def _get_table(self, custom_stats_format: bool = False) -> None:
if custom_stats_format:
test_exp_col = self.test_exp_col_stats_formatted
else:
test_exp_col = self.test_exp_col_stats_raw
ent_attrs = cast(dict, self.entity1['attributes'])
self._mock_get_table_entity()
self._create_mocked_report_entities_collection()
self.proxy._get_owners = MagicMock(
return_value=[User(email=ent_attrs['owner'])]) # type: ignore
self.proxy._driver.entity_bulk = MagicMock(
return_value=self.report_entity_collection)
response = self.proxy.get_table(table_uri=self.table_uri)
classif_name = self.classification_entity['classifications'][0][
'typeName']
col_attrs = cast(dict, self.test_column['attributes'])
exp_col_stats = list()
for stats in test_exp_col:
exp_col_stats.append(
Stat(
stat_type=stats['attributes']['stat_name'],
stat_val=stats['attributes']['stat_val'],
start_epoch=stats['attributes']['start_epoch'],
end_epoch=stats['attributes']['end_epoch'],
))
exp_col = Column(name=col_attrs['name'],
description='column description',
col_type='Managed',
sort_order=col_attrs['position'],
stats=exp_col_stats)
expected = Table(
database=self.entity_type,
cluster=self.cluster,
schema=self.db,
name=ent_attrs['name'],
tags=[Tag(tag_name=classif_name, tag_type="default")],
description=ent_attrs['description'],
owners=[User(email=ent_attrs['owner'])],
resource_reports=[
ResourceReport(name='test_report', url='http://test'),
ResourceReport(name='test_report3', url='http://test3')
],
last_updated_timestamp=int(str(self.entity1['updateTime'])[:10]),
columns=[exp_col] * self.active_columns,
programmatic_descriptions=[
ProgrammaticDescription(source='test parameter key a',
text='testParameterValueA'),
ProgrammaticDescription(source='test parameter key b',
text='testParameterValueB')
],
is_view=False)
self.assertEqual(str(expected), str(response))
def test_16_just_execute_next_descriptions(self) -> None:
descs = next_descriptions()
self.assertEqual(3, len(descs))
self.assertEqual([
ProgrammaticDescription(source='dedefghijk001233',
text='ijklmnopqrstuvwxyzab001224'),
ProgrammaticDescription(source='devwxyzabc001173',
text='abcdefghijklmnopqrst001164'),
ProgrammaticDescription(source='dezabcdefg001203',
text='efghijklmnopqrstuvwx001194')
], descs)
def test_get_table_view_only(self) -> None:
col_usage_return_value = copy.deepcopy(self.col_usage_return_value)
for col in col_usage_return_value:
col['tbl']['is_view'] = True
with patch.object(GraphDatabase, 'driver'), patch.object(Neo4jProxy, '_execute_cypher_query') as mock_execute:
mock_execute.side_effect = [col_usage_return_value, [], self.table_level_return_value]
neo4j_proxy = Neo4jProxy(host='DOES_NOT_MATTER', port=0000)
table = neo4j_proxy.get_table(table_uri='dummy_uri')
expected = Table(database='hive', cluster='gold', schema='foo_schema', name='foo_table',
tags=[Tag(tag_name='test', tag_type='default')],
badges=[Badge(badge_name='golden', category='table_status')],
table_readers=[], description='foo description',
watermarks=[Watermark(watermark_type='high_watermark',
partition_key='ds',
partition_value='fake_value',
create_time='fake_time'),
Watermark(watermark_type='low_watermark',
partition_key='ds',
partition_value='fake_value',
create_time='fake_time')],
columns=[Column(name='bar_id_1', description='bar col description', col_type='varchar',
sort_order=0, stats=[Stat(start_epoch=1,
end_epoch=1,
stat_type='avg',
stat_val='1')], badges=[]),
Column(name='bar_id_2', description='bar col2 description', col_type='bigint',
sort_order=1, stats=[Stat(start_epoch=2,
end_epoch=2,
stat_type='avg',
stat_val='2')],
badges=[Badge(badge_name='primary key', category='column')])],
owners=[User(email='*****@*****.**')],
table_writer=Application(application_url=self.table_writer['application_url'],
description=self.table_writer['description'],
name=self.table_writer['name'],
id=self.table_writer['id']),
last_updated_timestamp=1,
source=Source(source='/source_file_loc',
source_type='github'),
is_view=True,
programmatic_descriptions=[
ProgrammaticDescription(source='quality_report',
text='Test Test'),
ProgrammaticDescription(source='s3_crawler',
text='Test Test Test')
])
self.assertEqual(str(expected), str(table))
def _extract_programmatic_descriptions_from_query(self, raw_prog_descriptions: dict) -> list:
prog_descriptions = []
for prog_description in raw_prog_descriptions:
source = prog_description['description_source']
if source is None:
LOGGER.error("A programmatic description with no source was found... skipping.")
