Load only first file to a hash. Read second one - line after line and process one line only (search, store in output file etc.).