def test_api_known_invalid_swhid(api_client):
invalid_swhid_sha1 = ["swh:1:cnt:8068d0075010b590762c6cb5682ed53cb3c13de;"]
invalid_swhid_type = ["swh:1:cnn:8068d0075010b590762c6cb5682ed53cb3c13deb"]
url = reverse("api-1-known")
check_api_post_responses(api_client, url, data=invalid_swhid_sha1, status_code=400)
check_api_post_responses(api_client, url, data=invalid_swhid_type, status_code=400)
def test_api_known_swhid_some_present(api_client, content, directory):
content_ = gen_swhid(CONTENT, content["sha1_git"])
directory_ = gen_swhid(DIRECTORY, directory)
unknown_revision_ = gen_swhid(REVISION, random_sha1())
unknown_release_ = gen_swhid(RELEASE, random_sha1())
unknown_snapshot_ = gen_swhid(SNAPSHOT, random_sha1())
input_swhids = [
content_,
directory_,
unknown_revision_,
unknown_release_,
unknown_snapshot_,
]
url = reverse("api-1-known")
resp = check_api_post_responses(api_client, url, data=input_swhids, status_code=200)
assert resp.data == {
content_: {"known": True},
directory_: {"known": True},
unknown_revision_: {"known": False},
unknown_release_: {"known": False},
unknown_snapshot_: {"known": False},
}
def test_api_vault_cook_notfound(api_client, mocker, directory, revision,
unknown_directory, unknown_revision):
mock_vault = mocker.patch("swh.web.common.archive.vault")
mock_vault.cook.side_effect = NotFoundExc("object not found")
mock_vault.fetch.side_effect = NotFoundExc("cooked archive not found")
mock_vault.progress.side_effect = NotFoundExc("cooking request not found")
for obj_type, obj_id in (
("directory", directory),
("revision_gitfast", revision),
):
obj_name = obj_type.split("_")[0]
url = reverse(
f"api-1-vault-cook-{obj_type}",
url_args={f"{obj_type[:3]}_id": obj_id},
)
rv = check_api_get_responses(api_client, url, status_code=404)
assert rv.data["exception"] == "NotFoundExc"
assert (rv.data["reason"] ==
f"Cooking of {obj_name} '{obj_id}' was never requested.")
mock_vault.progress.assert_called_with(obj_type,
hashutil.hash_to_bytes(obj_id))
for obj_type, obj_id in (
("directory", unknown_directory),
("revision_gitfast", unknown_revision),
):
obj_name = obj_type.split("_")[0]
url = reverse(f"api-1-vault-cook-{obj_type}",
url_args={f"{obj_type[:3]}_id": obj_id})
rv = check_api_post_responses(api_client,
url,
data=None,
status_code=404)
assert rv.data["exception"] == "NotFoundExc"
assert rv.data["reason"] == f"{obj_name.title()} '{obj_id}' not found."
mock_vault.cook.assert_called_with(obj_type,
hashutil.hash_to_bytes(obj_id),
email=None)
fetch_url = reverse(
f"api-1-vault-fetch-{obj_type}",
url_args={f"{obj_type[:3]}_id": obj_id},
)
rv = check_api_get_responses(api_client, fetch_url, status_code=404)
assert rv.data["exception"] == "NotFoundExc"
assert (rv.data["reason"] ==
f"Cooked archive for {obj_name} '{obj_id}' not found.")
