How to Use __call__() Method Instead of __new__() of a Metaclass in Python?

In Python, metaclasses provide a way to customize class creation. The __new__ and __init__ methods are commonly used in metaclasses, but there’s another powerful tool at your disposal: the __call__ method. While __new__ is responsible for creating a new instance of a class, __call__ allows you to control what happens when an instance is called as a function. This article explores the __call__ method of a metaclass, its syntax, and advantages, and provides three code examples demonstrating how to use it instead of __new__.

What is __call__() Method of a Python Metaclass?

The __call__ method of a metaclass is invoked when an instance of that metaclass is called as a function. It provides a way to customize the behavior of instances when they are used in a callable context. The syntax for the __call__ method in a metaclass is as follows:

class YourMeta(type):

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

# Custom logic here

instance = super().__call__(*args, **kwargs)

# Additional customization if needed

return instance

Advantages of Using __call__ Method

• Dynamic Initialization: The __call__ method allows for dynamic initialization of instances based on the arguments passed during the call. This provides more flexibility compared to the static nature of __new__.
• Code Readability: Using __call__ can lead to more readable code, as it centralizes the logic related to instance creation and allows for a clearer separation of concerns.
• Easy Modification: With __call__, you can easily modify the behavior of instances without changing the class definition, making your code more maintainable and adaptable.
• Consistent Interface: By leveraging __call__, you maintain a consistent interface for creating instances across different metaclasses, enhancing code consistency.

Using __call__() Method Instead of __new__() of a Metaclass in Python

Below are some of the examples by which we can understand how to use the __call__() method of a Metaclass instead of __new__() in Python:

Example 1: Modifying Class Attributes

The code employs a custom metaclass, `CustomMeta`, with a `__call__` method, altering class instantiation. Instead of using `__new__`, it dynamically adds the `new_attribute` to instances of the class `MyClass`. This allows for flexible attribute assignment during instantiation, exemplified by the printed output “This is a new attribute.”

Python3

class CustomMeta(type):
    def __call__(cls, *args, **kwargs):
        instance = super().__call__(*args, **kwargs)
        instance.new_attribute = "This is a new attribute."
        return instance

class MyClass(metaclass=CustomMeta):
    pass

obj = MyClass()
print(obj.new_attribute) 

Output

This is a new attribute.

Example 2: Singleton Pattern

In this example, below code utilizes the __call__ method of the metaclass SingletonMeta instead of the __new__ method to implement the Singleton pattern. The __call__ method ensures that only one instance of the class SingletonClass is created. This approach offers a cleaner and more concise way to control instance creation and manage the Singleton pattern compared to using the __new__ method in the metaclass.

Python3

class SingletonMeta(type):
    _instances = {}

    def __call__(cls, *args, **kwargs):
        if cls not in cls._instances:
            instance = super().__call__(*args, **kwargs)
            cls._instances[cls] = instance
        return cls._instances[cls]

class SingletonClass(metaclass=SingletonMeta):
    pass

obj1 = SingletonClass()
obj2 = SingletonClass()

print(obj1 is obj2) 

Output

True

Example 3: Dynamic Class Modification

In this example, below code uses the `__call__` method in the metaclass `DynamicMeta` to dynamically add a method to the class `DynamicClass` during instantiation. This allows for on-the-fly customization of class behavior. The demonstration creates an instance of `DynamicClass` with the added method, producing the output “Dynamic method added!” when the method is called.

Python3

class DynamicMeta(type):
    def __call__(cls, *args, **kwargs):
        instance = super().__call__(*args, **kwargs)
        if "new_method" in kwargs:
            setattr(instance, kwargs["new_method"], lambda: print("Dynamic method added!"))
        return instance

class DynamicClass(metaclass=DynamicMeta):
    def __init__(self, *args, **kwargs):
        pass

obj = DynamicClass(new_method="dynamic_method")
obj.dynamic_method()

Output

Dynamic method added!