2022-10-13 03:24:04 +02:00
|
|
|
import datetime
|
2023-04-06 04:46:50 +02:00
|
|
|
from typing import Dict, List, Set
|
2022-10-13 03:24:04 +02:00
|
|
|
|
2023-08-31 12:44:24 -04:00
|
|
|
import pytest
|
|
|
|
|
2023-04-06 04:46:50 +02:00
|
|
|
from datahub.ingestion.source.bigquery_v2.bigquery_audit import (
|
|
|
|
BigQueryTableRef,
|
|
|
|
QueryEvent,
|
|
|
|
)
|
2022-10-13 03:24:04 +02:00
|
|
|
from datahub.ingestion.source.bigquery_v2.bigquery_config import BigQueryV2Config
|
|
|
|
from datahub.ingestion.source.bigquery_v2.bigquery_report import BigQueryV2Report
|
2024-08-26 11:21:00 +05:30
|
|
|
from datahub.ingestion.source.bigquery_v2.common import BigQueryIdentifierBuilder
|
2023-04-06 04:46:50 +02:00
|
|
|
from datahub.ingestion.source.bigquery_v2.lineage import (
|
|
|
|
BigqueryLineageExtractor,
|
|
|
|
LineageEdge,
|
|
|
|
)
|
2024-02-09 16:27:45 -05:00
|
|
|
from datahub.sql_parsing.schema_resolver import SchemaResolver
|
2023-04-06 04:46:50 +02:00
|
|
|
|
|
|
|
|
2023-08-31 12:44:24 -04:00
|
|
|
@pytest.fixture
|
|
|
|
def lineage_entries() -> List[QueryEvent]:
|
|
|
|
return [
|
2023-04-06 04:46:50 +02:00
|
|
|
QueryEvent(
|
|
|
|
timestamp=datetime.datetime.now(tz=datetime.timezone.utc),
|
|
|
|
actor_email="bla@bla.com",
|
2023-08-31 12:44:24 -04:00
|
|
|
query="""
|
|
|
|
INSERT INTO `my_project.my_dataset.my_table`
|
|
|
|
SELECT first.a, second.b FROM `my_project.my_dataset.my_source_table1` first
|
|
|
|
LEFT JOIN `my_project.my_dataset.my_source_table2` second ON first.id = second.id
|
|
|
|
""",
|
2024-04-02 22:13:05 +02:00
|
|
|
statementType="INSERT",
|
2023-04-06 04:46:50 +02:00
|
|
|
project_id="proj_12344",
|
|
|
|
end_time=None,
|
|
|
|
referencedTables=[
|
|
|
|
BigQueryTableRef.from_string_name(
|
|
|
|
"projects/my_project/datasets/my_dataset/tables/my_source_table1"
|
|
|
|
),
|
|
|
|
BigQueryTableRef.from_string_name(
|
|
|
|
"projects/my_project/datasets/my_dataset/tables/my_source_table2"
|
|
|
|
),
|
|
|
|
],
|
|
|
|
destinationTable=BigQueryTableRef.from_string_name(
|
|
|
|
"projects/my_project/datasets/my_dataset/tables/my_table"
|
|
|
|
),
|
|
|
|
),
|
|
|
|
QueryEvent(
|
|
|
|
timestamp=datetime.datetime.now(tz=datetime.timezone.utc),
|
|
|
|
actor_email="bla@bla.com",
|
|
|
|
query="testQuery",
|
|
|
|
statementType="SELECT",
|
|
|
|
project_id="proj_12344",
|
|
|
|
end_time=datetime.datetime.fromtimestamp(
|
|
|
|
1617295943.17321, tz=datetime.timezone.utc
|
|
|
|
),
|
|
|
|
referencedTables=[
|
|
|
|
BigQueryTableRef.from_string_name(
|
|
|
|
"projects/my_project/datasets/my_dataset/tables/my_source_table3"
|
|
|
|
),
|
|
|
|
],
|
|
|
|
destinationTable=BigQueryTableRef.from_string_name(
|
|
|
|
"projects/my_project/datasets/my_dataset/tables/my_table"
|
|
|
|
),
|
|
|
|
),
|
|
|
|
QueryEvent(
|
|
|
|
timestamp=datetime.datetime.now(tz=datetime.timezone.utc),
|
|
|
|
actor_email="bla@bla.com",
|
|
|
|
query="testQuery",
|
|
|
|
statementType="SELECT",
|
|
