# tesseract/cfg - "Build" settings for package

#---------------------------------------------------------------------

[technotes]

1. The source tarball used by this package has been renamed.

#---------------------------------------------------------------------

[buildtimes]

00.10 hours (or 006.27 minutes) - 2021 L.A. dedi box: 6x2 CPU 32GB RAM
00.17 hours (or 009.93 minutes) - ThinkPad W530 i7 4x2 32GB RAM

#---------------------------------------------------------------------

[settings]
build    = default
exepack  = yes
license  = See "license*.txt" in installed tree
licfile  = LICENSE
unpack   = relaxed
setpaths = cairo pango
tmpsize  = 63M

#---------------------------------------------------------------------

[depends]
actools
asciidoc
brotli
bzip2
cairo
curl
gawk
giflib5
grep
gtkcore
jpeg
leptonica
libarchive
libicu
libidn2
liblz4
libpng
libpsl
librtmp
libsasl
libunistring
libwebp
libxml2
libxslt
m4
nghttp2
openjpeg
openldap
openssl
pango
pkgconf
python3-core:strict
sed
tiff                            # May need static version of library
xzutils
zlib
zstd

#---------------------------------------------------------------------

[configure]
MANBASE=$PKGDIR_PROD/man
mkdir -p $MANBASE

export LIBTIFF_CFLAGS="-I$PRODTREE/tiff/include"
export LIBTIFF_LIBS="-ltiff -ljpeg"

./autogen.sh

CXXFLAGS="-fPIC -O2" \
bash ./configure \
    --prefix=$PKGDIR_PROD  \
    --build=$LACARCHBUILD  \
    --host=$LACARCHHOST    \
    --mandir=$MANBASE

#---------------------------------------------------------------------

[postbuild]
LIBDIR=$PKGDIR_PROD/lib
SHADIR=$PKGDIR_PROD/share

export LIBTIFF_CFLAGS="-I$PRODTREE/tiff/include"
export LIBTIFF_LIBS="-ltiff -ljpeg"

ldconfig
make training
make training-install

rm -f $LIBDIR/*.a
mkdir -p $SHADIR/tessdata
cd       $SHADIR/tessdata

BASE=tessdata_best-4.1.0
tar zxf $PKGDIR_SPECIAL/$BASE.tar.gz

rm -fr $BASE/{configs,tessconfigs,pdf.ttf}
mv     $BASE/* .
rm -fr $BASE

pip3install pytesseract

#---------------------------------------------------------------------

# Original URLs. These URLs were valid at one point, but may have died
# since then. If you download newer versions of tarballs [etc.], don't
# delete  the  original versions,  as you may not  be able  to replace
# them.

[urls]
url_debian  = Not checked yet
url_home    = http://code.google.com/p/tesseract-ocr/
url_lfs     = n/a

url_tarball = https://github.com/tesseract-ocr/tesseract/archive/\
5.5.0.tar.gz

#---------------------------------------------------------------------

[history]
220109 Updated from 3.00 to 5.0.1
220301 Updated to 5.1.0
221230 Updated to 5.3.0
230402 Updated to 5.3.1
230713 Updated to 5.3.2
240121 Updated to 5.3.4
240609 Updated to 5.4.0
240614 Updated to 5.4.1
241110 Updated to 5.5.0