPLUS	6	AC_LANG_CPLUSPLUS
Lines 9-20 Link Here
9		9
10	# --- iulib (required) ---	10	# --- iulib (required) ---
11	# NB: we can only use functions with C linkage here	11	# NB: we can only use functions with C linkage here
12	@@ -180,7 +182,7 @@	12	AC_CHECK_LIB(iulib, exit,,
		13	- AC_MSG_ERROR([no iulib; please install iulib first (see INSTALL)]))
		14	+ AC_CHECK_LIB(avcodec, avcodec_open, ,
		15	+ AC_MSG_ERROR([no iulib; please install iulib first (see INSTALL)]))
		16	+ AC_CHECK_LIB(avformat, url_fopen, ,
		17	+ AC_MSG_ERROR([no iulib; please install iulib first (see INSTALL)]))
		18	+ AC_CHECK_LIB(iulib, sleep, ,
		19	+ AC_MSG_ERROR([no iulib; please install iulib first (see INSTALL)]))
		20	+)
		21
		22	# --- libpthread (needed by tesseract) ---
		23	AC_CHECK_LIB(pthread, pthread_create,,)
		24	@@ -176,14 +184,14 @@
13	LDFLAGS="$LDFLAGS -L$leptheaders/../../lib"	25	LDFLAGS="$LDFLAGS -L$leptheaders/../../lib"
14	AC_CHECK_LIB(lept,pixCreate,,AC_MSG_ERROR([leptonica not found! Choose --without-leptonica if you don't want to use it.]))	26	AC_CHECK_LIB(lept,pixCreate,,AC_MSG_ERROR([leptonica not found! Choose --without-leptonica if you don't want to use it.]))
15	fi	27	fi
16	-AM_CONDITIONAL([use_leptonica], [test x$use_leptonica == xyes])	28	-AM_CONDITIONAL([use_leptonica], [test x$use_leptonica == xyes])
17	+AM_CONDITIONAL([use_leptonica], [test x$use_leptonica = xyes])	29	+AM_CONDITIONAL([use_leptonica], [test x$use_leptonica = xyes])
18		30
		31	# --- GSL (optional for glinerec) ---
		32	-AC_SUBST(use_gsl, "yes")
		33	+AC_SUBST(use_gsl, "no")
		34	AC_CHECK_LIB(gslcblas, abort,,AC_SUBST(use_gsl, "no"))
		35	AC_CHECK_LIB(gsl, gsl_error,,AC_SUBST(use_gsl, "no"))
		36	AC_CHECK_LIB(blas, exit,,AC_SUBST(use_gsl, "no"))
		37	-AM_CONDITIONAL([use_gsl], [test x$use_gsl == xyes])
		38	+AM_CONDITIONAL([use_gsl], [test x$use_gsl = xyes])
		39
19		40
20	# --- SDL (optional for graphical debugging in ocroscript) ---	41	# --- SDL (optional for graphical debugging) ---

Line 0 Link Here

(-)/usr/ports/graphics/ocropus/files/patch-genAM.py (+20 lines)
	1	--- ./genAM.py.orig 2009-06-01 05:18:41.000000000 +0900
	2	+++ ./genAM.py 2009-06-17 19:44:58.000000000 +0900
	3	@@ -62,7 +62,7 @@
	4	ocropusincludedir=$(includedir)/ocropus
	5
	6	AM_CPPFLAGS = -I$(srcdir)/include -I$(srcdir)/ocr-utils \
	7	--I@iulibheaders@ -I@tessheaders@
	8	+-I@iulibheaders@
	9
	10	AM_LDFLAGS =
	11
	12	@@ -180,7 +180,7 @@
	13	print """
	14	# run check-style everytime and give a hint about make check
	15	all:
	16	- $(srcdir)/utilities/check-style -f $(srcdir)
	17	+# $(srcdir)/utilities/check-style -f $(srcdir)
	18	@echo
	19	@echo "Use 'make check' to run tests!"
