def test_mand_conditions(table: str, mand_conditions: List[FunctionCall]) -> None:
query = Query(
Table(
table,
ColumnSet([]),
final=False,
sampling_rate=None,
mandatory_conditions=mand_conditions,
),
None,
None,
binary_condition(
BooleanFunctions.AND,
binary_condition(
OPERATOR_TO_FUNCTION["="], Column("d", None, "d"), Literal(None, "1"),
),
binary_condition(
OPERATOR_TO_FUNCTION["="], Column("c", None, "c"), Literal(None, "3"),
),
),
)
query_ast_copy = copy.deepcopy(query)
request_settings = HTTPRequestSettings(consistent=True)
processor = MandatoryConditionApplier()
processor.process_query(query, request_settings)
query_ast_copy.add_condition_to_ast(combine_and_conditions(mand_conditions))
assert query.get_condition_from_ast() == query_ast_copy.get_condition_from_ast()
def test_mand_conditions(table: str,
mand_conditions: List[MandatoryCondition]) -> None:
body = {"conditions": [["d", "=", "1"], ["c", "=", "3"]]}
query = Query(
copy.deepcopy(body),
TableSource(table, None, mand_conditions, ["c1"]),
None,
None,
binary_condition(
None,
BooleanFunctions.AND,
binary_condition(
None,
OPERATOR_TO_FUNCTION["="],
Column("d", None, "d"),
Literal(None, "1"),
),
binary_condition(
None,
OPERATOR_TO_FUNCTION["="],
Column("c", None, "c"),
Literal(None, "3"),
),
),
)
query_ast_copy = copy.deepcopy(query)
request_settings = HTTPRequestSettings(consistent=True)
processor = MandatoryConditionApplier()
processor.process_query(query, request_settings)
body["conditions"].extend([c.legacy for c in mand_conditions])
assert query.get_conditions() == body["conditions"]
query_ast_copy.add_condition_to_ast(
combine_and_conditions([c.ast for c in mand_conditions]))
assert query.get_condition_from_ast(
) == query_ast_copy.get_condition_from_ast()
def build_and_rank_plans(
self, query: LogicalQuery, settings: RequestSettings
) -> Sequence[ClickhouseQueryPlan]:
with sentry_sdk.start_span(
op="build_plan.selected_storage", description="select_storage"
):
storage, mappers = self.__selector.select_storage(query, settings)
with sentry_sdk.start_span(
op="build_plan.selected_storage", description="translate"
):
# The QueryTranslator class should be instantiated once for each call to build_plan,
# to avoid cache conflicts.
clickhouse_query = QueryTranslator(mappers).translate(query)
with sentry_sdk.start_span(
op="build_plan.selected_storage", description="set_from_clause"
):
clickhouse_query.set_from_clause(
get_query_data_source(
storage.get_schema().get_data_source(),
final=query.get_final(),
sampling_rate=query.get_sample(),
)
)
cluster = storage.get_cluster()
db_query_processors = [
*storage.get_query_processors(),
*self.__post_processors,
MandatoryConditionApplier(),
]
return [
ClickhouseQueryPlan(
query=clickhouse_query,
plan_query_processors=[],
db_query_processors=db_query_processors,
storage_set_key=storage.get_storage_set_key(),
execution_strategy=SimpleQueryPlanExecutionStrategy(
cluster=cluster,
db_query_processors=db_query_processors,
splitters=storage.get_query_splitters(),
),
)
]
def build_plan(self, request: Request) -> ClickhouseQueryPlan:
clickhouse_query = QueryTranslator(self.__mappers).translate(request.query)
clickhouse_query.set_from_clause(self.__storage.get_schema().get_data_source())
cluster = self.__storage.get_cluster()
return ClickhouseQueryPlan(
query=clickhouse_query,
plan_processors=[],
execution_strategy=SimpleQueryPlanExecutionStrategy(
cluster=cluster,
db_query_processors=[
*self.__storage.get_query_processors(),
*self.__post_processors,
MandatoryConditionApplier(),
],
splitters=self.__storage.get_query_splitters(),
),
)
SelectedExpression(
"average",
FunctionCall(
"average", "avg",
(Column(None, None, "count_environment"), )),
),
],
),
CompositeExecutionStrategy(get_cluster(StorageSetKey.EVENTS), [],
{}, []),
StorageSetKey.EVENTS,
SubqueryProcessors(
[],
[
*events_storage.get_query_processors(),
MandatoryConditionApplier(),
MandatoryConditionEnforcer([]),
],
),
None,
),
CompositeQuery(
from_clause=ClickhouseQuery(
from_clause=events_table,
selected_columns=[
SelectedExpression("project_id",
Column(None, None, "project_id")),
SelectedExpression(
"count_environment",
FunctionCall(
"count_environment",