ensure datatype present on each col

semantic_model_generator/sqlgen/generate_sql.py

@@ -2,7 +2,7 @@
 
 import sqlglot
 from sqlglot.dialects.snowflake import Snowflake
-
+from semantic_model_generator.validate.fields import validate_contains_datatype_for_each_col
 from semantic_model_generator.protos.semantic_model_pb2 import (
     Dimension,
     FullyQualifiedTable,
@@ -120,7 +120,10 @@ def generate_select_with_all_cols(table: Table, limit: int) -> str:
     Returns:
         str: A SQL statement formatted for Snowflake.
     """
+
 
     select = _create_select_statement(table, limit)
+
+    validate_contains_datatype_for_each_col(table)
 
     return _convert_to_snowflake_sql(select)

semantic_model_generator/tests/generate_sql_test.py

@@ -53,6 +53,15 @@
     sample_values=["1000.50", "2000.75", "1500.00"],
 )
 
+measure_example_no_data_type= Measure(
+    name="Total_Sales",
+    synonyms=["Sales", "Revenue"],
+    description="Total sales amount",
+    expr="sales_amount",
+    sample_values=["1000.50", "2000.75", "1500.00"],
+)
+
+
 time_dimension_example = TimeDimension(
     name="Date",
     synonyms=["Time"],
@@ -122,6 +131,15 @@
 )
 
 
+_TEST_TABLE_MISSING_DATATYPE = Table(
+    name="Transactions",
+    synonyms=["Transaction Records"],
+    description="Table containing transaction records",
+    base_table=fully_qualified_table_example,
+    measures=[measure_example_no_data_type],
+)
+
+
 def test_valid_table_sql_with_expr():
     want = "SELECT region_code AS Region, sales_amount - sales_total AS Total_Sales, transaction_date AS Date FROM SalesDB.public.transactions LIMIT 100"
     generated_sql = generate_select_with_all_cols(_TEST_VALID_TABLE, 100)
@@ -162,3 +180,14 @@ def test_table_invalid_col_expr():
         str(excinfo.value)
         == "Aggregations aren't allowed in columns yet. Please remove from SUM(sales_amount) as Total_Sales."
     )
+
+def test_table_missing_datatype():
+    with pytest.raises(ValueError) as excinfo:
+        _ = generate_select_with_all_cols(_TEST_TABLE_MISSING_DATATYPE, 100)
+
+    assert (
+        str(excinfo.value)
+        == "Your Semantic Model contains a col Total_Sales that does not have the `data_type` field. Please add."
+    )
+
+

semantic_model_generator/validate/fields.py

@@ -0,0 +1,13 @@
+from semantic_model_generator.protos.semantic_model_pb2 import Table
+
+def validate_contains_datatype_for_each_col(table: Table) -> None:
+    # Ensure every col for every table has 'data_type' present.
+    for dim_col in table.dimensions:
+        if dim_col.data_type is None or len(dim_col.data_type) < 2: # account for spaces
+            raise ValueError(f"Your Semantic Model contains a col {dim_col.name} that does not have the `data_type` field. Please add.")
+    for measure_col in table.measures:
+        if measure_col.data_type is None or len(measure_col.data_type) < 2: # account for spaces
+            raise ValueError(f"Your Semantic Model contains a col {measure_col.name} that does not have the `data_type` field. Please add.")
+    for time_dim_col in table.time_dimensions:
+        if time_dim_col.data_type is None or len(time_dim_col.data_type) < 2: # account for spaces
+            raise ValueError(f"Your Semantic Model contains a col {time_dim_col.name} that does not have the `data_type` field. Please add.")