Classes combine data with behavior.
Classes are used to create copies or
instances of bundled data and behavior, commonly known as
Objects can represent real world entities (such as cars or cats) - or more abstract concepts (such as integers, vehicles, or mammals).
Classes are integral to an object oriented programming (OOP) approach, which asks the programmer to think about modeling a problem as one or more objects that interact with one another, keep state, and act upon data.
Classes are the definitions of new object types, and from which new
instances of objects are created.
They often bundle data (known as
data members, or
variables) with code or functions (known as
methods) that operate on that data.
In this sense, a class is a blueprint or a set of instructions from which many objects of a similar type can be built and used.
A complex program can have many classes, each building many different flavors of objects.
The process of building and customizing an object from a class is known as
instantiation (or creating an instance of the class).
A class definition in Python is straightforward:
class MyClass: # Class body goes here
Class fields (otherwise known as properties, attributes, data members, or variables) can be added to the body of the class.
class MyClass: number = 5 string = "Hello!"
An instance (object) of
MyClass can be created and bound to a name:
>>> new_object = MyClass() # Class is instantiated and resulting object is bound to the "new_object" name (variable). # Note: the object address 'at 0x15adc55b0' will vary by computer. >>> new_object <__main__.MyClass at 0x15adc55b0>
Class attributes are shared across all objects (or instances) created from a class, and can be accessed via
dot notation - a
. placed after the object name and before the attribute name:
>>> new_object = MyClass() # Accessing the class attribute "number" via dot-notation. >>> new_object.number 5 # Accessing the class attribute "string" via dot-notation. >>> new_object.string 'Hello!' # Instantiating an additional object and binding it to the "second_new_object" name. >>> second_new_object = MyClass() >>> second_new_object <__main__.MyClass at 0x15ad99970> # Second_new_object shares the same class attributes as new_object. >>> new_object.number == second_new_object.number True
Class attributes are defined in the body of the class itself, before any other methods. They are owned by the class - allowing them to be shared across instances. Because these attributes are shared by all instances of the class, their value can be accessed and manipulated from the class directly. Altering the value of class attributes alters the value for all objects instantiated from the class:
>>> obj_one = MyClass() >>> obj_two = MyClass() # Accessing a class attribute from an object. >>> obj_two.number 5 # Accessing the class attribute from the class itself. >>> MyClass.number 5 # Modifying the value of the "number" class attribute. >>> MyClass.number = 27 # Modifying the "number" class attribute changes the "number" attribute for all objects. >>> obj_one.number 27 >>> obj_two.number 27
Having a bunch of objects with synchronized data at all times is not particularly useful.
Fortunately, objects created from a class can be customized with their own
instance attributes (or instance properties, variables, or fields).
As their name suggests, instance attributes are unique to each object, and can be modified independently.
The special "dunder method" (short for "double underscore method")
__init__() is used to customize class instances, and can be used to initialize instance attributes or properties for objects.
For its role in initializing instance attributes,
__init__() is also referred to as a
class constructor or
__init__() takes one required parameter called
self, which refers to the newly initiated or created object.
Data for instance attributes or properties can then be passed as arguments of
__init__(), following the
MyClass now has instance attributes called
As you can see, the two attributes have been assigned to the first and second indexes of the
location (a tuple) argument that has been passed to
location_y attributes for a class instance will now be initialized when you instantiate the class and an object is created:
class MyClass: # These are class attributes, variables, or fields. number = 5 string = "Hello!" # This is the class "constructor", with a "location" parameter that is a tuple. def __init__(self, location): # This is an instance or object property, attribute, or variable. # Note that we are unpacking the tuple argument into two seperate instance variables. self.location_x = location self.location_y = location # Create a new object "new_object_one", with object property (1, 2). >>> new_object_one = MyClass((1, 2)) # Create a second new object "new_object_two" with object property (-8, -9). >>> new_object_two = MyClass((-8, -9)) # Note that new_object_one.location_x and new_object_two.location_x two different values. >>> new_object_one.location_x 1 >>> new_object_two.location_x -8
Note that you only need to pass one argument when initializing
MyClass above -- Python takes care of passing
self when the class is called.
