Exemplo n.º 1
0
class Cookie(Model):
    """
    A "Cookie" is a representation of a data object's iteratively enriched metadata.
    """
    def __init__(self, identifier: str):
        from cookiemonster.common.collections import EnrichmentCollection
        self.identifier = identifier
        self.enrichments = EnrichmentCollection()

    def enrich(self, enrichment: Enrichment):
        """
        Enrich this cookie.
        :param enrichment: the enrichment
        """
        self.enrichments.add(enrichment)
Exemplo n.º 2
0
    def setUp(self):
        self.enrichment_collection = EnrichmentCollection()
        interesting_modifications = [
            DataObjectModification(IrodsMetadata({_METADATA_KEY: {0}})),
            DataObjectModification(IrodsMetadata({_METADATA_KEY: {1}}))
        ]
        encoder = DataObjectModificationJSONEncoder()
        self.interesting_modifications_as_metadata = [
            Metadata(encoder.default(modification)) for modification in interesting_modifications
        ]

        interesting_data_objects = [
            DataObject("/path-1", metadata=IrodsMetadata({_METADATA_KEY: {2}})),
            DataObject("/path-2", metadata=IrodsMetadata({_METADATA_KEY: {3}}))
        ]
        encoder = DataObjectJSONEncoder()
        self.interesting_data_objects_as_metadata = [
            Metadata(encoder.default(data_object)) for data_object in interesting_data_objects
        ]
Exemplo n.º 3
0
class TestWasCreationObserved(unittest.TestCase):
    """
    Tests for `was_creation_observed`.
    """
    def setUp(self):
        self.enrichment_collection = EnrichmentCollection()

    def test_no_enrichments(self):
        self.assertFalse(was_creation_observed(self.enrichment_collection))

    def test_unrelated_enrichment(self):
        self.enrichment_collection.add(Enrichment("other", datetime(1, 1, 1), Metadata()))
        self.assertFalse(was_creation_observed(self.enrichment_collection))

    def test_no_creation_observed(self):
        modified_replicas = DataObjectReplicaCollection(
            {DataObjectReplica(IRODS_FIRST_REPLICA_TO_BE_CREATED_VALUE + 1, "")})
        unrelated_modification = DataObjectModification(IrodsMetadata({"other": {"value"}}), modified_replicas)
        unrelated_modification_as_dict = DataObjectModificationJSONEncoder().default(unrelated_modification)
        self.enrichment_collection.add([
            Enrichment(IRODS_UPDATE_ENRICHMENT, datetime(1, 1, 1), UNINTERESTING_DATA_OBJECT_MODIFICATION_AS_METADATA),
            Enrichment(IRODS_UPDATE_ENRICHMENT, datetime(2, 2, 2), Metadata(unrelated_modification_as_dict)),
            Enrichment(IRODS_UPDATE_ENRICHMENT, datetime(3, 3, 3), UNINTERESTING_DATA_OBJECT_MODIFICATION_AS_METADATA),
        ])
        self.assertFalse(was_creation_observed(self.enrichment_collection))

    def test_creation_observed(self):
        modified_replicas = DataObjectReplicaCollection(
            {DataObjectReplica(IRODS_FIRST_REPLICA_TO_BE_CREATED_VALUE, "")})
        modification = DataObjectModification(IrodsMetadata({"other": {"value"}}), modified_replicas)
        modification_as_dict = DataObjectModificationJSONEncoder().default(modification)
        self.enrichment_collection.add([
            Enrichment(IRODS_UPDATE_ENRICHMENT, datetime(1, 1, 1), UNINTERESTING_DATA_OBJECT_MODIFICATION_AS_METADATA),
            Enrichment(IRODS_UPDATE_ENRICHMENT, datetime(2, 2, 2), Metadata(modification_as_dict)),
            Enrichment(IRODS_UPDATE_ENRICHMENT, datetime(3, 3, 3), UNINTERESTING_DATA_OBJECT_MODIFICATION_AS_METADATA),
        ])
        self.assertTrue(was_creation_observed(self.enrichment_collection))
Exemplo n.º 4
0
class TestExtractLatestMetadataKeyValueKnownInIrods(unittest.TestCase):
    """
    Tests for `extract_latest_metadata_key_value_known_in_irods`.
    """
    def setUp(self):
        self.enrichment_collection = EnrichmentCollection()
        interesting_modifications = [
            DataObjectModification(IrodsMetadata({_METADATA_KEY: {0}})),
            DataObjectModification(IrodsMetadata({_METADATA_KEY: {1}}))
        ]
        encoder = DataObjectModificationJSONEncoder()
        self.interesting_modifications_as_metadata = [
            Metadata(encoder.default(modification)) for modification in interesting_modifications
        ]

