app-text/tesseract: add live ebuild

AVX/SSE4.1 are detected at runtime, so no need for CPU USE flags

Package-Manager: Portage-2.3.3, Repoman-2.3.1

3 files changed, 169 insertions, 1 deletions

diff --git a/app-text/tesseract/files/tesseract-4.00.00-use-system-piccolo2d.patch b/app-text/tesseract/files/tesseract-4.00.00-use-system-piccolo2d.patch
new file mode 100644
index 000000000000..ad1f0f3df208
--- /dev/null
+++ b/app-text/tesseract/files/tesseract-4.00.00-use-system-piccolo2d.patch
@@ -0,0 +1,27 @@
+diff -Naur tesseract.orig/java/Makefile.am tesseract/java/Makefile.am
+--- tesseract.orig/java/Makefile.am	2017-01-30 17:20:23.135458366 +0100
++++ tesseract/java/Makefile.am	2017-01-30 17:21:29.834462557 +0100
+@@ -36,10 +36,9 @@
+ 	com/google/scrollview/ScrollView.class
+ 
+ SCROLLVIEW_LIBS = \
+-	piccolo2d-core-3.0.jar \
+-	piccolo2d-extras-3.0.jar
++	/usr/share/piccolo2d/lib/piccolo2d.jar
+ 
+-CLASSPATH = piccolo2d-core-3.0.jar:piccolo2d-extras-3.0.jar
++CLASSPATH = /usr/share/piccolo2d/lib/piccolo2d.jar
+ 
+ ScrollView.jar : $(SCROLLVIEW_CLASSES)
+ 	$(JAR) cfm $@ $(srcdir)/Manifest.txt com/google/scrollview/*.class \
+diff -Naur tesseract.orig/viewer/Makefile.am tesseract/viewer/Makefile.am
+--- tesseract.orig/viewer/Makefile.am	2017-01-30 17:20:23.176458368 +0100
++++ tesseract/viewer/Makefile.am	2017-01-30 17:22:39.281467163 +0100
+@@ -17,3 +17,7 @@
+ 
+ libtesseract_viewer_la_SOURCES = \
+     scrollview.cpp svmnode.cpp svutil.cpp svpaint.cpp
++
++if !GRAPHICS_DISABLED
++AM_CPPFLAGS += -DSCROLLVIEW_PATH=/usr/share/tessdata/
++endif
diff --git a/app-text/tesseract/tesseract-4.00.00_alpha.ebuild b/app-text/tesseract/tesseract-4.00.00_alpha.ebuild
index d5d4de1dcc18..029aee60eef5 100644
--- a/app-text/tesseract/tesseract-4.00.00_alpha.ebuild
+++ b/app-text/tesseract/tesseract-4.00.00_alpha.ebuild
@@ -21,7 +21,6 @@ SRC_URI="https://github.com/${MY_PN}/${PN}/archive/${PV/_}.tar.gz -> ${P}.tar.gz
 LICENSE="Apache-2.0"
 SLOT="0"
 KEYWORDS="~alpha ~amd64 ~arm ~mips ~ppc ~ppc64 ~sparc ~x86"
-# Next release will include SSE4.1/AVX
 IUSE="doc examples jpeg math opencl openmp osd png scrollview static-libs tiff training webp"
 
