2020-09-25 04:45:31 +00:00
|
|
|
{ stdenv, python37, fetchFromGitHub }:
|
2020-04-24 23:36:52 +00:00
|
|
|
|
2020-09-25 04:45:31 +00:00
|
|
|
let
|
|
|
|
python = python37.override {
|
|
|
|
self = python;
|
|
|
|
packageOverrides = self: super: {
|
|
|
|
tornado = super.tornado_4;
|
|
|
|
};
|
|
|
|
};
|
|
|
|
|
|
|
|
in with python.pkgs; buildPythonApplication rec {
|
2020-06-15 15:56:04 +00:00
|
|
|
version = "2.1.19";
|
2020-04-24 23:36:52 +00:00
|
|
|
name = "grab-site-${version}";
|
|
|
|
|
|
|
|
src = fetchFromGitHub {
|
|
|
|
rev = version;
|
|
|
|
owner = "ArchiveTeam";
|
|
|
|
repo = "grab-site";
|
2020-06-15 15:56:04 +00:00
|
|
|
sha256 = "1v1hnhv5knzdl0kj3574ccwlh171vcb7faddp095ycdmiiybalk4";
|
2020-04-24 23:36:52 +00:00
|
|
|
};
|
|
|
|
|
2020-09-25 04:45:31 +00:00
|
|
|
propagatedBuildInputs = [
|
2020-04-24 23:36:52 +00:00
|
|
|
click ludios_wpull manhole lmdb autobahn fb-re2 websockets cchardet
|
|
|
|
];
|
|
|
|
|
|
|
|
checkPhase = ''
|
|
|
|
export PATH=$PATH:$out/bin
|
|
|
|
bash ./tests/offline-tests
|
|
|
|
'';
|
|
|
|
|
|
|
|
meta = with stdenv.lib; {
|
|
|
|
description = "Crawler for web archiving with WARC output";
|
|
|
|
homepage = "https://github.com/ArchiveTeam/grab-site";
|
|
|
|
license = licenses.mit;
|
|
|
|
maintainers = with maintainers; [ ivan ];
|
|
|
|
platforms = platforms.all;
|
|
|
|
};
|
|
|
|
}
|