I originally didn't get this, until I realized that you meant a non-case-preserving case-insensitive filesystem, followed by a copy back to real-people filesystems. In which case, yes, that would be the easiest way to solve my problem (and is very clever).

Sadly, since the data are available on the web in their current, mis-cased form, dealing with it is the only option I have. Also, on another level, this wouldn't solve the problem of inconsistent filename casing IN the files - even after the files are all one case or the other, I'd still need to have a programming solution to match all the internal references.