Automatically Assessing Code Understandability: How Far Are We?