mdw@git.distorted.org.uk Git - termux-packages/blob - packages/tesseract/build.sh

   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"
   3 TERMUX_PKG_VERSION=3.04.01
   4 TERMUX_PKG_BUILD_REVISION=1
   5 TERMUX_PKG_DEPENDS="libtool, libuuid, leptonica"
   6 TERMUX_PKG_SRCURL=https://github.com/tesseract-ocr/tesseract/archive/${TERMUX_PKG_VERSION}.tar.gz
   7 TERMUX_PKG_FOLDERNAME=tesseract-${TERMUX_PKG_VERSION}
   8
   9 termux_step_pre_configure() {
  10         # http://blog.matt-swain.com/post/26419042500/installing-tesseract-ocr-on-mac-os-x-lion
  11         export LIBLEPT_HEADERSDIR=${TERMUX_PREFIX}/include/leptonica
  12
  13         cd $TERMUX_PKG_SRCDIR
  14         perl -p -i -e 's|ADD_RT, true|ADD_RT, false|g' configure.ac
  15         ./autogen.sh
  16 }
  17
  18 termux_step_post_make_install() {
  19         # download english trained data
  20         cd "${TERMUX_PREFIX}/share/tessdata"
  21         rm -f eng.*
  22         wget https://raw.githubusercontent.com/tesseract-ocr/tessdata/master/eng.cube.bigrams
  23         wget https://raw.githubusercontent.com/tesseract-ocr/tessdata/master/eng.cube.fold
  24         wget https://raw.githubusercontent.com/tesseract-ocr/tessdata/master/eng.cube.lm
  25         wget https://raw.githubusercontent.com/tesseract-ocr/tessdata/master/eng.cube.nn
  26         wget https://raw.githubusercontent.com/tesseract-ocr/tessdata/master/eng.cube.params
  27         wget https://raw.githubusercontent.com/tesseract-ocr/tessdata/master/eng.cube.size
  28         wget https://raw.githubusercontent.com/tesseract-ocr/tessdata/master/eng.cube.word-freq
  29         wget https://raw.githubusercontent.com/tesseract-ocr/tessdata/master/eng.tesseract_cube.nn
  30         wget https://raw.githubusercontent.com/tesseract-ocr/tessdata/master/eng.traineddata
  31 }