def test_get_table(self) -> None:
with patch.object(GraphDatabase, 'driver'), patch.object(
Neo4jProxy, '_execute_cypher_query') as mock_execute:
mock_execute.side_effect = [
self.col_usage_return_value, [], self.table_level_return_value
]
neo4j_proxy = Neo4jProxy(endpoint='bogus')
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')],
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=[
Statistics(start_epoch=1,
end_epoch=1,
stat_type='avg',
stat_val='1')
]),
Column(name='bar_id_2',
description='bar col2 description',
col_type='bigint',
sort_order=1,
stats=[
Statistics(start_epoch=2,
end_epoch=2,
stat_type='avg',
stat_val='2')
])
],
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'))
self.assertEqual(str(expected), str(table))
def test_get_table(self):
self._mock_get_table_entity()
response = self.proxy.get_table(table_uri=self.table_uri)
classif_name = self.classification_entity['classifications'][0]['typeName']
ent_attrs = self.entity1['attributes']
col_attrs = self.test_column['attributes']
exp_col_stats = list()
for stats in col_attrs['stats']:
exp_col_stats.append(
Statistics(
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['qualifiedName'],
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=self.name,
tags=[Tag(tag_name=classif_name, tag_type="default")],
description=ent_attrs['description'],
owners=[User(email=ent_attrs['owner'])],
columns=[exp_col],
last_updated_timestamp=self.entity1['updateTime'])
self.assertEqual(str(expected), str(response))
def get_table(self, *, table_id: str, table_info: Dict) -> Table:
"""
Gathers all the information needed for the Table Detail Page.
:param table_id:
:param table_info: Additional table information (entity, db, cluster, name)
:return: A Table object with all the information available
or gathered from different entities.
"""
table_entity = self._get_table_entity(table_id=table_id)
table_details = table_entity.entity
try:
attrs = table_details[self.ATTRS_KEY]
rel_attrs = table_details[self.REL_ATTRS_KEY]
tags = []
# Using or in case, if the key 'classifications' is there with a None
for classification in table_details.get(
"classifications") or list():
tags.append(
Tag(tag_name=classification.get('typeName'),
tag_type="default"))
columns = []
for column in rel_attrs.get('columns') or list():
col_entity = table_entity.referredEntities[column['guid']]
col_attrs = col_entity[self.ATTRS_KEY]
columns.append(
Column(
name=col_attrs.get(self.NAME_ATTRIBUTE),
description=col_attrs.get('description'),
col_type=col_attrs.get('type')
or col_attrs.get('dataType'),
sort_order=col_attrs.get('position'),
))
table = Table(
database=table_info['entity'],
cluster=table_info['cluster'],
schema=table_info['db'],
name=table_info['name'],
tags=tags,
description=attrs.get('description'),
owners=[User(email=attrs.get('owner'))],
columns=columns,
last_updated_timestamp=table_details.get('updateTime'))
return table
except KeyError as ex:
LOGGER.exception(
'Error while accessing table information. {}'.format(str(ex)))
raise BadRequest(
'Some of the required attributes '
'are missing in : ( {table_id} )'.format(table_id=table_id))
def _exec_col_query(self, table_uri: str) -> Tuple:
# Return Value: (Columns, Last Processed Record)
column_level_query = textwrap.dedent("""
MATCH (db:Database)-[:CLUSTER]->(clstr:Cluster)-[:SCHEMA]->(schema:Schema)
-[:TABLE]->(tbl:Table {key: $tbl_key})-[:COLUMN]->(col:Column)
OPTIONAL MATCH (tbl)-[:DESCRIPTION]->(tbl_dscrpt:Description)
OPTIONAL MATCH (col:Column)-[:DESCRIPTION]->(col_dscrpt:Description)
OPTIONAL MATCH (col:Column)-[:STAT]->(stat:Stat)
RETURN db, clstr, schema, tbl, tbl_dscrpt, col, col_dscrpt, collect(distinct stat) as col_stats
ORDER BY col.sort_order;""")
tbl_col_neo4j_records = self._execute_cypher_query(
statement=column_level_query, param_dict={'tbl_key': table_uri})
cols = []
last_neo4j_record = None
for tbl_col_neo4j_record in tbl_col_neo4j_records:
