Add tesseract-data 3.00, additional trained language data for the

Tesseract OCR engine.

4 files changed, 127 insertions, 0 deletions

diff --git a/graphics/Makefile b/graphics/Makefile
index 33ae92b65dc8..a0d4f8ad14c6 100644
--- a/graphics/Makefile
+++ b/graphics/Makefile
@@ -931,6 +931,7 @@
     SUBDIR += tcm
     SUBDIR += telak
     SUBDIR += tesseract
+    SUBDIR += tesseract-data
     SUBDIR += tgif
     SUBDIR += tif22pnm
     SUBDIR += tiff
diff --git a/graphics/tesseract-data/Makefile b/graphics/tesseract-data/Makefile
new file mode 100644
index 000000000000..7b69c9f10208
--- /dev/null
+++ b/graphics/tesseract-data/Makefile
@@ -0,0 +1,51 @@
+# New ports collection makefile for:	tesseract-data
+# Date created:		18 March 2011
+# Whom:			bf <bf@FreeBSD.org>
+#
+# $FreeBSD$
+#
+
+PORTNAME=	tesseract-data
+PORTVERSION=	3.00
+CATEGORIES=	graphics
+MASTER_SITES=	${MASTER_SITE_GOOGLE_CODE}
+DISTFILES=	${TESSERACT_LANGS:O:u:S|$|.traineddata.gz|}
+EXTRACT_ONLY=	#empty
+
+MAINTAINER=	bf@FreeBSD.org
+COMMENT=	Additional trained language data for the Tesseract OCR engine
+
+DIST_SUBDIR=	tesseract
+PROJECTHOST=	tesseract-ocr
+
+LICENSE=	ASL
+NO_BUILD=	yes
+
+EXTRA_LANGS=	bul chi_tra chi_sim cat ces 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
+
+.if make(makesum)
+TESSERACT_LANGS=	${EXTRA_LANGS}
+.else
+TESSERACT_LANGS ?=	${EXTRA_LANGS}
+.endif
+
+PLIST_DIRSTRY=	share/tessdata
+PLIST_FILES=	${TESSERACT_LANGS:O:u:S|$|.traineddata|:S|^|share/tessdata/|}
+
+pre-fetch:
+	@${ECHO_MSG} ""
+	@${ECHO_MSG} "The choice of language data to be installed may"
+	@${ECHO_MSG} "be overridden by defining TESSERACT_LANGS. "
+	@${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}
+.endfor
+
+.include <bsd.port.mk>
diff --git a/graphics/tesseract-data/distinfo b/graphics/tesseract-data/distinfo
new file mode 100644
index 000000000000..cb3feac81fff
--- /dev/null
+++ b/graphics/tesseract-data/distinfo
@@ -0,0 +1,70 @@
+SHA256 (tesseract/bul.traineddata.gz) = fc3c650568d014f2337052658e9ddd8efaff6ac74c9a90952d740eafdf400e2d
+SIZE (tesseract/bul.traineddata.gz) = 848731
+SHA256 (tesseract/cat.traineddata.gz) = dcc52db7522c67d2629117b7fde4c114ca51677f7f1efbedc12bf1e7319acde2
+SIZE (tesseract/cat.traineddata.gz) = 995008
+SHA256 (tesseract/ces.traineddata.gz) = 9848a4c137be45659e29f4aa07ccb6357fd4401307478cce0ed5f8e6686480d2
+SIZE (tesseract/ces.traineddata.gz) = 1059966
+SHA256 (tesseract/chi_sim.traineddata.gz) = 40c1cdf62106a0705807e8be5bee235a9e9efe716e22568e3885561a835d7d5b
+SIZE (tesseract/chi_sim.traineddata.gz) = 19732398
