在python中模拟一个int

Question

I want to define an object which has the exact same behavior as an int, with a few tiny differences. Consider the class

class MyInt:
  def __init__(self, num):
    self.num = num

I need to enable the following behavior: when MyInt is assigned to another variable, the operation results to assigning MyInt.num instead of the whole class, ie

i = MyInt(10)
a = i
print(type(a)) # <class 'int'>
print(a)       # 10

In other words, a=i results in a= i.num . All other functionality of MyInt should be exactly the same as int and applied on MyInt.num , eg i += 1 should be equivalent to i.num +=1

I guess that this can be accomplished by making MyInt inherit from int and making use of __getattr__ and __setattr__ . However, I am not sure what method is used to fetch the value of a variable in python so that I can achieve the desired functionality.

Disclaimer: I need this so that I can modify a variable globally inside a function. I am aware this is not recommended, but it is my only option. Also I cannot use global .

Answer 1

You can't do that. There are no hooks in Python that let you adjust assignments.

There are any number of ways of binding an object, and to Python, there is no difference, at all , between i = MyInt(10) and a = i . Both are assignments where the result of the expression on the right-hand side are executed first , then a reference to the result is stored under a specific name. Since i produces the a reference to the same object already produced by MyInt(10) , you end up with 2 references, with different names, to a single instance.

But [i] also stores a reference, now in a list object. So does for target_name in (i, i): (3 references, first in a tuple with two references, then target_name , which is bound twice like this when the loop executes and nothing exits the loop early). I count at least 8 different ways of binding objects in Python.

You can make MyInt work like an integer in most cases; there are a number of special methods to emulate numeric types that let you accomplish this. There is even a numbers.Integral abstract base class you can use as a starting point. These provide you with a number of those special methods pre-implemented, but you do have to still implement a number of abstract methods, however:

>>> from numbers import Integral
>>> sorted(Integral.__abstractmethods__)
['__abs__', '__add__', '__and__', '__ceil__', '__eq__', '__floor__', '__floordiv__', '__int__', '__invert__', '__le__', '__lshift__', '__lt__', '__mod__', '__mul__', '__neg__', '__or__', '__pos__', '__pow__', '__radd__', '__rand__', '__rfloordiv__', '__rlshift__', '__rmod__', '__rmul__', '__ror__', '__round__', '__rpow__', '__rrshift__', '__rshift__', '__rtruediv__', '__rxor__', '__truediv__', '__trunc__', '__xor__']

Since any subclass that doesn't can't be instantiated, these give you instant feedback on whether or not you missed any of those methods.

Methods already provided are:

>>> from numbers import Number
>>> sorted(n for n in dir(Integral) if n not in Integral.__abstractmethods__.union(dir(Number)))
['__bool__', '__complex__', '__divmod__', '__float__', '__index__', '__rdivmod__', '__rsub__', '__sub__', 'conjugate', 'denominator', 'imag', 'numerator', 'real']

and you really want to provide a suitable __hash__ method .