app-text/tesseract: Bump to 3.04.01 (bug 590788)

Package-Manager: portage-2.3.0

3 files changed, 178 insertions, 0 deletions

diff --git a/app-text/tesseract/Manifest b/app-text/tesseract/Manifest
index 6ec6df624bf0..d4b3c5d59953 100644
--- a/app-text/tesseract/Manifest
+++ b/app-text/tesseract/Manifest
@@ -70,6 +70,7 @@ DIST swe.traineddata-3.04.00 9460295 SHA256 2a6407bbf187eb1fcfd1633fae9abffcd622
 DIST swe.traineddata.gz 959911 SHA256 25331ca1f41378a86336604476049810a0c5350417205e300cb2d11048cec2c1 SHA512 e38fe9fd7d26420ee1aab3db7f700d58210dbfaef45d03114679d2671bc1942d7b031499e763204e15ebeadafa2f8f063f78941f2299b2c74f635425ddce856a WHIRLPOOL db1225d47e2e3b26372689f96164d8286701dbb2e0f1a82368c784c8d127a0d11248924936abd05d0092ba8ad5ce222b6d113ceb0f440b85faaca37396781480
 DIST tesseract-3.03-rc1.tar.gz 12812907 SHA256 d244956236f7491d74d7f342895f611a6c46c45fa9900173d5b7625d8461d2ea SHA512 4acd6a3c0db5568bd6cb199776a52fbf1643692cf2d9da7449644662461c0c3ef7e5941bb49be94135433a90de400413df062edd612ce9ec9c06e9206a57113f WHIRLPOOL 93f33f62aeb5a9f65a7642bc282138622076989618f787a044ad79463ab9cd9a9ebaeb5e34e52cf5c95c569b01fc552a44d2eda24e697d0c9a045fb626b9aa77
 DIST tesseract-3.04.00.tar.gz 2264427 SHA256 7e6e48b625e1fba9bc825a4ef8c39f12c60aae1084939133b3c6a00f8f8dc38c SHA512 60c31bad40b4c01395782fb98b3326407e75ebdbda962de4fb0e878824efb05140bdcb139dcf7d4f699ae5350e81a00320ebe7fbec083a0d374952488cc59e25 WHIRLPOOL 81b685e4d50d94056a4f99ce6beb4c73983e9449e1eed07a69d4649b0a8231b17fede430a8c181b37216a55e9f6102a578c85d84b2bbe6e9643ca98f16ae3260
+DIST tesseract-3.04.01.tar.gz 2269105 SHA256 57f63e1b14ae04c3932a2683e4be4954a2849e17edd638ffe91bc5a2156adc6a SHA512 816f6c29f8efeaf157891ddb022bd0daf8eb27c0cb8b6df5542e745d209f0ce617245e4583c59a561ed3216044d3afb38f8620ee94cdea656364923dcb140b5b WHIRLPOOL 995502084484d172becdfdaa38888cd21f0fe2b8845fe4a64726f9fbcd35310d64153be6325016dd50ca37b95069112b63d4f0528bd989a58c3b88dc6c6a6813
 DIST tesseract-ocr-3.01.heb-com.tar.gz 270977 SHA256 20a9d53172a8dbf1a4bbd7f7e9e3a523a9435c47c751533642e8a9becc8584ff SHA512 378dc49da535f70c860c29b9a43cae478f6d8403c93a623b1781194c02aad3ea12345f9ea6ebbac4fb2033179964d23fb075e6aa206ec79530b41947d5d44411 WHIRLPOOL cc6463928c05d5904af71ea6170e946e601639761eaab317f0570275e0eaee81cba2c0636dc5dec40f8ba6d16ad498e7177a0c2bf5a886f2598c99568f151fd6
 DIST tesseract-ocr-3.01.osd.tar.gz 4320544 SHA256 7861210fd0970ad30503e8c70d7841de6716bd293d8512fd8787a1a07219b7aa SHA512 ad0d6246574db10dbe5a439ba4266765114b1343228608c5ad35bea125c5e308e286a133ecf2fc9407287af56ea57bb48c32f445fef98256a119a2f432f6e2b6 WHIRLPOOL ffa5a9f209e54a36860e2a2cdee955a83df768fbb814717a2d1e0efe05b3aa2283bc14a33cfa72d7179ddeb684587fe04c0d37cefa944ba2182c9c7e6c9affd4
