I gotta complain about the idea that File::Find is the canonical way to traverse a directory. File::Find calls a subroutine of your choice on each file it traverses, but if you want to save the results from processing those files, you have to save them in a global variable, because there's no way to capture the return values from your subroutine. Not to mention there are a lot of things that are hard to do with it. Try to use File::Find just to construct a hash with the directory structure. You can do it, but you'll have to parse the full path to the file and traverse your hash structure to insert each element, using about 10 times as much code as just saying push(@{$hash{$dir}, $file). File::Find is hard to use and leads to bad code. I never use it anymore except for the simplest of tasks.