10 files changed, 136 insertions, 38 deletions

diff --git a/graphics/tesseract/Makefile b/graphics/tesseract/Makefile
index db5a02ec4196..9421193d75d4 100644
--- a/graphics/tesseract/Makefile
+++ b/graphics/tesseract/Makefile
@@ -6,7 +6,7 @@
 #
 
 PORTNAME=	tesseract
-DISTVERSION=	2.01
+DISTVERSION=	2.03
 CATEGORIES=	graphics
 MASTER_SITES=	${MASTER_SITE_GOOGLE_CODE} CENKES
 DISTFILES=	${DISTNAME}${EXTRACT_SUFX} ${LANGFILES}
@@ -15,23 +15,26 @@ MAINTAINER=	bf2006a@yahoo.com
 COMMENT=	Commercial quality open source OCR engine
 
 LIB_DEPENDS=	tiff.4:${PORTSDIR}/graphics/tiff
-RUN_DEPENDS=	xterm:${PORTSDIR}/x11/xterm
 
 PROJECTHOST=	tesseract-ocr
-LANGS=	deu eng fra ita nld spa
-LANGVER=	2.00
-LANGFILES=	${LANGS:S|^|${PORTNAME}-${LANGVER}.|:S|$|${EXTRACT_SUFX}|}
+LANGS2_00?=	deu eng fra ita nld spa
+LANGS2_01?=	deu-f por vie
+
+.for _N in 2_00 2_01
+LANGFILES+=	${LANGS${_N}:S|^|${PORTNAME}-${_N:S|_|.|}.|:S|$|${EXTRACT_SUFX}|}
+.endfor
+
 GNU_CONFIGURE=	yes
 CONFIGURE_ENV=	LIBTIFF_CFLAGS="-I${LOCALBASE}/include"\
 		LIBTIFF_LIBS="-L${LOCALBASE}/lib -ltiff"
 PORTDOCS=	AUTHORS ChangeLog NEWS README ReleaseNotes
 PORTEXAMPLES=	eurotext.tif phototest.tif
-PORT_VERBS=	cntraining mftraining tessdata unicharset_extractor \
-		wordlist2dawg ${PORTNAME}
 
 post-patch:
 	@${REINPLACE_CMD} -e 's|/usr/bin/X11/xterm|${LOCALBASE}/bin/xterm|'\
 		${WRKSRC}/ccutil/debugwin.cpp
+	@${REINPLACE_CMD} -e 's|-lpthread|${PTHREAD_LIBS}|'\
+		${WRKSRC}/configure
 
 post-install:
 	@cd ${WRKDIR}/tessdata/&&${COPYTREE_SHARE} . ${PREFIX}/share/tessdata/
diff --git a/graphics/tesseract/distinfo b/graphics/tesseract/distinfo
index ddc0243809ad..0e3ffa86a11f 100644
--- a/graphics/tesseract/distinfo
+++ b/graphics/tesseract/distinfo
@@ -1,6 +1,6 @@
-MD5 (tesseract-2.01.tar.gz) = fb0e6e7652b985049c11a4bc8e593885
-SHA256 (tesseract-2.01.tar.gz) = d6568b6080590ca7299a697ef8f87505232884c5dfe066d241ebca9f59a64154
-SIZE (tesseract-2.01.tar.gz) = 1050770
+MD5 (tesseract-2.03.tar.gz) = 5777b70b11df16c1ac9aa155d7cfc553
+SHA256 (tesseract-2.03.tar.gz) = f39097176b0b0ac83312b5994970d15283ad0e036bc174d98b87500f572acaae
+SIZE (tesseract-2.03.tar.gz) = 1075509
 MD5 (tesseract-2.00.deu.tar.gz) = 609d91b1ae3759a756b819b5d8403653
 SHA256 (tesseract-2.00.deu.tar.gz) = 6083f16373689ef55651f4dd7773426ab4c32a12ebf7f398af982ac2971fe404
 SIZE (tesseract-2.00.deu.tar.gz) = 1373777
@@ -19,3 +19,12 @@ SIZE (tesseract-2.00.nld.tar.gz) = 1408071
 MD5 (tesseract-2.00.spa.tar.gz) = bc26a777b2384613895677cb8e61ca75
 SHA256 (tesseract-2.00.spa.tar.gz) = b4b83e46be4602a361279c131da2c7f3ba8c51ac26c0e108f958406db6ae320e
 SIZE (tesseract-2.00.spa.tar.gz) = 1365863
+MD5 (tesseract-2.01.deu-f.tar.gz) = 62695819882e372cb86367c1b72cf229
+SHA256 (tesseract-2.01.deu-f.tar.gz) = 410143cf5029b10d274909439de2c407706e4d601cd5e1bb1309dbd46288757e
+SIZE (tesseract-2.01.deu-f.tar.gz) = 1312785
+MD5 (tesseract-2.01.por.tar.gz) = eca21fed3ce1f8ded8dfcedbd1351af7
+SHA256 (tesseract-2.01.por.tar.gz) = 0dc30b76814a7694246ca347ccb402bcd40e5e6f3bf70cca31e6d3ca06210baa
+SIZE (tesseract-2.01.por.tar.gz) = 2107696
+MD5 (tesseract-2.01.vie.tar.gz) = 34a05c2ae1ec8524e6b7647bdeecd408
