17. Class Methods


In addition to attributes, classes may have methods. A method is a function that exists inside of a class. Expanding the earlier example of a Dog class from the review problem 1 above, the code below adds a method for a dog barking.

1class Dog():
2    def __init__(self):
3        self.age = 0
4        self.name = ""
5        self.weight = 0
7    def bark(self):
8        print("Woof")

The method definition is contained in lines 7-8 above. Method definitions in a class look almost exactly like function definitions. The big difference is the addition of a parameter self on line 7. The first parameter of any method in a class must be self. This parameter is required even if the function does not use it.

Here are the important items to keep in mind when creating methods for classes:

  • Attributes should be listed first, methods after.

  • The first parameter of any method must be self.

  • Method definitions are indented exactly one tab stop.

Methods may be called in a manner similar to referencing attributes from an object. See the example code below.

1my_dog = Dog()
3my_dog.name = "Spot"
4my_dog.weight = 20
5my_dog.age = 3

Line 1 creates the dog. Lines 3-5 set the attributes of the object. Line 7 calls the bark function. Note that even through the bark function has one parameter, self, the call does not pass in anything. This is because the first parameter is assumed to be a reference to the dog object itself. Behind the scenes, Python makes a call that looks like:

# Example, not actually legal

If the bark function needs to make reference to any of the attributes, then it does so using the self reference variable. For example, we can change the Dog class so that when the dog barks, it also prints out the dog’s name. In the code below, the name attribute is accessed using a dot operator and the self reference.

def bark(self):
    print("Woof says", self.name)

Attributes are adjectives, and methods are verbs.

17.1. Example: Ball Class


This example code could be used as part of a program to draw and keep track of a ball. Having all the parameters contained in a class makes data management easier.

 1class Ball():
 2    def __init__(self):
 3        # --- Class Attributes ---
 4        # Ball position
 5        self.x = 0
 6        self.y = 0
 8        # Ball's vector
 9        self.change_x = 0
10        self.change_y = 0
12        # Ball size
13        self.size = 10
15        # Ball color
16        self.color = [255, 255, 255]
18    # --- Class Methods ---
19    def move(self):
20        self.x += self.change_x
21        self.y += self.change_y
23    def draw(self):
24        arcade.draw_circle_filled(self.x, self.y, self.size, self.color )

Below is the code that would go ahead of the main program loop to create a ball and set its attributes:

1the_ball = Ball()
2the_ball.x = 100
3the_ball.y = 100
4the_ball.change_x = 2
5the_ball.change_y = 1
6the_ball.color = [255, 0, 0]

This code would go inside the main loop to move and draw the ball:


17.1.1. References

Here’s where we separate the true programmers from the want-to-be’s. Understanding class references. Take a look at the following code:

 1class Person():
 2    def __init__(self):
 3        self.name = ""
 4        self.money = 0
 7def main():
 8    bob = Person()
 9    bob.name = "Bob"
10    bob.money = 100
12    nancy = Person()
13    nancy.name = "Nancy"
15    print(bob.name, "has", bob.money, "dollars.")
16    print(nancy.name, "has", nancy.money, "dollars.")

The code above creates two instances of the Person() class, and using www.pythontutor.com we can visualize the two classes in the figure.


Two Persons

The code above has nothing new. But the code below does:

 1class Person():
 2    def __init__(self):
 3        self.name = ""
 4        self.money = 0
 7def main():
 8    bob = Person()
 9    bob.name = "Bob"
10    bob.money = 100
12    nancy = bob
13    nancy.name = "Nancy"
15    print(bob.name, "has", bob.money, "dollars.")
16    print(nancy.name, "has", nancy.money, "dollars.")

See the difference on line 12?

A common misconception when working with objects is to assume that the variable bob is the Person object. This is not the case. The variable bob is a reference to the Person object. That is, it stores the memory address of where the object is, and not the object itself.

If bob actually was the object, then line 9 could create a copy of the object and there would be two objects in existence. The output of the program would show both Bob and Nancy having 100 dollars. But when run, the program outputs the following instead:

Nancy has 100 dollars.
Nancy has 100 dollars.

What bob stores is a reference to the object. Besides reference, one may call this address, pointer, or handle. A reference is an address in computer memory for where the object is stored. This address is a hexadecimal number which, if printed out, might look something like 0x1e504. When line 9 is run, the address is copied rather than the entire object the address points to. See the figure below.


