Exemple #1
0
def test_lister_cgit_run_populates_last_update(requests_mock_datadir,
                                               swh_scheduler):
    """cgit lister returns last updated date"""

    url = "https://git.tizen/cgit"

    urls_without_date = [
        f"https://git.tizen.org/cgit/{suffix_url}" for suffix_url in [
            "All-Projects",
            "All-Users",
            "Lock-Projects",
        ]
    ]

    lister_cgit = CGitLister(swh_scheduler, url=url)

    stats = lister_cgit.run()

    expected_nb_origins = 16
    assert stats == ListerStats(pages=3, origins=expected_nb_origins)

    # test page parsing
    scheduler_origins = swh_scheduler.get_listed_origins(
        lister_cgit.lister_obj.id).results
    assert len(scheduler_origins) == expected_nb_origins

    # test listed repositories
    for listed_origin in scheduler_origins:
        if listed_origin.url in urls_without_date:
            assert listed_origin.last_update is None
        else:
            assert listed_origin.last_update is not None
Exemple #2
0
def test_lister_cgit_with_base_git_url(url, base_git_url, expected_nb_origins,
                                       requests_mock_datadir, swh_scheduler):
    """With base git url provided, listed urls should be the computed origin urls

    """
    lister_cgit = CGitLister(
        swh_scheduler,
        url=url,
        base_git_url=base_git_url,
    )

    stats = lister_cgit.run()

    assert stats == ListerStats(pages=1, origins=expected_nb_origins)

    # test page parsing
    scheduler_origins = swh_scheduler.get_listed_origins(
        lister_cgit.lister_obj.id).results
    assert len(scheduler_origins) == expected_nb_origins

    # test listed repositories
    for listed_origin in scheduler_origins:
        assert listed_origin.visit_type == "git"
        assert listed_origin.url.startswith(base_git_url)
        assert (listed_origin.url.startswith(url) is
                False), f"url should be mapped to {base_git_url}"
Exemple #3
0
def test_lister_cgit_run_with_page(requests_mock_datadir, swh_scheduler):
    """cgit lister supports pagination"""

    url = "https://git.tizen/cgit/"
    lister_cgit = CGitLister(swh_scheduler, url=url)

    stats = lister_cgit.run()

    expected_nb_origins = 16
    assert stats == ListerStats(pages=3, origins=expected_nb_origins)

    # test page parsing
    scheduler_origins = swh_scheduler.get_listed_origins(
        lister_cgit.lister_obj.id).results
    assert len(scheduler_origins) == expected_nb_origins

    # test listed repositories
    for listed_origin in scheduler_origins:
        assert listed_origin.visit_type == "git"
        assert listed_origin.url.startswith("https://git.tizen")

    # test user agent content
    assert len(requests_mock_datadir.request_history) != 0
    for request in requests_mock_datadir.request_history:
        assert "User-Agent" in request.headers
        user_agent = request.headers["User-Agent"]
        assert "Software Heritage Lister" in user_agent
        assert __version__ in user_agent
Exemple #4
0
def test_lister_cgit_get_origin_from_repo_failing(
        requests_mock_datadir_missing_url, swh_scheduler):
    url = "https://git.tizen/cgit/"
    lister_cgit = CGitLister(swh_scheduler, url=url)

    stats = lister_cgit.run()

    expected_nb_origins = 15
    assert stats == ListerStats(pages=3, origins=expected_nb_origins)