fix: validate data_type for partition_by config (#5491)

Author: toriwei

sqlmesh/dbt/model.py

@@ -215,6 +215,14 @@ def _validate_partition_by(
             ):
                 granularity = v["granularity"]
                 raise ConfigError(f"Unexpected granularity '{granularity}' in partition_by '{v}'.")
+            if "data_type" in v and v["data_type"].lower() not in (
+                "timestamp",
+                "date",
+                "datetime",
+                "int64",
+            ):
+                data_type = v["data_type"]
+                raise ConfigError(f"Unexpected data_type '{data_type}' in partition_by '{v}'.")
             return {"data_type": "date", "granularity": "day", **v}
         raise ConfigError(f"Invalid format for partition_by '{v}'")
 

tests/dbt/test_transformation.py

@@ -1881,6 +1881,17 @@ def test_partition_by(sushi_test_project: Project):
     )
     assert model_config.to_sqlmesh(context).partitioned_by == []
 
+    with pytest.raises(ConfigError, match="Unexpected data_type 'string' in partition_by"):
+        ModelConfig(
+            name="model",
+            alias="model",
+            schema="test",
+            package_name="package",
+            materialized="table",
+            partition_by={"field": "ds", "data_type": "string"},
+            sql="""SELECT 1 AS one, ds FROM foo""",
+        )
+
 
 @pytest.mark.xdist_group("dbt_manifest")
 def test_partition_by_none(sushi_test_project: Project):