2016-02-06 13:57:02 +01:00
|
|
|
TERMUX_PKG_HOMEPAGE=https://github.com/tesseract-ocr/tesseract
|
|
|
|
|
TERMUX_PKG_DESCRIPTION="Tesseract is probably the most accurate open source OCR engine available"
|
2019-01-21 13:53:58 +01:00
|
|
|
TERMUX_PKG_LICENSE="Apache-2.0"
|
2019-02-04 00:14:50 +01:00
|
|
|
TERMUX_PKG_VERSION=4.0.0
|
2019-02-23 12:38:12 +01:00
|
|
|
TERMUX_PKG_REVISION=1
|
2019-02-04 00:14:50 +01:00
|
|
|
TERMUX_PKG_SHA256=a1f5422ca49a32e5f35c54dee5112b11b99928fc9f4ee6695cdc6768d69f61dd
|
2016-02-06 13:57:02 +01:00
|
|
|
TERMUX_PKG_SRCURL=https://github.com/tesseract-ocr/tesseract/archive/${TERMUX_PKG_VERSION}.tar.gz
|
2019-02-04 00:14:50 +01:00
|
|
|
TERMUX_PKG_DEPENDS="libtool, libuuid, leptonica, libandroid-glob"
|
2016-02-06 13:57:02 +01:00
|
|
|
|
|
|
|
|
termux_step_pre_configure() {
|
2016-03-27 13:21:42 +02:00
|
|
|
# http://blog.matt-swain.com/post/26419042500/installing-tesseract-ocr-on-mac-os-x-lion
|
2016-02-06 13:57:02 +01:00
|
|
|
export LIBLEPT_HEADERSDIR=${TERMUX_PREFIX}/include/leptonica
|
|
|
|
|
|
2017-03-27 05:27:36 +02:00
|
|
|
perl -p -i -e 's|ADD_RT], true|ADD_RT], false|g' configure.ac
|
2016-02-06 13:57:02 +01:00
|
|
|
./autogen.sh
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
termux_step_post_make_install() {
|
2016-03-27 13:21:42 +02:00
|
|
|
# download english trained data
|
2016-02-06 13:57:02 +01:00
|
|
|
cd "${TERMUX_PREFIX}/share/tessdata"
|
|
|
|
|
rm -f eng.*
|
2018-08-22 08:29:09 +02:00
|
|
|
|
|
|
|
|
local checksums
|
|
|
|
|
declare -A checksums
|
|
|
|
|
checksums[cube.bigrams]=64adf2cc0b2a6705368aa357224d1a6739035d5fe892cd0cc457016df5b4280f
|
|
|
|
|
checksums[cube.fold]=2b229895623934b493fe69c51fcc387295d91af8b4e43cc51748b3d269a95eed
|
|
|
|
|
checksums[cube.lm]=a6f769245b0a55f42a3ce157cd19d96828483c3384c6483433ed83579ea16e36
|
|
|
|
|
checksums[cube.nn]=8f345f1c19772dd71a5214bc94175ccf647c003ab77e4143fde48f11bf3cb0ef
|
|
|
|
|
checksums[cube.params]=c2aa2854951bd823d89cc86d53a6d9712a6a885de6fbaf650ff3df48bfed85d7
|
|
|
|
|
checksums[cube.size]=e5f95de7e2754eb2df03451885277ca4573b3770816043ae2e2f09d1f7232604
|
|
|
|
|
checksums[cube.word-freq]=8d612bef20ae3052fce0b8650575a80d87c94d772ec6d1f0c6a1ad591586ea44
|
|
|
|
|
checksums[tesseract_cube.nn]=196bedc8a5bc8c30361c2c9518f648b45b498759cb6041827ff6fbfb8da2a8d1
|
|
|
|
|
checksums[traineddata]=c0515c9f1e0c79e1069fcc05c2b2f6a6841fb5e1082d695db160333c1154f06d
|
|
|
|
|
|
|
|
|
|
mkdir -p $TERMUX_PKG_CACHEDIR/tessdata
|
|
|
|
|
|
2019-02-04 00:14:50 +01:00
|
|
|
termux_download \
|
|
|
|
|
https://raw.githubusercontent.com/tesseract-ocr/tessdata/$TERMUX_PKG_VERSION/eng.traineddata \
|
|
|
|
|
$TERMUX_PKG_CACHEDIR/tessdata/eng.traineddata \
|
|
|
|
|
daa0c97d651c19fba3b25e81317cd697e9908c8208090c94c3905381c23fc047
|
|
|
|
|
cp $TERMUX_PKG_CACHEDIR/tessdata/eng.traineddata .
|
2016-02-06 13:57:02 +01:00
|
|
|
}
|
