seems like textencoder is pretty fast even if you just use it to grab the utf8 length