If a SQL approach would help, you can leave the flatfile as is and handle it with DBD::CSV or DBD::AnyData, without having to change your files at all.