+SHA256 (tesseract-2.01.vie.tar.gz) = 8026240029341d45a6142cfccf6f0f98e5b1f35cae5725f8379ac4a8de34283f
+SIZE (tesseract-2.01.vie.tar.gz) = 964277
diff --git a/graphics/tesseract/files/patch-ccmain_applybox.cpp b/graphics/tesseract/files/patch-ccmain_applybox.cpp
new file mode 100644
index 000000000000..994d5a489fbc
--- /dev/null
+++ b/graphics/tesseract/files/patch-ccmain_applybox.cpp
@@ -0,0 +1,10 @@
+--- ccmain/applybox.cpp.orig	2009-04-14 09:59:39.000000000 +0000
++++ ccmain/applybox.cpp	2009-04-14 10:06:23.000000000 +0000
+@@ -195,6 +195,7 @@
+     bad_blobs);
+   tprintf ("				Final labelled words:     %6d\n",
+     final_labelled_blob_count);
++  fclose(box_file);
+ }
+ 
+ 
diff --git a/graphics/tesseract/files/patch-cutil__tordvars.h b/graphics/tesseract/files/patch-cutil__tordvars.h
deleted file mode 100644
index d0aaed146cbd..000000000000
--- a/graphics/tesseract/files/patch-cutil__tordvars.h
+++ /dev/null
@@ -1,13 +0,0 @@
---- ./cutil/tordvars.h.orig	2007-05-17 01:33:53.000000000 +0400
-+++ ./cutil/tordvars.h	2007-07-18 15:54:41.000000000 +0400
-@@ -45,8 +45,8 @@
- extern int similarity_debug;                /* Level of debug output */
- extern int write_raw_output;                /* Text before context */
- extern int write_output;                    /* Text file output */
--//extern "C" { extern int display_ratings; }  /* Show the ratings */
--extern int display_ratings;                 /* Show the ratings */
-+extern "C" { extern int display_ratings; }  /* Show the ratings */
-+//extern int display_ratings;                 /* Show the ratings */
- extern int show_bold;                       /* Use bold text */
- extern int display_text;                    /* Show word text */
- extern int display_blocks;                  /* Show word as boxes */
diff --git a/graphics/tesseract/files/patch-image_imgs.cpp b/graphics/tesseract/files/patch-image_imgs.cpp
new file mode 100644
index 000000000000..8f897c66de3b
--- /dev/null
+++ b/graphics/tesseract/files/patch-image_imgs.cpp
@@ -0,0 +1,20 @@
+--- image/imgs.cpp.orig	2009-04-14 10:20:27.000000000 +0000
++++ image/imgs.cpp	2009-04-14 10:25:19.000000000 +0000
+@@ -643,7 +643,7 @@
+                                  //put in destination
+     dest->put_line (xdest, ydest, destext, &copyline, 0);
+   }
+-  delete linesums;
++  delete[] linesums;
+ }
+ 
+ 
+@@ -784,7 +784,7 @@
+                                  //put in destination
+     dest->put_line (xdest, ydest, destext, &copyline, 0);
+   }
+-  delete linesums;
++  delete[] linesums;
+ }
+ 
+ 
diff --git a/graphics/tesseract/files/patch-java_makefile b/graphics/tesseract/files/patch-java_makefile
new file mode 100644
index 000000000000..45dca2dda177
--- /dev/null
+++ b/graphics/tesseract/files/patch-java_makefile
@@ -0,0 +1,11 @@
+--- java/makefile.orig	2009-04-14 12:55:04.000000000 +0000
++++ java/makefile	2009-04-14 12:55:14.000000000 +0000
+@@ -40,7 +40,7 @@
+ 	rm -f ScrollView.jar *.class
+ 
+ # all-am does nothing, to make the java part optional.
+-all all-am :
++all all-am install :
+ 
+ # dist runs the autoconf makefile to archive the files correctly.
