BoomMicrophone/nixpkgs-suyu
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

tesseract: Allow to specify a subset of languages.

Browse source 
Especially useful for our OCR based VM tests, where we only need the
english language. By default the argument is null so all languages are
included. If a list of language name is passed only those languages are
enabled, for example:

tesseract.override { enableLanguages = [ "eng" "spa" ]; };

To only enable support for English and Spanish languages.

Signed-off-by: aszlig <aszlig@redmoonstudios.org>
This commit is contained in:
aszlig 2015-05-22 07:45:59 +02:00
parent 8be00dc71d
commit adb7581459
No known key found for this signature in database
GPG key ID: D0EBD0EC8C2DC961
1 changed files with 23 additions and 18 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

41
pkgs/applications/graphics/tesseract/default.nix
View file

@ -1,26 +1,31 @@
{ stdenv, fetchurl, autoconf, automake, libtool, leptonica, libpng, libtiff }:
{ stdenv, fetchurl, autoconf, automake, libtool, leptonica, libpng, libtiff
, enableLanguages ? null
}:
with stdenv.lib;
let
majVersion = "3.02";
version = "${majVersion}.02";
f = lang : sha256 : let
src = fetchurl {
url = "http://tesseract-ocr.googlecode.com/files/tesseract-ocr-${majVersion}.${lang}.tar.gz";
inherit sha256;
};
in
"tar xfvz ${src} -C $out/share/ --strip=1";
mkLang = lang: sha256: let
src = fetchurl {
url = "http://tesseract-ocr.googlecode.com/files/tesseract-ocr-${majVersion}.${lang}.tar.gz";
inherit sha256;
};
in "tar xfvz ${src} -C $out/share/ --strip=1";
extraLanguages = ''
${f "cat" "0d1smiv1b3k9ay2s05sl7q08mb3ln4w5iiiymv2cs8g8333z8jl9"}
${f "rus" "059336mkhsj9m3hwfb818xjlxkcdpy7wfgr62qwz65cx914xl709"}
${f "spa" "1c9iza5mbahd9pa7znnq8yv09v5kz3gbd2sarcgcgc1ps1jc437l"}
${f "nld" "162acxp1yb6gyki2is3ay2msalmfcsnrlsd9wml2ja05k94m6bjy"}
${f "eng" "1y5xf794n832s3lymzlsdm2s9nlrd2v27jjjp0fd9xp7c2ah4461"}
${f "slv" "0rqng43435cly32idxm1lvxkcippvc3xpxbfizwq5j0155ym00dr"}
${f "jpn" "07v8pymd0iwyzh946lxylybda20gsw7p4fsb09jw147955x49gq9"}
'';
wantLang = name: const (enableLanguages == null || elem name enableLanguages);
extraLanguages = mapAttrsToList mkLang (filterAttrs wantLang {
cat = "0d1smiv1b3k9ay2s05sl7q08mb3ln4w5iiiymv2cs8g8333z8jl9";
rus = "059336mkhsj9m3hwfb818xjlxkcdpy7wfgr62qwz65cx914xl709";
spa = "1c9iza5mbahd9pa7znnq8yv09v5kz3gbd2sarcgcgc1ps1jc437l";
nld = "162acxp1yb6gyki2is3ay2msalmfcsnrlsd9wml2ja05k94m6bjy";
eng = "1y5xf794n832s3lymzlsdm2s9nlrd2v27jjjp0fd9xp7c2ah4461";
slv = "0rqng43435cly32idxm1lvxkcippvc3xpxbfizwq5j0155ym00dr";
jpn = "07v8pymd0iwyzh946lxylybda20gsw7p4fsb09jw147955x49gq9";
});
in
stdenv.mkDerivation rec {
@ -40,7 +45,7 @@ stdenv.mkDerivation rec {
'LIBLEPT_HEADERSDIR=${leptonica}/include'
'';
postInstall = extraLanguages;
postInstall = concatStrings extraLanguages;
meta = {
description = "OCR engine";