+SHA256 (tesseract/chi_tra.traineddata.gz) = 68ac2d83998b7f6e3fa26a58be8be5507e78185a16b3e106fe48a7b2667165fc
+SIZE (tesseract/chi_tra.traineddata.gz) = 27512772
+SHA256 (tesseract/dan.traineddata.gz) = 2104d8b1cfde6fa960e31db826a101a356c77c3efeae6e78cab5a97e62f3410c
+SIZE (tesseract/dan.traineddata.gz) = 958449
+SHA256 (tesseract/dan-frak.traineddata.gz) = ddc9f75b6104618d0682c9c98f5fed1ec52b1b0d5a54d38a20b20f0ab5d9a32b
+SIZE (tesseract/dan-frak.traineddata.gz) = 683525
+SHA256 (tesseract/deu.traineddata.gz) = 9e1845a69d5b6cf93d4fa05d5b8934e7cfaf8b088e6be9e8dac30c7859591ccc
+SIZE (tesseract/deu.traineddata.gz) = 965684
+SHA256 (tesseract/deu-frak.traineddata.gz) = 59f005f735a33039d416f699bff0bae7d42e5ab000dca6e1aae99d99f3776dc5
+SIZE (tesseract/deu-frak.traineddata.gz) = 827836
+SHA256 (tesseract/ell.traineddata.gz) = f8cb765c40733a677fa470370b935c34cfc53ba8de372b33e83ab59c7407195d
+SIZE (tesseract/ell.traineddata.gz) = 944284
+SHA256 (tesseract/fin.traineddata.gz) = 30c05d403295fb659048b5d82db0d87c9cbee5077314e2b52f6fdca9c9ec8520
+SIZE (tesseract/fin.traineddata.gz) = 959833
+SHA256 (tesseract/fra.traineddata.gz) = 1d795f1da1ee20671ffbbf4b2997b424f459861b217c9ffa7abbeb72abd760d6
+SIZE (tesseract/fra.traineddata.gz) = 933372
+SHA256 (tesseract/hun.traineddata.gz) = 7b4d80703067fc5a8bf3da80f7da699f26a665a3e5ca93004fe353a1d6e74f60
+SIZE (tesseract/hun.traineddata.gz) = 1008061
+SHA256 (tesseract/ind.traineddata.gz) = 9b2ee5ab2d33511f5fb8edbaee3d3a448cf8ab9efbf8e5c852d59481317e8218
+SIZE (tesseract/ind.traineddata.gz) = 836752
+SHA256 (tesseract/ita.traineddata.gz) = a6cbaa39fddee521090e48d6a83b6f62e77b5fe7d90ab8ce9ee4d8ada10730f5
+SIZE (tesseract/ita.traineddata.gz) = 939956
+SHA256 (tesseract/jpn.traineddata.gz) = ebbde8c607a05cdd97d492734896a24b2aa17d5b4fb00e9597a96b39c1f22aa3
+SIZE (tesseract/jpn.traineddata.gz) = 14604738
+SHA256 (tesseract/kor.traineddata.gz) = 8d4709341dbda7da9f42bb1d39c4d22d5ca767c2a30fbe36cb2ad152f092b01f
+SIZE (tesseract/kor.traineddata.gz) = 6032090
+SHA256 (tesseract/lav.traineddata.gz) = 305173b54d836dae2f22e4f488734440a83e683636c033cafe066083738ceebc
+SIZE (tesseract/lav.traineddata.gz) = 1018176
+SHA256 (tesseract/lit.traineddata.gz) = 1a04f9e952a76d430c2b9d16140f2d42f27c72a9bbd55b3e8f2b6e701ef0f399
+SIZE (tesseract/lit.traineddata.gz) = 1012936
+SHA256 (tesseract/nld.traineddata.gz) = 2826628b0ff22fc3bb5d6e9d6901a39141b805b78084e0c9ab61f12c28747831
+SIZE (tesseract/nld.traineddata.gz) = 954151
+SHA256 (tesseract/nor.traineddata.gz) = c97cbfd93f676b331296b729d5526d1accc1325474a6b6d91260d03a1c862606
