Add SQL rendering tests for querying a metric with a custom grain filter

metricflow/plan_conversion/dataflow_to_sql.py

@@ -1087,9 +1087,9 @@ def visit_metric_time_dimension_transform_node(self, node: MetricTimeDimensionTr
                     spec=metric_time_dimension_spec,
                 )
             )
-            output_column_to_input_column[metric_time_dimension_column_association.column_name] = (
-                matching_time_dimension_instance.associated_column.column_name
-            )
+            output_column_to_input_column[
+                metric_time_dimension_column_association.column_name
+            ] = matching_time_dimension_instance.associated_column.column_name
 
         output_instance_set = InstanceSet(
             measure_instances=tuple(output_measure_instances),
@@ -1324,11 +1324,11 @@ def visit_join_to_time_spine_node(self, node: JoinToTimeSpineNode) -> SqlDataSet
             and len(time_spine_dataset.checked_sql_select_node.select_columns) == 1
         ), "Time spine dataset not configured properly. Expected exactly one column."
         original_time_spine_dim_instance = time_spine_dataset.instance_set.time_dimension_instances[0]
-        time_spine_column_select_expr: Union[SqlColumnReferenceExpression, SqlDateTruncExpression] = (
-            SqlColumnReferenceExpression.create(
-                SqlColumnReference(
-                    table_alias=time_spine_alias, column_name=original_time_spine_dim_instance.spec.qualified_name
-                )
+        time_spine_column_select_expr: Union[
+            SqlColumnReferenceExpression, SqlDateTruncExpression
+        ] = SqlColumnReferenceExpression.create(
+            SqlColumnReference(
+                table_alias=time_spine_alias, column_name=original_time_spine_dim_instance.spec.qualified_name
             )
         )
 

tests_metricflow/query_rendering/test_custom_granularity.py

@@ -8,8 +8,10 @@
 
 import pytest
 from _pytest.fixtures import FixtureRequest
+from dbt_semantic_interfaces.implementations.filters.where_filter import PydanticWhereFilter
 from dbt_semantic_interfaces.references import EntityReference
 from dbt_semantic_interfaces.type_enums.time_granularity import TimeGranularity
+from metricflow_semantics.query.query_parser import MetricFlowQueryParser
 from metricflow_semantics.specs.metric_spec import MetricSpec
 from metricflow_semantics.specs.query_spec import MetricFlowQuerySpec
 from metricflow_semantics.specs.time_dimension_spec import TimeDimensionSpec
@@ -267,3 +269,59 @@ def test_simple_metric_with_custom_granularity_and_join(  # noqa: D103
         dataflow_plan_builder=dataflow_plan_builder,
         query_spec=query_spec,
     )
+
+
+# TODO: optimizer - could collaps subquery
+@pytest.mark.sql_engine_snapshot
+def test_simple_metric_with_custom_granularity_filter(
+    request: FixtureRequest,
+    mf_test_configuration: MetricFlowTestConfiguration,
+    dataflow_plan_builder: DataflowPlanBuilder,
+    dataflow_to_sql_converter: DataflowToSqlQueryPlanConverter,
+    sql_client: SqlClient,
+    query_parser: MetricFlowQueryParser,
+) -> None:
+    """Simple metric queried with a filter on a custom grain, where that grain is not used in the group by."""
+    query_spec = query_parser.parse_and_validate_query(
+        metric_names=("bookings",),
+        where_constraint=PydanticWhereFilter(
+            where_sql_template=("{{ TimeDimension('metric_time', 'martian_day') }} = '2020-01-01'")
+        ),
+    ).query_spec
+
+    render_and_check(
+        request=request,
+        mf_test_configuration=mf_test_configuration,
+        dataflow_to_sql_converter=dataflow_to_sql_converter,
+        sql_client=sql_client,
+        dataflow_plan_builder=dataflow_plan_builder,
+        query_spec=query_spec,
+    )
+
+
+@pytest.mark.sql_engine_snapshot
+def test_simple_metric_with_custom_granularity_in_filter_and_group_by(
+    request: FixtureRequest,
+    mf_test_configuration: MetricFlowTestConfiguration,
+    dataflow_plan_builder: DataflowPlanBuilder,
+    dataflow_to_sql_converter: DataflowToSqlQueryPlanConverter,
+    sql_client: SqlClient,
+    query_parser: MetricFlowQueryParser,
+) -> None:
+    """Simple metric queried with a filter on a custom grain, where that grain is also used in the group by."""
+    query_spec = query_parser.parse_and_validate_query(
+        metric_names=("bookings",),
+        group_by_names=("metric_time__martian_day",),
+        where_constraint=PydanticWhereFilter(
+            where_sql_template=("{{ TimeDimension('metric_time', 'martian_day') }} = '2020-01-01'")
+        ),
+    ).query_spec
+
+    render_and_check(
+        request=request,
+        mf_test_configuration=mf_test_configuration,
+        dataflow_to_sql_converter=dataflow_to_sql_converter,
+        sql_client=sql_client,
+        dataflow_plan_builder=dataflow_plan_builder,
+        query_spec=query_spec,
+    )

