fix(ingestion): correct trino datatype handling (#5541)

Co-authored-by: Ravindra Lanka <rlanka@acryl.io>

metadata-ingestion/src/datahub/ingestion/source/sql/sql_types.py

@@ -228,9 +228,10 @@ def resolve_postgres_modified_type(type_string: str) -> Any:
 
 
 def resolve_trino_modified_type(type_string: str) -> Any:
-    # for cases like timestamp(3)
-    if re.match(r"[a-zA-Z]+\([0-9]+\)", type_string):
-        modified_type_base = re.match(r"([a-zA-Z]+)\([0-9]+\)", type_string).group(1)  # type: ignore
+    # for cases like timestamp(3), decimal(10,0), row(...)
+    match = re.match(r"([a-zA-Z]+)\(.+\)", type_string)
+    if match:
+        modified_type_base: str = match.group(1)
         return TRINO_SQL_TYPES_MAP[modified_type_base]
     else:
         return TRINO_SQL_TYPES_MAP[type_string]
@@ -337,4 +338,5 @@ def resolve_trino_modified_type(type_string: str) -> Any:
     "date": DateType,
     "time": TimeType,
     "timestamp": TimeType,
+    "row": RecordType,
 }

metadata-ingestion/tests/integration/dbt/test_dbt.py

@@ -644,8 +644,25 @@ def test_dbt_stateful_tests(pytestconfig, tmp_path, mock_time, mock_datahub_grap
 @pytest.mark.parametrize(
     "data_type, expected_data_type",
     [
-        ("timestamp(3)", "timestamp"),
+        ("boolean", "boolean"),
+        ("tinyint", "tinyint"),
+        ("smallint", "smallint"),
+        ("int", "int"),
+        ("integer", "integer"),
+        ("bigint", "bigint"),
+        ("real", "real"),
+        ("double", "double"),
+        ("decimal(10,0)", "decimal"),
         ("varchar(20)", "varchar"),
+        ("char", "char"),
+        ("varbinary", "varbinary"),
+        ("json", "json"),
+        ("date", "date"),
+        ("time", "time"),
+        ("time(12)", "time"),
+        ("timestamp", "timestamp"),
+        ("timestamp(3)", "timestamp"),
+        ("row(x bigint, y double)", "row"),
     ],
 )
 def test_resolve_trino_modified_type(data_type, expected_data_type):