swh_storage = <swh.storage.retry.RetryingProxyStorage object at 0x7f4ccee069b0> deposit_client = <swh.loader.package.deposit.loader.ApiClient object at 0x7f4cced17e10> requests_mock_datadir = <requests_mock.mocker.Mocker object at 0x7f4cced17240> def test_deposit_loading_ok_2(swh_storage, deposit_client, requests_mock_datadir): """Field dates should be se appropriately """ external_id = "some-external-id" url = f"https://hal-test.archives-ouvertes.fr/{external_id}" deposit_id = 777 loader = DepositLoader( swh_storage, url, deposit_id, deposit_client, default_filename="archive.zip" ) actual_load_status = loader.load() expected_snapshot_id = "3e68440fdd7c81d283f8f3aebb6f0c8657864192" assert actual_load_status == { "status": "eventful", "snapshot_id": expected_snapshot_id, } assert_last_visit_matches(loader.storage, url, status="full", type="deposit") revision_id = "564d18943d71be80d0d73b43a77cfb205bcde96c" expected_snapshot = Snapshot( id=hash_to_bytes(expected_snapshot_id), branches={ b"HEAD": SnapshotBranch( target=hash_to_bytes(revision_id), target_type=TargetType.REVISION ) }, ) check_snapshot(expected_snapshot, storage=loader.storage) raw_meta = loader.client.metadata_get(deposit_id) # Ensure the date fields are set appropriately in the revision # Retrieve the revision revision = loader.storage.revision_get([hash_to_bytes(revision_id)])[0] assert revision assert revision.date.to_dict() == raw_meta["deposit"]["author_date"] assert revision.committer_date.to_dict() == raw_meta["deposit"]["committer_date"] read_api = f"{DEPOSIT_URL}/{deposit_id}/meta/" provider = { "provider_name": "hal", "provider_type": "deposit_client", "provider_url": "https://hal-test.archives-ouvertes.fr/", "metadata": None, } tool = { "name": "swh-deposit", "version": "0.0.1", "configuration": {"sword_version": "2"}, } assert revision.metadata == { "extrinsic": { "provider": read_api, "raw": { "origin": {"type": "deposit", "url": url,}, "origin_metadata": { "metadata": raw_meta["metadata_dict"], "provider": provider, "tool": tool, }, }, "when": revision.metadata["extrinsic"]["when"], # dynamic }, "original_artifact": [ { "checksums": { "sha1": "f8c63d7c890a7453498e6cf9fef215d85ec6801d", "sha256": "474bf646aeeff6d945eb752b1a9f8a40f3d81a88909ee7bd2d08cc822aa361e6", # noqa }, "filename": "archive.zip", "length": 956830, "url": "https://deposit.softwareheritage.org/1/private/777/raw/", } ], } fetcher = MetadataFetcher(name="swh-deposit", version="0.0.1",) authority = MetadataAuthority( type=MetadataAuthorityType.DEPOSIT_CLIENT, url="https://hal-test.archives-ouvertes.fr/", ) # Check the origin metadata swh side origin_extrinsic_metadata = loader.storage.raw_extrinsic_metadata_get( Origin(url).swhid(), authority ) assert origin_extrinsic_metadata.next_page_token is None all_metadata_raw: List[str] = raw_meta["metadata_raw"] # 1 raw metadata xml + 1 json dict assert len(origin_extrinsic_metadata.results) == len(all_metadata_raw) + 1 origin_swhid = Origin(url).swhid() expected_metadata = [] for idx, raw_meta in enumerate(all_metadata_raw): origin_meta = origin_extrinsic_metadata.results[idx] expected_metadata.append( RawExtrinsicMetadata( target=origin_swhid, discovery_date=origin_meta.discovery_date, metadata=raw_meta.encode(), format="sword-v2-atom-codemeta-v2", authority=authority, fetcher=fetcher, ) ) origin_metadata = { "metadata": all_metadata_raw, "provider": provider, "tool": tool, } expected_metadata.append( RawExtrinsicMetadata( target=origin_swhid, discovery_date=origin_extrinsic_metadata.results[-1].discovery_date, metadata=json.dumps(origin_metadata).encode(), format="original-artifacts-json", authority=authority, fetcher=fetcher, ) ) assert len(origin_extrinsic_metadata.results) == len(expected_metadata) for orig_meta in origin_extrinsic_metadata.results: assert orig_meta in expected_metadata # Check the revision metadata swh side directory_swhid = ExtendedSWHID( object_type=ExtendedObjectType.DIRECTORY, object_id=revision.directory ) actual_directory_metadata = loader.storage.raw_extrinsic_metadata_get( directory_swhid, authority ) assert actual_directory_metadata.next_page_token is None assert len(actual_directory_metadata.results) == len(all_metadata_raw) revision_swhid = CoreSWHID( object_type=ObjectType.REVISION, object_id=hash_to_bytes(revision_id) ) dir_metadata_template = RawExtrinsicMetadata( target=directory_swhid, format="sword-v2-atom-codemeta-v2", authority=authority, fetcher=fetcher, origin=url, revision=revision_swhid, # to satisfy the constructor discovery_date=now(), metadata=b"", ) expected_directory_metadata = [] for idx, raw_meta in enumerate(all_metadata_raw): dir_metadata = actual_directory_metadata.results[idx] expected_directory_metadata.append( RawExtrinsicMetadata.from_dict( { **{ k: v for (k, v) in dir_metadata_template.to_dict().items() if k != "id" }, "discovery_date": dir_metadata.discovery_date, "metadata": raw_meta.encode(), } ) ) > assert actual_directory_metadata.results == expected_directory_metadata E AssertionError: assert [RawExtrinsic...xc1\xc4\x12')] == [RawExtrinsic...\x80\t\x1c>')] E At index 0 diff: RawExtrinsicMetadata(target=ExtendedSWHID(namespace='swh', scheme_version=1, object_id=b'\xfd-\xf1-\xc5SL\x1d\xa1\xe9\x18\x0b\x91Q\x02\xfbo`\x1d\x19', object_type=<ExtendedObjectType.DIRECTORY: 'dir'>), discovery_date=datetime.datetime(2021, 3, 23, 9, 29, 12, tzinfo=datetime.timezone.utc), authority=MetadataAuthority(type=<MetadataAuthorityType.DEPOSIT_CLIENT: 'deposit_client'>, url='https://hal-test.archives-ouvertes.fr/', metadata=None), fetcher=MetadataFetcher(name='swh-deposit', version='0.0.1', metadata=None), format='sword-v2-atom-codemeta-v2', me... E E ...Full output truncated (7 lines hidden), use '-vv' to show .tox/py3/lib/python3.7/site-packages/swh/loader/package/deposit/tests/test_deposit.py:474: AssertionError