tests_metricflow/snapshots/test_custom_granularity.py/SqlQueryPlan/BigQuery/test_simple_metric_with_custom_granularity_filter__plan0.sql

@@ -0,0 +1,226 @@
+-- Compute Metrics via Expressions
+SELECT
+  subq_6.bookings
+FROM (
+  -- Aggregate Measures
+  SELECT
+    SUM(subq_5.bookings) AS bookings
+  FROM (
+    -- Pass Only Elements: ['bookings',]
+    SELECT
+      subq_4.bookings
+    FROM (
+      -- Constrain Output with WHERE
+      SELECT
+        subq_3.metric_time__martian_day
+        , subq_3.bookings
+      FROM (
+        -- Join to Custom Granularity Dataset
+        -- Pass Only Elements: ['bookings', 'metric_time__day']
+        SELECT
+          subq_1.bookings AS bookings
+          , subq_2.martian_day AS metric_time__martian_day
+        FROM (
+          -- Metric Time Dimension 'ds'
+          SELECT
+            subq_0.ds__day
+            , subq_0.ds__week
+            , subq_0.ds__month
+            , subq_0.ds__quarter
+            , subq_0.ds__year
+            , subq_0.ds__extract_year
+            , subq_0.ds__extract_quarter
+            , subq_0.ds__extract_month
+            , subq_0.ds__extract_day
+            , subq_0.ds__extract_dow
+            , subq_0.ds__extract_doy
+            , subq_0.ds_partitioned__day
+            , subq_0.ds_partitioned__week
+            , subq_0.ds_partitioned__month
+            , subq_0.ds_partitioned__quarter
+            , subq_0.ds_partitioned__year
+            , subq_0.ds_partitioned__extract_year
+            , subq_0.ds_partitioned__extract_quarter
+            , subq_0.ds_partitioned__extract_month
+            , subq_0.ds_partitioned__extract_day
+            , subq_0.ds_partitioned__extract_dow
+            , subq_0.ds_partitioned__extract_doy
+            , subq_0.paid_at__day
+            , subq_0.paid_at__week
+            , subq_0.paid_at__month
+            , subq_0.paid_at__quarter
+            , subq_0.paid_at__year
+            , subq_0.paid_at__extract_year
+            , subq_0.paid_at__extract_quarter
+            , subq_0.paid_at__extract_month
+            , subq_0.paid_at__extract_day
+            , subq_0.paid_at__extract_dow
+            , subq_0.paid_at__extract_doy
+            , subq_0.booking__ds__day
+            , subq_0.booking__ds__week
+            , subq_0.booking__ds__month
+            , subq_0.booking__ds__quarter
+            , subq_0.booking__ds__year
+            , subq_0.booking__ds__extract_year
+            , subq_0.booking__ds__extract_quarter
+            , subq_0.booking__ds__extract_month
+            , subq_0.booking__ds__extract_day
+            , subq_0.booking__ds__extract_dow
+            , subq_0.booking__ds__extract_doy
+            , subq_0.booking__ds_partitioned__day
+            , subq_0.booking__ds_partitioned__week
+            , subq_0.booking__ds_partitioned__month
+            , subq_0.booking__ds_partitioned__quarter
+            , subq_0.booking__ds_partitioned__year
+            , subq_0.booking__ds_partitioned__extract_year
+            , subq_0.booking__ds_partitioned__extract_quarter
+            , subq_0.booking__ds_partitioned__extract_month
+            , subq_0.booking__ds_partitioned__extract_day
+            , subq_0.booking__ds_partitioned__extract_dow
+            , subq_0.booking__ds_partitioned__extract_doy
+            , subq_0.booking__paid_at__day
+            , subq_0.booking__paid_at__week
+            , subq_0.booking__paid_at__month
+            , subq_0.booking__paid_at__quarter
+            , subq_0.booking__paid_at__year
+            , subq_0.booking__paid_at__extract_year
+            , subq_0.booking__paid_at__extract_quarter
+            , subq_0.booking__paid_at__extract_month
+            , subq_0.booking__paid_at__extract_day
+            , subq_0.booking__paid_at__extract_dow
+            , subq_0.booking__paid_at__extract_doy
