python: Avoid treating float as int