depot/third_party/nixpkgs/pkgs/development/python-modules/somajo/default.nix
Default email fa5436e0a7 Project import generated by Copybara.
GitOrigin-RevId: e8057b67ebf307f01bdcc8fba94d94f75039d1f6
2024-06-05 17:53:02 +02:00

41 lines
843 B
Nix

{
lib,
stdenv,
fetchFromGitHub,
buildPythonPackage,
pythonOlder,
setuptools,
regex,
}:
buildPythonPackage rec {
pname = "somajo";
version = "2.4.2";
pyproject = true;
disabled = pythonOlder "3.7";
src = fetchFromGitHub {
owner = "tsproisl";
repo = "SoMaJo";
rev = "refs/tags/v${version}";
hash = "sha256-5rlgDnPYTtuVMincG5CgVwNh/IGmZk6ItvzdB/wHmgg=";
};
nativeBuildInputs = [ setuptools ];
propagatedBuildInputs = [ regex ];
# loops forever
doCheck = !stdenv.isDarwin;
pythonImportsCheck = [ "somajo" ];
meta = with lib; {
description = "Tokenizer and sentence splitter for German and English web texts";
mainProgram = "somajo-tokenizer";
homepage = "https://github.com/tsproisl/SoMaJo";
license = licenses.gpl3Plus;
maintainers = with maintainers; [ ];
};
}