pyrox.dev
tesseract: make tessdata a fix output derivation (#41227)
the full tessdata is nearly a GB, so sparing a copy each time we need to
rebuild tesseract without updating tessdata is worth it.
authored by symphorien and committed by xeji b30d5290 88070395
+34
-30
pkgs/applications/graphics/tesseract/default.nix
+34
-30
pkgs/applications/graphics/tesseract/default.nix
···
1
1
{ stdenv, fetchFromGitHub, autoreconfHook, pkgconfig
2
2
, leptonica, libpng, libtiff, icu, pango, opencl-headers
3
-
4
3
# Supported list of languages or `null' for all available languages
5
4
, enableLanguages ? null
5
+
# if you want just a specific list of languages, optionally specify a hash
6
+
# to make tessdata a fixed output derivation.
7
+
, enableLanguagesHash ? (if enableLanguages == null # all languages
8
+
then "1h48xfzabhn0ldbx5ib67cp9607pr0zpblsy8z6fs4knn0zznfnw"
9
+
else null)
6
10
}:
7
11
12
+
let tessdata = stdenv.mkDerivation ({
13
+
name = "tessdata";
14
+
src = fetchFromGitHub {
15
+
owner = "tesseract-ocr";
16
+
repo = "tessdata";
17
+
rev = "3cf1e2df1fe1d1da29295c9ef0983796c7958b7d";
18
+
# when updating don't forget to update the default value fo enableLanguagesHash
19
+
sha256 = "1v4b63v5nzcxr2y3635r19l7lj5smjmc9vfk0wmxlryxncb4vpg7";
20
+
};
21
+
buildCommand = ''
22
+
cd $src;
23
+
for lang in ${if enableLanguages==null then "*.traineddata" else stdenv.lib.concatMapStringsSep " " (x: x+".traineddata") enableLanguages} ; do
24
+
install -Dt $out/share/tessdata $src/$lang ;
25
+
done;
26
+
'';
27
+
preferLocalBuild = true;
28
+
} // (stdenv.lib.optionalAttrs (enableLanguagesHash != null) {
29
+
# when a hash is given, we make this a fixed output derivation.
30
+
outputHashMode = "recursive";
31
+
outputHashAlgo = "sha256";
32
+
outputHash = enableLanguagesHash;
33
+
}));
34
+
in
35
+
8
36
stdenv.mkDerivation rec {
9
37
name = "tesseract-${version}";
10
38
version = "3.05.00";
···
16
44
sha256 = "11wrpcfl118wxsv2c3w2scznwb48c4547qml42s2bpdz079g8y30";
17
45
};
18
46
19
-
tessdata = fetchFromGitHub {
20
-
owner = "tesseract-ocr";
21
-
repo = "tessdata";
22
-
rev = "3cf1e2df1fe1d1da29295c9ef0983796c7958b7d";
23
-
sha256 = "1v4b63v5nzcxr2y3635r19l7lj5smjmc9vfk0wmxlryxncb4vpg7";
24
-
};
47
+
enableParallelBuilding = true;
25
48
26
49
nativeBuildInputs = [ pkgconfig autoreconfHook ];
27
50
buildInputs = [ leptonica libpng libtiff icu pango opencl-headers ];
28
51
29
52
LIBLEPT_HEADERSDIR = "${leptonica}/include";
30
53
31
-
# Copy the .traineddata files of the languages specified in enableLanguages
32
-
# into `$out/share/tessdata' and check afterwards if copying was successful.
33
-
postInstall = let
34
-
mkArg = lang: "-iname ${stdenv.lib.escapeShellArg "${lang}.traineddata"}";
35
-
mkFindArgs = stdenv.lib.concatMapStringsSep " -o " mkArg;
36
-
findLangArgs = if enableLanguages != null
37
-
then "\\( ${mkFindArgs enableLanguages} \\)"
38
-
else "-iname '*.traineddata'";
39
-
in ''
40
-
numLangs="$(find "$tessdata" -mindepth 1 -maxdepth 1 -type f \
41
-
${findLangArgs} -exec cp -t "$out/share/tessdata" {} + -print | wc -l)"
42
-
43
-
${if enableLanguages != null then ''
44
-
expected=${toString (builtins.length enableLanguages)}
45
-
'' else ''
46
-
expected="$(ls -1 "$tessdata/"*.traineddata | wc -l)"
47
-
''}
48
-
49
-
if [ "$numLangs" -ne "$expected" ]; then
50
-
echo "Expected $expected languages, but $numLangs" \
51
-
"were copied to \`$out/share/tessdata'" >&2
52
-
exit 1
53
-
fi
54
+
postInstall = ''
55
+
for i in ${tessdata}/share/tessdata/*; do
56
+
ln -s $i $out/share/tessdata;
57
+
done
54
58
'';
55
59
56
60
meta = {