2020-02-28 11:30:50 +01:00
|
|
|
{ pkgs, stdenv, fetchFromGitHub, buildPythonPackage, isPy3k, regex }:
|
|
|
|
|
|
|
|
buildPythonPackage rec {
|
|
|
|
pname = "SoMaJo";
|
2020-06-30 17:28:14 +02:00
|
|
|
version = "2.1.1";
|
2020-02-28 11:30:50 +01:00
|
|
|
disabled = !isPy3k;
|
|
|
|
|
|
|
|
src = fetchFromGitHub {
|
|
|
|
owner = "tsproisl";
|
|
|
|
repo = pname;
|
|
|
|
rev = "v${version}";
|
2020-06-30 17:28:14 +02:00
|
|
|
sha256 = "0yj3yb0qvfdv4wh7nzcsh9in8nzk7b59dvjj0x02rsmlv4kw9ah1";
|
2020-02-28 11:30:50 +01:00
|
|
|
};
|
|
|
|
|
|
|
|
propagatedBuildInputs = [ regex ];
|
|
|
|
|
|
|
|
meta = with stdenv.lib; {
|
|
|
|
description = "Tokenizer and sentence splitter for German and English web texts";
|
|
|
|
homepage = "https://github.com/tsproisl/SoMaJo";
|
|
|
|
license = licenses.gpl3;
|
|
|
|
maintainers = with maintainers; [ danieldk ];
|
|
|
|
};
|
|
|
|
}
|