The token rule was matching even when the quoted_string rule should have matched, which was preventing the " stripping to work. In fact quoted_string was never being used. Also improve formatting of long lines.
It will be used to detect received content types.
