type normalisation for SEA

Signed-off-by: varun-edachali-dbx <varun.edachali@databricks.com>

Author: varun-edachali-dbx

src/databricks/sql/backend/sea/backend.py

@@ -19,6 +19,7 @@
     WaitTimeout,
     MetadataCommands,
 )
+from databricks.sql.backend.sea.utils.normalize import normalize_sea_type_to_thrift
 from databricks.sql.thrift_api.TCLIService import ttypes
 
 if TYPE_CHECKING:
@@ -322,6 +323,11 @@ def _extract_description_from_manifest(
             # Format: (name, type_code, display_size, internal_size, precision, scale, null_ok)
             name = col_data.get("name", "")
             type_name = col_data.get("type_name", "")
+
+            # Normalize SEA type to Thrift conventions before any processing
+            type_name = normalize_sea_type_to_thrift(type_name, col_data)
+
+            # Now strip _TYPE suffix and convert to lowercase
             type_name = (
                 type_name[:-5] if type_name.endswith("_TYPE") else type_name
             ).lower()

src/databricks/sql/backend/sea/utils/normalize.py

@@ -0,0 +1,48 @@
+"""
+Type normalization utilities for SEA backend.
+
+This module provides functionality to normalize SEA type names to match
+Thrift type naming conventions.
+"""
+
+from typing import Dict, Any
+
+# Mapping of SEA types that need to be translated to Thrift conventions
+SEA_TO_THRIFT_TYPE_MAP = {
+    "BYTE": "TINYINT",
+    "SHORT": "SMALLINT",
+    "LONG": "BIGINT",
+    "INTERVAL": "INTERVAL",  # Default mapping, will be overridden if type_interval_type is present
+}
+
+
+def normalize_sea_type_to_thrift(type_name: str, col_data: Dict[str, Any]) -> str:
+    """
+    Normalize SEA type names to match Thrift type naming conventions.
+
+    Args:
+        type_name: The type name from SEA (e.g., "BYTE", "LONG", "INTERVAL")
+        col_data: The full column data dictionary from manifest (for accessing type_interval_type)
+
+    Returns:
+        Normalized type name matching Thrift conventions
+    """
+    # Early return if type doesn't need mapping
+    if type_name not in SEA_TO_THRIFT_TYPE_MAP:
+        return type_name
+
+    normalized_type = SEA_TO_THRIFT_TYPE_MAP[type_name]
+
+    # Special handling for interval types
+    if type_name == "INTERVAL":
+        type_interval_type = col_data.get("type_interval_type")
+        if type_interval_type:
+            if any(t in type_interval_type.upper() for t in ["YEAR", "MONTH"]):
+                return "INTERVAL_YEAR_MONTH"
+            elif any(
+                t in type_interval_type.upper()
+                for t in ["DAY", "HOUR", "MINUTE", "SECOND"]
+            ):
+                return "INTERVAL_DAY_TIME"
+
+    return normalized_type

tests/unit/test_sea_backend.py

@@ -550,6 +550,66 @@ def test_extract_description_from_manifest(self, sea_client):
         assert description[1][1] == "int"  # type_code
         assert description[1][6] is None  # null_ok
 
+    def test_extract_description_from_manifest_with_type_normalization(
+        self, sea_client
+    ):
+        """Test _extract_description_from_manifest with SEA to Thrift type normalization."""
+        manifest_obj = MagicMock()
+        manifest_obj.schema = {
+            "columns": [
+                {
+                    "name": "byte_col",
+                    "type_name": "BYTE",
+                },
+                {
+                    "name": "short_col",
+                    "type_name": "SHORT",
+                },
+                {
+                    "name": "long_col",
+                    "type_name": "LONG",
+                },
+                {
+                    "name": "interval_ym_col",
+                    "type_name": "INTERVAL",
+                    "type_interval_type": "YEAR TO MONTH",
+                },
+                {
+                    "name": "interval_dt_col",
+                    "type_name": "INTERVAL",
+                    "type_interval_type": "DAY TO SECOND",
+                },
+                {
+                    "name": "interval_default_col",
+                    "type_name": "INTERVAL",
+                    # No type_interval_type field
+                },
+            ]
+        }
+
+        description = sea_client._extract_description_from_manifest(manifest_obj)
+        assert description is not None
+        assert len(description) == 6
+
+        # Check normalized types
+        assert description[0][0] == "byte_col"
+        assert description[0][1] == "tinyint"  # BYTE -> tinyint
+
+        assert description[1][0] == "short_col"
+        assert description[1][1] == "smallint"  # SHORT -> smallint
+
+        assert description[2][0] == "long_col"
+        assert description[2][1] == "bigint"  # LONG -> bigint
+
+        assert description[3][0] == "interval_ym_col"
+        assert description[3][1] == "interval_year_month"  # INTERVAL with YEAR/MONTH
+
+        assert description[4][0] == "interval_dt_col"
+        assert description[4][1] == "interval_day_time"  # INTERVAL with DAY/TIME
+
+        assert description[5][0] == "interval_default_col"
+        assert description[5][1] == "interval"  # INTERVAL without subtype
+
     def test_filter_session_configuration(self):
         """Test that _filter_session_configuration converts all values to strings."""
         session_config = {