Intro to Python Classes

Learning Objectives

Set Up

Open a Python-based repl in repl.it.

Review of OOP

We covered several OOP concepts while learning JavaScript - hopefully some of the following will sound familiar...

Python is an object-oriented programming (OOP) language.

Object-oriented programming is characterized by programming with objects that represent the real-world objects of the application.

For example, the logged in user would be represented by a user object. Or a book object in a book-related app:

Objects bundle related properties (attributes) and methods (behavior) together. We discussed this principle of OOP in the JS Classes lesson and a few times in our React lessons - who remembers the name of this OOP principle?

Remember the analogy of a class being a blueprint for objects? In an OOP language, classes are used to create objects. We instantiate a class to obtain an instance (object) of that class.

Objects in Python

As you've already heard, everything in Python is an object. What this means is that every variable or piece of data has properties and/or methods encapsulated within the object.

Python provides a dir() function that can be used to list an objects attributes and methods:

# create a list
nums = [1, 2, 3]
print( dir(nums) )

You'll see a list printed containing strings representing the members (properties & methods) available on the object. Some of them like append & pop look familiar.

The other methods that start and end with double-underscores, are called magic, or dunder, methods. They are internal methods most commonly used to overload operators. If you would like to learn more about them, be sure to check out the link in the Further Study section.

We'll be using the __init__ dunder method shortly.

When we start working with Django, we'll be defining quite a few classes, so let's see how to we do it...

Writing a basic Python class

Like many of you, I like dogs - let's define a Dog class to create doggies from:

class Dog():
  def __init__(self, name, age = 0):
    self.name = name
    self.age = age

  def bark(self):
    print(f'{self.name} says woof!')

The naming convention for Python classes is UpperCamelCasing.

Python automatically calls the __init__ magic method when a new dog is created.

__init__ is short for "initialize" because the method is used to initialize the properties of the new object.

What method did we use in JS classes to perform the same thing?

The age = 0 in __init__'s parameter list is called a default parameter and will be assigned the the result of the expression to the right of the = if the function is called without an argument for that positional parameter.

The attributes for a dog instance are name and age.

bark is an instance method in this Dog class.

What's an instance method?

What's this self business?

In the JS lesson about this, it was mentioned that every OOP language must have the same or similar mechanism as this to be able to:

• Enable a method to access the other properties/methods in an object, and
• Enable a single-copy of a method in memory to serve any number of instances.

JavaScript, Java, C++, C#, and others call it this.

Ruby, Swift and others call it self.

However, in Python, only by convention is it called self because it's just a parameter name...

Take a look at the __init__ and bark method definitions, notice how the first parameter is named self. When we write code like spot.bark(), the object to the left of the dot is automatically assigned to the method's first parameter - which should be named self. This is how Python provides the "context" in both instance and class methods!

Creating Objects by Instantiating a Class

By defining the Dog class, we now know the structure that each of the pooches will have!

Let's make a doggie:

spot = Dog('Spot', 8)

print(spot) # -> similar to <__main__.Dog object at 0x7f27bad2c208>

# print the name and age attributes of the spot object
print(spot.name, spot.age) # -> Spot 8

# invoke the spot object's bark instance method
spot.bark() # -> Spot says woof!

Let's try out the default parameter for a new dog's age:

dog = Dog('Lassie')

print(dog.name, dog.age) # -> Lassie 0

Overriding Methods

As you saw above, when we used print(spot) to print the spot object, we got an unfriendly output.

We can change this behavior by overriding the __str__ method that the print function calls automatically to obtain the string to print out.

Let's modify the Dog class to override the __str__ method:

class Dog():
  def __init__(self, name, age = 0):
    self.name = name
    self.age = age

  def bark(self):
    print(f'{self.name} says woof!')
    
  def __str__(self):
    return f'Dog named {self.name} is {self.age} years old'

Let's try it out:

spot = Dog('Spot', 8)

print(spot) # -> Dog named Spot is 8 years old

💪 Exercise - Create a Class (15 min)

At the top of the repl, define a class named Vehicle with the following members:

1. vin: attribute for the vehicle's identification
2. make: attribute for the vehicle's make
3. model: attribute for the vehicle's model
4. running: attribute for maintaining whether or not the vehicle is running. This should be set to False within the __init__ method instead of being passed in at the time of instantiation.
5. start: method for changing running to True
6. stop: method for changing running to False

Test out the class by instantiating it a couple of times, calling start/stop methods and printing some of its attributes:

car = Vehicle('TS123', 'Tesla', 'Model S')
print(car.running) # -> False
car.start()
print(car.running) # -> True
plane = Vehicle('X99Y', 'Boeing', '747-B')
print(plane.vin, plane.make, plane.model)

Time permitting, override the __str__ method so that it returns a string formatted as:

Vehicle (<vin>) is a <make> model <model> 

Class vs. Instance Members

In Python, instance attributes & methods (members) are intended to be accessed/invoked by instances of the class, whereas, class members are intended to be accessible on the class only, not an instance.

Each object instance has its own copy of its attributes, e.g., name. However, all instances share class attributes.

