2 files changed, 39 insertions, 12 deletions

diff --git a/graphics/tesseract-data/Makefile b/graphics/tesseract-data/Makefile
index 75c15cd18709..8e61a03ec7d2 100644
--- a/graphics/tesseract-data/Makefile
+++ b/graphics/tesseract-data/Makefile
@@ -6,32 +6,43 @@
 #
 
 PORTNAME=	tesseract-data
-PORTVERSION=	3.00
-PORTREVISION=	1
+PORTVERSION=	3.01
 CATEGORIES=	graphics
 MASTER_SITES=	${MASTER_SITE_GOOGLE_CODE}
-DISTFILES=	${TESSERACT_LANGS:O:u:S|$|.traineddata.gz|}
+DISTFILES=	${TESSERACT_LANGS_300:O:u:S|$|.traineddata.gz|} \
+		${TESSERACT_LANGS_301:O:u:S|^|tesseract-ocr-3.01.|:S|$|.tar.gz|}
 EXTRACT_ONLY=	#empty
 
 MAINTAINER=	bf@FreeBSD.org
 COMMENT=	Additional trained language data for the Tesseract OCR engine
 
+LICENSE=	ASL
+
 DIST_SUBDIR=	tesseract
 PROJECTHOST=	tesseract-ocr
 
-LICENSE=	ASL
 NO_BUILD=	yes
 
-EXTRA_LANGS=	bul cat ces chi_sim chi_tra chr dan dan-frak deu deu-frak ell \
+LANGS_300=	bul cat ces chi_sim chi_tra chr dan dan-frak deu deu-frak ell \
 		fin fra hun ind ita jpn kor lav lit nld nor pol por ron rus slk \
 		slv spa srp swe swe-frak tgl tur ukr vie
 
+LANGS_301=	ara heb hin slk-frak tha
+
 .if make(makesum)
-TESSERACT_LANGS=	${EXTRA_LANGS}
+TESSERACT_LANGS=	${LANGS_300} ${LANGS_301}
 .else
-TESSERACT_LANGS ?=	${EXTRA_LANGS}
+TESSERACT_LANGS ?=	${LANGS_300} ${LANGS_301}
 .endif
 
+TESSERACT_LANGS_300=
+TESSERACT_LANGS_301=
+
+.for l in ${TESSERACT_LANGS}
+TESSERACT_LANGS_300+=	${LANGS_300:M${l}}
+TESSERACT_LANGS_301+=	${LANGS_301:M${l}}
+.endfor
+
 PLIST_DIRSTRY=	share/tessdata
 PLIST_FILES=	${TESSERACT_LANGS:O:u:S|$|.traineddata|:S|^|share/tessdata/|}
 
@@ -42,11 +53,15 @@ pre-fetch:
 	@${ECHO_MSG} ""
 
 do-install:
-	@${INSTALL} -d ${PREFIX}/share/tessdata/
-.for datafile in ${TESSERACT_LANGS:O:u:S|$|.traineddata|}
-	@${GUNZIP_CMD} -c ${_DISTDIR}/${datafile}.gz > ${PREFIX}/share/tessdata/${datafile}
-	@${CHOWN} ${SHAREOWN}:${SHAREGRP} ${PREFIX}/share/tessdata/${datafile}
-	@${CHMOD} ${SHAREMODE} ${PREFIX}/share/tessdata/${datafile}
+	@${INSTALL_DATA} -d ${PREFIX}/share/tessdata
+.for l in ${TESSERACT_LANGS_300:O:u:S|$|.traineddata|}
+	@${GUNZIP_CMD} -c ${_DISTDIR}/${l}.gz > ${PREFIX}/share/tessdata/${l}
+.endfor
+.for l in ${TESSERACT_LANGS_301:O:u}
+	@${TAR} --strip-components 2 -C ${PREFIX}/share/tessdata \
+	-xf ${_DISTDIR}/tesseract-ocr-3.01.${l}.tar.gz *.traineddata
 .endfor
+	@${CHOWN} -R ${SHAREOWN}:${SHAREGRP} ${PREFIX}/share/tessdata
+	@${CHMOD} -R ${SHAREMODE} ${PREFIX}/share/tessdata
 
 .include <bsd.port.mk>
diff --git a/graphics/tesseract-data/distinfo b/graphics/tesseract-data/distinfo
index 33cbfa526e2e..d8ec695fa02c 100644
--- a/graphics/tesseract-data/distinfo
+++ b/graphics/tesseract-data/distinfo
@@ -70,3 +70,15 @@ SHA256 (tesseract/ukr.traineddata.gz) = a54f0ce0843c863f102bfe135939200fc18702ce
 SIZE (tesseract/ukr.traineddata.gz) = 927741
 SHA256 (tesseract/vie.traineddata.gz) = 5f61c32daf9a7071ff0dc95415aed75276538813398e3debb3849bba70bda713
 SIZE (tesseract/vie.traineddata.gz) = 1575539
+SHA256 (tesseract/tesseract-ocr-3.01.ara.tar.gz) = bc0781e71ae688eec172500e70b603cb5277785475c466a3704e5b9e5a805934
+SIZE (tesseract/tesseract-ocr-3.01.ara.tar.gz) = 6611108
+SHA256 (tesseract/tesseract-ocr-3.01.heb.tar.gz) = d5cf025c43453bc5327b2631133b6695d85ad0f700546220fddb72929fb86fc1
+SIZE (tesseract/tesseract-ocr-3.01.heb.tar.gz) = 205379
+SHA256 (tesseract/tesseract-ocr-3.01.heb-com.tar.gz) = 20a9d53172a8dbf1a4bbd7f7e9e3a523a9435c47c751533642e8a9becc8584ff
+SIZE (tesseract/tesseract-ocr-3.01.heb-com.tar.gz) = 270977
+SHA256 (tesseract/tesseract-ocr-3.01.hin.tar.gz) = e97e5d4fcbce0edbd5d3973d603efe7072db17d20e4e5243fe58803710f2a549
+SIZE (tesseract/tesseract-ocr-3.01.hin.tar.gz) = 10104559
+SHA256 (tesseract/tesseract-ocr-3.01.slk-frak.tar.gz) = f35b8a858d509c540a5829d94e8f8800ed25db4c904661000f4c4fa528d5c156
+SIZE (tesseract/tesseract-ocr-3.01.slk-frak.tar.gz) = 292809
+SHA256 (tesseract/tesseract-ocr-3.01.tha.tar.gz) = 5f8af525ebdfa789e0741351e808858e1a613b895becb138352d51b9fbd31322
+SIZE (tesseract/tesseract-ocr-3.01.tha.tar.gz) = 3794178