tjh.dev
1{ lib
2, stdenv
3, fetchFromGitHub
4, buildPythonPackage
5, pythonOlder
6, regex
7}:
8
9buildPythonPackage rec {
10 pname = "somajo";
11 version = "2.2.3";
12 format = "setuptools";
13
14 disabled = pythonOlder "3.7";
15
16 src = fetchFromGitHub {
17 owner = "tsproisl";
18 repo = "SoMaJo";
19 rev = "refs/tags/v${version}";
20 hash = "sha256-EnYw8TSZLXgB4pZaZBgxaO13PpTDx4lGsdGJ+51A6wE=";
21 };
22
23 propagatedBuildInputs = [
24 regex
25 ];
26
27 # loops forever
28 doCheck = !stdenv.isDarwin;
29
30 pythonImportsCheck = [
31 "somajo"
32 ];
33
34 meta = with lib; {
35 description = "Tokenizer and sentence splitter for German and English web texts";
36 homepage = "https://github.com/tsproisl/SoMaJo";
37 license = licenses.gpl3Plus;
38 maintainers = with maintainers; [ ];
39 };
40}