This may be a silly question, but it's something that's bothered me ever since Scribd launched: How do you handle all the file format conversions?

It seems to be that you couldn't be wrangling all those formats yourself, as many are undocumented, and/or hideously complex, and you've got to do a lot more work than simply reading the files, and not much time to do it. So, are there some off-the-shelf components to do this work (with reasonable license terms), or is the problem easier than it seems?

I'm most interested in the MSFT formats, which seem the trickiest.