+ dist distdir :
diff --git a/graphics/tesseract/files/patch-viewer_scrollview.cpp b/graphics/tesseract/files/patch-viewer_scrollview.cpp
new file mode 100644
index 000000000000..5794366a818c
--- /dev/null
+++ b/graphics/tesseract/files/patch-viewer_scrollview.cpp
@@ -0,0 +1,11 @@
+--- viewer/scrollview.cpp.orig	2009-04-14 11:54:19.000000000 +0000
++++ viewer/scrollview.cpp	2009-04-14 11:56:15.000000000 +0000
+@@ -35,6 +35,8 @@
+ #include <algorithm>
+ #include <vector>
+ #include <string>
++#include <cstring>
++#include <climits>
+ 
+ #include "svutil.h"
+ 
diff --git a/graphics/tesseract/files/patch-viewer_svmnode.cpp b/graphics/tesseract/files/patch-viewer_svmnode.cpp
new file mode 100644
index 000000000000..1e8698321964
--- /dev/null
+++ b/graphics/tesseract/files/patch-viewer_svmnode.cpp
@@ -0,0 +1,10 @@
+--- viewer/svmnode.cpp.orig	2009-04-14 11:54:39.000000000 +0000
++++ viewer/svmnode.cpp	2009-04-14 11:57:02.000000000 +0000
+@@ -27,6 +27,7 @@
+ #include "svmnode.h"
+ 
+ #include <iostream>
++#include <cstring>
+ 
+ #include "scrollview.h"
+ 
diff --git a/graphics/tesseract/files/patch-viewer_svutil.cpp b/graphics/tesseract/files/patch-viewer_svutil.cpp
new file mode 100644
index 000000000000..07d8c07d0230
--- /dev/null
+++ b/graphics/tesseract/files/patch-viewer_svutil.cpp
@@ -0,0 +1,17 @@
+--- viewer/svutil.cpp.orig	2008-04-22 00:07:25.000000000 +0000
++++ viewer/svutil.cpp	2009-04-14 12:13:10.000000000 +0000
+@@ -34,10 +34,14 @@
+ #ifdef __linux__
+ #include <sys/prctl.h>
+ #endif
++#include <netinet/in.h>
+ #endif
+ 
+ #include <iostream>
+ #include <string>
++#include <cstring>
++#include <cstdio>
++#include <cstdlib>
+ 
+ const int kBufferSize = 65536;
+ const int kMaxMsgSize = 4096;
diff --git a/graphics/tesseract/pkg-plist b/graphics/tesseract/pkg-plist
index 92eb4eb1956b..1542e41e4e4c 100644
--- a/graphics/tesseract/pkg-plist
+++ b/graphics/tesseract/pkg-plist
@@ -38,7 +38,6 @@ include/tesseract/closed.h
 include/tesseract/clst.h
 include/tesseract/cluster.h
 include/tesseract/clusttool.h
-include/tesseract/cmndwin.h
 include/tesseract/const.h
 include/tesseract/context.h
 include/tesseract/control.h
@@ -65,8 +64,6 @@ include/tesseract/elst.h
 include/tesseract/elst2.h
 include/tesseract/emalloc.h
 include/tesseract/errcode.h
-include/tesseract/evntlst.h
-include/tesseract/evnts.h
 include/tesseract/expandblob.h
 include/tesseract/extern.h
 include/tesseract/extract.h
@@ -89,8 +86,6 @@ include/tesseract/general.h
 include/tesseract/globaloc.h
 include/tesseract/globals.h
 include/tesseract/gradechop.h
-include/tesseract/grphics.h
-include/tesseract/grphshm.h
 include/tesseract/hashfn.h
 include/tesseract/heuristic.h
 include/tesseract/hideedge.h
@@ -113,6 +108,8 @@ include/tesseract/intproto.h
 include/tesseract/ipoints.h
 include/tesseract/kdtree.h
 include/tesseract/labls.h
+include/tesseract/leptonica_pageseg.h
+include/tesseract/leptonica_pageseg_interface.h
 include/tesseract/linlsq.h
 include/tesseract/listio.h
 include/tesseract/lmedsq.h
@@ -163,15 +160,16 @@ include/tesseract/outlines.h
 include/tesseract/output.h
 include/tesseract/pageblk.h
 include/tesseract/pageres.h
+include/tesseract/pageseg.h
 include/tesseract/pagewalk.h
 include/tesseract/paircmp.h
 include/tesseract/pdblock.h
 include/tesseract/pdclass.h
 include/tesseract/permdawg.h
+include/tesseract/permngram.h
 include/tesseract/permnum.h
 include/tesseract/permute.h
 include/tesseract/pgedit.h