Class References

We can also run this in www.pythontutor.com to see how both of the variables are pointing to the same object.


One Person, Two Pointers

17.2. Functions and References

Look at the code example below. Line 1 creates a function that takes in a number as a parameter. The variable money is a variable that contains a copy of the data that was passed in. Adding 100 to that number does not change the number that was stored in bob.money on line 11. Thus, the print statement on line 14 prints out 100, and not 200.

 1def give_money1(money):
 2    money += 100
 5class Person():
 6    def __init__(self):
 7        self.name = ""
 8        self.money = 0
11def main():
12    bob = Person()
13    bob.name = "Bob"
14    bob.money = 100
16    give_money1(bob.money)
17    print(bob.money)

Running on PythonTutor we see that there are two instances of the money variable. One is a copy and local to the give_money1 function.


Function References

Look at the additional code below. This code does cause bob.money to increase and the print statement to print 200.

1def give_money2(person):
2    person.money += 100

Why is this? Because person contains a copy of the memory address of the object, not the actual object itself. One can think of it as a bank account number. The function has a copy of the bank account number, not a copy of the whole bank account. So using the copy of the bank account number to deposit 100 dollars causes Bob’s bank account balance to go up.


Function References

Arrays work the same way. A function that takes in an array (list) as a parameter and modifies values in that array will be modifying the same array that the calling code created. The address of the array is copied, not the entire array.

17.3. Review Questions

  1. Create a class called Cat. Give it attributes for name, color, and weight. Give it a method called meow.

  2. Create an instance of the cat class, set the attributes, and call the meow method.

  3. Create a class called Monster. Give it an attribute for name and an integer attribute for health. Create a method called decrease_health that takes in a parameter amount and decreases the health by that much. Inside that method, print that the animal died if health goes below zero.

17.4. Avoid This Mistake

Put everything for a method into just one definition. Don’t define it twice. For example:

1# Wrong:
2class Dog():
3    def __init__(self):
4        self.age = 0
5        self.name = ""
6        self.weight = 0
8    def __init__(self):
9        print("New dog!")

The computer will just ignore the first __init__ and go with the last definition. Instead do this:

1# Correct:
2class Dog():
3    def __init__(self):
4        self.age = 0
5        self.name = ""
6        self.weight = 0
7        print("New dog!")

A constructor can be used for initializing and setting data for the object. The example Dog class above still allows the name attribute to be left blank after the creation of the dog object. How do we keep this from happening? Many objects need to have values right when they are created. The constructor function can be used to make this happen. See the code below:

Constructor that takes in data to initialize the class
 1class Dog():
 3    def __init__(self, new_name):
 4        """ Constructor. """
 5        self.name = new_name
 8def main():
 9    # This creates the dog
10    my_dog = Dog("Spot")
12    # Print the name to verify it was set
13    print(my_dog.name)
15    # This line will give an error because
16    # a name is not passed in.
17    her_dog = Dog()

On line 3 the constructor function now has an additional parameter named new_name. The value of this parameter is used to set the name attribute in the Dog class on line 8. It is no longer possible to create a Dog class without a name. The code on line 15 tries this. It will cause a Python error and it will not run. A common mistake is to name the parameter of the __init__ function the same as the attribute and assume that the values will automatically synchronize. This does not happen.

17.5. Review Questions

  • Should class names begin with an upper or lower case letter?

  • Should method names begin with an upper or lower case letter?

  • Should attribute names begin with an upper or lower case letter?

  • Which should be listed first in a class, attributes or methods?

  • What are other names for a reference?

  • What is another name for instance variable?

  • What is the name for an instance of a class?

  • Create a class called Star that will print out “A star is born!” every time it is created.

  • Create a class called Monster with attributes for health and a name. Add a constructor to the class that sets the health and name of the object with data passed in as parameters.

17.5.1. Inheritance

Another powerful feature of using classes and objects is the ability to make use of inheritance. It is possible to create a class and inherit all of the attributes and methods of a parent class.

For example, a program may create a class called Boat which has all the attributes needed to represent a boat in a game:

Class definition for a boat
 1class Boat():
 2    def __init__(self):
 3        self.tonnage = 0
 4        self.name = ""
 5        self.is_docked = True
 7    def dock(self):
 8        if self.is_docked:
 9            print("You are already docked.")
10        else:
11            self.is_docked = True
12            print("Docking")
14    def undock(self):
15        if not self.is_docked:
16            print("You aren't docked.")
17        else:
18            self.is_docked = False
19            print("Undocking")

To test out our code:

Floating our boat
1b = Boat()

The output:

You are already docked.
You aren't docked.
You are already docked.