        interesting_data_objects = [
            DataObject("/path-1", metadata=IrodsMetadata({_METADATA_KEY: {2}})),
            DataObject("/path-2", metadata=IrodsMetadata({_METADATA_KEY: {3}}))
        ]
        encoder = DataObjectJSONEncoder()
        self.interesting_data_objects_as_metadata = [
            Metadata(encoder.default(data_object)) for data_object in interesting_data_objects
        ]

    def test_no_enrichments(self):
        self.assertIsNone(extract_latest_metadata_key_value_known_in_irods(self.enrichment_collection, _METADATA_KEY))

    def test_no_irods_related_enrichments(self):
        self.enrichment_collection.add(Enrichment("other", datetime(1, 1, 1), Metadata()))
        self.assertIsNone(extract_latest_metadata_key_value_known_in_irods(self.enrichment_collection, _METADATA_KEY))

    def test_not_in_irods_update_enrichment(self):
        enrichment = Enrichment(
            IRODS_UPDATE_ENRICHMENT, datetime(1, 1, 1), Metadata(UNINTERESTING_DATA_OBJECT_MODIFICATION_AS_METADATA))
        self.enrichment_collection.add(enrichment)
        self.assertIsNone(extract_latest_metadata_key_value_known_in_irods(self.enrichment_collection, _METADATA_KEY))

    def test_not_in_irods_enrichment(self):
        enrichment = Enrichment(IRODS_ENRICHMENT, datetime(1, 1, 1), UNINTERESTING_DATA_OBJECT_AS_METADATA)
        self.enrichment_collection.add(enrichment)
        self.assertIsNone(extract_latest_metadata_key_value_known_in_irods(self.enrichment_collection, _METADATA_KEY))

    def test_not_in_irods_enrichment_or_irods_update(self):
        enrichments = [
            Enrichment(IRODS_ENRICHMENT, datetime(1, 1, 1), UNINTERESTING_DATA_OBJECT_AS_METADATA),
            Enrichment(
                IRODS_UPDATE_ENRICHMENT, datetime(2, 2, 2),
                Metadata(UNINTERESTING_DATA_OBJECT_MODIFICATION_AS_METADATA))
        ]
        self.enrichment_collection.add(enrichments)
        self.assertIsNone(extract_latest_metadata_key_value_known_in_irods(self.enrichment_collection, _METADATA_KEY))

    def test_in_irods_update_enrichments(self):
        enrichments = [
            Enrichment(IRODS_UPDATE_ENRICHMENT, datetime(1, 1, 1), self.interesting_modifications_as_metadata[0]),
            Enrichment(IRODS_UPDATE_ENRICHMENT, datetime(2, 2, 2), self.interesting_modifications_as_metadata[1]),
            Enrichment(IRODS_UPDATE_ENRICHMENT, datetime(3, 3, 3), UNINTERESTING_DATA_OBJECT_MODIFICATION_AS_METADATA),
        ]
        self.enrichment_collection.add(enrichments)
        value = extract_latest_metadata_key_value_known_in_irods(self.enrichment_collection, _METADATA_KEY)
        self.assertEqual(len(list(value)), 1)
        self.assertIn(1, value)

    def test_in_irods_enrichments(self):
        enrichments = [
            Enrichment(IRODS_ENRICHMENT, datetime(1, 1, 1), self.interesting_data_objects_as_metadata[0]),
            Enrichment(IRODS_ENRICHMENT, datetime(2, 2, 2), self.interesting_data_objects_as_metadata[1]),
            Enrichment(IRODS_ENRICHMENT, datetime(3, 3, 3), UNINTERESTING_DATA_OBJECT_AS_METADATA),
        ]
        self.enrichment_collection.add(enrichments)
        value = extract_latest_metadata_key_value_known_in_irods(self.enrichment_collection, _METADATA_KEY)
        self.assertEqual(len(list(value)), 1)
        self.assertIn(3, value)

