aboutsummaryrefslogtreecommitdiffstats
path: root/community/tesseract-ocr
diff options
context:
space:
mode:
Diffstat (limited to 'community/tesseract-ocr')
-rw-r--r--community/tesseract-ocr/APKBUILD119
1 files changed, 97 insertions, 22 deletions
diff --git a/community/tesseract-ocr/APKBUILD b/community/tesseract-ocr/APKBUILD
index cb338a431ed..0c4e18ec104 100644
--- a/community/tesseract-ocr/APKBUILD
+++ b/community/tesseract-ocr/APKBUILD
@@ -1,32 +1,107 @@
# Contributor: Carlo Landmeter <clandmeter@alpinelinux.org>
# Maintainer: Francesco Colista <fcolista@alpinelinux.org>
pkgname=tesseract-ocr
-pkgver=5.0.1
+pkgver=5.3.4
_tdver=4.1.0
pkgrel=0
pkgdesc="open source OCR engine"
url="https://github.com/tesseract-ocr/tesseract/releases"
-arch="all"
+# gigantic package that will most likely never be used here
+arch="all !s390x !armhf !x86 !ppc64le"
license="Apache-2.0"
-makedepends="automake autoconf libtool leptonica-dev pango-dev icu-dev
- cairo-dev"
+makedepends="
+ autoconf
+ automake
+ cairo-dev
+ icu-dev
+ leptonica-dev
+ libtool
+ opencl-dev
+ pango-dev
+ "
subpackages="$pkgname-dev"
source="$pkgname-$pkgver.tar.gz::https://github.com/tesseract-ocr/tesseract/archive/$pkgver.tar.gz
- https://github.com/tesseract-ocr/tessdata/archive/${_tdver}/tessdata-${_tdver}.tar.gz
+ https://github.com/tesseract-ocr/tessdata/archive/$_tdver/tessdata-$_tdver.tar.gz
+ "
+builddir="$srcdir"/tesseract-$pkgver
+options="!check" # todo
+
+_langs="
+ afr
+ ara
+ aze
+ bel
+ ben
+ bul
+ cat
+ ces
+ chi_sim
+ chi_tra
+ chr
+ dan
+ deu
+ eng
+ enm
+ epo
+ equ
+ est
+ eus
+ fin
+ fra
+ frk
+ frm
+ glg
+ grc
+ heb
+ hin
+ hrv
+ hun
+ ind
+ isl
+ ita
+ ita_old
+ jpn
+ kan
+ kat
+ khm
+ kor
+ lav
+ lit
+ mal
+ mkd
+ mlt
+ msa
+ nld
+ nor
+ osd
+ pol
+ por
+ ron
+ rus
+ slk
+ slv
+ spa
+ spa_old
+ sqi
+ srp
+ swa
+ swe
+ tam
+ tel
+ tgl
+ tha
+ tur
+ ukr
+ vie
"
-_langs="afr ara aze bel ben bul cat ces chi_sim chi_tra chr dan deu ell
-enm epo equ est eus fin fra frk frm glg grc heb hin hrv hun ind isl ita
-ita_old jpn kan kat kor lav lit mal mkd mlt msa nld nor pol por ron rus slk slv spa spa_old sqi srp swa swe tam tel tgl tha tur ukr vie"
for _lang in $_langs; do
subpackages="$subpackages $pkgname-data-$_lang:_lang_data:noarch"
done
-builddir="$srcdir"/${pkgname/-*}-$pkgver
-
prepare() {
default_prepare
- ./autogen.sh
+ GIT_DIR=. ./autogen.sh
}
build() {
@@ -38,7 +113,8 @@ build() {
--mandir=/usr/share/man \
--infodir=/usr/share/info \
--localstatedir=/var \
- --disable-static
+ --disable-static \
+ --enable-opencl
make
make training
}
@@ -46,22 +122,21 @@ build() {
package() {
make DESTDIR="$pkgdir" install
make DESTDIR="$pkgdir" training-install
- install -D "$srcdir"/tessdata-$_tdver/eng.* \
- "$srcdir"/tessdata-$_tdver/osd.* \
- "$srcdir"/tessdata-$_tdver/equ.* \
- "$pkgdir"/usr/share/tessdata/
+ local lang
+ for lang in $_langs; do
+ install -Dm644 -t "$pkgdir"/usr/share/tessdata \
+ "$srcdir"/tessdata-$_tdver/$lang.traineddata
+ done
}
_lang_data() {
- local lang="${subpkgname#$pkgname-data-}"
+ local lang=${subpkgname#"$pkgname"-data-}
pkgdesc="Tesseract language data for $lang"
- depends="$pkgname"
- mkdir -p "$subpkgdir"/usr/share/tessdata
- mv "$srcdir"/tessdata-$_tdver/$lang.* \
- "$subpkgdir"/usr/share/tessdata/
+ depends="$pkgname=$pkgver-r$pkgrel"
+ amove usr/share/tessdata/$lang.*
}
sha512sums="
-0a4824f71eb76ab022f3b9258d12b9a059adcf2f37fac6442ae1096d0efc80404ba1bfd9c90e8263178c1af606c2aad2b277ceb93efc3b9e93c125f931ff08c1 tesseract-ocr-5.0.1.tar.gz
+a81c98c3754a71093df7b51390ccd43d05f661352b4cb564e403b96d81909664c2ecbf2eb6f37614c4639e6dadbf2329b926d09271dbbdaa302f2d7b6b0d628a tesseract-ocr-5.3.4.tar.gz
c0b55fb5542d25ebd4b56d25155cc8254027c9503af298641686388886403ee26ebf9ef47d21d530d372deeba6a01f6eb17705e1ab46bb02f9f7bf4f63805cd0 tessdata-4.1.0.tar.gz
"