// Tutorial //

Understanding Class Inheritance in Python 3

Published on April 5, 2017 · Updated on August 20, 2021
Default avatar
By Lisa Tagliaferri
Developer and author at DigitalOcean.
Understanding Class Inheritance in Python 3

Introduction

Object-oriented programming creates reusable patterns of code to curtail redundancy in development projects. One way that object-oriented programming achieves recyclable code is through inheritance, when one subclass can leverage code from another base class.

This tutorial will go through some of the major aspects of inheritance in Python, including how parent classes and child classes work, how to override methods and attributes, how to use the super() function, and how to make use of multiple inheritance.

Prerequisites

You should have Python 3 installed and a programming environment set up on your computer or server. If you don’t have a programming environment set up, you can refer to the installation and setup guides for a local programming environment or for a programming environment on your server appropriate for your operating system (Ubuntu, CentOS, Debian, etc.)

What Is Inheritance?

Inheritance is when a class uses code constructed within another class. If we think of inheritance in terms of biology, we can think of a child inheriting certain traits from their parent. That is, a child can inherit a parent’s height or eye color. Children also may share the same last name with their parents.

Classes called child classes or subclasses inherit methods and variables from parent classes or base classes.

We can think of a parent class called Parent that has class variables for last_name, height, and eye_color that the child class Child will inherit from the Parent.

Because the Child subclass is inheriting from the Parent base class, the Child class can reuse the code of Parent, allowing the programmer to use fewer lines of code and decrease redundancy.

Parent Classes

Parent or base classes create a pattern out of which child or subclasses can be based on. Parent classes allow us to create child classes through inheritance without having to write the same code over again each time. Any class can be made into a parent class, so they are each fully functional classes in their own right, rather than just templates.

Let’s say we have a general Bank_account parent class that has Personal_account and Business_account child classes. Many of the methods between personal and business accounts will be similar, such as methods to withdraw and deposit money, so those can belong to the parent class of Bank_account. The Business_account subclass would have methods specific to it, including perhaps a way to collect business records and forms, as well as an employee_identification_number variable.

Similarly, an Animal class may have eating() and sleeping() methods, and a Snake subclass may include its own specific hissing() and slithering() methods.

Let’s create a Fish parent class that we will later use to construct types of fish as its subclasses. Each of these fish will have first names and last names in addition to characteristics.

Info: To follow along with the example code in this tutorial, open a Python interactive shell on your local system by running the python3 command. Then you can copy, paste, or edit the examples by adding them after the >>> prompt.

We’ll create a new file called fish.py and start with the __init__() constructor method, which we’ll populate with first_name and last_name class variables for each Fish object or subclass.

fish.py
class Fish:
    def __init__(self, first_name, last_name="Fish"):
        self.first_name = first_name
        self.last_name = last_name

We have initialized our last_name variable with the string "Fish" because we know that most fish will have this as their last name.

Let’s also add some other methods:

fish.py
class Fish:
    def __init__(self, first_name, last_name="Fish"):
        self.first_name = first_name
        self.last_name = last_name

    def swim(self):
        print("The fish is swimming.")

    def swim_backwards(self):
        print("The fish can swim backwards.")

We have added the methods swim() and swim_backwards() to the Fish class, so that every subclass will also be able to make use of these methods.

Since most of the fish we’ll be creating are considered to be bony fish (as in they have a skeleton made out of bone) rather than cartilaginous fish (as in they have a skeleton made out of cartilage), we can add a few more attributes to the __init__() method:

fish.py
class Fish:
    def __init__(self, first_name, last_name="Fish",
                 skeleton="bone", eyelids=False):
        self.first_name = first_name
        self.last_name = last_name
        self.skeleton = skeleton
        self.eyelids = eyelids

    def swim(self):
        print("The fish is swimming.")

    def swim_backwards(self):
        print("The fish can swim backwards.")

Building a parent class follows the same methodology as building any other class, except we are thinking about what methods the child classes will be able to make use of once we create those.

Child Classes

Child or subclasses are classes that will inherit from the parent class. That means that each child class will be able to make use of the methods and variables of the parent class.

For example, a Goldfish child class that subclasses the Fish class will be able to make use of the swim() method declared in Fish without needing to declare it.

We can think of each child class as being a class of the parent class. That is, if we have a child class called Rhombus and a parent class called Parallelogram, we can say that a Rhombus is a Parallelogram, just as a Goldfish is a Fish.

The first line of a child class looks a little different than non-child classes as you must pass the parent class into the child class as a parameter:

class Trout(Fish):

The Trout class is a child of the Fish class. We know this because of the inclusion of the word Fish in parentheses.

