Things get even weirder when you throw non-Latin characters in the mix. May such parsers predate widespread use of UTF-8.