	20	@echo

Lines 1-96 Link Here

(-)/usr/ports/graphics/ocropus/files/patch-ocroscript__scripts__rec-tess-complete.lua (-96 lines)
1	--- ./ocroscript/scripts/rec-tess-complete.lua.orig 2008-10-16 05:40:35.000000000 +0900
2	+++ ./ocroscript/scripts/rec-tess-complete.lua 2009-05-26 21:22:11.000000000 +0900
3	@@ -20,11 +20,20 @@
4	-- Reviewer:
5	-- Primary Repository:
6	-- Web Sites: www.iupr.org, www.dfki.de, www.ocropus.org
7	+--
8	+-- Patch applied:
9	+-- http://code.google.com/p/ocropus/issues/detail?id=137
10
11
12	require 'lib.util'
13	require 'lib.headings'
14	require 'lib.paragraphs'
15	+require 'lib.path'
16	+require 'lib.hocr'
17	+import_all(ocr)
18	+import_all(graphics)
19	+import_all(iulib)
20	+import_all(nustring)
21
22	remove_hyphens = true
23
24	@@ -74,7 +83,7 @@
25	-- RecognizedPage is a transport object of tesseract_recognize_blockwise().
26	-- This function will convert it to a DOM.
27	function convert_RecognizedPage_to_DOM(p, image_path, keep_char_boxes)
28	- page_DOM = get_page_DOM(p, image_path)
29	+ page_DOM = hocr.get_page_DOM(p, image_path)
30	for i = 0, p:linesCount() - 1 do
31	local bbox = p:bbox(i)
32	local text = nustring()
33	@@ -85,13 +94,12 @@
34	bboxes = narray_to_table(r)
35	end
36	p:text(text, i)
37	- line_DOM = get_line_DOM(bbox, text, bboxes, p)
38	+ line_DOM = hocr.get_line_DOM(bbox, text, bboxes, p)
39	table.insert(page_DOM, line_DOM)
40	end
41	return page_DOM
42	end
43
44	-
45	function get_images_DOM(tiseg_image, html_path, images_dir, page_image)
46	os.execute('mkdir -p "'..images_dir..'"')
47	local rects = rectarray()
48	@@ -102,12 +110,11 @@
49	local dom = {{tag = 'hr', size = '0'}}
50	for i = 0, rects:length() - 1 do
51	local src = images_dir .. ('/%04d.png'):format(i + 1)
52	- local img_path = util.combine_paths(html_path, src)
53	img = bytearray()
54	r = rects:at(i)
55	extract_subimage(img, page_image, r.x0, r.y0, r.x1, r.y1)
56	- write_image_gray(img_path, img)
57	- local props = {bbox = bbox_to_string(page_image, r)}
58	+ iulib.write_image_gray(src, img)
59	+ local props = {bbox = hocr.bbox_to_string(page_image, r)}
60	local link = {tag = 'a', href=src}
61	local width = r.x1 - r.x0
62	local height = r.y1 -r.y0
63	@@ -119,7 +126,7 @@
64	height = "200px"
65	end
66	local tag = {tag = 'img', src = src, width=width, height=height,
67	- class = 'ocr_image', title = hocr_properties_attribute(props)}
68	+ class = 'ocr_image', title = hocr.properties_attribute(props)}
69	table.insert(link, tag)
70	table.insert(dom, link)
71	table.insert(dom, '\n')
72	@@ -146,8 +153,8 @@
73	get_nontext_mask(nontext_mask,tiseg_image)
74	remove_masked_region(text_image,nontext_mask,clean_image)
75	segmenter:segment(page_segmentation,text_image)
76	- local p = RecognizedPage()
77	- tesseract_recognize_blockwise(p, page_image, page_segmentation)
78	+ local p = tesseract.RecognizedPage()
79	+ tesseract.recognize_blockwise(p, page_image, page_segmentation)
80	page_DOM = convert_RecognizedPage_to_DOM(p, pages:getFileName(),
81	option("charboxes"))
82	page_DOM = detect_headings(page_DOM, page_image)
83	@@ -157,10 +164,10 @@
