But I'm unsure of how much error checking uncompress done when uncompressing.

Frinstance, when I ran uncompress on the corrupted file, it gave me the first 70 characters or so, quite correctly, until it returned a 'file corrupted' messages. Which lead me to think uncompress knows when it's bad, so I could make it good, and let it continue on to the next error.

I wasn't going use this woolly thinking as the means to turn my script into reality, however! :)