+            , subq_0.ds__day AS metric_time__day
+            , subq_0.ds__week AS metric_time__week
+            , subq_0.ds__month AS metric_time__month
+            , subq_0.ds__quarter AS metric_time__quarter
+            , subq_0.ds__year AS metric_time__year
+            , subq_0.ds__extract_year AS metric_time__extract_year
+            , subq_0.ds__extract_quarter AS metric_time__extract_quarter
+            , subq_0.ds__extract_month AS metric_time__extract_month
+            , subq_0.ds__extract_day AS metric_time__extract_day
+            , subq_0.ds__extract_dow AS metric_time__extract_dow
+            , subq_0.ds__extract_doy AS metric_time__extract_doy
+            , subq_0.listing
+            , subq_0.guest
+            , subq_0.host
+            , subq_0.booking__listing
+            , subq_0.booking__guest
+            , subq_0.booking__host
+            , subq_0.is_instant
+            , subq_0.booking__is_instant
+            , subq_0.bookings
+            , subq_0.instant_bookings
+            , subq_0.booking_value
+            , subq_0.max_booking_value
+            , subq_0.min_booking_value
+            , subq_0.bookers
+            , subq_0.average_booking_value
+            , subq_0.referred_bookings
+            , subq_0.median_booking_value
+            , subq_0.booking_value_p99
+            , subq_0.discrete_booking_value_p99
+            , subq_0.approximate_continuous_booking_value_p99
+            , subq_0.approximate_discrete_booking_value_p99
+          FROM (
+            -- Read Elements From Semantic Model 'bookings_source'
+            SELECT
+              1 AS bookings
+              , CASE WHEN is_instant THEN 1 ELSE 0 END AS instant_bookings
+              , bookings_source_src_28000.booking_value
+              , bookings_source_src_28000.booking_value AS max_booking_value
+              , bookings_source_src_28000.booking_value AS min_booking_value
+              , bookings_source_src_28000.guest_id AS bookers
+              , bookings_source_src_28000.booking_value AS average_booking_value
+              , bookings_source_src_28000.booking_value AS booking_payments
+              , CASE WHEN referrer_id IS NOT NULL THEN 1 ELSE 0 END AS referred_bookings
+              , bookings_source_src_28000.booking_value AS median_booking_value
+              , bookings_source_src_28000.booking_value AS booking_value_p99
+              , bookings_source_src_28000.booking_value AS discrete_booking_value_p99
+              , bookings_source_src_28000.booking_value AS approximate_continuous_booking_value_p99
+              , bookings_source_src_28000.booking_value AS approximate_discrete_booking_value_p99
+              , bookings_source_src_28000.is_instant
+              , DATETIME_TRUNC(bookings_source_src_28000.ds, day) AS ds__day
+              , DATETIME_TRUNC(bookings_source_src_28000.ds, isoweek) AS ds__week
+              , DATETIME_TRUNC(bookings_source_src_28000.ds, month) AS ds__month
+              , DATETIME_TRUNC(bookings_source_src_28000.ds, quarter) AS ds__quarter
+              , DATETIME_TRUNC(bookings_source_src_28000.ds, year) AS ds__year
+              , EXTRACT(year FROM bookings_source_src_28000.ds) AS ds__extract_year
+              , EXTRACT(quarter FROM bookings_source_src_28000.ds) AS ds__extract_quarter
+              , EXTRACT(month FROM bookings_source_src_28000.ds) AS ds__extract_month
+              , EXTRACT(day FROM bookings_source_src_28000.ds) AS ds__extract_day
+              , IF(EXTRACT(dayofweek FROM bookings_source_src_28000.ds) = 1, 7, EXTRACT(dayofweek FROM bookings_source_src_28000.ds) - 1) AS ds__extract_dow
+              , EXTRACT(dayofyear FROM bookings_source_src_28000.ds) AS ds__extract_doy
+              , DATETIME_TRUNC(bookings_source_src_28000.ds_partitioned, day) AS ds_partitioned__day
+              , DATETIME_TRUNC(bookings_source_src_28000.ds_partitioned, isoweek) AS ds_partitioned__week