84	table.insert(body_DOM, page_DOM)
85	end
86	--end
87	-doc_DOM = get_html_tag()
88	-table.insert(doc_DOM, get_head_tag())
89	+doc_DOM = hocr.get_html_tag()
90	+table.insert(doc_DOM, hocr.get_head_tag())
91	table.insert(doc_DOM, '\n')
92	table.insert(doc_DOM, body_DOM)
93	file = io.open(output_file, 'w')
94	-dump_DOM(file, doc_DOM, html_preamble)
95	+hocr.dump(file, doc_DOM, html_preamble)
96	file:close()

Lines 1-119 Link Here

(-)/usr/ports/graphics/ocropus/pkg-plist (-106 / +35 lines)
1	bin/ocroscript
2	lib/libocropus.a	1	lib/libocropus.a
3	lib/libocroscript.a	2	bin/ocr-distance
4	include/ocropus/extern.h	3	bin/ocropus
5	include/ocropus/read_image.h	4	include/ocropus/ocr-openfst.h
6	include/ocropus/function.h	5	include/ocropus/glclass.h
7	include/ocropus/defs.h
8	include/ocropus/voronoi-ocropus.h
9	include/ocropus/const.h
10	include/ocropus/beam-search.h
11	include/ocropus/langmod-shortest-path.h
12	include/ocropus/lattice.h
13	include/ocropus/ocr-binarize-sauvola.h
14	include/ocropus/ocr-binarize-otsu.h
15	include/ocropus/grouping.h
16	include/ocropus/make-garbage.h
17	include/ocropus/charlib.h
18	include/ocropus/feature-extractor.h
19	include/ocropus/confusion-matrix.h
20	include/ocropus/bpnet.h
21	include/ocropus/additions.h
22	include/ocropus/feature-stream.h
23	include/ocropus/mnist.h
24	include/ocropus/classmap.h
25	include/ocropus/classify-chars.h
26	include/ocropus/bpnetline.h
27	include/ocropus/ocr-deskew-rast.h
28	include/ocropus/ocr-noisefilter.h
29	include/ocropus/ocr-doc-clean.h
30	include/ocropus/ocr-doc-clean-concomp.h
31	include/ocropus/ocr-pageframe-rast.h
32	include/ocropus/ocr-whitespace-cover.h
33	include/ocropus/ocr-char-stats.h
34	include/ocropus/ocr-layout-rast.h
35	include/ocropus/line-info.h	6	include/ocropus/line-info.h
36	include/ocropus/ocr-extract-gutters.h	7	include/ocropus/glcuts.h
37	include/ocropus/ocr-ctextline-rast.h	8	include/ocropus/tesseract.h
38	include/ocropus/ocr-ctextline-rast-extended.h	9	include/ocropus/ocr-pfst.h
39	include/ocropus/ocr-classify-zones.h	10	include/ocropus/glfmaps.h
40	include/ocropus/ocr-pageseg-wcuts.h	11	include/ocropus/gldataset.h
41	include/ocropus/ocr-text-image-seg.h	12	include/ocropus/glutils.h
42	include/ocropus/log-reg-data.h	13	include/ocropus/grouper.h
43	include/ocropus/ocr-pageseg-xycut.h	14	include/ocropus/ocropus.h
44	include/ocropus/ocr-word-segmentation.h	15	include/ocropus/gsl.h
45	include/ocropus/ocrcomponents.h
46	include/ocropus/kmeans.h
47	include/ocropus/glinerec.h	16	include/ocropus/glinerec.h
48	include/ocropus/idmap.h	17	include/ocropus/ocr-layout.h
49	include/ocropus/narray-io.h	18	include/ocropus/narray-io.h
50	include/ocropus/ocr-segmentations.h	19	include/ocropus/queue.h
51	include/ocropus/eigens.h	20	include/ocropus/grid.h
52	include/ocropus/sysutil.h
53	include/ocropus/logger.h	21	include/ocropus/logger.h
54	include/ocropus/grouper.h
55	include/ocropus/segmentation.h
56	include/ocropus/pages.h	22	include/ocropus/pages.h
57	include/ocropus/didegrade.h	23	include/ocropus/linesegs.h
58	include/ocropus/lines.h	24	include/ocropus/components.h