-include/tesseract/pgeditx.h
 include/tesseract/picofeat.h
 include/tesseract/pieces.h
 include/tesseract/pithsync.h
@@ -196,18 +194,14 @@ include/tesseract/rejctmap.h
 include/tesseract/reject.h
 include/tesseract/render.h
 include/tesseract/rwpoly.h
-include/tesseract/sbdmenu.h
-include/tesseract/sbgconst.h
-include/tesseract/sbgdefs.h
-include/tesseract/sbgtypes.h
 include/tesseract/scaleimg.h
 include/tesseract/scanedg.h
 include/tesseract/scanutils.cpp
 include/tesseract/scanutils.h
+include/tesseract/scrollview.h
 include/tesseract/seam.h
 include/tesseract/secname.h
 include/tesseract/serialis.h
-include/tesseract/showim.h
 include/tesseract/sigmenu.h
 include/tesseract/sortflts.h
 include/tesseract/speckle.h
@@ -219,7 +213,10 @@ include/tesseract/stepblob.h
 include/tesseract/stopper.h
 include/tesseract/strngs.h
 include/tesseract/structures.h
-include/tesseract/submen.h
+include/tesseract/svmnode.h
+include/tesseract/svpaint.cpp
+include/tesseract/svshowim.h
+include/tesseract/svutil.h
 include/tesseract/tally.h
 include/tesseract/tessarray.h
 include/tesseract/tessbox.h
@@ -251,8 +248,6 @@ include/tesseract/unicharmap.h
 include/tesseract/unicharset.h
 include/tesseract/varable.h
 include/tesseract/varabled.h
-include/tesseract/varblmen.h
-include/tesseract/varblwin.h
 include/tesseract/variables.h
 include/tesseract/vecfuncs.h
 include/tesseract/werd.h
@@ -265,10 +260,10 @@ lib/libtesseract_ccutil.a
 lib/libtesseract_classify.a
 lib/libtesseract_cutil.a
 lib/libtesseract_dict.a
-lib/libtesseract_display.a
 lib/libtesseract_full.a
 lib/libtesseract_image.a
 lib/libtesseract_main.a
+lib/libtesseract_pageseg.a
 lib/libtesseract_textord.a
 lib/libtesseract_training.a
 lib/libtesseract_viewer.a
@@ -276,6 +271,7 @@ lib/libtesseract_wordrec.a
 share/tessdata/configs/api_config
 share/tessdata/configs/box.train
 share/tessdata/configs/inter
+share/tessdata/configs/kannada
 share/tessdata/configs/makebox
 share/tessdata/configs/unlv
 share/tessdata/confsets
@@ -287,6 +283,14 @@ share/tessdata/deu.pffmtable
 share/tessdata/deu.unicharset
 share/tessdata/deu.user-words
 share/tessdata/deu.word-dawg
+share/tessdata/deu-f.DangAmbigs
+share/tessdata/deu-f.freq-dawg
+share/tessdata/deu-f.inttemp
+share/tessdata/deu-f.normproto
+share/tessdata/deu-f.pffmtable
+share/tessdata/deu-f.unicharset
+share/tessdata/deu-f.user-words
+share/tessdata/deu-f.word-dawg
 share/tessdata/eng.DangAmbigs
 share/tessdata/eng.freq-dawg
 share/tessdata/eng.inttemp
@@ -319,6 +323,14 @@ share/tessdata/nld.pffmtable
 share/tessdata/nld.unicharset
 share/tessdata/nld.user-words
 share/tessdata/nld.word-dawg
+share/tessdata/por.DangAmbigs
+share/tessdata/por.freq-dawg
+share/tessdata/por.inttemp
+share/tessdata/por.normproto
+share/tessdata/por.pffmtable
+share/tessdata/por.unicharset
+share/tessdata/por.user-words
+share/tessdata/por.word-dawg
 share/tessdata/spa.DangAmbigs
 share/tessdata/spa.freq-dawg
 share/tessdata/spa.inttemp
@@ -327,6 +339,14 @@ share/tessdata/spa.pffmtable
 share/tessdata/spa.unicharset
 share/tessdata/spa.user-words
 share/tessdata/spa.word-dawg
+share/tessdata/vie.DangAmbigs
+share/tessdata/vie.freq-dawg
+share/tessdata/vie.inttemp
+share/tessdata/vie.normproto
+share/tessdata/vie.pffmtable
+share/tessdata/vie.unicharset
+share/tessdata/vie.user-words
+share/tessdata/vie.word-dawg
 share/tessdata/tessconfigs/batch
 share/tessdata/tessconfigs/batch.nochop
 share/tessdata/tessconfigs/matdemo