+              , DATETIME_TRUNC(bookings_source_src_28000.ds_partitioned, month) AS ds_partitioned__month
+              , DATETIME_TRUNC(bookings_source_src_28000.ds_partitioned, quarter) AS ds_partitioned__quarter
+              , DATETIME_TRUNC(bookings_source_src_28000.ds_partitioned, year) AS ds_partitioned__year
+              , EXTRACT(year FROM bookings_source_src_28000.ds_partitioned) AS ds_partitioned__extract_year
+              , EXTRACT(quarter FROM bookings_source_src_28000.ds_partitioned) AS ds_partitioned__extract_quarter
+              , EXTRACT(month FROM bookings_source_src_28000.ds_partitioned) AS ds_partitioned__extract_month
+              , EXTRACT(day FROM bookings_source_src_28000.ds_partitioned) AS ds_partitioned__extract_day
+              , IF(EXTRACT(dayofweek FROM bookings_source_src_28000.ds_partitioned) = 1, 7, EXTRACT(dayofweek FROM bookings_source_src_28000.ds_partitioned) - 1) AS ds_partitioned__extract_dow
+              , EXTRACT(dayofyear FROM bookings_source_src_28000.ds_partitioned) AS ds_partitioned__extract_doy
+              , DATETIME_TRUNC(bookings_source_src_28000.paid_at, day) AS paid_at__day
+              , DATETIME_TRUNC(bookings_source_src_28000.paid_at, isoweek) AS paid_at__week
+              , DATETIME_TRUNC(bookings_source_src_28000.paid_at, month) AS paid_at__month
+              , DATETIME_TRUNC(bookings_source_src_28000.paid_at, quarter) AS paid_at__quarter
+              , DATETIME_TRUNC(bookings_source_src_28000.paid_at, year) AS paid_at__year
+              , EXTRACT(year FROM bookings_source_src_28000.paid_at) AS paid_at__extract_year
+              , EXTRACT(quarter FROM bookings_source_src_28000.paid_at) AS paid_at__extract_quarter
+              , EXTRACT(month FROM bookings_source_src_28000.paid_at) AS paid_at__extract_month
+              , EXTRACT(day FROM bookings_source_src_28000.paid_at) AS paid_at__extract_day
+              , IF(EXTRACT(dayofweek FROM bookings_source_src_28000.paid_at) = 1, 7, EXTRACT(dayofweek FROM bookings_source_src_28000.paid_at) - 1) AS paid_at__extract_dow
+              , EXTRACT(dayofyear FROM bookings_source_src_28000.paid_at) AS paid_at__extract_doy
+              , bookings_source_src_28000.is_instant AS booking__is_instant
+              , DATETIME_TRUNC(bookings_source_src_28000.ds, day) AS booking__ds__day
+              , DATETIME_TRUNC(bookings_source_src_28000.ds, isoweek) AS booking__ds__week
+              , DATETIME_TRUNC(bookings_source_src_28000.ds, month) AS booking__ds__month
+              , DATETIME_TRUNC(bookings_source_src_28000.ds, quarter) AS booking__ds__quarter
+              , DATETIME_TRUNC(bookings_source_src_28000.ds, year) AS booking__ds__year
+              , EXTRACT(year FROM bookings_source_src_28000.ds) AS booking__ds__extract_year
+              , EXTRACT(quarter FROM bookings_source_src_28000.ds) AS booking__ds__extract_quarter
+              , EXTRACT(month FROM bookings_source_src_28000.ds) AS booking__ds__extract_month
+              , EXTRACT(day FROM bookings_source_src_28000.ds) AS booking__ds__extract_day
+              , IF(EXTRACT(dayofweek FROM bookings_source_src_28000.ds) = 1, 7, EXTRACT(dayofweek FROM bookings_source_src_28000.ds) - 1) AS booking__ds__extract_dow
+              , EXTRACT(dayofyear FROM bookings_source_src_28000.ds) AS booking__ds__extract_doy
+              , DATETIME_TRUNC(bookings_source_src_28000.ds_partitioned, day) AS booking__ds_partitioned__day
+              , DATETIME_TRUNC(bookings_source_src_28000.ds_partitioned, isoweek) AS booking__ds_partitioned__week
+              , DATETIME_TRUNC(bookings_source_src_28000.ds_partitioned, month) AS booking__ds_partitioned__month
+              , DATETIME_TRUNC(bookings_source_src_28000.ds_partitioned, quarter) AS booking__ds_partitioned__quarter
