Add test for ID enumeration.

metricflow/test/integration/test_rendered_query.py

@@ -2,14 +2,19 @@
 
 import pytest
 from _pytest.fixtures import FixtureRequest
+from dbt_semantic_interfaces.test_utils import as_datetime
 
 from metricflow.dataflow.sql_table import SqlTable
-from metricflow.engine.metricflow_engine import MetricFlowQueryRequest
+from metricflow.engine.metricflow_engine import MetricFlowEngine, MetricFlowQueryRequest
+from metricflow.model.semantic_manifest_lookup import SemanticManifestLookup
+from metricflow.plan_conversion.column_resolver import DunderColumnAssociationResolver
+from metricflow.protocols.sql_client import SqlClient
 from metricflow.test.fixtures.setup_fixtures import MetricFlowTestSessionState
 from metricflow.test.integration.conftest import IntegrationTestHelpers
 from metricflow.test.snapshot_utils import (
     assert_sql_snapshot_equal,
 )
+from metricflow.test.time.configurable_time_source import ConfigurableTimeSource
 
 
 @pytest.mark.sql_engine_snapshot
@@ -51,3 +56,102 @@ def test_render_write_to_table_query(  # noqa: D
         sql=result.rendered_sql.sql_query,
         sql_engine=it_helpers.sql_client.sql_engine_type,
     )
+
+
+@pytest.mark.sql_engine_snapshot
+def test_id_enumeration(  # noqa: D
+    request: FixtureRequest,
+    mf_test_session_state: MetricFlowTestSessionState,
+    simple_semantic_manifest_lookup: SemanticManifestLookup,
+    sql_client: SqlClient,
+) -> None:
+    mf_engine = MetricFlowEngine(
+        semantic_manifest_lookup=simple_semantic_manifest_lookup,
+        sql_client=sql_client,
+        column_association_resolver=DunderColumnAssociationResolver(
+            semantic_manifest_lookup=simple_semantic_manifest_lookup
+        ),
+        time_source=ConfigurableTimeSource(as_datetime("2020-01-01")),
+        consistent_id_enumeration=True,
+    )
+
+    result = mf_engine.explain(
+        MetricFlowQueryRequest.create_with_random_request_id(
+            metric_names=["bookings", "listings"],
+            group_by_names=["metric_time"],
+        )
+    )
+
+    assert_sql_snapshot_equal(
+        request=request,
+        mf_test_session_state=mf_test_session_state,
+        snapshot_id="query",
+        sql=result.rendered_sql.sql_query,
+        sql_engine=sql_client.sql_engine_type,
+    )
+
+    # The resulting snapshot should be the same since mf_engine was created with consistent_id_enumeration=True
+    result = mf_engine.explain(
+        MetricFlowQueryRequest.create_with_random_request_id(
+            metric_names=["bookings", "listings"],
+            group_by_names=["metric_time"],
+        )
+    )
+
+    assert_sql_snapshot_equal(
+        request=request,
+        mf_test_session_state=mf_test_session_state,
+        snapshot_id="query",
+        sql=result.rendered_sql.sql_query,
+        sql_engine=sql_client.sql_engine_type,
+    )
+
+
+@pytest.mark.sql_engine_snapshot
+def test_id_enumeration_without_consistency_flag(
+    request: FixtureRequest,
+    mf_test_session_state: MetricFlowTestSessionState,
+    simple_semantic_manifest_lookup: SemanticManifestLookup,
+    sql_client: SqlClient,
+) -> None:
+    """Test generated SQL with consistent_id_enumeration=False."""
+    mf_engine = MetricFlowEngine(
+        semantic_manifest_lookup=simple_semantic_manifest_lookup,
+        sql_client=sql_client,
+        column_association_resolver=DunderColumnAssociationResolver(
+            semantic_manifest_lookup=simple_semantic_manifest_lookup
+        ),
+        time_source=ConfigurableTimeSource(as_datetime("2020-01-01")),
+        consistent_id_enumeration=False,
+    )
+
+    result = mf_engine.explain(
+        MetricFlowQueryRequest.create_with_random_request_id(
+            metric_names=["bookings", "listings"],
+            group_by_names=["metric_time"],
+        )
+    )
+
+    assert_sql_snapshot_equal(
+        request=request,
+        mf_test_session_state=mf_test_session_state,
+        snapshot_id="first_query",
+        sql=result.rendered_sql.sql_query,
+        sql_engine=sql_client.sql_engine_type,
+    )
+
+    # The second set of queries should have IDs that are sequentially numbered with respect to the previous query.
+    result = mf_engine.explain(
+        MetricFlowQueryRequest.create_with_random_request_id(
+            metric_names=["bookings", "listings"],
+            group_by_names=["metric_time"],
+        )
+    )
+
+    assert_sql_snapshot_equal(
+        request=request,
+        mf_test_session_state=mf_test_session_state,
+        snapshot_id="second_query",
+        sql=result.rendered_sql.sql_query,
+        sql_engine=sql_client.sql_engine_type,
+    )