With child classes, we can choose to add more methods, override existing parent methods, or accept the default parent methods with the pass keyword, which we’ll do in this case:

fish.py
...
class Trout(Fish):
    pass

We can now create a Trout object without having to define any additional methods.

fish.py
...
class Trout(Fish):
    pass

terry = Trout("Terry")
print(terry.first_name + " " + terry.last_name)
print(terry.skeleton)
print(terry.eyelids)
terry.swim()
terry.swim_backwards()

We have created a Trout object terry that makes use of each of the methods of the Fish class even though we did not define those methods in the Trout child class. We only needed to pass the value of "Terry" to the first_name variable because all of the other variables were initialized.

When we run the program, we’ll receive the following output:

Output
Terry Fish bone False The fish is swimming. The fish can swim backwards.

Next, let’s create another child class that includes its own method. We’ll call this class Clownfish, and its special method will permit it to live with sea anemone:

fish.py
...
class Clownfish(Fish):

    def live_with_anemone(self):
        print("The clownfish is coexisting with sea anemone.")

Next, let’s create a Clownfish object to see how this works:

fish.py
...
casey = Clownfish("Casey")
print(casey.first_name + " " + casey.last_name)
casey.swim()
casey.live_with_anemone()

When we run the program, we’ll receive the following output:

Output
Casey Fish The fish is swimming. The clownfish is coexisting with sea anemone.

The output shows that the Clownfish object casey is able to use the Fish methods __init__() and swim() as well as its child class method of live_with_anemone().

If we try to use the live_with_anemone() method in a Trout object, we’ll receive an error:

Output
terry.live_with_anemone() AttributeError: 'Trout' object has no attribute 'live_with_anemone'

This is because the method live_with_anemone() belongs only to the Clownfish child class, and not the Fish parent class.

Child classes inherit the methods of the parent class it belongs to, so each child class can make use of those methods within programs.

Overriding Parent Methods

So far, we have looked at the child class Trout that made use of the pass keyword to inherit all of the parent class Fish behaviors, and another child class Clownfish that inherited all of the parent class behaviors and also created its own unique method that is specific to the child class. Sometimes, however, we will want to make use of some of the parent class behaviors but not all of them. When we change parent class methods we override them.

When constructing parent and child classes, it is important to keep program design in mind so that overriding does not produce unnecessary or redundant code.

We’ll create a Shark child class of the Fish parent class. Because we created the Fish class with the idea that we would be creating primarily bony fish, we’ll have to make adjustments for the Shark class that is instead a cartilaginous fish. In terms of program design, if we had more than one non-bony fish, we would most likely want to make separate classes for each of these two types of fish.

Sharks, unlike bony fish, have skeletons made of cartilage instead of bone. They also have eyelids and are unable to swim backwards. Sharks can, however, maneuver themselves backwards by sinking.

In light of this, we’ll be overriding the __init__() constructor method and the swim_backwards() method. We don’t need to modify the swim() method since sharks are fish that can swim. Let’s review this child class:

fish.py
...
class Shark(Fish):
    def __init__(self, first_name, last_name="Shark",
                 skeleton="cartilage", eyelids=True):
        self.first_name = first_name
        self.last_name = last_name
        self.skeleton = skeleton
        self.eyelids = eyelids

    def swim_backwards(self):
        print("The shark cannot swim backwards, but can sink backwards.")

We have overridden the initialized parameters in the __init__() method, so that the last_name variable is now set equal to the string "Shark", the skeleton variable is now set equal to the string "cartilage", and the eyelids variable is now set to the Boolean value True. Each instance of the class can also override these parameters.

The method swim_backwards() now prints a different string than the one in the Fish parent class because sharks are not able to swim backwards in the way that bony fish can.

We can now create an instance of the Shark child class, which will still make use of the swim() method of the Fish parent class:

fish.py
...
sammy = Shark("Sammy")
print(sammy.first_name + " " + sammy.last_name)
sammy.swim()
sammy.swim_backwards()
print(sammy.eyelids)
print(sammy.skeleton)

When we run this code, we’ll receive the following output:

Output
Sammy Shark The fish is swimming. The shark cannot swim backwards, but can sink backwards. True cartilage

The Shark child class successfully overrode the __init__() and swim_backwards() methods of the Fish parent class, while also inheriting the swim() method of the parent class.

When there will be a limited number of child classes that are more unique than others, overriding parent class methods can prove to be useful.

The super() Function

With the super() function, you can gain access to inherited methods that have been overwritten in a class object.

When we use the super() function, we are calling a parent method into a child method to make use of it. For example, we may want to override one aspect of the parent method with certain functionality, but then call the rest of the original parent method to finish the method.