(If you watch the video for this section of the class, you’ll note that the “Boat” class in the video doesn’t actually run. The code above has been corrected but I haven’t fixed the video. Use this as a reminder, no matter how simple the code and how experienced the developer, test your code before you deliver it!)

Our program also needs a submarine. Our submarine can do everything a boat can, plus we need a command for submerge. Without inheritance we have two options.

  • One, add the submerge() command to our boat. This isn’t a great idea because we don’t want to give the impression that our boats normally submerge.

  • Two, we could create a copy of the Boat class and call it Submarine. In this class we’d add the submerge() command. This is easy at first, but things become harder if we change the Boat class. A programmer would need to remember that we’d need to change not only the Boat class, but also make the same changes to the Submarine class. Keeping this code synchronized is time consuming and error-prone.

Luckily, there is a better way. Our program can create child classes that will inherit all the attributes and methods of the parent class. The child classes may then add fields and methods that correspond to their needs. For example:

1class Submarine(Boat):
2    def submerge(self):
3        print("Submerge!")

Line 1 is the important part. Just by putting Boat in between the parentheses during the class declaration, we have automatically picked up every attribute and method that is in the Boat class. If we update Boat, then the child class Submarine will automatically get these updates. Inheritance is that easy!

The next code example is diagrammed out in the figure below.


Class Diagram

Person, Employee, Customer Classes Examples
 1class Person():
 2    def __init__(self):
 3        self.name = ""
 5class Employee(Person):
 6    def __init__(self):
 7        # Call the parent/super class constructor first
 8        super().__init__()
10        # Now set up our variables
11        self.job_title = ""
13class Customer(Person):
14    def __init__(self):
15        super().__init__()
16        self.email = ""
18def main():
19    john_smith = Person()
20    john_smith.name = "John Smith"
22    jane_employee = Employee()
23    jane_employee.name = "Jane Employee"
24    jane_employee.job_title = "Web Developer"
26    bob_customer = Customer()
27    bob_customer.name = "Bob Customer"
28    bob_customer.email = "send_me@spam.com"

By placing Person between the parentheses on lines 5 and 13, the programmer has told the computer that Person is a parent class to both Employee and Customer. This allows the program to set the name attribute on lines 19 and 22.

Methods are also inherited. Any method the parent has, the child class will have too. But what if we have a method in both the child and parent class?

We have two options. We can run them both with super() keyword. Using super() followed by a dot operator, and then finally a method name allows you to call the parent’s version of the method.

The code above shows the first option using super where we run not only the child constructor but also the parent constructor.

If you are writing a method for a child and want to call a parent method, normally it will be the first statement in the child method. Notice how it is in the example above.

All constructors should call the parent constructor because then you’d have a child without a parent and that is just sad. In fact, some languages force this rule, but Python doesn’t.

The second option? Methods may be overridden by a child class to provide different functionality. The example below shows both options. The Employee.report overrides the Person.report because it never calls and runs the parent report method. The Customer report does call the parent and the report method in Customer adds to the Person functionality.

Overriding constructors
 1class Person():
 2    def __init__(self):
 3        self.name = ""
 5    def report(self):
 6        # Basic report
 7        print("Report for", self.name)
 9class Employee(Person):
10    def __init__(self):
11        # Call the parent/super class constructor first
12        super().__init__()
14        # Now set up our variables
15        self.job_title = ""
17    def report(self):
18        # Here we override report and just do this:
19        print("Employee report for", self.name)
21class Customer(Person):
22    def __init__(self):
23        super().__init__()
24        self.email = ""
26    def report(self):
27        # Run the parent report:
28        super().report()
29        # Now add our own stuff to the end so we do both
30        print("Customer e-mail:", self.email)
32def main():
33    john_smith = Person()
34    john_smith.name = "John Smith"
36    jane_employee = Employee()
37    jane_employee.name = "Jane Employee"
38    jane_employee.job_title = "Web Developer"
40    bob_customer = Customer()
41    bob_customer.name = "Bob Customer"
42    bob_customer.email = "send_me@spam.com"
44    john_smith.report()
45    jane_employee.report()
46    bob_customer.report()

17.6. Is-A and Has-A Relationships

Classes have two main types of relationships. They are “is a” and “has a” relationships.

