2eacddf0dc
* pgadmin: use https homepage * msn-pecan: move homepage to github google code is now unavailable * pidgin-latex: use https for homepage * pidgin-opensteamworks: use github for homepage google code is unavailable * putty: use https for homepage * ponylang: use https for homepage * picolisp: use https for homepage * phonon: use https for homepage * pugixml: use https for homepage * pioneer: use https for homepage * packer: use https for homepage * pokerth: usee https for homepage * procps-ng: use https for homepage * pycaml: use https for homepage * proot: move homepage to .github.io * pius: use https for homepage * pdfread: use https for homepage * postgresql: use https for homepage * ponysay: move homepage to new site * prometheus: use https for homepage * powerdns: use https for homepage * pm-utils: use https for homepage * patchelf: move homepage to https * tesseract: move homepage to github * quodlibet: move homepage from google code * jbrout: move homepage from google code * eiskaltdcpp: move homepage to github * nodejs: use https to homepage * nix: use https for homepage * pdf2djvu: move homepage from google code * game-music-emu: move homepage from google code * vacuum: move homepae from google code
62 lines
2.1 KiB
Nix
62 lines
2.1 KiB
Nix
{ stdenv, fetchFromGitHub, autoreconfHook, autoconf-archive, pkgconfig
|
|
, leptonica, libpng, libtiff, icu, pango, opencl-headers
|
|
|
|
# Supported list of languages or `null' for all available languages
|
|
, enableLanguages ? null
|
|
}:
|
|
|
|
stdenv.mkDerivation rec {
|
|
name = "tesseract-${version}";
|
|
version = "4.00.00alpha-git-20170410";
|
|
|
|
src = fetchFromGitHub {
|
|
owner = "tesseract-ocr";
|
|
repo = "tesseract";
|
|
rev = "36a995bdc92eb2dd8bc5a63205708944a3f990a1";
|
|
sha256 = "0xz3krvap8sdm27v1dyb34lcdmx11wzvxyszpppfsfmjgkvg19bq";
|
|
};
|
|
|
|
tessdata = fetchFromGitHub {
|
|
owner = "tesseract-ocr";
|
|
repo = "tessdata";
|
|
rev = "8bf2e7ad08db9ca174ae2b0b3a7498c9f1f71d40";
|
|
sha256 = "0idwkv4qsmmqhrxcgyhy32yldl3vk054m7dkv4fjswfnalgsx794";
|
|
};
|
|
|
|
nativeBuildInputs = [ pkgconfig autoreconfHook autoconf-archive ];
|
|
buildInputs = [ leptonica libpng libtiff icu pango opencl-headers ];
|
|
|
|
# Copy the .traineddata files of the languages specified in enableLanguages
|
|
# into `$out/share/tessdata' and check afterwards if copying was successful.
|
|
postInstall = let
|
|
mkArg = lang: "-iname ${stdenv.lib.escapeShellArg "${lang}.traineddata"}";
|
|
mkFindArgs = stdenv.lib.concatMapStringsSep " -o " mkArg;
|
|
findLangArgs = if enableLanguages != null
|
|
then "\\( ${mkFindArgs enableLanguages} \\)"
|
|
else "-iname '*.traineddata'";
|
|
in ''
|
|
numLangs="$(find "$tessdata" -mindepth 1 -maxdepth 1 -type f \
|
|
${findLangArgs} -exec cp -t "$out/share/tessdata" {} + -print | wc -l)"
|
|
|
|
${if enableLanguages != null then ''
|
|
expected=${toString (builtins.length enableLanguages)}
|
|
'' else ''
|
|
expected="$(ls -1 "$tessdata/"*.traineddata | wc -l)"
|
|
''}
|
|
|
|
if [ "$numLangs" -ne "$expected" ]; then
|
|
echo "Expected $expected languages, but $numLangs" \
|
|
"were copied to \`$out/share/tessdata'" >&2
|
|
exit 1
|
|
fi
|
|
'';
|
|
|
|
meta = {
|
|
description = "OCR engine";
|
|
homepage = https://github.com/tesseract-ocr/tesseract;
|
|
license = stdenv.lib.licenses.asl20;
|
|
maintainers = with stdenv.lib.maintainers; [viric];
|
|
platforms = with stdenv.lib.platforms; linux;
|
|
};
|
|
}
|