|
|
project_id="proj_12344",
|
|
|
|
referencedViews=[
|
|
|
|
BigQueryTableRef.from_string_name(
|
|
|
|
"projects/my_project/datasets/my_dataset/tables/my_source_view1"
|
|
|
|
),
|
|
|
|
],
|
|
|
|
destinationTable=BigQueryTableRef.from_string_name(
|
|
|
|
"projects/my_project/datasets/my_dataset/tables/my_table"
|
|
|
|
),
|
|
|
|
),
|
|
|
|
]
|
|
|
|
|
2023-08-31 12:44:24 -04:00
|
|
|
|
|
|
|
def test_lineage_with_timestamps(lineage_entries: List[QueryEvent]) -> None:
|
|
|
|
config = BigQueryV2Config()
|
|
|
|
report = BigQueryV2Report()
|
2023-09-16 00:25:39 +05:30
|
|
|
extractor: BigqueryLineageExtractor = BigqueryLineageExtractor(
|
2024-08-26 11:21:00 +05:30
|
|
|
config, report, BigQueryIdentifierBuilder(config, report)
|
2023-09-16 00:25:39 +05:30
|
|
|
)
|
2023-08-31 12:44:24 -04:00
|
|
|
|
2023-04-06 04:46:50 +02:00
|
|
|
bq_table = BigQueryTableRef.from_string_name(
|
|
|
|
"projects/my_project/datasets/my_dataset/tables/my_table"
|
|
|
|
)
|
|
|
|
|
|
|
|
lineage_map: Dict[str, Set[LineageEdge]] = extractor._create_lineage_map(
|
2023-07-11 11:12:51 -07:00
|
|
|
iter(lineage_entries),
|
|
|
|
sql_parser_schema_resolver=SchemaResolver(platform="bigquery"),
|
2023-04-06 04:46:50 +02:00
|
|
|
)
|
|
|
|
|
|
|
|
upstream_lineage = extractor.get_lineage_for_table(
|
|
|
|
bq_table=bq_table,
|
2023-07-11 11:12:51 -07:00
|
|
|
bq_table_urn="urn:li:dataset:(urn:li:dataPlatform:bigquery,my_project.my_dataset.my_table,PROD)",
|
2023-04-06 04:46:50 +02:00
|
|
|
lineage_metadata=lineage_map,
|
|
|
|
)
|
|
|
|
assert upstream_lineage
|
2023-07-11 11:12:51 -07:00
|
|
|
assert len(upstream_lineage.upstreams) == 4
|
2023-08-31 12:44:24 -04:00
|
|
|
|
|
|
|
|
|
|
|
def test_column_level_lineage(lineage_entries: List[QueryEvent]) -> None:
|
|
|
|
config = BigQueryV2Config(extract_column_lineage=True, incremental_lineage=False)
|
|
|
|
report = BigQueryV2Report()
|
2023-09-16 00:25:39 +05:30
|
|
|
extractor: BigqueryLineageExtractor = BigqueryLineageExtractor(
|
2024-08-26 11:21:00 +05:30
|
|
|
config, report, BigQueryIdentifierBuilder(config, report)
|
2023-09-16 00:25:39 +05:30
|
|
|
)
|
2023-08-31 12:44:24 -04:00
|
|
|
|
|
|
|
bq_table = BigQueryTableRef.from_string_name(
|
|
|
|
"projects/my_project/datasets/my_dataset/tables/my_table"
|
|
|
|
)
|
|
|
|
|
|
|
|
lineage_map: Dict[str, Set[LineageEdge]] = extractor._create_lineage_map(
|
|
|
|
lineage_entries[:1],
|
|
|
|
sql_parser_schema_resolver=SchemaResolver(platform="bigquery"),
|
|
|
|
)
|
|
|
|
|
|
|
|
upstream_lineage = extractor.get_lineage_for_table(
|
|
|
|
bq_table=bq_table,
|
|
|
|
bq_table_urn="urn:li:dataset:(urn:li:dataPlatform:bigquery,my_project.my_dataset.my_table,PROD)",
|
|
|
|
lineage_metadata=lineage_map,
|
|
|
|
)
|
|
|
|
assert upstream_lineage
|
|
|
|
assert len(upstream_lineage.upstreams) == 2
|
|
|
|
assert (
|
|
|
|
upstream_lineage.fineGrainedLineages
|
|
|
|
and len(upstream_lineage.fineGrainedLineages) == 2
|
|
|
|
)
|