In a program that grades students, we may want to have a child class for Weighted_grade that inherits from the Grade parent class. In the child class Weighted_grade, we may want to override a calculate_grade() method of the parent class in order to include functionality to calculate a weighted grade, but still keep the rest of the functionality of the original class. By invoking the super() function we would be able to achieve this.

The super() function is most commonly used within the __init__() method because that is where you will most likely need to add some uniqueness to the child class and then complete initialization from the parent.

To see how this works, let’s modify our Trout child class. Since trout are typically freshwater fish, let’s add a water variable to the __init__() method and set it equal to the string "freshwater", but then maintain the rest of the parent class’s variables and parameters:

fish.py
...
class Trout(Fish):
    def __init__(self, water = "freshwater"):
        self.water = water
        super().__init__(self)
...

We have overridden the __init__() method in the Trout child class, providing a different implementation of the __init__() that is already defined by its parent class Fish. Within the __init__() method of our Trout class we have explicitly invoked the __init__() method of the Fish class.

Because we have overridden the method, we no longer need to pass first_name in as a parameter to Trout, and if we did pass in a parameter, we would reset freshwater instead. We will therefore initialize the first_name by calling the variable in our object instance.

Now we can invoke the initialized variables of the parent class and also make use of the unique child variable. Let’s use this in an instance of Trout:

fish.py
...
terry = Trout()

# Initialize first name
terry.first_name = "Terry"

# Use parent __init__() through super()
print(terry.first_name + " " + terry.last_name)
print(terry.eyelids)

# Use child __init__() override
print(terry.water)

# Use parent swim() method
terry.swim()
Output
Terry Fish False freshwater The fish is swimming.

The output shows that the object terry of the Trout child class is able to make use of both the child-specific __init__() variable water while also being able to call the Fish parent __init__() variables of first_name, last_name, and eyelids.

The built-in Python function super() allows us to utilize parent class methods even when overriding certain aspects of those methods in our child classes.

Multiple Inheritance

Multiple inheritance is when a class can inherit attributes and methods from more than one parent class. This can allow programs to reduce redundancy, but it can also introduce a certain amount of complexity as well as ambiguity, so it should be done with thought to overall program design.

To show how multiple inheritance works, let’s create a Coral_reef child class than inherits from a Coral class and a Sea_anemone class. We can create a method in each and then use the pass keyword in the Coral_reef child class:

coral_reef.py
class Coral:

    def community(self):
        print("Coral lives in a community.")


class Anemone:

    def protect_clownfish(self):
        print("The anemone is protecting the clownfish.")


class CoralReef(Coral, Anemone):
    pass

The Coral class has a method called community() that prints one line, and the Anemone class has a method called protect_clownfish() that prints another line. Then we call both classes into the inheritance tuple. This means that CoralReef is inheriting from two parent classes.

Let’s now instantiate a CoralReef object:

coral_reef.py
...
great_barrier = CoralReef()
great_barrier.community()
great_barrier.protect_clownfish()

The object great_barrier is set as a CoralReef object, and can use the methods in both parent classes. When we run the program, we’ll see the following output:

Output
Coral lives in a community. The anemone is protecting the clownfish.

The output shows that methods from both parent classes were effectively used in the child class.

Multiple inheritance allows us to use the code from more than one parent class in a child class. If the same method is defined in multiple parent methods, the child class will use the method of the first parent declared in its tuple list.

Though it can be used effectively, multiple inheritance should be done with care so that our programs do not become ambiguous and difficult for other programmers to understand.

Conclusion

This tutorial went through constructing parent classes and child classes, overriding parent methods and attributes within child classes, using the super() function, and allowing for child classes to inherit from multiple parent classes.

Inheritance in object-oriented coding can allow for adherence to the DRY (don’t repeat yourself) principle of software development, allowing for more to be done with less code and repetition. Inheritance also compels programmers to think about how they are designing the programs they are creating to ensure that code is effective and clear.

If you’ve enjoyed this tutorial and our broader community, consider checking out our DigitalOcean products which can also help you achieve your development goals.

Learn more here


Tutorial Series: Object-Oriented Programming in Python 3

Object-oriented programming (OOP) focuses on creating reusable patterns of code, in contrast to procedural programming, which focuses on explicit sequenced instructions. When working on complex programs in particular, object-oriented programming lets you reuse code and write code that is more readable, which in turn makes it more maintainable.

Tutorial Series: How To Code in Python

Python banner image

Introduction

Python is a flexible and versatile programming language that can be leveraged for many use cases, with strengths in scripting, automation, data analysis, machine learning, and back-end development. It is a great tool for both new learners and experienced developers alike.