Another way of creating an instance variable is to simply access a class variable via an object, and change it to something else:
>>> obj_one = MyClass() >>> obj_two = MyClass() >>> MyClass.string 'Hello!' >>> obj_two.string = "Hi!" >>> obj_two.string 'Hi!' >>> obj_one.string 'Hello!'
Now, watch what happens when the class variable changes:
>>> MyClass.string = "World!" >>> obj_two.string # obj_two.string has not changed 'Hi!' >>> obj_one.string # obj_one.string changed 'World!'
string is now an instance variable in the case of
obj_two, but remains a class variable in
This can also be done from within the class:
class Demo: new_var = 3 def add_two(self): self.new_var += 2
The moment that
<object>.add_two() is called, and
self.new_var += 2 is read,
new_var changes from a class variable to an instance variable of the same name.
This can be useful during initialization when all instances of a class will need some attribute(s) to start with the same value.
However, the instance variable then shadows* the class variable, making the class variable inaccessible from the instance where it is shadowed.
Given this situation, it may be safer and clearer to set instance attributes from the
__init__() method as
method is a
function that is bound to either the class itself (known as a class method, which will be discussed in depth in a later exercise) or an instance of the class (object).
Methods that operate on an object or instance need to be defined with
self as the first parameter.
You can then define the rest of the parameters as you would for a "normal" or non-bound function.
Methods that operate on a class need to be defined with the
@classmethod decorator and (by convention) with
cls as the first parameter.
Class methods are called on a class directly without first creating an object from the class.
class MyClass: number = 5 string = "Hello!" # Class constructor. def __init__(self, location): # Instance properties self.location_x = location self.location_y = location # Class method. Uses the @classmethod decorator, and cls as the first parameter. # Can be called without first making an instance of the class. @classmethod def change_string(cls, new_string): #Class attributes are referred to with cls. cls.string = new_string return cls.string # Instance method. Note "self" as first parameter. def change_location(self, amount): self.location_x += amount self.location_y += amount return self.location_x, self.location_y
Like attribute access, calling a method is as simple as putting a
. after the object name and before the method name.
The called method does not need a copy of the object as a first parameter -- Python fills in
class MyClass: number = 5 string = "Hello!" def __init__(self, location): self.location_x = location self.location_y = location def change_location(self, amount): self.location_x += amount self.location_y += amount return self.location_x, self.location_y # Make a new test_object with location (3,7) >>> test_object = MyClass((3,7)) >>> (test_object.location_x, test_object.location_y) (3,7) # Call change_location to increase location_x and location_y by 7. >>> test_object.change_location(7) (10, 14)
Class attributes can be accessed from within instance methods in the same way that they are accessed outside of the class:
class MyClass: number = 5 string = "Hello!" def __init__(self, location): self.location_x = location self.location_y = location # Alter instance variable location_x and location_y def change_location(self, amount): self.location_x += amount self.location_y += amount return self.location_x, self.location_y # Alter class variable number for all instances from within an instance. def increment_number(self): # Increment the 'number' class variable by 1. MyClass.number += 1 >>> test_object_one = MyClass((0,0)) >>> test_object_one.number 5 >>> test_object_two = MyClass((13, -3)) >>> test_object_two.increment_number() >>> test_object_one.number 6
In previous concept exercises and practice exercise stubs, you will have seen the
pass keyword used within the body of functions in place of actual code.
pass keyword is a syntactically valid placeholder - it prevents Python from throwing a syntax error for an empty function or class definition.
Essentially, it is a way to say to the Python interpreter, 'Don't worry! I will put code here eventually, I just haven't done it yet.'
class MyClass: number = 5 string = "Hello!" def __init__(self, location): self.location_x = location self.location_y = location # Alter instance variable location_x and location_y. def change_location(self, amount): self.location_x += amount self.location_y += amount return self.location_x, self.location_y # Alter class variable number for all instances. def increment_number(self): # Increment the 'number' class variable by 1. MyClass.number += 1 # This will compile and run without error, but has no current functionality. def pending_functionality(self): # Stubbing or placholding the body of this method. pass