fix: fix time_data_type serialization/deserialization (#2342)

* fix: fix time_data_type serialization/deserialization

* bump sqlglot

Author: eakmanrq

setup.py

@@ -45,7 +45,7 @@
         "requests",
         "rich[jupyter]",
         "ruamel.yaml",
-        "sqlglot[rs]~=23.0.3",
+        "sqlglot[rs]~=23.2.0",
     ],
     extras_require={
         "bigquery": [

sqlmesh/core/model/kind.py

@@ -370,20 +370,23 @@ class _SCDType2Kind(_ModelKind):
     valid_from_name: SQLGlotString = "valid_from"
     valid_to_name: SQLGlotString = "valid_to"
     invalidate_hard_deletes: SQLGlotBool = False
-    time_data_type: exp.DataType = exp.DataType.build("TIMESTAMP")
+    time_data_type: exp.DataType = Field(exp.DataType.build("TIMESTAMP"), validate_default=True)
 
     forward_only: SQLGlotBool = True
     disable_restatement: SQLGlotBool = True
 
-    @field_validator("time_data_type", mode="before")
+    # always=True can be removed once Pydantic 1 is deprecated
+    @field_validator("time_data_type", mode="before", always=True)
     @classmethod
     def _time_data_type_validator(
         cls, v: t.Union[str, exp.Expression], values: t.Any
     ) -> exp.Expression:
         values = values if isinstance(values, dict) else values.data
         if isinstance(v, exp.Expression) and not isinstance(v, exp.DataType):
             v = v.name
-        return exp.DataType.build(v, dialect=values.get("dialect"))
+        data_type = exp.DataType.build(v, dialect=values.get("dialect"))
+        data_type.meta["dialect"] = values.get("dialect")
+        return data_type
 
     @property
     def managed_columns(self) -> t.Dict[str, exp.DataType]:

tests/core/test_model.py

@@ -444,6 +444,40 @@ def test_json_serde():
     assert deserialized_model == model
 
 
+def test_scd_type_2_by_col_serde():
+    expressions = parse(
+        """
+        MODEL(
+            name test_model,
+            KIND SCD_TYPE_2_BY_COLUMN (
+                unique_key a,
+                columns *,
+            ),
+            cron '@daily',
+            owner 'reakman',
+            grain a,
+            dialect bigquery
+        );
+
+        SELECT a, ds FROM `tbl`
+    """,
+        default_dialect="bigquery",
+    )
+
+    model = load_sql_based_model(expressions)
+
+    model_json = model.json()
+    model_json_parsed = json.loads(model.json())
+    assert model_json_parsed["kind"]["dialect"] == "bigquery"
+    assert model_json_parsed["kind"]["unique_key"] == ["`a`"]
+    assert model_json_parsed["kind"]["columns"] == "*"
+    # Bigquery converts TIMESTAMP -> DATETIME
+    assert model_json_parsed["kind"]["time_data_type"] == "DATETIME"
+
+    deserialized_model = SqlModel.parse_raw(model_json)
+    assert deserialized_model.dict() == model.dict()
+
+
 def test_column_descriptions(sushi_context, assert_exp_eq):
     assert sushi_context.models[
         '"memory"."sushi"."customer_revenue_by_day"'