metricflow/test/snapshots/test_rendered_query.py/str/BigQuery/test_id_enumeration__query.sql

@@ -0,0 +1,45 @@
+-- Combine Aggregated Outputs
+SELECT
+  COALESCE(subq_4.metric_time__day, subq_9.metric_time__day) AS metric_time__day
+  , MAX(subq_4.bookings) AS bookings
+  , MAX(subq_9.listings) AS listings
+FROM (
+  -- Aggregate Measures
+  -- Compute Metrics via Expressions
+  SELECT
+    metric_time__day
+    , SUM(bookings) AS bookings
+  FROM (
+    -- Read Elements From Semantic Model 'bookings_source'
+    -- Metric Time Dimension 'ds'
+    -- Pass Only Elements: ['bookings', 'metric_time__day']
+    SELECT
+      DATE_TRUNC(ds, day) AS metric_time__day
+      , 1 AS bookings
+    FROM ***************************.fct_bookings bookings_source_src_10000
+  ) subq_2
+  GROUP BY
+    metric_time__day
+) subq_4
+FULL OUTER JOIN (
+  -- Aggregate Measures
+  -- Compute Metrics via Expressions
+  SELECT
+    metric_time__day
+    , SUM(listings) AS listings
+  FROM (
+    -- Read Elements From Semantic Model 'listings_latest'
+    -- Metric Time Dimension 'ds'
+    -- Pass Only Elements: ['listings', 'metric_time__day']
+    SELECT
+      DATE_TRUNC(created_at, day) AS metric_time__day
+      , 1 AS listings
+    FROM ***************************.dim_listings_latest listings_latest_src_10000
+  ) subq_7
+  GROUP BY
+    metric_time__day
+) subq_9
+ON
+  subq_4.metric_time__day = subq_9.metric_time__day
+GROUP BY
+  metric_time__day

metricflow/test/snapshots/test_rendered_query.py/str/BigQuery/test_id_enumeration_without_consistency_flag__first_query.sql

@@ -0,0 +1,45 @@
+-- Combine Aggregated Outputs
+SELECT
+  COALESCE(subq_10013.metric_time__day, subq_10018.metric_time__day) AS metric_time__day
+  , MAX(subq_10013.bookings) AS bookings
+  , MAX(subq_10018.listings) AS listings
+FROM (
+  -- Aggregate Measures
+  -- Compute Metrics via Expressions
+  SELECT
+    metric_time__day
+    , SUM(bookings) AS bookings
+  FROM (
+    -- Read Elements From Semantic Model 'bookings_source'
+    -- Metric Time Dimension 'ds'
+    -- Pass Only Elements: ['bookings', 'metric_time__day']
+    SELECT
+      DATE_TRUNC(ds, day) AS metric_time__day
+      , 1 AS bookings
+    FROM ***************************.fct_bookings bookings_source_src_10000
+  ) subq_10011
+  GROUP BY
+    metric_time__day
+) subq_10013
+FULL OUTER JOIN (
+  -- Aggregate Measures
+  -- Compute Metrics via Expressions
+  SELECT
+    metric_time__day
+    , SUM(listings) AS listings
+  FROM (
+    -- Read Elements From Semantic Model 'listings_latest'
+    -- Metric Time Dimension 'ds'
+    -- Pass Only Elements: ['listings', 'metric_time__day']
+    SELECT
+      DATE_TRUNC(created_at, day) AS metric_time__day
+      , 1 AS listings
+    FROM ***************************.dim_listings_latest listings_latest_src_10000
+  ) subq_10016
+  GROUP BY
+    metric_time__day
+) subq_10018
+ON
+  subq_10013.metric_time__day = subq_10018.metric_time__day
+GROUP BY
+  metric_time__day