+              , DATETIME_TRUNC(bookings_source_src_28000.ds_partitioned, year) AS booking__ds_partitioned__year
+              , EXTRACT(year FROM bookings_source_src_28000.ds_partitioned) AS booking__ds_partitioned__extract_year
+              , EXTRACT(quarter FROM bookings_source_src_28000.ds_partitioned) AS booking__ds_partitioned__extract_quarter
+              , EXTRACT(month FROM bookings_source_src_28000.ds_partitioned) AS booking__ds_partitioned__extract_month
+              , EXTRACT(day FROM bookings_source_src_28000.ds_partitioned) AS booking__ds_partitioned__extract_day
+              , IF(EXTRACT(dayofweek FROM bookings_source_src_28000.ds_partitioned) = 1, 7, EXTRACT(dayofweek FROM bookings_source_src_28000.ds_partitioned) - 1) AS booking__ds_partitioned__extract_dow
+              , EXTRACT(dayofyear FROM bookings_source_src_28000.ds_partitioned) AS booking__ds_partitioned__extract_doy
+              , DATETIME_TRUNC(bookings_source_src_28000.paid_at, day) AS booking__paid_at__day
+              , DATETIME_TRUNC(bookings_source_src_28000.paid_at, isoweek) AS booking__paid_at__week
+              , DATETIME_TRUNC(bookings_source_src_28000.paid_at, month) AS booking__paid_at__month
+              , DATETIME_TRUNC(bookings_source_src_28000.paid_at, quarter) AS booking__paid_at__quarter
+              , DATETIME_TRUNC(bookings_source_src_28000.paid_at, year) AS booking__paid_at__year
+              , EXTRACT(year FROM bookings_source_src_28000.paid_at) AS booking__paid_at__extract_year
+              , EXTRACT(quarter FROM bookings_source_src_28000.paid_at) AS booking__paid_at__extract_quarter
+              , EXTRACT(month FROM bookings_source_src_28000.paid_at) AS booking__paid_at__extract_month
+              , EXTRACT(day FROM bookings_source_src_28000.paid_at) AS booking__paid_at__extract_day
+              , IF(EXTRACT(dayofweek FROM bookings_source_src_28000.paid_at) = 1, 7, EXTRACT(dayofweek FROM bookings_source_src_28000.paid_at) - 1) AS booking__paid_at__extract_dow
+              , EXTRACT(dayofyear FROM bookings_source_src_28000.paid_at) AS booking__paid_at__extract_doy
+              , bookings_source_src_28000.listing_id AS listing
+              , bookings_source_src_28000.guest_id AS guest
+              , bookings_source_src_28000.host_id AS host
+              , bookings_source_src_28000.listing_id AS booking__listing
+              , bookings_source_src_28000.guest_id AS booking__guest
+              , bookings_source_src_28000.host_id AS booking__host
+            FROM ***************************.fct_bookings bookings_source_src_28000
+          ) subq_0
+        ) subq_1
+        LEFT OUTER JOIN
+          ***************************.mf_time_spine subq_2
+        ON
+          subq_1.metric_time__day = subq_2.ds
+      ) subq_3
+      WHERE metric_time__martian_day = '2020-01-01'
+    ) subq_4
+  ) subq_5
+) subq_6

tests_metricflow/snapshots/test_custom_granularity.py/SqlQueryPlan/BigQuery/test_simple_metric_with_custom_granularity_filter__plan0_optimized.sql

@@ -0,0 +1,26 @@
+-- Constrain Output with WHERE
+-- Pass Only Elements: ['bookings',]
+-- Aggregate Measures
+-- Compute Metrics via Expressions
+SELECT
+  SUM(bookings) AS bookings
+FROM (
+  -- Join to Custom Granularity Dataset
+  -- Pass Only Elements: ['bookings', 'metric_time__day']
+  SELECT
+    subq_8.bookings AS bookings
+    , subq_9.martian_day AS metric_time__martian_day
+  FROM (
+    -- Read Elements From Semantic Model 'bookings_source'
+    -- Metric Time Dimension 'ds'
+    SELECT
+      DATETIME_TRUNC(ds, day) AS metric_time__day
+      , 1 AS bookings
+    FROM ***************************.fct_bookings bookings_source_src_28000
+  ) subq_8
+  LEFT OUTER JOIN
+    ***************************.mf_time_spine subq_9
+  ON
+    subq_8.metric_time__day = subq_9.ds
+) subq_10
+WHERE metric_time__martian_day = '2020-01-01'