termux-packages/packages/tesseract/build.sh

TERMUX_PKG_HOMEPAGE=https://github.com/tesseract-ocr/tesseract
TERMUX_PKG_DESCRIPTION="Tesseract is probably the most accurate open source OCR engine available"
TERMUX_PKG_LICENSE="Apache-2.0"
TERMUX_PKG_MAINTAINER="@termux"
TERMUX_PKG_VERSION=5.0.0-rc1
TERMUX_PKG_REVISION=3
TERMUX_PKG_SRCURL=https://github.com/tesseract-ocr/tesseract/archive/${TERMUX_PKG_VERSION}.tar.gz
TERMUX_PKG_SHA256=4bb7bc253edd77dc2bade54855e91d0858efac73df9f0117437337f358196987
TERMUX_PKG_AUTO_UPDATE=true
TERMUX_PKG_DEPENDS="libc++, libtool, libuuid, leptonica, libandroid-glob, zstd, libicu, pango"
TERMUX_PKG_BREAKS="tesseract-dev"
TERMUX_PKG_REPLACES="tesseract-dev"
TERMUX_PKG_FORCE_CMAKE=true
TERMUX_PKG_EXTRA_CONFIGURE_ARGS="
-DUSE_SYSTEM_ICU=on
-DTESSDATA_PREFIX=$TERMUX_PREFIX/share
"
# NEON checks in src/arch/simddetect.cpp don't work for i686 currently
TERMUX_PKG_BLACKLISTED_ARCHES="i686"

termux_step_post_make_install() {
	# download english trained data
	mkdir -p "${TERMUX_PREFIX}"/share/tessdata
	cd "${TERMUX_PREFIX}"/share/tessdata
	rm -f eng.*

	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

	termux_download \
		https://raw.githubusercontent.com/tesseract-ocr/tessdata/4.0.0/eng.traineddata \
		$TERMUX_PKG_CACHEDIR/tessdata/eng.traineddata \
		daa0c97d651c19fba3b25e81317cd697e9908c8208090c94c3905381c23fc047
	cp $TERMUX_PKG_CACHEDIR/tessdata/eng.traineddata .
}
Added Leptonica and Tesseract packages 2016-02-06 13:57:02 +01:00			`TERMUX_PKG_HOMEPAGE=https://github.com/tesseract-ocr/tesseract`
			`TERMUX_PKG_DESCRIPTION="Tesseract is probably the most accurate open source OCR engine available"`
add licenses for more packages 2019-01-21 13:53:58 +01:00			`TERMUX_PKG_LICENSE="Apache-2.0"`
explicitly set maintainer for each package In issue https://github.com/termux/termux-packages/issues/6160 I have found that community repo "its-pointless.github.io" specifies us as maintainer for its packages. This is NOT TRUE and potentially misleads people using these packages. Now TERMUX_PKG_MAINTAINER will contain a default value which is neutral and not specify maintainer. So all packages now have to override it to the correct value. [skip ci] %ci:no-build 2020-12-20 14:16:34 +01:00			`TERMUX_PKG_MAINTAINER="@termux"`
tesseract: update to 5.0.0-rc1 This commit has been automatically submitted by Github Actions. 2021-10-30 02:51:36 +02:00			`TERMUX_PKG_VERSION=5.0.0-rc1`
tesseract: set TESSDATA_PREFIX when compiling So that users do not need to set it manually. 2021-11-10 09:41:11 +01:00			`TERMUX_PKG_REVISION=3`
Added Leptonica and Tesseract packages 2016-02-06 13:57:02 +01:00			`TERMUX_PKG_SRCURL=https://github.com/tesseract-ocr/tesseract/archive/${TERMUX_PKG_VERSION}.tar.gz`
tesseract: update to 5.0.0-rc1 This commit has been automatically submitted by Github Actions. 2021-10-30 02:51:36 +02:00			`TERMUX_PKG_SHA256=4bb7bc253edd77dc2bade54855e91d0858efac73df9f0117437337f358196987`
auto updates: migrate config from github-projects.txt to build.sh-specific Re-enable automatic updates after deleting github-projects.txt. %ci:no-build 2021-10-04 12:43:31 +02:00			`TERMUX_PKG_AUTO_UPDATE=true`
tesseract: build with cmake to fix missing symbol errors For whatever reason building with configure and make leads to symbols not being found (see https://github.com/termux/termux-packages/issues/7953). When switching to cmake it seems that libicu and pango are not considered optional dependencies anymore, so add these to dependecies. 2021-11-09 10:54:31 +01:00			`TERMUX_PKG_DEPENDS="libc++, libtool, libuuid, leptonica, libandroid-glob, zstd, libicu, pango"`
no more develsplit + provide static libraries as separate package %ci:no-build 2019-07-20 10:42:33 +02:00			`TERMUX_PKG_BREAKS="tesseract-dev"`
			`TERMUX_PKG_REPLACES="tesseract-dev"`
tesseract: build with cmake to fix missing symbol errors For whatever reason building with configure and make leads to symbols not being found (see https://github.com/termux/termux-packages/issues/7953). When switching to cmake it seems that libicu and pango are not considered optional dependencies anymore, so add these to dependecies. 2021-11-09 10:54:31 +01:00			`TERMUX_PKG_FORCE_CMAKE=true`
tesseract: set TESSDATA_PREFIX when compiling So that users do not need to set it manually. 2021-11-10 09:41:11 +01:00			`TERMUX_PKG_EXTRA_CONFIGURE_ARGS="`
			`-DUSE_SYSTEM_ICU=on`
			`-DTESSDATA_PREFIX=$TERMUX_PREFIX/share`
			`"`
tesseract: blacklist i686 build for now Needs some work to get code in src/arch/simddetect.cpp for all arches. 2021-11-09 12:11:07 +01:00			`# NEON checks in src/arch/simddetect.cpp don't work for i686 currently`
			`TERMUX_PKG_BLACKLISTED_ARCHES="i686"`
Added Leptonica and Tesseract packages 2016-02-06 13:57:02 +01:00
			`termux_step_post_make_install() {`
Added fmemopen support to Leptonica 2016-03-27 13:21:42 +02:00			`# download english trained data`
tesseract: build with cmake to fix missing symbol errors For whatever reason building with configure and make leads to symbols not being found (see https://github.com/termux/termux-packages/issues/7953). When switching to cmake it seems that libicu and pango are not considered optional dependencies anymore, so add these to dependecies. 2021-11-09 10:54:31 +01:00			`mkdir -p "${TERMUX_PREFIX}"/share/tessdata`
			`cd "${TERMUX_PREFIX}"/share/tessdata`
Added Leptonica and Tesseract packages 2016-02-06 13:57:02 +01:00			`rm -f eng.*`
tesseract: Use checksums when downloading Also cache the downloaded files instead of downloading them on each build. 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`

tesseract: Update from 3.05.02 to 4.0.0 2019-02-04 00:14:50 +01:00			`termux_download \`
tesseract: update to 4.1.0 2019-07-24 14:56:45 +02:00			`https://raw.githubusercontent.com/tesseract-ocr/tessdata/4.0.0/eng.traineddata \`
tesseract: Update from 3.05.02 to 4.0.0 2019-02-04 00:14:50 +01:00			`$TERMUX_PKG_CACHEDIR/tessdata/eng.traineddata \`
			`daa0c97d651c19fba3b25e81317cd697e9908c8208090c94c3905381c23fc047`
			`cp $TERMUX_PKG_CACHEDIR/tessdata/eng.traineddata .`
Added Leptonica and Tesseract packages 2016-02-06 13:57:02 +01:00			`}`