OCR engine for Japanese language

WWW: http://code.google.com/p/nhocr

PR:		134720
Submitted by:	Hiroto Kagotani <hiroto.kagotani@gmail.com>

9 files changed, 84 insertions, 0 deletions

diff --git a/japanese/Makefile b/japanese/Makefile
index 70229f349836..10578a08829f 100644
--- a/japanese/Makefile
+++ b/japanese/Makefile
@@ -204,6 +204,7 @@
     SUBDIR += ng-canna
     SUBDIR += ng-devel
     SUBDIR += ngraph-fonts
+    SUBDIR += nhocr
     SUBDIR += nkf
     SUBDIR += nvi-euc-jp
     SUBDIR += nvi-iso-2022-jp
diff --git a/japanese/nhocr/Makefile b/japanese/nhocr/Makefile
new file mode 100644
index 000000000000..7cc7d2424089
--- /dev/null
+++ b/japanese/nhocr/Makefile
@@ -0,0 +1,27 @@
+# New ports collection makefile for:	nhocr
+# Date created:		2009-05-20
+# Whom:			Hiroto Kagotani <hiroto.kagotani@gmail.com>
+#
+# $FreeBSD$
+#
+
+PORTNAME=	nhocr
+PORTVERSION=	0.16
+CATEGORIES=	japanese graphics
+MASTER_SITES=	${MASTER_SITE_GOOGLE_CODE}
+
+MAINTAINER=	hiroto.kagotani@gmail.com
+COMMENT=	OCR engine for Japanese language
+
+BUILD_DEPENDS=	${LOCALBASE}/lib/libsip.a:${PORTSDIR}/graphics/O2-tools
+
+MAKE_JOBS_SAFE=	yes
+GNU_CONFIGURE=	yes
+CONFIGURE_ARGS=	--with-O2tools=${LOCALBASE} --datadir=${DATADIR}
+
+post-install:
+	${MKDIR} ${PREFIX}/share/ocropus/scripts
+	${INSTALL_DATA} ${WRKSRC}/ocropus/rec-nhocr-0.3.lua ${PREFIX}/share/ocropus/scripts/rec-nhocr.lua
+	@${CAT} ${PKGMESSAGE}
+
+.include <bsd.port.mk>
diff --git a/japanese/nhocr/distinfo b/japanese/nhocr/distinfo
new file mode 100644
index 000000000000..5e236c882f36
--- /dev/null
+++ b/japanese/nhocr/distinfo
@@ -0,0 +1,3 @@
+MD5 (nhocr-0.16.tar.gz) = 3b3f08b4ccc052fa7e0f7a7838e3c89d
+SHA256 (nhocr-0.16.tar.gz) = 68369747a0f5989df5a69969f3933e919d70ddd4e8dbdd70629bb06512d459f2
+SIZE (nhocr-0.16.tar.gz) = 5775971
diff --git a/japanese/nhocr/files/patch-libnhocr__nhrec.cpp b/japanese/nhocr/files/patch-libnhocr__nhrec.cpp
new file mode 100644
index 000000000000..9ac795bb14c2
--- /dev/null
+++ b/japanese/nhocr/files/patch-libnhocr__nhrec.cpp
@@ -0,0 +1,11 @@
+--- ./libnhocr/nhrec.cpp.orig	2009-05-10 23:56:11.000000000 +0900
++++ ./libnhocr/nhrec.cpp	2009-05-19 19:24:48.000000000 +0900
+@@ -45,7 +45,7 @@
+ 
+ 
+ // default file/path names
+-static char	*nhocrlibdir0 = "/opt/nhocr/share";
++static char	*nhocrlibdir0 = "/usr/local/share/nhocr";
+ static char	*cctablefile0 = "cctable.utf-8";
+ static char	*dicfile0 = "PLM.dic";
+ 
diff --git a/japanese/nhocr/files/patch-libnhocr__ocrbase.cpp b/japanese/nhocr/files/patch-libnhocr__ocrbase.cpp
new file mode 100644
index 000000000000..9d1c7599adab
--- /dev/null
+++ b/japanese/nhocr/files/patch-libnhocr__ocrbase.cpp
@@ -0,0 +1,11 @@
+--- ./libnhocr/ocrbase.cpp.orig	2009-05-15 15:55:20.000000000 +0900
++++ ./libnhocr/ocrbase.cpp	2009-05-19 19:24:48.000000000 +0900
+@@ -9,7 +9,7 @@
+ #include	<stdlib.h>
+ #include	<string.h>
+ #include	<math.h>
+-#include	<values.h>
++#include	<float.h>
+ 
+ #if defined (__SVR4) && defined (__sun)
+ #include	<float.h>
diff --git a/japanese/nhocr/files/patch-ocropus__rec-nhocr-0.3.lua b/japanese/nhocr/files/patch-ocropus__rec-nhocr-0.3.lua
new file mode 100644
index 000000000000..cdcc23383add
--- /dev/null
+++ b/japanese/nhocr/files/patch-ocropus__rec-nhocr-0.3.lua
@@ -0,0 +1,10 @@
+--- ./ocropus/rec-nhocr-0.3.lua.orig	2009-05-15 17:17:07.000000000 +0900
++++ ./ocropus/rec-nhocr-0.3.lua	2009-05-19 19:34:47.000000000 +0900
+@@ -41,6 +41,6 @@
+    for i = 1,regions:length()-1 do
+       regions:extract(line_image,page_image,i,1)
+       write_image_gray("line.pgm", line_image)
+-      os.execute("/opt/nhocr/bin/nhocr -line -o - line.pgm ; rm line.pgm")
++      os.execute("nhocr -line -o - line.pgm ; rm line.pgm")
+    end
+ end
diff --git a/japanese/nhocr/pkg-descr b/japanese/nhocr/pkg-descr
new file mode 100644
index 000000000000..63cebd41f6c7
--- /dev/null
+++ b/japanese/nhocr/pkg-descr
@@ -0,0 +1,3 @@
+OCR engine for Japanese language
+
+WWW: http://code.google.com/p/nhocr
diff --git a/japanese/nhocr/pkg-message b/japanese/nhocr/pkg-message
new file mode 100644
index 000000000000..d64c5a3d8398
--- /dev/null
+++ b/japanese/nhocr/pkg-message
@@ -0,0 +1,7 @@
+********************************************************************
+The current version of nhocr supports only "-line" recognition mode.
+You may want to install graphics/ocropus, document layout analyzer,
+and use nhocr as a plugin recognizer for ocropus. 
+
+% ocroscript rec-nhocr input.pgm >output.txt
+********************************************************************
diff --git a/japanese/nhocr/pkg-plist b/japanese/nhocr/pkg-plist
new file mode 100644
index 000000000000..4b4d9275764a
--- /dev/null
+++ b/japanese/nhocr/pkg-plist
@@ -0,0 +1,11 @@
+bin/nhocr
+%%DATADIR%%/PLM.dic
+%%DATADIR%%/cctable.utf-8
+include/nhocr.h
+include/ocrbase.h
+include/codelist.h
+lib/libnhocr.a
+share/ocropus/scripts/rec-nhocr.lua
+@dirrm %%DATADIR%%
+@dirrmtry share/ocropus/scripts
+@dirrmtry share/ocropus