metricflow/test/snapshots/test_rendered_query.py/str/BigQuery/test_id_enumeration_without_consistency_flag__second_query.sql

@@ -0,0 +1,45 @@
+-- Combine Aggregated Outputs
+SELECT
+  COALESCE(subq_10023.metric_time__day, subq_10028.metric_time__day) AS metric_time__day
+  , MAX(subq_10023.bookings) AS bookings
+  , MAX(subq_10028.listings) AS listings
+FROM (
+  -- Aggregate Measures
+  -- Compute Metrics via Expressions
+  SELECT
+    metric_time__day
+    , SUM(bookings) AS bookings
+  FROM (
+    -- Read Elements From Semantic Model 'bookings_source'
+    -- Metric Time Dimension 'ds'
+    -- Pass Only Elements: ['bookings', 'metric_time__day']
+    SELECT
+      DATE_TRUNC(ds, day) AS metric_time__day
+      , 1 AS bookings
+    FROM ***************************.fct_bookings bookings_source_src_10000
+  ) subq_10021
+  GROUP BY
+    metric_time__day
+) subq_10023
+FULL OUTER JOIN (
+  -- Aggregate Measures
+  -- Compute Metrics via Expressions
+  SELECT
+    metric_time__day
+    , SUM(listings) AS listings
+  FROM (
+    -- Read Elements From Semantic Model 'listings_latest'
+    -- Metric Time Dimension 'ds'
+    -- Pass Only Elements: ['listings', 'metric_time__day']
+    SELECT
+      DATE_TRUNC(created_at, day) AS metric_time__day
+      , 1 AS listings
+    FROM ***************************.dim_listings_latest listings_latest_src_10000
+  ) subq_10026
+  GROUP BY
+    metric_time__day
+) subq_10028
+ON
+  subq_10023.metric_time__day = subq_10028.metric_time__day
+GROUP BY
+  metric_time__day

metricflow/test/snapshots/test_rendered_query.py/str/Databricks/test_id_enumeration__query.sql

@@ -0,0 +1,45 @@
+-- Combine Aggregated Outputs
+SELECT
+  COALESCE(subq_4.metric_time__day, subq_9.metric_time__day) AS metric_time__day
+  , MAX(subq_4.bookings) AS bookings
+  , MAX(subq_9.listings) AS listings
+FROM (
+  -- Aggregate Measures
+  -- Compute Metrics via Expressions
+  SELECT
+    metric_time__day
+    , SUM(bookings) AS bookings
+  FROM (
+    -- Read Elements From Semantic Model 'bookings_source'
+    -- Metric Time Dimension 'ds'
+    -- Pass Only Elements: ['bookings', 'metric_time__day']
+    SELECT
+      DATE_TRUNC('day', ds) AS metric_time__day
+      , 1 AS bookings
+    FROM ***************************.fct_bookings bookings_source_src_10000
+  ) subq_2
+  GROUP BY
+    metric_time__day
+) subq_4
+FULL OUTER JOIN (
+  -- Aggregate Measures
+  -- Compute Metrics via Expressions
+  SELECT
+    metric_time__day
+    , SUM(listings) AS listings
+  FROM (
+    -- Read Elements From Semantic Model 'listings_latest'
+    -- Metric Time Dimension 'ds'
+    -- Pass Only Elements: ['listings', 'metric_time__day']
+    SELECT
+      DATE_TRUNC('day', created_at) AS metric_time__day
+      , 1 AS listings
+    FROM ***************************.dim_listings_latest listings_latest_src_10000
+  ) subq_7
+  GROUP BY
+    metric_time__day
+) subq_9
+ON
+  subq_4.metric_time__day = subq_9.metric_time__day
+GROUP BY
+  COALESCE(subq_4.metric_time__day, subq_9.metric_time__day)

metricflow/test/snapshots/test_rendered_query.py/str/Databricks/test_id_enumeration_without_consistency_flag__first_query.sql

