Validate Semantic Manifests Used in Tests

metricflow/test/fixtures/model_fixtures.py

@@ -4,10 +4,12 @@
 import os
 from collections import OrderedDict
 from dataclasses import dataclass
-from typing import Dict, List, Sequence
+from typing import Dict, List, Optional, Sequence
 
 import pytest
+from dbt_semantic_interfaces.implementations.semantic_manifest import PydanticSemanticManifest
 from dbt_semantic_interfaces.parsing.dir_to_model import (
+    SemanticManifestBuildResult,
     parse_directory_of_yaml_files_to_semantic_manifest,
     parse_yaml_files_to_validation_ready_semantic_manifest,
 )
@@ -148,6 +150,20 @@ def create_data_sets(
     return data_sets
 
 
+def load_semantic_manifest(
+    relative_manifest_path: str,
+    template_mapping: Optional[Dict[str, str]] = None,
+) -> SemanticManifestBuildResult:
+    """Reads the manifest YAMLs from the standard location, applies transformations, runs validations."""
+    yaml_file_directory = os.path.join(os.path.dirname(__file__), f"semantic_manifest_yamls/{relative_manifest_path}")
+    build_result = parse_directory_of_yaml_files_to_semantic_manifest(
+        yaml_file_directory, template_mapping=template_mapping
+    )
+    validator = SemanticManifestValidator[PydanticSemanticManifest]()
+    validator.checked_validations(build_result.semantic_manifest)
+    return build_result
+
+
 @pytest.fixture(scope="session")
 def template_mapping(mf_test_session_state: MetricFlowTestSessionState) -> Dict[str, str]:
     """Mapping for template variables in the model YAML files."""
@@ -156,64 +172,41 @@ def template_mapping(mf_test_session_state: MetricFlowTestSessionState) -> Dict[
 
 @pytest.fixture(scope="session")
 def simple_semantic_manifest_lookup_non_ds(template_mapping: Dict[str, str]) -> SemanticManifestLookup:  # noqa: D
-    build_result = parse_directory_of_yaml_files_to_semantic_manifest(
-        os.path.join(os.path.dirname(__file__), "semantic_manifest_yamls/non_sm_manifest"),
-        template_mapping=template_mapping,
-    )
+    build_result = load_semantic_manifest("non_sm_manifest", template_mapping)
     return SemanticManifestLookup(build_result.semantic_manifest)
 
 
 @pytest.fixture(scope="session")
 def simple_semantic_manifest_lookup(template_mapping: Dict[str, str]) -> SemanticManifestLookup:  # noqa: D
-    build_result = parse_directory_of_yaml_files_to_semantic_manifest(
-        os.path.join(os.path.dirname(__file__), "semantic_manifest_yamls/simple_manifest"),
-        template_mapping=template_mapping,
-    )
+    build_result = load_semantic_manifest("simple_manifest", template_mapping)
     return SemanticManifestLookup(build_result.semantic_manifest)
 
 
 @pytest.fixture(scope="session")
 def multi_hop_join_semantic_manifest_lookup(template_mapping: Dict[str, str]) -> SemanticManifestLookup:  # noqa: D
-    build_result = parse_directory_of_yaml_files_to_semantic_manifest(
-        os.path.join(
-            os.path.dirname(__file__), "semantic_manifest_yamls/multi_hop_join_manifest/partitioned_semantic_models"
-        ),
-        template_mapping=template_mapping,
-    )
+    build_result = load_semantic_manifest("multi_hop_join_manifest/partitioned_semantic_models", template_mapping)
     return SemanticManifestLookup(build_result.semantic_manifest)
 
 
 @pytest.fixture(scope="session")
 def unpartitioned_multi_hop_join_semantic_manifest_lookup(  # noqa: D
     template_mapping: Dict[str, str]
 ) -> SemanticManifestLookup:
-    build_result = parse_directory_of_yaml_files_to_semantic_manifest(
-        os.path.join(
-            os.path.dirname(__file__), "semantic_manifest_yamls/multi_hop_join_manifest/unpartitioned_semantic_models"
-        ),
-        template_mapping=template_mapping,
-    )
+    build_result = load_semantic_manifest("multi_hop_join_manifest/unpartitioned_semantic_models", template_mapping)
     return SemanticManifestLookup(build_result.semantic_manifest)
 
 
 @pytest.fixture(scope="session")
 def simple_semantic_manifest(template_mapping: Dict[str, str]) -> SemanticManifest:
     """Model used for many tests."""
-    build_result = parse_directory_of_yaml_files_to_semantic_manifest(
-        os.path.join(os.path.dirname(__file__), "semantic_manifest_yamls/simple_manifest"),
-        template_mapping=template_mapping,
-    )
+    build_result = load_semantic_manifest("simple_manifest", template_mapping)
     return build_result.semantic_manifest
 
