You are of course correct. This also ties into my sibling comment - the cleverness in clever low-level parsing code is based on assumptions that may be wrong for some part of the huge input space.