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 get_table(self, *, table_uri: str) -> Table:
"""
Gathers all the information needed for the Table Detail Page.
:param table_uri:
:return: A Table object with all the information available
or gathered from different entities.
"""
entity, table_info = self._get_table_entity(table_uri=table_uri)
table_details = entity.entity
try:
attrs = table_details[self.ATTRS_KEY]
table_qn = parse_table_qualified_name(
qualified_name=attrs.get(self.QN_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 = self._serialize_columns(entity=entity)
table = Table(
database=table_details.get('typeName'),
cluster=table_qn.get('cluster_name', ''),
schema=table_qn.get('db_name', ''),
name=attrs.get('name') or table_qn.get("table_name", ''),
tags=tags,
description=attrs.get('description') or attrs.get('comment'),
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_uri} )'
.format(table_uri=table_uri))
def _exec_usage_query(self, table_uri: str) -> List[Reader]:
# Return Value: List[Reader]
usage_query = textwrap.dedent("""\
MATCH (user:User)-[read:READ]->(table:Table {key: $tbl_key})
RETURN user.email as email, read.read_count as read_count, table.name as table_name
ORDER BY read.read_count DESC LIMIT 5;
""")
usage_neo4j_records = self._execute_cypher_query(
statement=usage_query, param_dict={'tbl_key': table_uri})
readers = [] # type: List[Reader]
for usage_neo4j_record in usage_neo4j_records:
reader = Reader(user=User(email=usage_neo4j_record['email']),
read_count=usage_neo4j_record['read_count'])
readers.append(reader)
return readers
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')
def _exec_table_query(self, table_uri: str) -> Tuple:
"""
Queries one Cypher record with watermark list, Application,
,timestamp, owner records and tag records.
"""
# Return Value: (Watermark Results, Table Writer, Last Updated Timestamp, owner records, tag records)
table_level_query = textwrap.dedent("""\
MATCH (tbl:Table {key: $tbl_key})
OPTIONAL MATCH (wmk:Watermark)-[:BELONG_TO_TABLE]->(tbl)
OPTIONAL MATCH (application:Application)-[:GENERATES]->(tbl)
OPTIONAL MATCH (tbl)-[:LAST_UPDATED_AT]->(t:Timestamp)
OPTIONAL MATCH (owner:User)-[:OWNER_OF]->(tbl)
OPTIONAL MATCH (tbl)-[:TAGGED_BY]->(tag:Tag)
OPTIONAL MATCH (tbl)-[:SOURCE]->(src:Source)
RETURN collect(distinct wmk) as wmk_records,
application,
t.last_updated_timestamp as last_updated_timestamp,
collect(distinct owner) as owner_records,
collect(distinct tag) as tag_records,
src
""")
table_records = self._execute_cypher_query(
statement=table_level_query, param_dict={'tbl_key': table_uri})
table_records = table_records.single()
wmk_results = []
table_writer = None
wmk_records = table_records['wmk_records']
for record in wmk_records:
if record['key'] is not None:
watermark_type = record['key'].split('/')[-2]
wmk_result = Watermark(
watermark_type=watermark_type,
partition_key=record['partition_key'],
partition_value=record['partition_value'],
create_time=record['create_time'])
wmk_results.append(wmk_result)
tags = []
if table_records.get('tag_records'):
tag_records = table_records['tag_records']
for record in tag_records:
tag_result = Tag(tag_name=record['key'],
tag_type=record['tag_type'])
tags.append(tag_result)
application_record = table_records['application']
if application_record is not None:
table_writer = Application(
application_url=application_record['application_url'],
description=application_record['description'],
name=application_record['name'],
id=application_record.get('id', ''))
timestamp_value = table_records['last_updated_timestamp']
owner_record = []
for owner in table_records.get('owner_records', []):
owner_record.append(User(email=owner['email']))
src = None
if table_records['src']:
src = Source(source_type=table_records['src']['source_type'],
source=table_records['src']['source'])
return wmk_results, table_writer, timestamp_value, owner_record, tags, src