I assumed that leap seconds could be determined algorithmically, it appears I assumed badly. This is a bit of a can of worms...