If I undertand you correctly you want to have multilpe input file names with meta characters that you'll expand using glob dump into the same output file? With your algorithm is it possible to make multiple runs for each glob expansion and then dump the output to the same file or to all these file HAVE to be processed together?