mock_vault.fetch.assert_called_with(obj_type,
hashutil.hash_to_bytes(obj_id))
def test_api_known_raises_large_payload_error(api_client):
random_swhid = "swh:1:cnt:8068d0075010b590762c6cb5682ed53cb3c13deb"
limit = 10000
err_msg = "The maximum number of SWHIDs this endpoint can receive is 1000"
swhids = [random_swhid for i in range(limit)]
url = reverse("api-1-known")
resp = check_api_post_responses(api_client, url, data=swhids, status_code=413)
assert resp.data == {"exception": "LargePayloadExc", "reason": err_msg}
def test_api_vault_cook(api_client, mocker, directory, revision):
mock_archive = mocker.patch("swh.web.api.views.vault.archive")
for obj_type, obj_id in (
("directory", directory),
("revision_gitfast", revision),
):
fetch_url = reverse(
f"api-1-vault-fetch-{obj_type}",
url_args={f"{obj_type[:3]}_id": obj_id},
)
stub_cook = {
"type": obj_type,
"progress_msg": None,
"task_id": 1,
"task_status": "done",
"object_id": obj_id,
}
stub_fetch = b"content"
mock_archive.vault_cook.return_value = stub_cook
mock_archive.vault_fetch.return_value = stub_fetch
email = "*****@*****.**"
url = reverse(
f"api-1-vault-cook-{obj_type}",
url_args={f"{obj_type[:3]}_id": obj_id},
query_params={"email": email},
)
rv = check_api_post_responses(api_client,
url,
data=None,
status_code=200)
assert rv.data == {
"fetch_url": rv.wsgi_request.build_absolute_uri(fetch_url),
"obj_type": obj_type,
"progress_message": None,
"id": 1,
"status": "done",
"obj_id": obj_id,
}
mock_archive.vault_cook.assert_called_with(
obj_type, hashutil.hash_to_bytes(obj_id), email)
rv = check_http_get_response(api_client, fetch_url, status_code=200)
assert rv["Content-Type"] == "application/gzip"
assert rv.content == stub_fetch
mock_archive.vault_fetch.assert_called_with(
obj_type, hashutil.hash_to_bytes(obj_id))
def test_api_known_swhid_all_present(
api_client, content, directory, release, revision, snapshot
):
input_swhids = [
gen_swhid(CONTENT, content["sha1_git"]),
gen_swhid(DIRECTORY, directory),
gen_swhid(REVISION, revision),
gen_swhid(RELEASE, release),
gen_swhid(SNAPSHOT, snapshot),
]
url = reverse("api-1-known")
resp = check_api_post_responses(api_client, url, data=input_swhids, status_code=200)
assert resp.data == {swhid: {"known": True} for swhid in input_swhids}
def test_api_check_content_known_post(api_client, content):
url = reverse("api-1-content-known")
rv = check_api_post_responses(api_client,
url,
data={"q": content["sha1"]},
status_code=200)
assert rv.data == {
"search_res": [{
"found": True,
"sha1": content["sha1"]
}],
"search_stats": {
"nbfiles": 1,
"pct": 100.0
},
}
def check_created_save_request_status(
api_client,
mocker,
origin_url,
expected_request_status,
scheduler_task_status=None,
scheduler_task_run_status=None,
expected_task_status=None,
visit_date=None,
):
mock_scheduler = mocker.patch("swh.web.common.origin_save.scheduler")
if scheduler_task_status is None:
mock_scheduler.get_tasks.return_value = []
else:
mock_scheduler.get_tasks.return_value = [
{
"priority": "high",
"policy": "oneshot",
"type": "load-git",
"arguments": {"kwargs": {"repo_url": origin_url}, "args": []},
"status": scheduler_task_status,
"id": 1,
}
]
if scheduler_task_run_status is None:
mock_scheduler.get_task_runs.return_value = []
else:
mock_scheduler.get_task_runs.return_value = [
{
"backend_id": "f00c712c-e820-41ce-a07c-9bf8df914205",
"ended": datetime.now(tz=timezone.utc) + timedelta(minutes=5),
"id": 1,
"metadata": {},
"scheduled": datetime.now(tz=timezone.utc),
"started": None,
"status": scheduler_task_run_status,
"task": 1,
}
]
mock_scheduler.create_tasks.return_value = [
{
"priority": "high",
"policy": "oneshot",
"type": "load-git",
"arguments": {"kwargs": {"repo_url": origin_url}, "args": []},
"status": "next_run_not_scheduled",
"id": 1,
}
]
url = reverse(
"api-1-save-origin", url_args={"visit_type": "git", "origin_url": origin_url}
)
mock_visit_date = mocker.patch(
("swh.web.common.origin_save." "_get_visit_info_for_save_request")
)
mock_visit_date.return_value = (visit_date, None)
if expected_request_status != SAVE_REQUEST_REJECTED:
response = check_api_post_responses(api_client, url, data=None, status_code=200)
assert response.data["save_request_status"] == expected_request_status
assert response.data["save_task_status"] == expected_task_status
else:
check_api_post_responses(api_client, url, data=None, status_code=403)