def test_empty_dataset(self, mock_build):
mock_build.return_value = MockBigQueryClient(ONE_DATASET, NO_TABLES, None)
extractor = BigQueryMetadataExtractor()
extractor.init(Scoped.get_scoped_conf(conf=self.conf,
scope=extractor.get_scope()))
result = extractor.extract()
self.assertIsNone(result)
def test_can_handle_datasets(self, mock_build):
mock_build.return_value = MockBigQueryClient(NO_DATASETS, None, None)
extractor = BigQueryMetadataExtractor()
extractor.init(Scoped.get_scoped_conf(conf=self.conf,
scope=extractor.get_scope()))
result = extractor.extract()
self.assertIsNone(result)
def test_view(self, mock_build):
mock_build.return_value = MockBigQueryClient(ONE_DATASET, ONE_VIEW, VIEW_DATA)
extractor = BigQueryMetadataExtractor()
extractor.init(Scoped.get_scoped_conf(conf=self.conf,
scope=extractor.get_scope()))
result = extractor.extract()
self.assertIsInstance(result, TableMetadata)
self.assertEqual(result.is_view, True)
def test_empty_dataset(self, mock_datacatalogue, mock_bigquery):
mock_bigquery.return_value = MockBigQueryClient(
ONE_DATASET, NO_TABLES, None)
mock_datacatalogue.DataCatalogClient.return_value = MockDataCatalogClient(
ENTRY, TAGS)
extractor = BigQueryMetadataExtractor()
extractor.init(
Scoped.get_scoped_conf(conf=self.conf,
scope=extractor.get_scope()))
result = extractor.extract()
self.assertIsNone(result)
def test_view(self, mock_datacatalogue, mock_bigquery):
mock_bigquery.return_value = MockBigQueryClient(
ONE_DATASET, ONE_VIEW, VIEW_DATA)
mock_datacatalogue.DataCatalogClient.return_value = MockDataCatalogClient(
ENTRY, TAGS)
extractor = BigQueryMetadataExtractor()
extractor.init(
Scoped.get_scoped_conf(conf=self.conf,
scope=extractor.get_scope()))
result = extractor.extract()
self.assertIsInstance(result, TableMetadata)
self.assertEqual(result.is_view, True)
def test_table_without_tags(self, mock_datacatalogue, mock_bigquery):
mock_bigquery.return_value = MockBigQueryClient(
ONE_DATASET, ONE_TABLE, TABLE_DATA)
mock_datacatalogue.DataCatalogClient.return_value = MockDataCatalogClient(
ENTRY, NO_TAGS)
extractor = BigQueryMetadataExtractor()
extractor.init(
Scoped.get_scoped_conf(conf=self.conf,
scope=extractor.get_scope()))
result = extractor.extract()
self.assertEqual(result.tags, None)
def test_table_without_columns(self, mock_build):
mock_build.return_value = MockBigQueryClient(ONE_DATASET, ONE_TABLE, NO_COLS)
extractor = BigQueryMetadataExtractor()
extractor.init(Scoped.get_scoped_conf(conf=self.conf,
scope=extractor.get_scope()))
result = extractor.extract()
self.assertEqual(result.database, 'bigquery')
self.assertEqual(result.cluster, 'your-project-here')
self.assertEqual(result.schema, 'fdgdfgh')
self.assertEqual(result.name, 'nested_recs')
self.assertEqual(result.description, "")
self.assertEqual(result.columns, [])
self.assertEqual(result.is_view, False)
def test_accepts_dataset_filter_by_label(self, mock_build):
config_dict = {
'extractor.bigquery_table_metadata.{}'.format(BigQueryMetadataExtractor.PROJECT_ID_KEY):
'your-project-here',
'extractor.bigquery_table_metadata.{}'.format(BigQueryMetadataExtractor.FILTER_KEY):
'label.key:value'
}
conf = ConfigFactory.from_dict(config_dict)
mock_build.return_value = MockBigQueryClient(ONE_DATASET, ONE_TABLE, TABLE_DATA)
extractor = BigQueryMetadataExtractor()
extractor.init(Scoped.get_scoped_conf(conf=conf,
scope=extractor.get_scope()))
result = extractor.extract()
self.assertIsInstance(result, TableMetadata)
def test_table_part_of_table_date_range(self, mock_build):
mock_build.return_value = MockBigQueryClient(ONE_DATASET, TABLE_DATE_RANGE, TABLE_DATA)
extractor = BigQueryMetadataExtractor()
extractor.init(Scoped.get_scoped_conf(conf=self.conf,
scope=extractor.get_scope()))
count = 0
result = extractor.extract()
table_name = result.name
while result:
count += 1
result = extractor.extract()
