fix(csv-upload): log detailed errors during chunk concatenation for debugging

superset/commands/database/uploaders/csv_reader.py

@@ -414,7 +414,17 @@ def _read_csv(  # noqa: C901
                         break
 
                 if chunks:
-                    result = pd.concat(chunks, ignore_index=False)
+                    try:
+                        result = pd.concat(chunks, ignore_index=False)
+                    except Exception as ex:
+                        logger.warning(
+                            "Error concatenating CSV chunks: %s. "
+                            "This may be due to inconsistent date parsing "
+                            "across chunks.",
+                            str(ex),
+                        )
+                        raise
+
                     # When using chunking, we need to reset and rebuild the index
                     if kwargs.get("index_col") is not None:
                         # The index was already set by pandas during read_csv

tests/unit_tests/commands/databases/csv_reader_test.py

@@ -1352,3 +1352,68 @@ def track_read(size):
 
     # Test that the method handles the sample sizes properly
     assert all(size > 0 for size in read_sizes), "All sample sizes should be positive"
+
+
+def test_csv_reader_chunk_concatenation_error_logging():
+    """Test that pd.concat errors during chunking are logged and re-raised."""
+    from unittest.mock import patch
+
+    # Create a large CSV that will trigger chunking (>100k rows)
+    large_data = [["col1", "col2"]]
+    for i in range(100001):
+        large_data.append([f"val{i}", str(i)])
+
+    csv_reader = CSVReader(options=CSVReaderOptions())
+
+    # Mock pd.concat to raise an exception
+    with patch(
+        "superset.commands.database.uploaders.csv_reader.pd.concat"
+    ) as mock_concat:
+        mock_concat.side_effect = ValueError(
+            "Cannot concatenate chunks with different dtypes"
+        )
+
+        with pytest.raises(DatabaseUploadFailed) as exc_info:
+            csv_reader.file_to_dataframe(create_csv_file(large_data))
+
+        # Verify the exception is still raised (wrapped as DatabaseUploadFailed)
+        assert "Cannot concatenate chunks with different dtypes" in str(exc_info.value)
+
+        # Verify concat was called (meaning chunking happened)
+        assert mock_concat.called
+
+
+def test_csv_reader_chunk_concatenation_error_warning(caplog):
+    """Test that pd.concat errors during chunking log a warning message."""
+    from unittest.mock import patch
+
+    # Create a large CSV that will trigger chunking (>100k rows)
+    large_data = [["col1", "col2"]]
+    for i in range(100001):
+        large_data.append([f"val{i}", str(i)])
+
+    csv_reader = CSVReader(options=CSVReaderOptions())
+
+    # Mock pd.concat to raise an exception
+    with patch(
+        "superset.commands.database.uploaders.csv_reader.pd.concat"
+    ) as mock_concat:
+        mock_concat.side_effect = ValueError(
+            "Cannot concatenate chunks with different dtypes"
+        )
+
+        import logging
+
+        with caplog.at_level(logging.WARNING):
+            with pytest.raises(DatabaseUploadFailed):
+                csv_reader.file_to_dataframe(create_csv_file(large_data))
+
+        # Verify warning was logged
+        assert any(
+            "Error concatenating CSV chunks" in record.message
+            for record in caplog.records
+        )
+        assert any(
+            "inconsistent date parsing across chunks" in record.message
+            for record in caplog.records
+        )