I cannot believe I did not notice that. I will rerun all of my testing with a valid UTF-8 byte sequence :)