Added a really fragile UTF-8 parser. Really needs a lot of work: