We already had one place that checked to make sure that we weren't trying to

format un-decoded bytes into a unicode string, but there were a couple other

places where we didn't. I've just copied and pasted the check into all the

relevant places.

Ran unit tests.