 
 @pytest.fixture(scope="session")
 def simple_model__with_primary_transforms(template_mapping: Dict[str, str]) -> SemanticManifest:
     """Model used for tests pre-transformations."""
-    build_result = parse_directory_of_yaml_files_to_semantic_manifest(
-        os.path.join(os.path.dirname(__file__), "semantic_manifest_yamls/simple_manifest"),
-        template_mapping=template_mapping,
-        apply_transformations=False,
-    )
+    build_result = load_semantic_manifest("simple_manifest", template_mapping)
     transformed_model = PydanticSemanticManifestTransformer().transform(
         model=build_result.semantic_manifest,
         ordered_rule_sequences=(PydanticSemanticManifestTransformRuleSet().primary_rules,),
@@ -223,38 +216,26 @@ def simple_model__with_primary_transforms(template_mapping: Dict[str, str]) -> S
 
 @pytest.fixture(scope="session")
 def extended_date_semantic_manifest_lookup(template_mapping: Dict[str, str]) -> SemanticManifestLookup:  # noqa: D
-    build_result = parse_directory_of_yaml_files_to_semantic_manifest(
-        os.path.join(os.path.dirname(__file__), "semantic_manifest_yamls/extended_date_manifest"),
-        template_mapping=template_mapping,
-    )
+    build_result = load_semantic_manifest("extended_date_manifest", template_mapping)
     return SemanticManifestLookup(build_result.semantic_manifest)
 
 
 @pytest.fixture(scope="session")
 def scd_semantic_manifest_lookup(template_mapping: Dict[str, str]) -> SemanticManifestLookup:
     """Initialize semantic model for SCD tests."""
-    build_result = parse_directory_of_yaml_files_to_semantic_manifest(
-        os.path.join(os.path.dirname(__file__), "semantic_manifest_yamls/scd_manifest"),
-        template_mapping=template_mapping,
-    )
+    build_result = load_semantic_manifest("scd_manifest", template_mapping)
     return SemanticManifestLookup(build_result.semantic_manifest)
 
 
 @pytest.fixture(scope="session")
 def data_warehouse_validation_model(template_mapping: Dict[str, str]) -> SemanticManifest:
     """Model used for data warehouse validation tests."""
-    build_result = parse_directory_of_yaml_files_to_semantic_manifest(
-        os.path.join(os.path.dirname(__file__), "semantic_manifest_yamls/data_warehouse_validation_manifest"),
-        template_mapping=template_mapping,
-    )
+    build_result = load_semantic_manifest("data_warehouse_validation_manifest", template_mapping)
     return build_result.semantic_manifest
 
 
 @pytest.fixture(scope="session")
 def cyclic_join_semantic_manifest_lookup(template_mapping: Dict[str, str]) -> SemanticManifestLookup:
     """Manifest that contains a potential cycle in the join graph (if not handled properly)."""
-    build_result = parse_directory_of_yaml_files_to_semantic_manifest(
-        os.path.join(os.path.dirname(__file__), "semantic_manifest_yamls/cyclic_join_manifest"),
-        template_mapping=template_mapping,
-    )
+    build_result = load_semantic_manifest("cyclic_join_manifest", template_mapping)
     return SemanticManifestLookup(build_result.semantic_manifest)

metricflow/test/fixtures/semantic_manifest_yamls/extended_date_manifest/semantic_models/bookings_monthly_source.yaml

@@ -8,19 +8,23 @@ semantic_model:
     alias: fct_bookings_extended_monthly
 
   defaults:
-    agg_time_dimension: ds
+    agg_time_dimension: monthly_ds
 
   measures:
     - name: bookings_monthly
       agg: sum
       create_metric: true
 
-  primary_entity: booking_id
+  primary_entity: booking
 
   dimensions:
-    - name: ds
+    - name: monthly_ds
+      expr: ds
       type: time
       type_params:
         time_granularity: month
-    - name: is_instant
-      type: categorical
+
+  entities:
+    - name: listing
+      expr: listing_id
+      type: foreign

metricflow/test/fixtures/semantic_manifest_yamls/extended_date_manifest/semantic_models/extended_bookings_source.yaml

@@ -25,11 +25,11 @@ semantic_model:
       type: time
       type_params:
         time_granularity: day
-    - name: is_instant
-      type: categorical
 
   entities:
-    - name: booking_id
+    - name: booking
+      expr: booking_id
       type: primary
-    - name: listing_id
+    - name: listing
+      expr: listing_id
       type: foreign

metricflow/test/fixtures/semantic_manifest_yamls/extended_date_manifest/semantic_models/listings_extended_source.yaml

