I like revdiablo's approach (++).

This could be taken one step furthur - Make an object class out of it.

What you are operating on looks like a File, so, the myFile package could hage a get_data method. You would create an instance of "myFile", provide it with a file name, filter conditions, pre and post processing anon subrefs, then simply call get_data on each instance of the myFile class.