self.assertEqual(count, 1)
self.assertEqual(table_name, 'date_range_')
def test_table_with_nested_records(self, mock_build):
mock_build.return_value = MockBigQueryClient(ONE_DATASET, ONE_TABLE, NESTED_DATA)
extractor = BigQueryMetadataExtractor()
extractor.init(Scoped.get_scoped_conf(conf=self.conf,
scope=extractor.get_scope()))
result = extractor.extract()
first_col = result.columns[0]
self.assertEqual(first_col.name, 'nested')
self.assertEqual(first_col.type, 'RECORD')
second_col = result.columns[1]
self.assertEqual(second_col.name, 'nested.nested2')
self.assertEqual(second_col.type, 'RECORD')
third_col = result.columns[2]
self.assertEqual(third_col.name, 'nested.nested2.ahah')
self.assertEqual(third_col.type, 'STRING')
def test_keypath_and_pagesize_can_be_set(self, mock_build):
config_dict = {
'extractor.bigquery_table_metadata.{}'.format(BigQueryMetadataExtractor.PROJECT_ID_KEY):
'your-project-here',
'extractor.bigquery_table_metadata.{}'.format(BigQueryMetadataExtractor.PAGE_SIZE_KEY):
200,
'extractor.bigquery_table_metadata.{}'.format(BigQueryMetadataExtractor.KEY_PATH_KEY):
'/tmp/doesnotexist',
}
conf = ConfigFactory.from_dict(config_dict)
mock_build.return_value = MockBigQueryClient(ONE_DATASET, ONE_TABLE, TABLE_DATA)
extractor = BigQueryMetadataExtractor()
with self.assertRaises(FileNotFoundError):
extractor.init(Scoped.get_scoped_conf(conf=conf,
scope=extractor.get_scope()))
def test_table_without_columns(self, mock_datacatalogue, mock_bigquery):
mock_bigquery.return_value = MockBigQueryClient(
ONE_DATASET, ONE_TABLE, NO_COLS)
mock_datacatalogue.DataCatalogClient.return_value = MockDataCatalogClient(
ENTRY, TAGS)
extractor = BigQueryMetadataExtractor()
extractor.init(
Scoped.get_scoped_conf(conf=self.conf,
scope=extractor.get_scope()))
result = extractor.extract()
self.assertEqual(result.database, "bigquery")
self.assertEqual(result.cluster, "your-project-here")
self.assertEqual(result.schema, "fdgdfgh")
self.assertEqual(result.name, "nested_recs")
self.assertEqual(result.description, "")
self.assertEqual(result.columns, [])
self.assertEqual(result.is_view, False)
def test_normal_table(self, mock_build):
mock_build.return_value = MockBigQueryClient(ONE_DATASET, ONE_TABLE, TABLE_DATA)
extractor = BigQueryMetadataExtractor()
extractor.init(Scoped.get_scoped_conf(conf=self.conf,
scope=extractor.get_scope()))
result = extractor.extract()
self.assertEqual(result.database, 'bigquery')
self.assertEqual(result.cluster, 'your-project-here')
self.assertEqual(result.schema, 'fdgdfgh')
self.assertEqual(result.name, 'nested_recs')
self.assertEqual(result.description, "")
first_col = result.columns[0]
self.assertEqual(first_col.name, 'test')
self.assertEqual(first_col.type, 'STRING')
self.assertEqual(first_col.description, 'some_description')
self.assertEqual(result.is_view, False)
def test_accepts_dataset_filter_by_label(self, mock_datacatalogue,
mock_bigquery):
config_dict = {
"extractor.bigquery_table_metadata.{}".format(BigQueryMetadataExtractor.PROJECT_ID_KEY):
"your-project-here",
"extractor.bigquery_table_metadata.{}".format(BigQueryMetadataExtractor.FILTER_KEY):
"label.key:value",
}
conf = ConfigFactory.from_dict(config_dict)
mock_bigquery.return_value = MockBigQueryClient(
ONE_DATASET, ONE_TABLE, TABLE_DATA)
mock_datacatalogue.DataCatalogClient.return_value = MockDataCatalogClient(
ENTRY, TAGS)
extractor = BigQueryMetadataExtractor()
extractor.init(
Scoped.get_scoped_conf(conf=conf, scope=extractor.get_scope()))
result = extractor.extract()
self.assertIsInstance(result, TableMetadata)
def test_table_part_of_table_date_range(self, mock_datacatalogue,
mock_bigquery):
mock_bigquery.return_value = MockBigQueryClient(
ONE_DATASET, TABLE_DATE_RANGE, TABLE_DATA)
mock_datacatalogue.DataCatalogClient.return_value = MockDataCatalogClient(
ENTRY, TAGS)
extractor = BigQueryMetadataExtractor()