@@ -8,11 +8,13 @@ semantic_model:
     alias: dim_listings_extended
 
   dimensions:
-    - name: listing_creation_ds
+    - name: ds
+      expr: listing_creation_ds
       type: time
       type_params:
         time_granularity: day
 
   entities:
-    - name: listing_id
+    - name: listing
+      expr: listing_id
       type: primary

metricflow/test/fixtures/semantic_manifest_yamls/multi_hop_join_manifest/partitioned_semantic_models/bridge_table.yaml

@@ -16,8 +16,10 @@ semantic_model:
         time_granularity: day
       is_partition: true
 
+  primary_entity: bridge_account
+
   entities:
     - name: account_id
-      type: primary
+      type: unique
     - name: customer_id
       type: foreign

metricflow/test/fixtures/semantic_manifest_yamls/scd_manifest/scd_accounts.yaml

@@ -25,6 +25,8 @@ semantic_model:
     - name: account_type
       type: categorical
 
+  primary_entity: scd_user
+
   entities:
     - name: user
       type: natural

metricflow/test/fixtures/semantic_manifest_yamls/simple_manifest/semantic_models/users_latest.yaml

@@ -8,8 +8,9 @@ semantic_model:
     alias: dim_users_latest
 
   dimensions:
-    - name: ds
+    - name: ds_latest
       type: time
+      expr: ds
       type_params:
         time_granularity: day
     - name: home_state_latest

metricflow/test/fixtures/source_table_snapshots/extended_date_model/dim_listings_extended.yaml

@@ -4,7 +4,7 @@ table_snapshot:
     - name: listing_creation_ds
       type: TIME
     - name: listing_id
-      type: INT
+      type: STRING
   rows:
     - ["2020-01-01", "0"]
     - ["2020-01-02", "1"]

metricflow/test/fixtures/source_table_snapshots/extended_date_model/fct_bookings_extended.yaml

@@ -8,7 +8,7 @@ table_snapshot:
     - name: booking
       type: INT
     - name: listing_id
-      type: INT
+      type: STRING
     - name: is_instant
       type: BOOLEAN
   rows:

metricflow/test/fixtures/source_table_snapshots/extended_date_model/fct_bookings_extended_monthly.yaml

@@ -7,28 +7,30 @@ table_snapshot:
       type: INT
     - name: is_instant
       type: BOOLEAN
+    - name: listing_id
+      type: STRING
   rows:
-    - ["2020-01-01", "10", "True"]
-    - ["2020-01-01", "5", "False"]
-    - ["2020-02-01", "10", "True"]
-    - ["2020-02-01", "5", "False"]
-    - ["2020-03-01", "10", "True"]
-    - ["2020-03-01", "5", "False"]
-    - ["2020-04-01", "10", "True"]
-    - ["2020-04-01", "5", "False"]
-    - ["2020-05-01", "10", "True"]
-    - ["2020-05-01", "5", "False"]
-    - ["2020-06-01", "10", "True"]
-    - ["2020-06-01", "5", "False"]
-    - ["2020-07-01", "10", "True"]
-    - ["2020-07-01", "5", "False"]
-    - ["2020-08-01", "10", "True"]
-    - ["2020-08-01", "5", "False"]
-    - ["2020-09-01", "10", "True"]
-    - ["2020-09-01", "5", "False"]
-    - ["2020-10-01", "10", "True"]
-    - ["2020-10-01", "5", "False"]
-    - ["2020-11-01", "10", "True"]
-    - ["2020-11-01", "5", "False"]
-    - ["2020-12-01", "10", "True"]
-    - ["2020-12-01", "5", "False"]
+    - ["2020-01-01", "10", "True", "0"]
+    - ["2020-01-01", "5", "False", "1"]
+    - ["2020-02-01", "10", "True", "2"]
+    - ["2020-02-01", "5", "False", "3"]
+    - ["2020-03-01", "10", "True", "4"]
+    - ["2020-03-01", "5", "False", "5"]
+    - ["2020-04-01", "10", "True", "6"]
+    - ["2020-04-01", "5", "False", "7"]
+    - ["2020-05-01", "10", "True", "8"]
+    - ["2020-05-01", "5", "False", "9"]
+    - ["2020-06-01", "10", "True", "0"]
+    - ["2020-06-01", "5", "False", "0"]
+    - ["2020-07-01", "10", "True", "1"]
+    - ["2020-07-01", "5", "False", "2"]
+    - ["2020-08-01", "10", "True", "3"]
+    - ["2020-08-01", "5", "False", "4"]
+    - ["2020-09-01", "10", "True", "5"]
+    - ["2020-09-01", "5", "False", "6"]
+    - ["2020-10-01", "10", "True", "7"]
+    - ["2020-10-01", "5", "False", "8"]
+    - ["2020-11-01", "10", "True", "9"]
+    - ["2020-11-01", "5", "False", "0"]
+    - ["2020-12-01", "10", "True", "0"]
+    - ["2020-12-01", "5", "False", "1"]