else:
prog_descriptions.append(ProgrammaticDescription(source=source, text=prog_description['description']))
prog_descriptions.sort(key=lambda x: x.source)
return prog_descriptions
def _get_programmatic_descriptions(self, parameters: dict) -> List[ProgrammaticDescription]:
programmatic_descriptions: Dict[str, ProgrammaticDescription] = {}
for source, text in parameters.items():
use_parameter = True
for regex_filter in app.config['PROGRAMMATIC_DESCRIPTIONS_EXCLUDE_FILTERS']:
pattern = re.compile(regex_filter)
if pattern.match(source):
use_parameter = False
break
if use_parameter:
source = re.sub("([a-z])([A-Z])", "\g<1> \g<2>", source).lower()
programmatic_descriptions[source] = ProgrammaticDescription(source=source, text=text)
result = dict(sorted(programmatic_descriptions.items()))
return list(result.values())
def test_get_table(self, mock_rds_client: Any) -> None:
database = RDSDatabase(name='hive')
cluster = RDSCluster(name='gold')
schema = RDSSchema(name='foo_schema')
schema.cluster = cluster
cluster.database = database
table = RDSTable(name='foo_table')
table.schema = schema
table.description = RDSTableDescription(description='foo description')
col1 = RDSColumn(name='bar_id_1', type='varchar', sort_order=0)
col1.description = RDSColumnDescription(
description='bar col description')
col1.stats = [
RDSColumnStat(stat_type='avg',
start_epoch='1',
end_epoch='1',
stat_val='1')
]
col2 = RDSColumn(name='bar_id_2', type='bigint', sort_order=1)
col2.description = RDSColumnDescription(
description='bar col2 description')
col2.stats = [
RDSColumnStat(stat_type='avg',
start_epoch='2',
end_epoch='2',
stat_val='2')
]
col2.badges = [RDSBadge(rk='primary key', category='column')]
columns = [col1, col2]
table.watermarks = [
RDSTableWatermark(
rk='hive://gold.test_schema/test_table/high_watermark/',
partition_key='ds',
partition_value='fake_value',
create_time='fake_time'),
RDSTableWatermark(
rk='hive://gold.test_schema/test_table/low_watermark/',
partition_key='ds',
partition_value='fake_value',
create_time='fake_time')
]
table.application = RDSApplication(
application_url='airflow_host/admin/airflow/tree?dag_id=test_table',
description='DAG generating a table',
name='Airflow',
id='dag/task_id')
table.timestamp = RDSTableTimestamp(last_updated_timestamp=1)
table.owners = [
RDSUser(rk='*****@*****.**', email='*****@*****.**')
]
table.tags = [RDSTag(rk='test', tag_type='default')]
table.badges = [RDSBadge(rk='golden', category='table_status')]
table.source = RDSTableSource(rk='some key',
source_type='github',
source='/source_file_loc')
table.programmatic_descriptions = [
RDSTableProgrammaticDescription(description_source='s3_crawler',
description='Test Test Test'),
RDSTableProgrammaticDescription(
description_source='quality_report', description='Test Test')
]
readers = [RDSTableUsage(user_rk='*****@*****.**', read_count=5)]
mock_client = MagicMock()
mock_rds_client.return_value = mock_client
mock_create_session = MagicMock()
mock_client.create_session.return_value = mock_create_session
mock_session = MagicMock()
mock_create_session.__enter__.return_value = mock_session
mock_session_query = MagicMock()
mock_session.query.return_value = mock_session_query
mock_session_query_filter = MagicMock()
mock_session_query.filter.return_value = mock_session_query_filter
mock_session_query_filter.first.return_value = table
mock_session_query_filter_orderby = MagicMock()
mock_session_query_filter.order_by.return_value = mock_session_query_filter_orderby
mock_session_query_filter_orderby_limit = MagicMock()
mock_session_query_filter_orderby.limit.return_value = mock_session_query_filter_orderby_limit
mock_session_query_filter_orderby_limit.all.return_value = readers
mock_session_query_filter_options = MagicMock()
mock_session_query_filter.options.return_value = mock_session_query_filter_options
mock_session_query_filter_options.all.return_value = columns
proxy = MySQLProxy()
actual_table = proxy.get_table(table_uri='dummy_uri')
expected = Table(
database='hive',
cluster='gold',
schema='foo_schema',
name='foo_table',
tags=[Tag(tag_name='test', tag_type='default')],
badges=[Badge(badge_name='golden', category='table_status')],
table_readers=[
Reader(user=User(email='*****@*****.**'), read_count=5)
],
description='foo description',
watermarks=[
Watermark(watermark_type='high_watermark',
partition_key='ds',
partition_value='fake_value',
create_time='fake_time'),
Watermark(watermark_type='low_watermark',
partition_key='ds',
partition_value='fake_value',
create_time='fake_time')
],
columns=[
Column(name='bar_id_1',
description='bar col description',
col_type='varchar',
sort_order=0,
stats=[
Stat(start_epoch=1,
end_epoch=1,
stat_type='avg',
stat_val='1')
],
badges=[]),
Column(name='bar_id_2',
description='bar col2 description',
col_type='bigint',
sort_order=1,
stats=[
Stat(start_epoch=2,
end_epoch=2,
stat_type='avg',
stat_val='2')
],
badges=[
Badge(badge_name='primary key', category='column')
])
],
owners=[User(email='*****@*****.**')],
table_writer=Application(
application_url=
'airflow_host/admin/airflow/tree?dag_id=test_table',
description='DAG generating a table',
name='Airflow',
id='dag/task_id'),
last_updated_timestamp=1,
source=Source(source='/source_file_loc', source_type='github'),
is_view=False,
programmatic_descriptions=[
ProgrammaticDescription(source='quality_report',
text='Test Test'),
ProgrammaticDescription(source='s3_crawler',
text='Test Test Test')
])
self.assertEqual(str(expected), str(actual_table))
def test_13_next_description(self) -> None:
self.assertEqual(
ProgrammaticDescription(text='ijklmnopqrstuvwxyzab001120',
source='dedefghijk001129'),
next_description())