 # List of supported Gentoo linguas and their upstream mapping
diff --git a/app-text/tesseract/tesseract-9999.ebuild b/app-text/tesseract/tesseract-9999.ebuild
new file mode 100644
index 000000000000..883c5c3bdb0d
--- /dev/null
+++ b/app-text/tesseract/tesseract-9999.ebuild
@@ -0,0 +1,142 @@
+# Copyright 1999-2017 Gentoo Foundation
+# Distributed under the terms of the GNU General Public License v2
+# $Id$
+
+EAPI=6
+
+MY_PN="tesseract-ocr"
+LANGPACKV="4.00"
+URI_PREFIX="https://github.com/${MY_PN}/tessdata/raw/${LANGPACKV}/"
+JAVA_PKG_OPT_USE="scrollview"
+
+inherit autotools git-r3 java-pkg-opt-2 toolchain-funcs
+
+DESCRIPTION="An OCR Engine, orginally developed at HP, now open source."
+HOMEPAGE="https://github.com/tesseract-ocr"
+SRC_URI="${URI_PREFIX}eng.traineddata -> eng.traineddata-${LANGPACKV}
+	math? ( ${URI_PREFIX}equ.traineddata -> equ.traineddata-${LANGPACKV} )
+	osd? ( ${URI_PREFIX}osd.traineddata -> osd.traineddata-${LANGPACKV} )"
+EGIT_REPO_URI="https://github.com/${MY_PN}/${PN}.git"
+
+LICENSE="Apache-2.0"
+SLOT="0"
+KEYWORDS=""
+IUSE="doc examples jpeg math opencl openmp osd png scrollview static-libs tiff training webp"
+
+# List of supported Gentoo linguas and their upstream mapping
+# https://github.com/tesseract-ocr/tesseract/wiki/Data-Files
+# "old" variants were regrouped in the matching modern locale
+LANGUAGES="af:afr am:amh ar:ara as:asm az:aze,aze_cyrl be:bel bn:ben bo:bod bs:bos bg:bul ca:cat cs:ces zh:chi_sim,chi_tra cy:cym da:dan de:deu,frk dz:dzo el:ell,grc en:enm eo:epo et:est eu:eus fa:fas fi:fin fr:fra,frm ga:gle gl:glg gu:guj he:heb hi:hin hr:hrv hu:hun id:ind is:isl it:ita,ita_old ja:jpn kn:kan ka:kat,kat_old kk:kaz km:khm ky:kir ko:kor ku:kur lo:lao la:lat lv:lav lt:lit ml:mal mr:mar mk:mkd ms:msa my:mya ne:nep nl:nld no:nor or:ori pa:pan pl:pol pt:por ro:ron ru:rus sa:san si:sin sk:slk sl:slv es:spa,spa_old sq:sqi sr:srp,srp_latn sw:swa sv:swe syc:syr ta:tam te:tel tg:tgk tl:tgl th:tha tr:tur ug:uig uk:ukr uz:uzb,uzb_cyrl vi:vie"
+# Missing matches:
+#	ceb 	Cebuano
+#	chr 	Cherokee
+#	hat 	Haitian; Haitian Creole
+#	iku 	Inuktitut
+#	jav 	Javanese
+#	mlt 	Maltese
+#	pus 	Pushto; Pashto
+#	tir 	Tigrinya
+#	urd 	Urdu
+#	yid 	Yiddish
+# l10n_en provides the additional data:
+#	enm 	English, Middle (1100-1500)
+
+for lang in ${LANGUAGES}; do
+	gentoo_lang=${lang%:*}
+	tess_langs=${lang#*:}
+	for tess_lang in ${tess_langs//,/ }; do
+		SRC_URI+=" l10n_${gentoo_lang}? ( ${URI_PREFIX}${tess_lang}.traineddata -> ${tess_lang}.traineddata-${LANGPACKV} )"
+	done
+	IUSE+=" l10n_${gentoo_lang}"
+done
+
+# With opencl USE=tiff is necessary in leptonica
+CDEPEND=">=media-libs/leptonica-1.74:=[zlib,tiff?,jpeg?,png?,webp?]
+	opencl? (
+		virtual/opencl
+		media-libs/tiff:0=
+		media-libs/leptonica:=[tiff]
+	)
+	scrollview? (
+		>=dev-java/piccolo2d-3.0:0
+	)
+	training? (
+		dev-libs/icu:=
+		x11-libs/pango:=
+		x11-libs/cairo:=
+	)"
+
+DEPEND="${CDEPEND}
+	doc? ( app-doc/doxygen )
+	scrollview? ( >=virtual/jdk-1.7 )"
+
+RDEPEND="${CDEPEND}
+	scrollview? ( >=virtual/jre-1.7 )"
+
+DOCS=( AUTHORS ChangeLog NEWS README.md )
+
+PATCHES=(
+	"${FILESDIR}/${PN}-4.00.00-use-system-piccolo2d.patch"
+)
+
+pkg_pretend() {
+	[[ ${MERGE_TYPE} != binary ]] && use openmp && tc-check-openmp
+}
+
+pkg_setup() {
+	[[ ${MERGE_TYPE} != binary ]] && use openmp && tc-check-openmp
+}
+
+src_unpack() {
+	git-r3_src_unpack
+	for file in ${A}; do
+		if [[ "${file}" == *traineddata* ]]; then
+			cp "${DISTDIR}/${file}" "${S}/tessdata/${file%-*}" || die
+		fi
+	done
+}
+
+src_prepare() {
+	default
+	eautoreconf
+
+	java-pkg-opt-2_src_prepare
+}
+
+src_configure() {
+	local myeconfargs=(
+		--enable-shared
+		$(use_enable opencl)
+		$(use_enable openmp)
+		$(use_enable scrollview graphics)
+		$(use_enable static-libs static)
+	)
+
+	econf "${myeconfargs[@]}"
+}
+
+src_compile() {
+	default
+	use doc && emake doc
+	use scrollview && emake ScrollView.jar JAVAC="javac $(java-pkg_javac-args)"
+	use training && emake training
+}
+
+src_install() {
+	use doc && HTML_DOCS=( doc/html/. )
+	default
+	prune_libtool_files
+
+	if use training; then
+		emake DESTDIR="${D}" training-install
+	fi
+
+	if use examples; then
+		insinto /usr/share/doc/${PF}/examples
+		doins testing/eurotext.tif testing/phototest.tif
+	fi
+
+	insinto /usr/share/tessdata
+	doins tessdata/*traineddata* # language files
+	use scrollview && doins java/ScrollView.jar # scrollview
+}