well, i was asked theoretically how to do this as records are parsed. however, i generally deal with 5-10k lines of csv files and it would sometimes be good to count dupes of certain fields and not have to worry about sorting the count of 30+ columns when looking for these duplicates at the end of processing them would be cool. it only takes a few seconds. but, the theoretical question and noticing the processing time at the end was enough to make me want to look into another way to deal with this.