Parser improvements - phase 1

• Provide meaningful message for users on token recognition error
• Fix document comment start position for the first line
• Fix parser NPE when CommonToken.getText() is null