As someone who spends a lot of time on file processing I appreciate the link. Althought I have to selfishly admit that had I stumbled on to this myself I would have avoided it just because of SourceForge.

How well does it handle the oddball cases? Not that you could call CSV a rigorously-defined spec, but we have a vendor that sends us a file where they sometimes don't escape double-quotes correctly.

That is where I've found it's strengths to be. I've used it to process affiliate feed data. The quality is pretty poor and it has been pretty robust.