Commit | Line | Data |
---|---|---|
93779b4a DM |
1 | TERMUX_PKG_HOMEPAGE=https://github.com/tesseract-ocr/tesseract |
2 | TERMUX_PKG_DESCRIPTION="Tesseract is probably the most accurate open source OCR engine available" | |
f2730837 | 3 | TERMUX_PKG_VERSION=3.04.01 |
93779b4a DM |
4 | TERMUX_PKG_DEPENDS="libtool, libuuid, leptonica" |
5 | TERMUX_PKG_SRCURL=https://github.com/tesseract-ocr/tesseract/archive/${TERMUX_PKG_VERSION}.tar.gz | |
6 | TERMUX_PKG_FOLDERNAME=tesseract-${TERMUX_PKG_VERSION} | |
7 | ||
8 | termux_step_pre_configure() { | |
f2730837 | 9 | # http://blog.matt-swain.com/post/26419042500/installing-tesseract-ocr-on-mac-os-x-lion |
93779b4a DM |
10 | export LIBLEPT_HEADERSDIR=${TERMUX_PREFIX}/include/leptonica |
11 | ||
12 | cd $TERMUX_PKG_SRCDIR | |
13 | perl -p -i -e 's|ADD_RT, true|ADD_RT, false|g' configure.ac | |
14 | ./autogen.sh | |
15 | } | |
16 | ||
17 | termux_step_post_make_install() { | |
f2730837 | 18 | # download english trained data |
93779b4a DM |
19 | cd "${TERMUX_PREFIX}/share/tessdata" |
20 | rm -f eng.* | |
21 | wget https://raw.githubusercontent.com/tesseract-ocr/tessdata/master/eng.cube.bigrams | |
22 | wget https://raw.githubusercontent.com/tesseract-ocr/tessdata/master/eng.cube.fold | |
23 | wget https://raw.githubusercontent.com/tesseract-ocr/tessdata/master/eng.cube.lm | |
24 | wget https://raw.githubusercontent.com/tesseract-ocr/tessdata/master/eng.cube.nn | |
25 | wget https://raw.githubusercontent.com/tesseract-ocr/tessdata/master/eng.cube.params | |
26 | wget https://raw.githubusercontent.com/tesseract-ocr/tessdata/master/eng.cube.size | |
27 | wget https://raw.githubusercontent.com/tesseract-ocr/tessdata/master/eng.cube.word-freq | |
28 | wget https://raw.githubusercontent.com/tesseract-ocr/tessdata/master/eng.tesseract_cube.nn | |
29 | wget https://raw.githubusercontent.com/tesseract-ocr/tessdata/master/eng.traineddata | |
30 | } |