 DIST tesseract-ocr-3.01.slk-frak.tar.gz 292809 SHA256 f35b8a858d509c540a5829d94e8f8800ed25db4c904661000f4c4fa528d5c156 SHA512 a015719fe72b5463b6e9849de0ba5cdb9ed996ac60a47a7fab360c1028cc52b3ad41872b8b35e988f4430f991791ae0837c6823e479cb1549b65d5f530f62fe6 WHIRLPOOL 79419ea93b9d69d4be3ef8399656a5adb6dc383b9496434c99db8f5f0d6fca70314d68e96f59940674a51b79448e43b5ed59389e382528bf050134e6a6edb00d
diff --git a/app-text/tesseract/files/tesseract-3.04.01-use-system-piccolo2d.patch b/app-text/tesseract/files/tesseract-3.04.01-use-system-piccolo2d.patch
new file mode 100644
index 000000000000..a9dd00419347
--- /dev/null
+++ b/app-text/tesseract/files/tesseract-3.04.01-use-system-piccolo2d.patch
@@ -0,0 +1,27 @@
+diff -U 3 -dHrN tesseract-3.04.01/java/Makefile.am tesseract-3.04.01-scrollview/java/Makefile.am
+--- tesseract-3.04.01/java/Makefile.am	2016-02-16 22:27:01.000000000 +0100
++++ tesseract-3.04.01-scrollview/java/Makefile.am	2016-08-09 13:18:12.810503862 +0200
+@@ -36,10 +36,9 @@
+ 	com/google/scrollview/ScrollView.class
+ 
+ SCROLLVIEW_LIBS = \
+-	$(srcdir)/piccolo2d-core-3.0.jar \
+-	$(srcdir)/piccolo2d-extras-3.0.jar
++	/usr/share/piccolo2d/lib/piccolo2d.jar
+ 
+-CLASSPATH = $(srcdir)/piccolo2d-core-3.0.jar:$(srcdir)/piccolo2d-extras-3.0.jar
++CLASSPATH = /usr/share/piccolo2d/lib/piccolo2d.jar
+ 
+ ScrollView.jar : $(SCROLLVIEW_CLASSES)
+ 	$(JAR) cfm $@ Manifest.txt com/google/scrollview/*.class \
+diff -U 3 -dHrN tesseract-3.04.01/viewer/Makefile.am tesseract-3.04.01-scrollview/viewer/Makefile.am
+--- tesseract-3.04.01/viewer/Makefile.am	2016-02-16 22:27:01.000000000 +0100
++++ tesseract-3.04.01-scrollview/viewer/Makefile.am	2016-08-09 13:20:57.788318147 +0200
+@@ -17,3 +17,7 @@
+ 
+ libtesseract_viewer_la_SOURCES = \
+     scrollview.cpp svmnode.cpp svutil.cpp svpaint.cpp
++
++if !GRAPHICS_DISABLED
++AM_CPPFLAGS += -DSCROLLVIEW_PATH=/usr/share/tessdata/
++endif
diff --git a/app-text/tesseract/tesseract-3.04.01.ebuild b/app-text/tesseract/tesseract-3.04.01.ebuild
new file mode 100644
index 000000000000..ee55a2589d5e
--- /dev/null
+++ b/app-text/tesseract/tesseract-3.04.01.ebuild
@@ -0,0 +1,150 @@
+# Copyright 1999-2016 Gentoo Foundation
+# Distributed under the terms of the GNU General Public License v2
+# $Id$
+
+EAPI=5
+
+MY_PN="tesseract-ocr"
+LANGPACKV="3.04.00"
+URI_PREFIX="https://github.com/${MY_PN}/tessdata/raw/${LANGPACKV}/"
+JAVA_PKG_OPT_USE="scrollview"
+
+inherit eutils autotools java-pkg-opt-2
+
+DESCRIPTION="An OCR Engine, orginally developed at HP, now open source."
