20. Objects II — More on Methods

  • The Circle class discussed in the previous topic introduced:

    • How to write classes

    • How to write and make use of a constructor

    • Attributes, both how to set them and how to access them

    • How to write and use methods

    • How to make instances of the class

  • In this topic, we will create more classes containing more complex functionality

  • Additionally, we will see objects interacting with one another and how objects can provide a nice level of abstraction

  • The goal is to define a Sphere class

    • Like the Circle class, it will know its radius

    • It will also know its position within some three dimensional space

    • It will provide functionality to measure distances between Sphere objects and check if they overlap/collide

20.1. Sphere Class

  • In order to define a Sphere, all we really need is a radius

  • With a radius, can calculate some values about the Sphere

    • Diameter

    • Surface Area

    • Volume

  • But we have an extra requirement — we need to know where the Sphere is in a three dimensional coordinate space

  • Therefore, in addition to a radius, we need to keep track of x, y, and z coordinates

  • With this information, we can start to perform some more sophisticated calculations

    • How far away are two Sphere objects from one another?

    • Do two Sphere objects overlap/collide?

  • There is nothing stopping us from adding more functionality to our Sphere class, but we will keep it simple for now

20.1.1. Constructor and Attributes

  • Below is the start of the Sphere class, including the constructor and the assignment of attributes

  • It follows the same pattern as the Circle class discussed in the previous topic

  • The only differences here with the Sphere are trivial

    • An import to help with math calculations

    • The name of the class is different

    • The parameters and attributes are for a Sphere

 1import math
 2
 3
 4class Sphere:
 5    """
 6    Class for managing Spheres within a 3D space. This includes tracking it's location in three dimensional space and
 7    radius. Additionally, it allows for some basic geometry calculations, distance measurements between Spheres, and
 8    checking if two Spheres overlap.
 9    """
10
11    def __init__(self, x: float, y: float, z: float, radius: float):
12        self.x = x
13        self.y = y
14        self.z = z
15        self.radius = radius

  • That’s it — that is all we need to get started with the Sphere class

  • Like before, we can even start making instances of a Sphere

  • However, like before, the class will not be particularly useful here without the needed functionality

20.1.2. Methods

  • The methods we want are

    • Calculate the diameter, surface_area and volume

    • Measure the distance_between_centres of two Sphere objects

    • Measure the distance_between_edges of two Sphere objects

    • Check if a Sphere overlaps another in the three dimensional space

    • A way to check if two Sphere objects are equivalent (__eq__)

    • A way to generate a human readable string representation of a Sphere (__repr__)

 1import math
 2
 3
 4class Sphere:
 5
 6    # init and/or other methods not shown for brevity
 7
 8    def diameter(self) -> float:
 9        return 2 * self.radius
10
11    def surface_area(self) -> float:
12        return 4 * math.pi * self.radius**2
13
14    def volume(self) -> float:
15        return (4 / 3) * math.pi * self.radius**3

  • The above three methods follow the same pattern we saw with the Circle

  • These look like regular functions, but the difference is

    • They are associated with an instance of a Sphere

    • They have a self parameter, which is a reference variable to the Sphere instance

    • Accessing any of the object’s attributes are done through the use of the self reference variable

  • Below is the distance_between_centres method, where we see some things that may feel odd

 1import math
 2
 3
 4class Sphere:
 5
 6    # init and/or other methods not shown for brevity
 7
 8    def distance_between_centres(self, other: "Sphere") -> float:
 9        """
10        Calculate and return the distance between the centres of two Spheres.
11
12        :param other: Sphere whose centre to find the distance to from the self Sphere.
13        :type other: Sphere
14        :return: Distance between the Sphere centres.
15        :rtype: float
16        """
17        return math.sqrt((self.x - other.x) ** 2 + (self.y - other.y) ** 2 + (self.z - other.z) ** 2)

  • The method takes a parameter, other, that should be of type Sphere — the class we are writing

  • But this does not break any rules — we are writing a method that can be invoked on an instance of the Sphere class that takes an instance of a Sphere as a parameter

  • This is OK since the intended functionality is to find the distance between two Sphere objects

    • The distance from the Sphere the method was invoked on to the Sphere that was passed as a parameter

  • If this still makes you uneasy, consider how we would use this method

1sphere_a = Sphere(1, 1, 1, 10)
2sphere_b = Sphere(2, 2, 2, 15)
3distance = sphere_a.distance_between_centres(sphere_b)
4print(distance)                                         # Results in 1.732051