+SIZE (tesseract/nor.traineddata.gz) = 951018
+SHA256 (tesseract/pol.traineddata.gz) = 708e2c59cc4e6451e90fc1ef57b83b809bc354b9e0ef3935e9f181d2a718d5ce
+SIZE (tesseract/pol.traineddata.gz) = 1060352
+SHA256 (tesseract/por.traineddata.gz) = 25df33e4f6c55749d04a5307f1143e31a71a76b7417b91a531c686bf49308b85
+SIZE (tesseract/por.traineddata.gz) = 911645
+SHA256 (tesseract/ron.traineddata.gz) = 367ccb440283e57e4e3f87cd0b97a59a07ceb920e09da8778feafe5e086a9892
+SIZE (tesseract/ron.traineddata.gz) = 929925
+SHA256 (tesseract/rus.traineddata.gz) = 8ced9431d2b1d544fbdd362c36786e3804451da06093bc45acecad84829e5f7a
+SIZE (tesseract/rus.traineddata.gz) = 848490
+SHA256 (tesseract/slk.traineddata.gz) = 037839ad756b9177f7d8f7bd3f01a3a1012094360c5c62a49abbde0a33389511
+SIZE (tesseract/slk.traineddata.gz) = 1091624
+SHA256 (tesseract/slv.traineddata.gz) = bb7318f24972abc380688c67be86d07193e0294b252b60c648720978ad4a8b04
+SIZE (tesseract/slv.traineddata.gz) = 930221
+SHA256 (tesseract/spa.traineddata.gz) = 5de0748b068f35c941e86bba622e23a376a6b084cb094007e7059714f1e030e0
+SIZE (tesseract/spa.traineddata.gz) = 910992
+SHA256 (tesseract/srp.traineddata.gz) = 542b83f62389ff8cc34746dd765345566ab368b63dcf275c223ecb52c2cb9291
+SIZE (tesseract/srp.traineddata.gz) = 977674
+SHA256 (tesseract/swe.traineddata.gz) = 25331ca1f41378a86336604476049810a0c5350417205e300cb2d11048cec2c1
+SIZE (tesseract/swe.traineddata.gz) = 959911
+SHA256 (tesseract/swe-frak.traineddata.gz) = 2cf1a0e1eb8a0e3f8477ebcde44d99a9229f2c20b2d96ecb1ca6c25db5c6fdec
+SIZE (tesseract/swe-frak.traineddata.gz) = 1503568
+SHA256 (tesseract/tgl.traineddata.gz) = d4d88e1fb31771d0b42e65291c03da3a167541df7c9682b81d833a4bd6c9e8cf
+SIZE (tesseract/tgl.traineddata.gz) = 978138
+SHA256 (tesseract/tur.traineddata.gz) = cb4e187f6c25c145252c88a576f3e3c9204c1d77e1c012172a85980c5c01088b
+SIZE (tesseract/tur.traineddata.gz) = 933401
+SHA256 (tesseract/ukr.traineddata.gz) = a54f0ce0843c863f102bfe135939200fc18702ce61ae6ebc571ab49460849365
+SIZE (tesseract/ukr.traineddata.gz) = 927741
+SHA256 (tesseract/vie.traineddata.gz) = 5f61c32daf9a7071ff0dc95415aed75276538813398e3debb3849bba70bda713
+SIZE (tesseract/vie.traineddata.gz) = 1575539
diff --git a/graphics/tesseract-data/pkg-descr b/graphics/tesseract-data/pkg-descr
new file mode 100644
index 000000000000..ea9f224c92e2
--- /dev/null
+++ b/graphics/tesseract-data/pkg-descr
@@ -0,0 +1,5 @@
+This port furnishes various non-English
+trained language data intended for use with
+the Tesseract OCR engine.
+
+WWW: http://code.google.com/p/tesseract-ocr/