If you're just processing each line separately, I would think the easiest way, especially for large files, would be

1. open the first file
2. open an output file
3. then read a line from the first file,
4. process the line, including converting to fields or editing the fields in place
5. write the line to the new file.

If no errors, the new file contains all the processed data.

Text::CSV shows an example of testing a field in place right at the top of the perldoc. It's also the best tool for CSV files.

--marmot