@@ -0,0 +1,45 @@
+-- Combine Aggregated Outputs
+SELECT
+  COALESCE(subq_10013.metric_time__day, subq_10018.metric_time__day) AS metric_time__day
+  , MAX(subq_10013.bookings) AS bookings
+  , MAX(subq_10018.listings) AS listings
+FROM (
+  -- Aggregate Measures
+  -- Compute Metrics via Expressions
+  SELECT
+    metric_time__day
+    , SUM(bookings) AS bookings
+  FROM (
+    -- Read Elements From Semantic Model 'bookings_source'
+    -- Metric Time Dimension 'ds'
+    -- Pass Only Elements: ['bookings', 'metric_time__day']
+    SELECT
+      DATE_TRUNC('day', ds) AS metric_time__day
+      , 1 AS bookings
+    FROM ***************************.fct_bookings bookings_source_src_10000
+  ) subq_10011
+  GROUP BY
+    metric_time__day
+) subq_10013
+FULL OUTER JOIN (
+  -- Aggregate Measures
+  -- Compute Metrics via Expressions
+  SELECT
+    metric_time__day
+    , SUM(listings) AS listings
+  FROM (
+    -- Read Elements From Semantic Model 'listings_latest'
+    -- Metric Time Dimension 'ds'
+    -- Pass Only Elements: ['listings', 'metric_time__day']
+    SELECT
+      DATE_TRUNC('day', created_at) AS metric_time__day
+      , 1 AS listings
+    FROM ***************************.dim_listings_latest listings_latest_src_10000
+  ) subq_10016
+  GROUP BY
+    metric_time__day
+) subq_10018
+ON
+  subq_10013.metric_time__day = subq_10018.metric_time__day
+GROUP BY
+  COALESCE(subq_10013.metric_time__day, subq_10018.metric_time__day)

metricflow/test/snapshots/test_rendered_query.py/str/Databricks/test_id_enumeration_without_consistency_flag__second_query.sql

@@ -0,0 +1,45 @@
+-- Combine Aggregated Outputs
+SELECT
+  COALESCE(subq_10023.metric_time__day, subq_10028.metric_time__day) AS metric_time__day
+  , MAX(subq_10023.bookings) AS bookings
+  , MAX(subq_10028.listings) AS listings
+FROM (
+  -- Aggregate Measures
+  -- Compute Metrics via Expressions
+  SELECT
+    metric_time__day
+    , SUM(bookings) AS bookings
+  FROM (
+    -- Read Elements From Semantic Model 'bookings_source'
+    -- Metric Time Dimension 'ds'
+    -- Pass Only Elements: ['bookings', 'metric_time__day']
+    SELECT
+      DATE_TRUNC('day', ds) AS metric_time__day
+      , 1 AS bookings
+    FROM ***************************.fct_bookings bookings_source_src_10000
+  ) subq_10021
+  GROUP BY
+    metric_time__day
+) subq_10023
+FULL OUTER JOIN (
+  -- Aggregate Measures
+  -- Compute Metrics via Expressions
+  SELECT
+    metric_time__day
+    , SUM(listings) AS listings
+  FROM (
+    -- Read Elements From Semantic Model 'listings_latest'
+    -- Metric Time Dimension 'ds'
+    -- Pass Only Elements: ['listings', 'metric_time__day']
+    SELECT
+      DATE_TRUNC('day', created_at) AS metric_time__day
+      , 1 AS listings
+    FROM ***************************.dim_listings_latest listings_latest_src_10000
+  ) subq_10026
+  GROUP BY
+    metric_time__day
+) subq_10028
+ON
+  subq_10023.metric_time__day = subq_10028.metric_time__day
+GROUP BY
+  COALESCE(subq_10023.metric_time__day, subq_10028.metric_time__day)