To demonstrate class attributes, let's add a nextId class attribute to the Dog class that can be used to assign an id to each dog instance:

class Dog():
  # class attribute
  next_id = 1

  # updated __init__
  def __init__(self, name, age = 0):
    self.name = name
    self.age = age
    self.id = Dog.next_id
    Dog.next_id += 1

  def bark(self):
    print(f'{self.name} says woof!')

  # updated __str__
  def __str__(self):
    return f'Dog ({self.id}) named {self.name} is {self.age} years old'

Note how the Dog.next_id class attribute is being accessed within the __init__ method.

Note: Technically, instances can also access class members via self due to the fact that if the instance does not have an accessed member, Python will check the class and provide the class version of the member if it exists.

Now let's make sure it worked :)

spot = Dog('Spot', 8)
print(spot)
pup = Dog('Lassie')
print(pup)

Cool, now let's see how class methods are created by adding a get_total_dogs method.

Add this to the bottom of the Dog class:

  def __str__(self):
    return f'Dog ({self.id}) named {self.name} is {self.age} years old'
  
  # new code below

  @classmethod
  def get_total_dogs(cls):
    # cls represents the Dog class
    return cls.next_id - 1

There's only two differences when defining a class method:

1. The @classmethod decorator
2. The naming convention of the first parameter is to use cls instead of self

Decorators in programming are used to implement metaprogramming (when a program has knowledge or manipulates itself). In Python, decorators are used to modify the behavior of a function or class. They are not very common, but there's a link in the Further Study section if you'd like to learn more about decorators in Python.

Let's test out the new class method:

spot = Dog('Spot', 8)
pup = Dog('Lassie')

# class methods are called on the class, not an instance
print(Dog.get_total_dogs())  # -> 2

Inheritance

Maybe the following graphic will jog your memory in regards to what inheritance is:

Using inheritance, a subclass automatically inherits all of the attributes and methods of its superclass.

The subclass can then define additional attributes and/or methods to make a more specialized class than the superclass.

For example, in the JS Classes lesson, we specialized the Vehicle class by extending it to create a Plane subclass.

Let's see how inheritance is implemented in Python by creating a ShowDog class that specializes the Dog class:

# Pass in superclass as argument
class ShowDog(Dog):
  # Add additional parameters AFTER those in the superclass
  def __init__(self, name, age = 0, total_earnings = 0):
    # Always call the superclass's __init__ first
    Dog.__init__(self, name, age)
    # Now add any new attributes
    self.total_earnings = total_earnings
  
  # Add additional methods
  def add_prize_money(self, amount):
    self.total_earnings += amount

It's show time!

winky = ShowDog('Winky', 3, 1000)
print(winky) # Yay, inherited the overriden __str__
winky.bark() # Yay, inherited the bark method
print(winky.total_earnings) # -> 1000
winky.add_prize_money(500) # New method that 'Dogs' don't have
print(winky.total_earnings) # -> 1500

Inheritance is critical to OOP languages. In fact, they even have their own object hierarchies. Check this out:

Frameworks like Django and Rails have elaborate object hierarchies of their own. For example, when we move on to Django, we'll be defining Models by inheriting from a Django class like this:

class Person(models.Model):

Essential Questions

Take a couple of minutes to review these...

❓ What's the difference between a class and an object?

❓ What Python keyword is used to define a class?

❓ Another word for an object is an _.

❓ How do we use a class to create objects?

❓ True or False: Class attributes are shared by all instances of that class.

❓ What OOP principle refers to subclasses specializing superclasses?

💪 Practice Exercise

Looking for some practice building an object hierarchy in Python? Good!

In a separate Python repl...

Create a BankAccount class with the following members:

1. owner: (attribute) The owner's name as a string
2. balance: (attribute) The amount of money in the account
3. account_no: (attribute) A number to be randomly generated and assigned within __init__ 1. not passed in at time of instantiation
4. deposit(amount): (method) When called on an instance, increases the balance by the amount argument and returns the new balance
5. withdraw(amount): (method) When called on an instance, decreases the balance by the amount argument and returns the new balance

Here's how to generate a random integer for the in Python:

# Put this line at the top of the repl
import random

# Use this inside of BankAccount's __init__ to generate
# a random account number from 111111111 to 999999999
self.account_no = random.randint(111111111, 999999999)

Create two instances, make both deposits and withdrawals, and print the attributes to test them out.

Bonus 1

Override the __str__ method to return the following formatted string:

Account <account_no> / Balance: xxxxx.xx

Bonus 2

Create a SavingsAccount class that subclasses BankAccount and specializes it so that the withdraw method no longer accepts any argument, does not change the balance, and returns a string of No withdrawals permitted.

Bonus 3

Add an additional has_overdraft attribute to the BankAccount class that accepts True or False at the time of instantiation, but defaults to False if not passed in (hint: review default parameters discussed above).

When the withdraw method is called, do not allow the withdraw if the amount being withdrawn is greater than balance, unless has_overdraft is True. withdraw should continue to return the balance.

Further Study

Classes Tutorial

Python Inheritance

Learn more about magic methods here

Learn more about Python's self here

Learn more about metaprogramming here

Decorators in Python