Fair point, sorry. I misread the sample data and gave an alternative to the (also incorrect) perl version above. Thanks for the catch.