Magic attributes and methods

Magic attributes and methods

Builtin class attributes:

• __name__

• __doc__

• __module__

• __dict__

• __class__

class A:
    "Our great class" 
    x = 1 

Magic methods

__<method>__

• __init__

  • Initializing instance

• __new__

  • Instance constructor

• __repr__

  • Representation of an object

• __str__

  • String representation

• __del__

  • Deletion of object(called when del used and count to obj is 0)

Instance constructor

__new__()

• First method for creating new object

• It is static method

• Invoked before __init__(), need to return object

• Works with cls (class object, as instance is not created yet)

• Old style class don't have this method!

def __new__(cls):
    return super().__new__(cls)

Instance initializer

__init__()

• It is not a constructor (there is __new__() for that)

• __init__() takes an already created object and fills required attributes according to declared logic first arg is self

• Doesn't return anything (it returns None), just assigning attributes to an object.

def __init__(self, name): 
    self.name = name

Representations

__str__(self)

• String representation – for humans

• Used when str(obj) or print(obj)

__repr__(self)

• Representation for "machines"

• Ideal case: eval(repr(obj)) --> obj

• Used when calling repr(obj) or just obj in Python session

print(A())
A()

Hash

__hash__(self)

• Called by hashed collections:

  • set

  • frozenset

  • dict

• In simplest case as for instances it is based on id(self) so all objects are different

A().__hash__()

Bases

__bases__ (attribute)

• Tuple (possibly empty or a singleton) containing the base classes, in the order of their occurrence in the base class list

class A: pass

A.__bases__

Docstring

__doc__ (attribute)

• Class’s documentation string, or None if undefined.

print(A.__doc__)
help(A)

Module

__module__ (attribute)

• Module name in which the class is defined. This attribute is "__main__" in interactive mode.

A.__module__

Boolean value

__bool__(self) (__nonzero__ for Python 2)

• Used when bool(obj)

• By default - True

bool(A())

Comparison

__eq__(self, other)

• Return True if objects are equal

__gt__(self, other)

• Return True if self > other

Other methods:

• __ne__(self, other) # !=

• __ge__(self, other) # >=

• __lt__(self, other) # <

• __le__(self, other) # <=

It is really a pain to fill all of those!

Battery: functools.total_ordering

import functools

@functools.total_ordering
class Something:
    value = 1
    
    def __eq__(self, other):
        return self.value == other.value
    
    def __gt__(self, other):
        return self.value > other.value
    
b1 = Something()
b1.value = 5
b2 = Something()
print(b1 >= b2)
print(b1 <= b2)

Iterator protocol

__iter__(self)

• Returns iterator for given object (or self if it is iterator)

__next__(self) (next(self) for Python 2.x)

• Return next value for iterator-like object

Call

Object can be callable (function-like object).

__call__(self, *args)

This is very useful to call the method which used very often.

Slots and __dict__

All instance's attributes are hold in __dict__ attribute which is a dictionary with keys as names bound to the instance.

class C:
    a = 1


c = C()
c.b = 2
c.__dict__

__slots__ is a list with instance attributes that can be assigned

This is needed to suppress automatic creation of __dict__ - this can be useful when we have a lot of simple objects and their dictionaries are taking too much space. Also slots are limiting attributes that can be assigned to an object.

class D:
    __slots__ = ("a", )


d = D()
print(hasattr(d, "__dict__"))
d.a = 5  # OK
try:
    d.b = 5
except AttributeError:
    print("We can't access attribute b as it is not listed in __slots__")

__slots__ are meaninless when the class is inherited from the one which doesn't have __slots__.