I did something similar. I had some code with special characters in it (accented chars, mostly) which would lead to problems with some of our tools. I got bored with it and used a perl script to remove every non ascii char from the code. No problem there.
A colleague had the same issue on a different branch, so I told him I would do the same operation again. Except this time I called the script on the root directory of my drive rather than the code folder. Turns out there are a lot of non ascii characters in zipped files, executable files, and other non plain text files :P (It was inconvenient, but all important files were local copies of files backed up somewhere though, so I didn't lose anything)