Replace ai_list_files truncated with last_modified column

Author: BohuTANG

databend_aiserver/udfs/stage.py

@@ -114,7 +114,7 @@ def _collect_stage_files(
         ("mode", "VARCHAR"),
         ("content_type", "VARCHAR"),
         ("etag", "VARCHAR"),
-        ("truncated", "BOOLEAN"),
+        ("last_modified", "TIMESTAMP"),
     ],
     name="ai_list_files",
 )
@@ -162,6 +162,7 @@ def ai_list_files(
 
             # Convert mode to string if it exists, otherwise None
             mode_str = str(metadata.mode) if metadata.mode is not None else None
+            last_modified = getattr(metadata, "last_modified", None)
 
             yield {
                 "stage_name": stage_location.stage_name,
@@ -172,10 +173,9 @@ def ai_list_files(
                 "mode": mode_str,
                 "content_type": metadata.content_type,
                 "etag": metadata.etag,
-                "truncated": False,
+                "last_modified": last_modified,
             }
 
     except Exception as e:
         logging.getLogger(__name__).error("Error listing files: %s", e)
         # Stop yielding
-

tests/integration/test_listing_integration.py

@@ -53,12 +53,10 @@ def test_list_stage_files_schema(running_server, memory_stage):
         assert "mode" in row
         assert "content_type" in row  # May be None
         assert "etag" in row  # May be None
-        assert row["truncated"] is False
+        assert "last_modified" in row
 
 
 def test_list_stage_files_truncation(running_server, memory_stage):
     rows = _get_listing(running_server, memory_stage, max_files=1)
     assert len(rows) == 1
-    # With lazy iteration, we can't easily know if result is truncated for all rows
-    # without buffering. Prioritizing performance over this flag.
-    assert rows[0]["truncated"] is False
+    assert "last_modified" in rows[0]