+HOMEPAGE="https://github.com/tesseract-ocr"
+SRC_URI="https://github.com/${MY_PN}/${PN}/archive/${PV}.tar.gz -> ${P}.tar.gz
+	${URI_PREFIX}eng.traineddata -> eng.traineddata-${LANGPACKV}
+	doc? ( https://tesseract-ocr.googlecode.com/files/tesseract-ocr-3.02.02-doc-html.tar.gz )
+	math? ( ${URI_PREFIX}equ.traineddata -> equ.traineddata-${LANGPACKV} )
+	linguas_ar? ( ${URI_PREFIX}ara.traineddata -> ara.traineddata-${LANGPACKV} )
+	linguas_bg? ( ${URI_PREFIX}bul.traineddata -> bul.traineddata-${LANGPACKV} )
+	linguas_ca? ( ${URI_PREFIX}cat.traineddata -> cat.traineddata-${LANGPACKV} )
+	linguas_chr? ( ${URI_PREFIX}chr.traineddata -> chr.traineddata-${LANGPACKV} )
+	linguas_cs? ( ${URI_PREFIX}ces.traineddata -> ces.traineddata-${LANGPACKV} )
+	linguas_de? ( ${URI_PREFIX}deu.traineddata -> deu.traineddata-${LANGPACKV}
+				  ${URI_PREFIX}deu_frak.traineddata -> deu_frak.traineddata-${LANGPACKV} )
+	linguas_da? ( ${URI_PREFIX}dan.traineddata -> dan.traineddata-${LANGPACKV}
+				  ${URI_PREFIX}dan_frak.traineddata -> dan_frak.traineddata-${LANGPACKV} )
+	linguas_el? ( ${URI_PREFIX}ell.traineddata -> ell.traineddata-${LANGPACKV} )
+	linguas_es? ( ${URI_PREFIX}spa.traineddata -> spa.traineddata-${LANGPACKV} )
+	linguas_fi? ( ${URI_PREFIX}fin.traineddata -> fin.traineddata-${LANGPACKV} )
+	linguas_fr? ( ${URI_PREFIX}fra.traineddata -> fra.traineddata-${LANGPACKV} )
+	linguas_he? ( ${URI_PREFIX}heb.traineddata -> heb.traineddata-${LANGPACKV} )
+	linguas_hi? ( ${URI_PREFIX}hin.traineddata -> hin.traineddata-${LANGPACKV} )
+	linguas_hu? ( ${URI_PREFIX}hun.traineddata -> hun.traineddata-${LANGPACKV} )
+	linguas_id? ( ${URI_PREFIX}ind.traineddata -> ind.traineddata-${LANGPACKV} )
+	linguas_it? ( ${URI_PREFIX}ita.traineddata -> ita.traineddata-${LANGPACKV} )
+	linguas_ja? ( ${URI_PREFIX}jpn.traineddata -> jpn.traineddata-${LANGPACKV} )
+	linguas_ko? ( ${URI_PREFIX}kor.traineddata -> kor.traineddata-${LANGPACKV} )
+	linguas_lt? ( ${URI_PREFIX}lit.traineddata -> lit.traineddata-${LANGPACKV} )
+	linguas_lv? ( ${URI_PREFIX}lav.traineddata -> lav.traineddata-${LANGPACKV} )
+	linguas_nl? ( ${URI_PREFIX}nld.traineddata -> nld.traineddata-${LANGPACKV} )
+	linguas_no? ( ${URI_PREFIX}nor.traineddata -> nor.traineddata-${LANGPACKV} )
+	linguas_pl? ( ${URI_PREFIX}pol.traineddata -> pol.traineddata-${LANGPACKV} )
+	linguas_pt? ( ${URI_PREFIX}por.traineddata -> por.traineddata-${LANGPACKV} )
+	linguas_ro? ( ${URI_PREFIX}ron.traineddata -> ron.traineddata-${LANGPACKV} )
+	linguas_ru? ( ${URI_PREFIX}rus.traineddata -> rus.traineddata-${LANGPACKV} )
+	linguas_sk? ( ${URI_PREFIX}slk.traineddata -> slk.traineddata-${LANGPACKV}
+				  ${URI_PREFIX}slk_frak.traineddata -> slk_frak.traineddata-${LANGPACKV} )
+	linguas_sl? ( ${URI_PREFIX}slv.traineddata -> slv.traineddata-${LANGPACKV} )
+	linguas_sr? ( ${URI_PREFIX}srp.traineddata -> srp.traineddata-${LANGPACKV} )
+	linguas_sv? ( ${URI_PREFIX}swe.traineddata -> swe.traineddata-${LANGPACKV} )
