The Python standard library routine math.isfinite() is the key to filtering floating-point exceptions out of the results of float(). So if you're parsing text that you expect to contain only actual numbers, you need isfinite() to detect when you do something like float("nan"), float("inf"), and float("-inf"). Also, if you want to break naive parsers, putting nan, inf, or -inf in places where a real number is expected might have amusingly awful consequences. #UseYourPowersForGood