A parent class should always be a more general, abstract version of the child class. This type of child to parent relationship is called an is a relationship. For example, a parent class Animal could have a child class Dog. The dog is an animal. The Dog class could have a child class Poodle. The poodle is a dog, and is an animal.

It does not work the other way! A dolphin is a mammal, but a mammal is not always a dolphin. So the class Dolphin should never be a parent to a class Mammal.

Unrelated items that do not pass the is a test should not form parent/child relationships. For example, a class Table should not be a parent to a class Chair because a chair is not a table.

The other type of relationship is the has a relationship. These relationships are implemented in code by class attributes. A dog has a name, and so the Dog class has an attribute for name. Likewise a person could have a dog, and that would be implemented by having the Person class have an attribute for Dog. The Person class would not derive from Dog because that would be some kind of insult.

Looking at the prior code example we can see:

  • Employee is a person.

  • Customer is a person.

  • Person has a name.

  • Employee has a job title.

  • Customer has an e-mail.

17.6.1. Static Variables vs. Instance Variables

The difference between static and instance variables is confusing. Thankfully it isn’t necessary to completely understand the difference right now. But if you stick with programming, it will be. Therefore we will briefly introduce it here.

There are also some oddities with Python that kept me confused the first several years I’ve made this book available. So you might see older videos and examples where I get it wrong.

An instance variable is the type of class variable we’ve used so far. Each instance of the class gets its own value. For example, in a room full of people each person will have their own age. Some of the ages may be the same, but we still need to track each age individually.

With instance variables, we can’t just say “age” with a room full of people. We need to specify whose age we are talking about. Also, if there are no people in the room, then referring to an age when there are no people to have an age makes no sense.

With static variables the value is the same for every single instance of the class. Even if there are no instances, there still is a value for a static variable. For example, we might have a count static variable for the number of Human classes in existence. No humans? The value is zero, but the count variable still exists.

In the example below, ClassA creates an instance variable. ClassB creates a static variable.

 1# Example of an instance variable
 2class ClassA():
 3    def __init__(self):
 4        self.y = 3
 6# Example of a static variable
 7class ClassB():
 8    x = 7
10def main():
11    # Create class instances
12    a = ClassA()
13    b = ClassB()
15    # Two ways to print the static variable.
16    # The second way is the proper way to do it.
17    print(b.x)
18    print(ClassB.x)
20    # One way to print an instance variable.
21    # The second generates an error, because we don't know what instance
22    # to reference.
23    print(a.y)
24    print(ClassA.y)

In the example above, lines 16 and 17 print out the static variable. Line 17 is the “proper” way to do so. Unlike before, we can refer to the class name when using static variables, rather than a variable that points to a particular instance. Because we are working with the class name, by looking at line 17 we instantly can tell we are working with a static variable. Line 16 could be either an instance or static variable. That confusion makes line 17 the better choice.

Line 22 prints out the instance variable, just like we’ve done in prior examples. Line 23 will generate an error because each instance of y is different (it is an instance variable after all) and we aren’t telling the computer what instance of ClassA we are talking about.

17.7. Instance Variables Hiding Static Variables

This is one “feature” of Python I dislike. It is possible to have a static variable, and an instance variable with the same name. Look at the example below:

 1# Class with a static variable
 2class ClassB():
 3    x = 7
 5def main():
 6    # Create a class instance
 7    b = ClassB()
 9    # This prints 7
10    print(b.x)
12    # This also prints 7
13    print(ClassB.x)
15    # Set x to a new value using the class name
16    ClassB.x = 8
18    # This also prints 8
19    print(b.x)
21    # This prints 8
22    print(ClassB.x)
24    # Set x to a new value using the instance.
25    # Wait! Actually, it doesn't set x to a new value!
26    # It creates a brand new variable, x. This x
27    # is an instance variable. The static variable is
28    # also called x. But they are two different
29    # variables. This is super-confusing and is bad
30    # practice.
31    b.x = 9
33    # This prints 9
34    print(b.x)
36    # This prints 8. NOT 9!!!
37    print(ClassB.x)

Allowing instance variables to hide static variable caused confusion for me for many years!