    def test_in_irods_enrichments_and_irods_update_enrichments_when_most_recent_from_irods_enrichment(self):
        enrichments = [
            Enrichment(IRODS_ENRICHMENT, datetime(1, 1, 1), self.interesting_data_objects_as_metadata[0]),
            Enrichment(IRODS_UPDATE_ENRICHMENT, datetime(2, 2, 2), self.interesting_modifications_as_metadata[0]),
            Enrichment(IRODS_ENRICHMENT, datetime(3, 3, 3), UNINTERESTING_DATA_OBJECT_AS_METADATA),
            Enrichment(IRODS_UPDATE_ENRICHMENT, datetime(4, 4, 4), self.interesting_modifications_as_metadata[1]),
            Enrichment(IRODS_ENRICHMENT, datetime(5, 5, 5), self.interesting_data_objects_as_metadata[1])
        ]
        self.enrichment_collection.add(enrichments)
        value = extract_latest_metadata_key_value_known_in_irods(self.enrichment_collection, _METADATA_KEY)
        self.assertEqual(len(list(value)), 1)
        self.assertIn(3, value)

    def test_in_irods_enrichments_and_irods_update_enrichments_when_most_recent_from_irods_update_enrichment(self):
        enrichments = [
            Enrichment(IRODS_UPDATE_ENRICHMENT, datetime(1, 1, 1), self.interesting_modifications_as_metadata[0]),
            Enrichment(IRODS_ENRICHMENT, datetime(2, 2, 2), self.interesting_data_objects_as_metadata[0]),
            Enrichment(IRODS_ENRICHMENT, datetime(3, 3, 3), UNINTERESTING_DATA_OBJECT_AS_METADATA),
            Enrichment(IRODS_ENRICHMENT, datetime(4, 4, 4), self.interesting_data_objects_as_metadata[1]),
            Enrichment(IRODS_UPDATE_ENRICHMENT, datetime(5, 5, 5), self.interesting_modifications_as_metadata[1])
        ]
        self.enrichment_collection.add(enrichments)
        value = extract_latest_metadata_key_value_known_in_irods(self.enrichment_collection, _METADATA_KEY)
        self.assertEqual(len(list(value)), 1)
        self.assertIn(1, value)
Exemplo n.º 5
0
 def setUp(self):
     self.enrichment_collection = EnrichmentCollection()
Exemplo n.º 6
0
class TestEnrichmentCollection(unittest.TestCase):
    """
    Tests for `EnrichmentCollection`.
    """
    def setUp(self):
        self.enrichments = EnrichmentCollection()

    def test_enrichment_no_diff(self):
        self.enrichments.add(Enrichment('source', datetime(1, 1, 1), Metadata()))
        self.enrichments.add(Enrichment('source', datetime(2, 2, 2), Metadata()))

        diffs = get_enrichment_changes_from_source(self.enrichments, 'source')
        self.assertEqual(len(diffs), 0)

    def test_enrichment_diff(self):
        self.enrichments.add(Enrichment('source', datetime(1, 1, 1), Metadata({'foo': 123, 'bar': 456, 'quux': 789})))
        self.enrichments.add(Enrichment('source', datetime(2, 2, 2), Metadata({'xyz': 123, 'bar': 456, 'quux': 999})))

        diffs = get_enrichment_changes_from_source(self.enrichments, 'source')
        self.assertEqual(len(diffs), 1)

        diff = diffs[0]
        self.assertIsInstance(diff, EnrichmentDiff)
        self.assertEqual(diff.source, 'source')
        self.assertEqual(diff.timestamp, datetime(2, 2, 2))
        self.assertTrue(diff.is_different())
        self.assertEqual(diff.additions, Metadata({'xyz': 123, 'quux': 999}))
        self.assertEqual(diff.deletions, Metadata({'foo': 123, 'quux': 789}))

    def test_enrichment_diff_by_key(self):
        self.enrichments.add(Enrichment('source', datetime(1, 1, 1), Metadata({'foo': 123, 'bar': 456, 'quux': 789})))
        self.enrichments.add(Enrichment('source', datetime(2, 2, 2), Metadata({'xyz': 123, 'bar': 456, 'quux': 999})))

