Trying to encode the input to UTF8 before matching would cause errors
when the string includes incomplete multibyte characters or any
characters ruby can't convert to UTF8.