59	include/ocropus/regionextractor.h	25	include/ocropus/stringutil.h
60	include/ocropus/ocr-utils.h	26	include/ocropus/init-ocropus.h
		27	include/ocropus/editdist.h
		28	include/ocropus/ocrinterfaces.h
		29	include/ocropus/arraypaint.h
		30	include/ocropus/narray-binio.h
		31	include/ocropus/sysutil.h
		32	include/ocropus/segmentation.h
61	include/ocropus/resource-path.h	33	include/ocropus/resource-path.h
62	include/ocropus/queue.h	34	include/ocropus/xml-entities.h
63	include/ocropus/grid.h	35	include/ocropus/pagesegs.h
64	include/ocropus/seg-eval.h	36	include/ocropus/ocr-utils.h
		37	include/ocropus/didegrade.h
		38	include/ocropus/docproc.h
65	include/ocropus/enumerator.h	39	include/ocropus/enumerator.h
66	include/ocropus/editdist.h	40	%%DATADIR%%/models/default.fst.gz
67	include/ocropus/tesseract.h	41	%%DATADIR%%/models/default.model
68	include/ocropus/recognized-page.h	42	%%DATADIR%%/models/latin3-full.model
69	%%LEPTONICA%%include/ocropus/ocr-text-image-seg-leptonica.h	43	%%DATADIR%%/models/ocr-dict-case.fst.gz
70	%%DATADIR%%/models/neural-net-file.nn
71	%%DATADIR%%/words/en-us	44	%%DATADIR%%/words/en-us
72	%%DATADIR%%/scripts/lib/align.lua
73	%%DATADIR%%/scripts/lib/datasets.lua
74	%%DATADIR%%/scripts/lib/editdist.lua
75	%%DATADIR%%/scripts/lib/getopt.lua
76	%%DATADIR%%/scripts/lib/headings.lua
77	%%DATADIR%%/scripts/lib/hocr.lua
78	%%DATADIR%%/scripts/lib/paragraphs.lua
79	%%DATADIR%%/scripts/lib/path.lua
80	%%DATADIR%%/scripts/lib/util.lua
81	%%DATADIR%%/scripts/lib/xml.lua
82	%%DATADIR%%/scripts/align-lines-wordwise.lua
83	%%DATADIR%%/scripts/align-transcription.lua
84	%%DATADIR%%/scripts/align.lua
85	%%DATADIR%%/scripts/build-ngram-model.lua
86	%%DATADIR%%/scripts/check-train-valid-bpnet-feature.lua
87	%%DATADIR%%/scripts/degrade.lua
88	%%DATADIR%%/scripts/deskew.lua
89	%%DATADIR%%/scripts/editdist.lua
90	%%DATADIR%%/scripts/erode3.lua
91	%%DATADIR%%/scripts/eval-bpnet-on-words.lua
92	%%DATADIR%%/scripts/eval-editdist-layout.lua
93	%%DATADIR%%/scripts/eval-on-word-list.lua
94	%%DATADIR%%/scripts/hocr-to-text.lua
95	%%DATADIR%%/scripts/line-clean.lua
96	%%DATADIR%%/scripts/matra-clipping.lua
97	%%DATADIR%%/scripts/rec-bpnet-isolated.lua
98	%%DATADIR%%/scripts/rec-bpnet.lua
99	%%DATADIR%%/scripts/rec-guided.lua
100	%%DATADIR%%/scripts/rec-line.lua
101	%%DATADIR%%/scripts/rec-ltess.lua
102	%%DATADIR%%/scripts/rec-minimal.lua
103	%%DATADIR%%/scripts/rec-tess-complete.lua
104	%%DATADIR%%/scripts/recognize.lua
105	%%DATADIR%%/scripts/reflow.lua
106	%%DATADIR%%/scripts/sauvola.lua
107	%%DATADIR%%/scripts/segment-line.lua
108	%%DATADIR%%/scripts/show.lua
109	%%DATADIR%%/scripts/showseg.lua
110	%%DATADIR%%/scripts/strict.lua
111	%%DATADIR%%/scripts/text-to-hocr.lua
112	%%DATADIR%%/scripts/train-bpnet-isolated.lua
113	%%DATADIR%%/scripts/train-bpnet-lines.lua
114	@dirrm include/ocropus	45	@dirrm include/ocropus
115	@dirrm %%DATADIR%%/models	46	@dirrm %%DATADIR%%/models
116	@dirrm %%DATADIR%%/scripts/lib
117	@dirrmtry %%DATADIR%%/scripts
118	@dirrm %%DATADIR%%/words	47	@dirrm %%DATADIR%%/words
119	@dirrmtry %%DATADIR%%	48	@dirrmtry %%DATADIR%%