  • In the above example, I invoked the method distance_between_centres on sphere_a and passed sphere_b as the argument

  • If we take a moment to analyze the code within the function, we may get a better sense of the self reference variable

    • Below is the relevant code from the distance_between_centres method

1def distance_between_centres(self, other: "Sphere") -> float:
2    return math.sqrt((self.x - other.x) ** 2 + (self.y - other.y) ** 2 + (self.z - other.z) ** 2)

  • This code calculates the Euclidean distance between the centres in three dimensional space

  • But notice that we are making use of two reference variables — self and other

    • This may be where self starts to make a little more sense

  • Again, consider sphere_a.distance_between_centres(sphere_b)

    • In this context, sphere_a would be the self Sphere object reference

    • And sphere_b would be the other reference

Note

You may have also noticed how the type hint for other is the string "Sphere", as opposed to just Sphere, like how the function’s return type hint is just float instead of the string "float". This is because the Sphere class, as far as Python is concerned, is not defined yet as the method distance_between_centres is being defined within the class Sphere that is currently being defined.

 1import math
 2
 3
 4class Sphere:
 5
 6    # init and/or other methods not shown for brevity
 7
 8    def distance_between_edges(self, other: "Sphere") -> float:
 9        """
10        Calculate and return the distance between the edges of two Spheres. If the value is negative, the two Spheres
11        overlap.
12
13        :param other: Sphere whose edge to find the distance to from the self Sphere.
14        :type other: Sphere
15        :return: Distance between the Sphere edges.
16        :rtype: float
17        """
18        return self.distance_between_centres(other) - self.radius - other.radius

  • In distance_between_edges above, notice how the method makes a call to the method distance_between_centres

  • Since the distance_between_edges needs the distance between centres in order to complete it’s calculation, there is no need to re-write that code — just call distance_between_centres

  • But, like the attributes, if we want to access the instance’s methods, we must access them via the reference variable self

 1import math
 2
 3
 4class Sphere:
 5
 6    # init and/or other methods not shown for brevity
 7
 8    def overlaps(self, other: "Sphere") -> bool:
 9        """
10        Determine if two Sphere objects overlap within the 3D space. Two Spheres that are touching (distance of 0
11        between edges) are considered overlapping.
12
13        :param other: Sphere to check if it overlaps the self Sphere overlaps
14        :type other: Sphere
15        :return: Boolean indicating if the two Spheres overlap
16        :rtype: bool
17        """
18        return self.distance_between_edges(other) <= 0

  • Similarly, the overlaps method can be written by making use of the already existing method distance_between_edges

20.1.2.1. Magic Methods

  • There exist a number of special, or magic methods within Python

  • What makes these methods magic is that you do not call them directly; you call them indirectly through some other syntax

  • In fact, the __init__ method, the constructor, is a magic method

    • You never actually directly call __init__ in your code

    • Instead, the constructor gets invoked when instantiating an instance of the class

      • some_sphere = Sphere(1, 2, 3, 4)

  • There are many of these special methods

  • In addition to the constructor, we will focus on two very important ones here

    • __eq__ — a method for checking object equality

    • __repr__ — a method for generating a human readable string representation of the object

20.1.2.1.1. __eq__

  • It is common to want to check if two things are equal

  • For example, like numbers — if some_number == 10:

  • With numbers, strings, booleans, and other types, Python already knows what equality is

  • However, with custom classes, Python will not know what it means for instances of that class to be equal, unless you tell it

  • In the context of the Sphere class, you may have a good idea of what it means for two instances of this object to be equal

  • But Python cannot read your mind; you need to tell Python what it means for two Sphere objects to be equal

  • By default, Python will try to be helpful if you ask it if two objects of a custom class are equal

    • The default equality check is checking if two reference variables are referencing literally the exact same object in memory (aliases)

  • A more reasonable equality check for Sphere objects would be if they are the same size and exist in the same location within the three dimensional space

    • That is, if the radius, x, y, and z attributes are equal

1import math
2
3
4class Sphere:
5
6    # init and/or other methods not shown for brevity
7
8    def __eq__(self, other) -> bool:
9        return self.x == other.x and self.y == other.y and self.z == other.z and self.radius == other.radius