        diffs = get_enrichment_changes_from_source(self.enrichments, 'source', 'foo')
        self.assertEqual(len(diffs), 1)

        diff = diffs[0]
        self.assertEqual(diff.additions, Metadata())
        self.assertEqual(diff.deletions, Metadata({'foo': 123}))

    def test_enrichment_diff_by_keys(self):
        self.enrichments.add(Enrichment('source', datetime(1, 1, 1), Metadata({'foo': 123, 'bar': 456, 'quux': 789})))
        self.enrichments.add(Enrichment('source', datetime(2, 2, 2), Metadata({'xyz': 123, 'bar': 456, 'quux': 999})))

        diffs = get_enrichment_changes_from_source(self.enrichments, 'source', ['bar', 'quux'])
        self.assertEqual(len(diffs), 1)

        diff = diffs[0]
        self.assertEqual(diff.additions, Metadata({'quux': 999}))
        self.assertEqual(diff.deletions, Metadata({'quux': 789}))

    def test_enrichment_diff_from_timestamp(self):
        self.enrichments.add(Enrichment('source', datetime(1, 1, 1), Metadata()))
        self.enrichments.add(Enrichment('source', datetime(2, 2, 2), Metadata({'foo': 123})))
        self.enrichments.add(Enrichment('source', datetime(3, 3, 3), Metadata({'bar': 123})))

        all_diffs = get_enrichment_changes_from_source(self.enrichments, 'source')
        self.assertEqual(len(all_diffs), 2)

        since_diffs = get_enrichment_changes_from_source(self.enrichments, 'source', since=datetime(3, 3, 3))
        self.assertEqual(len(since_diffs), 1)

        diff = since_diffs[0]
        self.assertEqual(diff.timestamp, datetime(3, 3, 3))
        self.assertEqual(diff.additions, Metadata({'bar': 123}))
        self.assertEqual(diff.deletions, Metadata({'foo': 123}))
Exemplo n.º 7
0
 def __init__(self, identifier: str):
     from cookiemonster.common.collections import EnrichmentCollection
     self.identifier = identifier
     self.enrichments = EnrichmentCollection()
Exemplo n.º 8
0
class TestEnrichmentCollection(unittest.TestCase):
    """
    Tests for `EnrichmentCollection`.
    """
    assert _ENRICHMENT_1.timestamp < _ENRICHMENT_2.timestamp < _ENRICHMENT_3.timestamp

    def setUp(self):
        self.enrichments = EnrichmentCollection()

    def test_eq_with_none(self):
        self.assertNotEqual(_ENRICHMENT_1, None)

    def test_eq_with_other_object_type(self):
        self.assertNotEqual(_ENRICHMENT_1, object())

    def test_eq_with_different_enrichment(self):
        self.assertNotEqual(_ENRICHMENT_1, _ENRICHMENT_2)

    def test_eq_with_same_enrichment(self):
        self.assertEqual(_ENRICHMENT_1, deepcopy(_ENRICHMENT_1))

    def test_eq_with_self(self):
        self.assertEqual(_ENRICHMENT_1, _ENRICHMENT_1)

    def test_init_with_enrichments(self):
        enrichments = EnrichmentCollection([_ENRICHMENT_2, _ENRICHMENT_3, _ENRICHMENT_1])
        self.assertSequenceEqual(enrichments, [_ENRICHMENT_1, _ENRICHMENT_2, _ENRICHMENT_3])

    def test_add_with_no_previous(self):
        self.enrichments.add(_ENRICHMENT_1)
        self.assertSequenceEqual(self.enrichments, [_ENRICHMENT_1])

    def test_add_with_older_previous(self):
        self.enrichments.add(_ENRICHMENT_1)
        self.enrichments.add(_ENRICHMENT_2)
        self.assertSequenceEqual(self.enrichments, [_ENRICHMENT_1, _ENRICHMENT_2])

    def test_add_with_newer_previous(self):
        self.enrichments.add(_ENRICHMENT_2)
        self.enrichments.add(_ENRICHMENT_1)
        self.assertSequenceEqual(self.enrichments, [_ENRICHMENT_1, _ENRICHMENT_2])

