tokenize_to_columns or parse_to_columns results in a single column we shouldn't add the 1#6607