I had a similar problem: differently formatted CSV-files to be unified and entered into one DB.

Best approach (for me) was to write a script for each type of CSV-file (using Text::CSV_XS) and dumping the data from these CSV-files into a temporary table (either through a loader or with DBI and then finally transfer the data from the temporary table to the main DB (again with DBI.

A dedicated loader is indeed faster than DBI, but if you take into account the time to write out the "unified" CSV-file as well, I'm not so sure that you win much time in the end.