extractor.init(
Scoped.get_scoped_conf(conf=self.conf,
scope=extractor.get_scope()))
count = 0
result = extractor.extract()
table_name = result.name
while result:
count += 1
result = extractor.extract()
self.assertEqual(count, 1)
self.assertEqual(table_name, "date_range_")
def test_keypath_and_pagesize_can_be_set(self, mock_datacatalogue,
mock_bigquery):
config_dict = {
"extractor.bigquery_table_metadata.{}".format(BigQueryMetadataExtractor.PROJECT_ID_KEY):
"your-project-here",
"extractor.bigquery_table_metadata.{}".format(BigQueryMetadataExtractor.PAGE_SIZE_KEY):
200,
"extractor.bigquery_table_metadata.{}".format(BigQueryMetadataExtractor.KEY_PATH_KEY):
"/tmp/doesnotexist",
}
conf = ConfigFactory.from_dict(config_dict)
mock_bigquery.return_value = MockBigQueryClient(
ONE_DATASET, ONE_TABLE, TABLE_DATA)
mock_datacatalogue.DataCatalogClient.return_value = MockDataCatalogClient(
ENTRY, TAGS)
extractor = BigQueryMetadataExtractor()
with self.assertRaises(FileNotFoundError):
extractor.init(
Scoped.get_scoped_conf(conf=conf, scope=extractor.get_scope()))
def test_table_with_nested_records(self, mock_datacatalogue,
mock_bigquery):
mock_bigquery.return_value = MockBigQueryClient(
ONE_DATASET, ONE_TABLE, NESTED_DATA)
mock_datacatalogue.DataCatalogClient.return_value = MockDataCatalogClient(
ENTRY, TAGS)
extractor = BigQueryMetadataExtractor()
extractor.init(
Scoped.get_scoped_conf(conf=self.conf,
scope=extractor.get_scope()))
result = extractor.extract()
first_col = result.columns[0]
self.assertEqual(first_col.name, "nested")
self.assertEqual(first_col.type, "RECORD")
second_col = result.columns[1]
self.assertEqual(second_col.name, "nested.nested2")
self.assertEqual(second_col.type, "RECORD")
third_col = result.columns[2]
self.assertEqual(third_col.name, "nested.nested2.ahah")
self.assertEqual(third_col.type, "STRING")
def test_normal_table(self, mock_datacatalogue, mock_bigquery):
mock_bigquery.return_value = MockBigQueryClient(
ONE_DATASET, ONE_TABLE, TABLE_DATA)
mock_datacatalogue.DataCatalogClient.return_value = MockDataCatalogClient(
ENTRY, TAGS)
extractor = BigQueryMetadataExtractor()
extractor.init(
Scoped.get_scoped_conf(conf=self.conf,
scope=extractor.get_scope()))
result = extractor.extract()
self.assertEqual(result.database, "bigquery")
self.assertEqual(result.cluster, "your-project-here")
self.assertEqual(result.schema, "fdgdfgh")
self.assertEqual(result.name, "nested_recs")
self.assertEqual(result.description, "")
self.assertEqual(result.is_view, False)
self.assertEqual(
result.tags,
{
"name":
"projects/your-project-here/locations/us/entryGroups/@bigquery/entries/cHJvamVjdHMvd2hhbGUtZGV2LTI5NDgxMi9kYXRhc2V0cy90ZXN0aW5nL3RhYmxlcy90YWJsZTE/tags/CXy_PbcgFLIaW",
"template":
"projects/your-project-here/locations/europe-west2/tagTemplates/demo_tag",
"fields": {
"demo2": {
"displayName": "demo2",
"boolValue": "true"
},
"demo1": {
"displayName": "demo1",
"stringValue": "test1",
"order": 1,
},
},
"templateDisplayName": "demo-tag",
},
)
self.assertEqual(result.labels, {
"label_1": "test_label_1",
"label_2": "test_label_2"
})
first_col = result.columns[0]
self.assertEqual(first_col.name, "test")
self.assertEqual(first_col.type, "STRING")
self.assertEqual(first_col.description, "some_description")
self.assertEqual(first_col.tags, None)
fourth_col = result.columns[3]
self.assertEqual(
fourth_col.tags,
{
"name":
"projects/your-project-here/locations/us/entryGroups/@bigquery/entries/cHJvamVjdHMvd2hhbGUtZGV2LTI5NDgxMi9kYXRhc2V0cy90ZXN0aW5nL3RhYmxlcy90YWJsZTE/tags/CYPqPyHt4oku",
"template":
"projects/your-project-here/locations/europe-west2/tagTemplates/demo_tag",
"fields": {
"demo1": {
"displayName": "demo1",
"stringValue": "test1",
"order": 1,
},
"demo2": {
"displayName": "demo2",
"boolValue": "true"
},
},
"column": "test4",
"templateDisplayName": "demo-tag",
},
)