Stuck in encoding

So I started working on encoding detection... but I ran into some conceptual problems. Specifically, some languages can be written in multiple scripts, which in turn may be written in different encodings. For that matter, virtually any language can be written in any phonetic script. What to do about that?