Methods

You’ve just seen the difference between class and instance variables. Classes can also have class methods - methods that are shared among all instances of a certain type. As with variables, they can be overriden in a specific instance or subclass.

Let’s add a class method to our Car class:

class Car:
    runs = True
    number_of_wheels = 4

    @classmethod
    def get_number_of_wheels(cls):
        return cls.number_of_wheels

    def start(self):
        if self.runs:
            print("Car is started. Vroom vroom!")
        else:
            print("Car is broken :(")

And call it:

>>> my_car = Car()
>>> print(f"Cars have {Car.get_number_of_wheels()} wheels.")
Cars have 4 wheels.
# Of course, we can override this in our instance:
>>> my_car.number_of_wheels = 6
# And when we access our new instance variable:
>>> print(f"My car has {my_car.number_of_wheels} wheels.")
My car has 6 wheels.
# But, when we call our class method on our instance:
>>> print(f"My car has {my_car.get_number_of_wheels()} wheels.")
My car has 4 wheels.

Why? Because get_number_of_wheels() is a class method, and when it’s called, the class (Car) gets passed in, and the value of Car.number_of_wheels is returned. Although we can access the instance variable (with a value of 6), the get_number_of_wheels() class method still returns the class variable, which is 4.

type, isinstance, and issubclass

Python comes with some built-in functions for inspecting classes and types:

As we’ve seen throughout the workshop, the type() function returns the type of the object you pass it, or it’s class. For example:

>>> type(42)
<class 'int'>
>>> type("Hello world!")
<class 'str'>
>>> type(my_car)
<class '__main__.Car'>

The isinstance() function takes an object and a class, and returns True if the object you pass it is an instance of the class. For example:

>>> isinstance(42, int)
True
>>> isinstance("Hello world!", str)
True
>>> isinstance(my_car, float)
False
>>> isinstance(my_car, Car)
True

The issubclass function takes two classes, and returns True if the first class is a subclass of the second. For example:

# bool is a subclass of int
>>> issubclass(bool, int)
True
# int is a subclass of object
>>> issubclass(int, object)
True
# technically, everything is a subclass of object
>>> issubclass(bool, object)
True

__init__

Classes can have an optional magic method called __init__() that gets run when you instantiate an instance of a class. You can use the __init__() method to do any special thing you want to happen when your instance is instantiated, including setting instance variables. __init__ can take arguments, too.

Methods that are bracketed by underscores are sometimes called “magic methods.” We won’t be covering magic methods in this class, but we will point out a few of the interesting ones.

For example:

class Car:
    runs = True

    def __init__(self, make, model):
        self.make = make
        self.model = model

    def start(self):
        if self.runs:
            print(f"Your {self.make} {self.model} is started. Vroom vroom!")
        else:
            print(f"Your {self.make} {self.model} is broken :("

>>> my_car = Car("Ford", "Thunderbird")
>>> my_car.start()
Your Ford Thunderbird is started. Vroom vroom!

Here, we accept two required variables, make and model in our __init__() method, and set instance variables of the same names using self. Later, when we call start(), we can grab self.make and self.model from the bound instance and use them in our string.

__str__ and __repr__

Classes have two other magic methods that come in handy for debugging, __str__() and __repr__(). Both functions return a string representation of an object. __str__() should return readable end-user output, and __repr__() should return the Python code necessary to rebuild the object. __str__() maps to the built-in function str() and __repr__() maps to the built-in function repr().

For example, we’ll use the datetime library to generate a datetime object for right now:

>>> import datetime
>>> now = datetime.datetime.now()
>>> str(now)
'2019-03-16 21:04:01.396256'
>>> repr(now)
'datetime.datetime(2019, 3, 16, 21, 4, 1, 396256)'

You can see that str() has returned a human-readable date/time, and repr() has returned a string that represents the Python code we would need to run to recreate this object.

We can, of course, set our own __str__() and __repr__() methods in our custom classes:

class Car:
    def __init__(self, make, model):
        self.make = make
        self.model = model

    def __str__(self):
        return f"<<Car object: {self.make} {self.model}>>"

    def __repr__(self):
        return f"Car('{self.make}', '{self.model}')"

>>> my_car = Car("Ford", "Thunderbird")
>>> print(f"This object is a {str(my_car)}")
This object is a <<Car object: Ford Thunderbird>>
>>> print(f"To reproduce it, type: {repr(my_car)}")
To reproduce it, type: Car('Ford', 'Thunderbird')

Bonus

You don’t have to instantiate everything by hand, you can instantiate objects in for loops or even comprehensions. This is useful for running a function on a list of objects. For example, to convert a list of number-strings into a list of integers, you could do:

>>> my_ints = [int(str_num) for str_num in ["1", "2", "3"]]
>>> print(my_ints)
[1, 2, 3]