+	linguas_th? ( ${URI_PREFIX}tha.traineddata -> tha.traineddata-${LANGPACKV} )
+	linguas_tl? ( ${URI_PREFIX}tgl.traineddata -> tgl.traineddata-${LANGPACKV} )
+	linguas_tr? ( ${URI_PREFIX}tur.traineddata -> tur.traineddata-${LANGPACKV} )
+	linguas_uk? ( ${URI_PREFIX}ukr.traineddata -> ukr.traineddata-${LANGPACKV} )
+	linguas_vi? ( ${URI_PREFIX}vie.traineddata -> vie.traineddata-${LANGPACKV} )
+	linguas_zh_CN? ( ${URI_PREFIX}chi_sim.traineddata -> chi_sim.traineddata-${LANGPACKV} )
+	linguas_zh_TW? ( ${URI_PREFIX}chi_tra.traineddata -> chi_tra.traineddata-${LANGPACKV} )
+	osd? ( ${URI_PREFIX}osd.traineddata -> osd.traineddata-${LANGPACKV} )
+"
+
+LICENSE="Apache-2.0"
+SLOT="0"
+KEYWORDS="~alpha ~amd64 ~arm ~ppc ~ppc64 ~sparc ~x86"
+
+IUSE="doc examples jpeg math opencl osd png +scrollview static-libs tiff training -webp linguas_ar linguas_bg linguas_ca linguas_chr linguas_cs linguas_de linguas_da linguas_el linguas_es linguas_fi linguas_fr linguas_he linguas_hi linguas_hu linguas_id linguas_it linguas_ja linguas_ko linguas_lt linguas_lv linguas_nl linguas_no linguas_pl linguas_pt linguas_ro linguas_ru linguas_sk linguas_sl linguas_sr linguas_sv linguas_th linguas_tl linguas_tr linguas_uk linguas_vi linguas_zh_CN linguas_zh_TW"
+
+# With opencl USE=tiff is necessary in leptonica
+CDEPEND=">=media-libs/leptonica-1.71:=[zlib,tiff?,jpeg?,png?,webp?]
+	opencl? (
+		virtual/opencl
+		media-libs/tiff:0=
+		media-libs/leptonica:=[tiff]
+	)
+	scrollview? (
+		>=dev-java/piccolo2d-3.0:0
+	)
+	training? (
+		dev-libs/icu:=
+		x11-libs/pango:=
+		x11-libs/cairo:=
+	)
+"
+
+DEPEND="${CDEPEND}
+	scrollview? ( >=virtual/jdk-1.7 )"
+
+RDEPEND="${CDEPEND}
+	scrollview? ( >=virtual/jre-1.7 )"
+
+DOCS=(AUTHORS ChangeLog NEWS README.md ReleaseNotes )
+
+PATCHES=(
+	"${FILESDIR}/tesseract-2.04-gcc47.patch"
+	"${FILESDIR}/${P}-use-system-piccolo2d.patch"
+)
+
+src_unpack() {
+	unpack ${P}.tar.gz
+	use doc && unpack tesseract-ocr-3.02.02-doc-html.tar.gz
+	find "${DISTDIR}/" -name "*traineddata-${LANGPACKV}" \
+		 -execdir sh -c 'cp -- "$0" "${S}/tessdata/${0%-*}"' '{}' ';' || die
+}
+
+src_prepare() {
+	epatch "${PATCHES[@]}"
+	eautoreconf
+
+	java-pkg-opt-2_src_prepare
+}
+
+src_configure() {
+	local myeconfargs=(
+		--enable-shared
+		$(use_enable opencl)
+		$(use_enable scrollview graphics)
+		$(use_enable static-libs static)
+	)
+
+	econf "${myeconfargs[@]}"
+}
+
+src_compile() {
+	default
+	use scrollview && emake ScrollView.jar JAVAC="javac $(java-pkg_javac-args)"
+	use training && emake training
+}
+
+src_install() {
+	default
+	prune_libtool_files
+
+	if use training; then
+		emake DESTDIR="${D}" training-install
+	fi
+
+	if use examples; then
+		insinto /usr/share/doc/${PF}/examples
+		doins testing/eurotext.tif testing/phototest.tif
+	fi
+
+	if use doc; then
+		dohtml -r "${WORKDIR}/${MY_PN}"/doc/html/*
+	fi
+
+	insinto /usr/share/tessdata
+	doins tessdata/*traineddata* # language files
+	use scrollview && doins java/ScrollView.jar # scrollview
+}