Prerequisites

You should have Python 3 installed and a programming environment set up on your computer or server. If you don’t have a programming environment set up, you can refer to the installation and setup guides for a local programming environment or for a programming environment on your server appropriate for your operating system (Ubuntu, CentOS, Debian, etc.)

About the authors
Default avatar
Developer and author at DigitalOcean.

Still looking for an answer?

Was this helpful?
9 Comments

This textbox defaults to using Markdown to format your answer.

You can type !ref in this text area to quickly search our full set of tutorials, documentation & marketplace offerings and insert the link!

Thanks, Lisa, that’s exactly what I was searching for. You are super() :)

Hello Lisa,

I came across your tutorial and I really like it, but I noticed some problems, for example like the one ehmatthes pointed out. I’ve taken the time to outline what the problems are, and how to correct them here: https://github.com/svetarosemond/digital-ocean-inheritance-correction

Take a look and let me know what you think.

Amazing tutorial. Make me understand the inheritance in OOP, very clearly, for first time . Hats off to Lisa

Hey Lisa,

I have some concerns over the super function, even if we pass in variables as parameters for super function init , it do not reset the child init param freshwater, as it works the following way when ever we call a variable on object, it first checks inside its class ie Trout and then if its not available it checks in parent class. Let me write code for better understanding

class Trout(Fish):
    def __init__(self, first_name, water="freshwater"):
        self.water = water
        super().__init__(self, first_name)

I believe we can do the above way. Please let me know if i am doing anything wrong in my way of coding

Hi. Thanks for the tutorial, it was well written.

One part that doesn’t seem right, is where you create a Trout object terry = Trout() and then set the name: terry.first_name = “Terry”

Most programming designs frown on this as if anything happens between those 2 lines, you’ve created an indeterminate state, where a Trout may or may not have a name.

A better solution is to do this:

class Trout(Fish): def init(self, *args, water = “freshwater”): self.water = water super().init(*args) # Notice you don’t need to pass self into the parent init!

Now you can set the name when you create the object as before: terry = Trout(“Terry”)

Everything else is as before.

Thank you so much for this tutorial! I’m a beginner in Python and this made learning so fun and simple, great job!! :) I genuinely picked up more in this tutorial than in tons of others!

Could I also trouble you to explain a piece of code I found elsewhere in your own words? (I’ve used the shark example provided in the tutorial to replace the context of the code I found elsewhere so it might be easier)

Class Shark(Layer):
      def __init__(self, **kwargs):
             super(Shark, self).__init__(**kwargs)

I have read your other tutorial on the use of *args and **kwargs, but I still can’t seem to piece them together.

Perhaps my main confusion would be the use of Shark,self within the parentheses in super(Shark,self) and the use of **kwargs

Thank you so much and I appreciate your tutorial nonetheless! :)

Wow am I Impressed

I’m a programming beginner learning Python, trying to get clear on Classes and Objects. I’m also a life-long professional writer and former investment industry quant (an odd pairing to be sure).

This tutorial and the two that precede it are some of the clearest, most direct and lucid examples of expository instructional writing I’ve ever seen.

Absolutely amazing job! I’m old enough to have seen a lot and learned a lot. Nothing is complicated if it’s explained well by an expert instructor skilled at illuminating the inner simplicity that lies hidden underneath apparent outer complexity.

These tutorials achieve that and more. Hat’s off and bows, Lisa, in your general direction.

Hi Lisa,

After being filed as spam for putting up some useful links, and then a second time for no apparent reason, I’m going to keep my message short(er):

Your tutorial was really helpful to me, but I do agree with ehmatthes. Overriding init doesn’t mean that the parent’s init changes-- it still maintains the same arguments. The child just has a different init that is called when a new child instance is created. When you call a parent’s init, you need to pass any non-‘self’ and non-default arguments as specified in the parent init’s definition; otherwise, an error occurs. The solution that ehmatthes proposed is the conventional/correct way to go about making sure the initialization goes smoothly.

That being said, I thought much of content was beneficial. The fun and intuitive examples were especially appreciated. Thanks for the help!

Best, V

I think there’s an issue in the section about super(). In the __init__() method for Trout, self is being passed to the parent __init__() method. But this argument is being used as the value for first_name. It’s not causing an issue because the next line in the example sets terry.first_name to “Terry”.

If you call super().__init__() from a child class, I think you need to include all the required arguments for the parent class’ __init__() method. I think the fix for this example is to add a parameter for first_name in the child class __init__() method, and pass that as an argument to super().__init__()?

Thank you for this tutorial, this is a really helpful resource!