1{
2 lib,
3 buildPythonPackage,
4 fetchurl,
5 protobuf,
6 pymorphy3,
7 pymorphy3-dicts-uk,
8 sentencepiece,
9 setuptools,
10 spacy,
11 spacy-pkuseg,
12 spacy-curated-transformers,
13 sudachipy,
14 sudachidict-core,
15 transformers,
16 writeScript,
17 stdenv,
18 jq,
19 nix,
20 moreutils,
21}:
22let
23 buildModelPackage =
24 {
25 pname,
26 version,
27 sha256,
28 license,
29 }:
30
31 let
32 lang = builtins.substring 0 2 pname;
33 requires-protobuf =
34 pname == "fr_dep_news_trf" || pname == "sl_core_news_trf" || pname == "uk_core_news_trf";
35 requires-sentencepiece = pname == "fr_dep_news_trf" || pname == "sl_core_news_trf";
36 requires-transformers = pname == "uk_core_news_trf";
37 in
38 buildPythonPackage {
39 inherit pname version;
40 pyproject = true;
41
42 src = fetchurl {
43 url = "https://github.com/explosion/spacy-models/releases/download/${pname}-${version}/${pname}-${version}.tar.gz";
44 inherit sha256;
45 };
46
47 propagatedBuildInputs =
48 [ spacy ]
49 ++ lib.optionals (lib.hasSuffix "_trf" pname) [ spacy-curated-transformers ]
50 ++ lib.optionals requires-transformers [ transformers ]
51 ++ lib.optionals (lang == "ja") [
52 sudachidict-core
53 sudachipy
54 ]
55 ++ lib.optionals (lang == "ru") [ pymorphy3 ]
56 ++ lib.optionals (lang == "uk") [
57 pymorphy3
58 pymorphy3-dicts-uk
59 ]
60 ++ lib.optionals (lang == "zh") [ spacy-pkuseg ]
61 ++ lib.optionals requires-sentencepiece [ sentencepiece ];
62
63 postPatch =
64 lib.optionalString requires-protobuf ''
65 substituteInPlace meta.json \
66 --replace-fail "protobuf<3.21.0" "protobuf"
67 ''
68 + lib.optionalString (lang == "zh") ''
69 # Uses numpy 2.x, while the rest of the dependencies still uses
70 # numpy 1.x. Remove once all spaCy packages are updated for
71 # numpy 2.x.
72 substituteInPlace meta.json \
73 --replace-fail "spacy-pkuseg>=1.0.0,<2.0.0" "spacy-pkuseg"
74 '';
75
76 nativeBuildInputs = [ setuptools ] ++ lib.optionals requires-protobuf [ protobuf ];
77
78 pythonImportsCheck = [ pname ];
79
80 passthru.updateScript = writeScript "update-spacy-models" ''
81 #!${stdenv.shell}
82 set -eou pipefail
83 PATH=${
84 lib.makeBinPath [
85 jq
86 nix
87 moreutils
88 ]
89 }
90
91 IFS=. read -r major minor patch <<<"${spacy.version}"
92 spacyVersion="$(echo "$major.$minor.0")"
93
94 pushd pkgs/development/python-modules/spacy/ || exit
95
96 jq -r '.[] | .pname' models.json | while IFS= read -r pname; do
97 if [ "$(jq --arg pname "$pname" -r '.[] | select(.pname == $pname) | .version' models.json)" == "$spacyVersion" ]; then
98 continue
99 fi
100
101 newHash="$(nix-prefetch-url "https://github.com/explosion/spacy-models/releases/download/$pname-$spacyVersion/$pname-$spacyVersion.tar.gz")"
102 jq --arg newHash "$newHash" --arg pname "$pname" --arg spacyVersion "$spacyVersion" \
103 '[(.[] | select(.pname != $pname)), (.[] | select(.pname == $pname) | .sha256 = $newHash | .version = $spacyVersion)] | sort_by(.pname)' \
104 models.json | sponge models.json
105 done
106
107 popd || exit
108 '';
109
110 meta = {
111 description = "Models for the spaCy NLP library";
112 homepage = "https://github.com/explosion/spacy-models";
113 license = lib.licenses.${license};
114 };
115 };
116
117 makeModelSet =
118 models: lib.listToAttrs (map (m: lib.nameValuePair m.pname (buildModelPackage m)) models);
119in
120makeModelSet (lib.importJSON ./models.json)