Python Attribute Reference

The . (dot) operator in Python is the attribute reference operator, used to access, mutate, or delete attributes and methods bound to an object’s namespace. It serves as the primary syntactic mechanism for traversing object hierarchies and resolving names within the scope of a specific instance, class, or module.

class Namespace:
    attribute_name = "value"

object_reference = Namespace()
object_reference.attribute_name

Internal Mechanics

When the Python interpreter encounters the dot operator, it translates the syntax into underlying C API calls or specific magic (dunder) method invocations on the object. The behavior changes depending on the context of the operation (access, assignment, or deletion).

1. Attribute Access

Evaluating obj.attr invokes the object’s __getattribute__ method. If __getattribute__ fails to find the attribute, it raises an AttributeError. Python will then catch this exception and invoke the __getattr__ method as a fallback, provided the class implements it.

class MyClass:
    attr = "value"
    
    def __getattr__(self, name):
        return "fallback_value"

obj = MyClass()

# Standard dot operator syntax accessing an existing attribute
value = obj.attr  # Returns "value" via __getattribute__

# Standard dot operator syntax triggering the fallback
missing = obj.missing_attr  # Returns "fallback_value" via __getattr__

# Equivalent built-in function
missing = getattr(obj, 'missing_attr')

# Note: Explicitly calling obj.__getattribute__('missing_attr') is NOT strictly 

# equivalent to obj.missing_attr because it bypasses the __getattr__ fallback mechanism.

2. Attribute Assignment

Using the dot operator on the left side of an assignment (obj.attr = value) binds a value to a name within the object’s namespace. This translates to the __setattr__ method.

class MyClass:
    pass

obj = MyClass()

# Standard dot operator syntax
obj.attr = "new_value"

# Equivalent built-in function
setattr(obj, 'attr', "new_value")

# Equivalent underlying dunder method invocation
obj.__setattr__('attr', "new_value")

3. Attribute Deletion

Using the del statement with the dot operator removes the attribute from the object’s namespace, invoking the __delattr__ method.

class MyClass:
    pass

obj = MyClass()


# Standard dot operator syntax
obj.attr = "value"
del obj.attr


# Equivalent built-in function
obj.attr = "value"  # Recreate the attribute before deleting again
delattr(obj, 'attr')


# Equivalent underlying dunder method invocation
obj.attr = "value"  # Recreate the attribute before deleting again
obj.__delattr__('attr')

Attribute Resolution Order

When the dot operator is used for access, Python follows a strict lookup chain to resolve the attribute name. The dot operator searches in the following order:

Data Descriptors: Objects in the class hierarchy that implement both __get__ and __set__ (e.g., @property), resolved by traversing the Method Resolution Order (MRO).
Instance Dictionary: The instance’s local namespace, stored in obj.__dict__.
Class and Base Class Dictionaries: Attributes found in the class or its base classes, traversed together according to the MRO. If the retrieved object is a Non-Data Descriptor (an object implementing only __get__, such as a standard function), the dot operator invokes its __get__ method to return the result. Otherwise, it returns the raw attribute value.
Fallback: The __getattr__ method, if defined on the class.

Method Binding

When the dot operator accesses a function defined within a class hierarchy, the function itself acts as a descriptor. The dot operator triggers the descriptor protocol (__get__), which automatically binds the instance to the first parameter (self) of the function, returning a bound method object rather than a raw function.

class BaseClass:
    def method(self):
        pass

class DerivedClass(BaseClass):
    pass

obj = DerivedClass()


# The dot operator binds 'obj' to 'method'
bound_method = obj.method


# Equivalent manual binding using the descriptor protocol.

# getattr() is used on the class to safely traverse the MRO, 

# preventing a KeyError if the method is inherited from a base class.
unbound_func = getattr(type(obj), 'method')
bound_method = unbound_func.__get__(obj, type(obj))