The resulting database would be enormous, though. It might be more efficient to just use `grep` across outputs.