How do I overload `float()` for a custom class?

Question

How do I implement a method that overloads float(), i.e. it is called when an instance of a class I've written is passed to float()?

I am starting to write my own Fraction class:

class Fraction:
    def __init__(self, num = 0, denom = 1):
        self.num = float(num)
        self.denom = float(denom)

Now I want to be able to do this:

float(Fraction())

I tried using the below method, but that didn't work.

def float(self):
    return self.num / self.denom

@DavidHeffernan Could you do me a favour? My question here: http://stackoverflow.com/questions/9938329/why-use-a-flat-list-in-heapsort had gotten closed for being to vague. I edited it and now it seems like a perfectly legit question for SO. — Yatharth Agarwal, Aug 19 '12 at 15:14
Thing is, my account just got automatically banned because of that and I can't ask any more questions, so could you please upvote it if you think my question is good enough? — Yatharth Agarwal, Aug 19 '12 at 15:15
@DavidHeffernan Thanks a lot. If it wouldn't be too much, could you also flag the question for moderator's attention so it can be reopened? ( Apparently, that privilege has been stripped too :< ) — Yatharth Agarwal, Aug 19 '12 at 16:07
@DavidHeffernan Do you mean that you had enough rep to cast a reopen vote? — Yatharth Agarwal, Aug 19 '12 at 16:30
@DavidHeffernan Cool... BTW, how many votes are needed to reopen the question? And how many have been cast? I need to know how many more moderators I have to hunt before I regain my former privileges. — Yatharth Agarwal, Aug 19 '12 at 16:39
Oh, and thanks a lot BTW. When do these election things happen? — Yatharth Agarwal, Aug 19 '12 at 16:40

kindall · Accepted Answer · 2012-04-05T19:36:01.567

27

Define the __float__() special method on your class.

class MyClass(object):
    def __float__(self):
         return 0.0

float(MyClass())   # 0.0

Note that this method must return a float! The calculation self.num / self.denom, returns an int by default in versions of Python prior to 3.0 assuming both operands are integers. In this case you'd just make sure one of the operands is a float: float(self.num) / self.denom for example.

edited Apr 05 '12 at 19:36

answered Apr 05 '12 at 18:44

kindall

178,883
35
278
309

I _am_ using ints throughout my code for `self.num` and `self.denom`. 10x for the answer. – Yatharth Agarwal Apr 05 '12 at 19:31
What _would_ happen if you didn't return a float? – Yatharth Agarwal Apr 05 '12 at 19:32
1

You get a `TypeError` exception. – kindall Apr 05 '12 at 19:35
1

@YatharthROCK: Basically, you will create an unstable black hole inside your python interpreter and the universe will be destroyed. I think kindall is just pointing out the fact that `/` denotes integer division by default in <= python 2.x and floating point division in python 3.x. You should be aware of this. – Joel Cornett Apr 05 '12 at 19:38
1

@kindall: So when you call one of the special methods, it doesn't just execute the overloaded method, it still passes through the special method? – Joel Cornett Apr 05 '12 at 19:46
if you *are* using Python <3.0 , you can always use a future import to get float-returning division, i.e.: `from __future__ import division`. Then `self.num / self.denom` will work fine. – CrepeGoat Jan 04 '18 at 17:59

How do I overload `float()` for a custom class?

1 Answers1

Linked