fix(ingest/json-schema): convert non-string enums to strings (#8479)

Author: benjamin-awd

metadata-ingestion/src/datahub/ingestion/extractor/json_schema_util.py

@@ -254,11 +254,13 @@ def _field_from_primitive(
                 isPartOfKey=field_path.is_key_schema,
             )
         elif datahub_field_type in [EnumTypeClass]:
+            # Convert enums to string representation
+            schema_enums = list(map(json.dumps, schema["enum"]))
             yield SchemaField(
                 fieldPath=field_path.expand_type("enum", schema).as_string(),
                 type=type_override or SchemaFieldDataTypeClass(type=EnumTypeClass()),
                 nativeDataType="Enum",
-                description=f"one of {','.join(schema['enum'])}",
+                description=f"One of: {', '.join(schema_enums)}",
                 nullable=nullable,
                 jsonProps=JsonSchemaTranslator._get_jsonprops_for_any_schema(
                     schema, required=required

metadata-ingestion/tests/unit/schema/test_json_schema_util.py

@@ -712,6 +712,19 @@ def test_required_field():
     assert json.loads(fields[0].jsonProps or "{}")["required"] is False
 
 
+def test_non_str_enums():
+    schema = {
+        "$id": "test",
+        "$schema": "http://json-schema.org/draft-06/schema#",
+        "properties": {"bar": {"description": "Mixed enum", "enum": ["baz", 1, None]}},
+    }
+
+    fields = list(JsonSchemaTranslator.get_fields_from_schema(schema))
+    expected_field_paths: List[str] = ["[version=2.0].[type=object].[type=enum].bar"]
+    assert_field_paths_match(fields, expected_field_paths)
+    assert fields[0].description == 'One of: "baz", 1, null'
+
+
 def test_anyof_with_properties():
     # We expect the event / timestamp fields to be included in both branches of the anyOf.