automatically detecting dtype and column separations could potentially be done by using some pandas libraries. If that's possible, it should be done