# Getting last record from this for loop as Neo4j's result's random access is O(n) operation.
col_stats = []
for stat in tbl_col_neo4j_record['col_stats']:
col_stat = Statistics(stat_type=stat['stat_name'],
stat_val=stat['stat_val'],
start_epoch=int(
float(stat['start_epoch'])),
end_epoch=int(float(stat['end_epoch'])))
col_stats.append(col_stat)
last_neo4j_record = tbl_col_neo4j_record
col = Column(
name=tbl_col_neo4j_record['col']['name'],
description=self._safe_get(tbl_col_neo4j_record, 'col_dscrpt',
'description'),
col_type=tbl_col_neo4j_record['col']['type'],
sort_order=int(tbl_col_neo4j_record['col']['sort_order']),
stats=col_stats)
cols.append(col)
if not cols:
raise NotFoundException(
'Table URI( {table_uri} ) does not exist'.format(
table_uri=table_uri))
return sorted(cols,
key=lambda item: item.sort_order), last_neo4j_record
def _serialize_columns(self, *, entity: EntityUniqueAttribute) -> \
Union[List[Column], List]:
"""
Helper function to fetch the columns from entity and serialize them
using Column and Statistics model.
:param entity: EntityUniqueAttribute object,
along with relationshipAttributes
:return: A list of Column objects, if there are any columns available,
else an empty list.
"""
columns = list()
for column in entity.entity[self.REL_ATTRS_KEY].get('columns') or list():
col_entity = entity.referredEntities[column['guid']]
col_attrs = col_entity[self.ATTRS_KEY]
col_rel_attrs = col_entity[self.REL_ATTRS_KEY]
col_metadata = col_rel_attrs.get('metadata')
statistics = list()
if col_metadata:
col_metadata = entity.referredEntities.get(col_metadata.get('guid'))
for stats in col_metadata['attributes'].get('statistics') or list():
stats_attrs = stats['attributes']
statistics.append(
Statistics(
stat_type=stats_attrs.get('stat_name'),
stat_val=stats_attrs.get('stat_val'),
start_epoch=stats_attrs.get('start_epoch'),
end_epoch=stats_attrs.get('end_epoch'),
)
)
columns.append(
Column(
name=col_attrs.get('name'),
description=col_attrs.get('description') or col_attrs.get('comment'),
col_type=col_attrs.get('type') or col_attrs.get('dataType'),
sort_order=col_attrs.get('position'),
stats=statistics,
)
)
return sorted(columns, key=lambda item: item.sort_order)
def test_get_resources_by_user_relation(self) -> None:
with patch.object(GraphDatabase, 'driver'), \
patch.object(Neo4jProxy, '_execute_cypher_query') as mock_execute, \
patch.object(Neo4jProxy, 'get_table') as mock_get_table:
mock_execute.return_value.single.return_value = {
'table_records': [{
'key': 'table_uri',
}]
}
mock_get_table.return_value = Table(
database='hive',
cluster='gold',
schema='foo_schema',
name='foo_table',
tags=[Tag(tag_name='test', tag_type='default')],
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=[
Statistics(start_epoch=1,
end_epoch=1,
stat_type='avg',
stat_val='1')
]),
Column(name='bar_id_2',
description='bar col2 description',
col_type='bigint',
sort_order=1,
stats=[
Statistics(start_epoch=2,
end_epoch=2,
stat_type='avg',
stat_val='2')
])
],
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'))
neo4j_proxy = Neo4jProxy(host='DOES_NOT_MATTER', port=0000)
result = neo4j_proxy.get_table_by_user_relation(
user_email='test_user', relation_type=UserResourceRel.follow)
self.assertEqual(len(result['table']), 1)
self.assertEqual(result['table'][0].name, 'foo_table')
