how would you do this? you'd put a for each file looping task in there and stop the data flow (read) operation when you hit the right number.
Why would you do this? I have no idea why you wouldn't simply query the original data source in the DATABASE and get a result so much faster.