__new__ in Python

Python is an object-oriented programming language, where everything is an object. Python provides several special methods, known as magic methods or dunder methods (dunder means double underscores), which enable powerful and flexible object-oriented capabilities. These methods start and end with double underscores (e.g., __init__, __add__, __len__, __repr__, etc.). One such powerful magic method is __new__, which plays an important role in instance creation.

What is __new__ in Python?

Whenever a class is instantiated, two methods are called:

• __new__ : Responsible for creating a new instance of the class.
• __init__ : Initializes the instance.

Unlike __init__, which is used for setting up an object after it has been created, __new__ is responsible for creating and returning the new instance itself.

Example:

class A:
    def __new__(cls):
        print("Creating instance")
        return super(A, cls).__new__(cls)
    
    def __init__(self):
        print("Initializing instance")

A()

Creating instance
Initializing instance

Explanation: __new__ method is called first to create an instance of class A and then returns the newly created instance. After that, the __init__ method initializes the instance.

Syntax

class ClassName:

def __new__(cls, *args, **kwargs):

# Custom instance creation logic

instance = super(ClassName, cls).__new__(cls, *args, **kwargs)

return instance

Parameters:

• cls : The class itself.
• *args : Positional arguments passed to __new__ .
• **kwargs : Keyword arguments passed to __new__ .

Return Value:

• Must return an instance of the class (cls) or another class.
• If __new__ returns an object of another type, __init__ will not be called.

Alternative method

Instances can be created in __new__ using either:

instance = super(ClassName, cls).__new__(cls, *args, **kwargs)

or

instance = object.__new__(cls, *args, **kwargs)

If both __new__ and __init__ exist in a class, __new__ executes first. It determines whether __init__ should be executed by either returning an instance of the class or bypassing __init__ by returning an instance of another class.

Examples of __new__ method

Example 1: What happens if __new__ does not return an instance?

class A:
    def __new__(cls):
        print("Creating instance")

    def __init__(self):
        print("Initializing instance")

print(A())

Creating instance
None

Explanation: __new__ method does not return an instance of the class. Since __new__ must return an instance, but it lacks a return statement, it implicitly returns None.

Example 2: Returning a different type from __new__

class A:
    def __new__(cls):
        print("Creating instance")
        return "Hello, World!"

print(A())

Creating instance
Hello, World!

Explanation : __new__ method should return a new object of class A, but here it returns the string "Hello, World!". Because of this, Python does not create an instance of A, so the __init__ method is never called. Instead, print(A()) simply prints "Hello, World!".

Example 3: Returning an instances of another class

class GeeksforGeeks:
    def __str__(self):
        return "GeeksforGeeks Instance"

class Geek:
    def __new__(cls):
        return GeeksforGeeks()
    
    def __init__(self):
        print("Inside init")

print(Geek())

GeeksforGeeks Instance

Explanation :__new__ method of Geek returns an instance of GeeksforGeeks instead of Geek, so the __init__ method is never called. When print(Geek()) is executed, it prints "GeeksforGeeks Instance" from the __str__ method of GeeksforGeeks.

Example 4: Returning values from __new__ and __init__

class A(object):
    def __new__(cls):
        print("Creating instance")
        return "GeeksforGeeks"

class B(object):
    def __init__(self):
        print("Initializing instance")
        return "GeeksforGeeks"

print(A())
print(B())

Output

Hangup (SIGHUP)
Traceback (most recent call last):
  File "/home/guest/sandbox/Solution.py", line 12, in <module>
    print(B())
          ~^^
TypeError: __init__() should return None, not 'str'

Explanation: In class A, the __new__ method prints "Creating instance" and returns the string "GeeksforGeeks", so the __init__ method is never called. This results in "GeeksforGeeks" being printed. In class B, the __init__ method incorrectly tries to return a string, but since __init__ must return None, it raises a TypeError.

When to use __new__

__new__ is rarely overridden, but it is useful in specific scenarios, such as:

• Implementing Singleton Pattern: Ensures only one instance of a class exists.
• Returning Cached Objects: Helps in memory optimization by reusing existing objects instead of creating new ones.
• Immutable Object Creation: Used in classes like str and tuple since they are immutable.
• Subclassing Immutable Types: When extending built-in immutable types like int, float or str.