    def test_add_with_older_and_newer_previous(self):
        self.enrichments.add(_ENRICHMENT_1)
        self.enrichments.add(_ENRICHMENT_3)
        self.enrichments.add(_ENRICHMENT_2)
        self.assertSequenceEqual(self.enrichments, [_ENRICHMENT_1, _ENRICHMENT_2, _ENRICHMENT_3])

    def test_add_multiple(self):
        self.enrichments.add([_ENRICHMENT_3, _ENRICHMENT_1, _ENRICHMENT_2])
        self.assertSequenceEqual(self.enrichments, [_ENRICHMENT_1, _ENRICHMENT_2, _ENRICHMENT_3])

    def test_get_most_recent_from_source_when_none_from_source(self):
        self.assertIsNone(self.enrichments.get_most_recent_from_source("other_source"))

    def test_get_most_recent_from_source_when_multiple_from_source(self):
        assert _ENRICHMENT_1.source == _ENRICHMENT_2.source == _ENRICHMENT_3.source
        self.enrichments.add([_ENRICHMENT_1, _ENRICHMENT_2, _ENRICHMENT_3])
        self.assertEqual(self.enrichments.get_most_recent_from_source(_ENRICHMENT_1.source), _ENRICHMENT_3)

    def test_get_all_since_enrichment_from_source_when_no_enrichments(self):
        enrichments = self.enrichments.get_all_since_enrichment_from_source(_ENRICHMENT_1.source)
        self.assertEqual(len(enrichments), 0)

    def test_get_all_since_enrichment_from_source_when_no_enrichments_from_source(self):
        self.enrichments.add(_ENRICHMENT_1)
        enrichments = self.enrichments.get_all_since_enrichment_from_source("other_source")
        self.assertCountEqual(enrichments, [_ENRICHMENT_1])

    def test_get_all_since_enrichment_from_source_when_only_enrichment_from_source(self):
        assert _ENRICHMENT_1.source == _ENRICHMENT_2.source
        self.enrichments.add([_ENRICHMENT_1, _ENRICHMENT_2])
        enrichments = self.enrichments.get_all_since_enrichment_from_source(_ENRICHMENT_1.source)
        self.assertEqual(len(enrichments), 0)

    def test_get_all_since_enrichment_from_source_when_no_after_enrichment_from_source(self):
        assert _ENRICHMENT_1.source == _ENRICHMENT_2.source
        self.enrichments.add([_ENRICHMENT_1, Enrichment("other_source", _ENRICHMENT_2.timestamp, Metadata())])
        enrichments = self.enrichments.get_all_since_enrichment_from_source("other_source")
        self.assertEqual(len(enrichments), 0)

    def test_get_all_since_enrichment_from_source_when_one_enrichment_from_source_and_enrichments_afterwards(self):
        delta = timedelta(seconds=1)
        assert _ENRICHMENT_2.timestamp - _ENRICHMENT_1.timestamp > delta
        self.enrichments.add([
            _ENRICHMENT_1,
            Enrichment("other_source", _ENRICHMENT_1.timestamp + delta, Metadata()),
            _ENRICHMENT_2,
            _ENRICHMENT_3
        ])
        enrichments = self.enrichments.get_all_since_enrichment_from_source("other_source")
        self.assertCountEqual(enrichments, [_ENRICHMENT_2, _ENRICHMENT_3])

    def test_get_all_since_enrichment_from_source_when_multiple_enrichment_from_source_and_enrichments_afterwards(self):
        delta = timedelta(seconds=1)
        assert _ENRICHMENT_2.timestamp - _ENRICHMENT_1.timestamp > delta
        assert _ENRICHMENT_3.timestamp - _ENRICHMENT_2.timestamp > delta
        source = "other_source"
        self.enrichments.add([
            _ENRICHMENT_1,
            Enrichment(source, _ENRICHMENT_1.timestamp + delta, Metadata()),
            _ENRICHMENT_2,
            Enrichment(source, _ENRICHMENT_2.timestamp + delta, Metadata()),
            _ENRICHMENT_3
        ])
        enrichments = self.enrichments.get_all_since_enrichment_from_source(source)
        self.assertCountEqual(enrichments, [_ENRICHMENT_3])