  • The above code showing the __eq__ method is how we define our equals magic method

    • For our needs, check if self and other have all their attributes being the same

  • You may be tempted to then check equality by calling the __eq__ method explicitly

    • sphere_a.__eq__(sphere_b)

  • Although this will work, it is a little clunky and bad style

  • Instead, we will indirectly invoke the equality method by using == like we have used for every other equality check

    • sphere_a == sphere_b

  • There is, however, one problem with the way we wrote our equality method

  • Consider the below example

1some_sphere = Sphere(1, 2, 3, 4)
2some_circle = Circle(10)
3
4print(some_sphere == some_circle)

  • Running this code results in AttributeError: 'Circle' object has no attribute 'x'

  • The trouble is that the Circle instance, which would be other in the Sphere's equality method, does not have an x, y, or z attribute

  • A simple way to fix this is to check if the other reference variable is even referencing something that can be properly compared to

 1import math
 2
 3
 4class Sphere:
 5
 6    # init and/or other methods not shown for brevity
 7
 8    def __eq__(self, other) -> bool:
 9        if isinstance(other, Sphere):
10            return self.x == other.x and self.y == other.y and self.z == other.z and self.radius == other.radius
11        return False

Note

There was nothing stopping us from defining __eq__ for our Circle class. In fact, it is arguably something we should do. Below is an example of an equality check for the Circle class.

 1import math
 2
 3
 4class Circle:
 5
 6    # init and/or other methods not shown for brevity
 7
 8    def __eq__(self, other) -> bool:
 9        if isinstance(other, Circle):
10            return self.radius == other.radius
11        return False
20.1.2.1.2. __repr__

  • It is nice to have a good, human readable representation of the values within our program

  • For example, think of the number of times you have printed the values of variables when doing some quick tests of your code

1some_list = ["a", "b", "c"]
2if condition:
3    some_list.append("x")
4print(some_list)

  • If you were to try this with our newly created Sphere class, we would get something not overly helpful

1sphere = Sphere(1, 2, 3, 4)
2print(sphere)                   # Results in <__main__.Sphere object at 0x7f99a2edac10>

  • The default behaviour will be the name of the class along with the memory address of where the object is

  • Chances are, this is not overly helpful to you

  • To address this, we write another magic method — __repr__

  • __repr__ is the representation function, which is for getting a nice string representation of the instance of the class

  • This __repr__ method is called whenever we need a string representation of our object

    • print(some_sphere) will automatically call it

    • str(some_sphere) will call it

    • repr(some_sphere) will call it too

  • Based on what the object is, there may be a very natural way one would want to represent the object as a string

    • For example, the List class’ __repr__ returns a string of the form ["a", "b", "c", "d"]

  • But sometimes, like with a Sphere, it may not be obvious and we just want to get enough information about the Sphere to be helpful for us

  • If this is the case, a common representation is Sphere(x=1, y=2, z=3, radius=4) – class name, and then relevant attribute values within parentheses

1import math
2
3
4class Sphere:
5
6    # init and/or other methods not shown for brevity
7
8    def __repr__(self) -> str:
9        return f"Sphere(x={self.x}, y={self.y}, z={self.z}, radius={self.radius})"

  • With the __repr__ written, if I were to call print, str, or repr on an instance of the class, I would see the values specified

Note

Like with the __eq__ method, we could go back and write a __repr__ for the Circle class.

1import math
2
3
4class Circle:
5
6    # init and/or other methods not shown for brevity
7
8    def __repr__(self) -> str:
9        return f"Circle(radius={self.radius})"

20.1.3. Testing

  • Below is a collection of assert tests for the Sphere class

  • Though, you may feel that at this stage it’s getting harder to get a feel for how complete your tests are

 1sphere_origin_0 = Sphere(0, 0, 0, 0)
 2assert 0 == sphere_origin_0.x
 3assert 0 == sphere_origin_0.y
 4assert 0 == sphere_origin_0.z
 5assert 0 == sphere_origin_0.radius
 6assert 0 == sphere_origin_0.diameter()
 7assert 0 == sphere_origin_0.surface_area()
 8assert 0 == sphere_origin_0.volume()
 9assert 0 == sphere_origin_0.distance_between_centres(Sphere(0, 0, 0, 0))
10assert 0 == sphere_origin_0.distance_between_edges(Sphere(0, 0, 0, 0))
11assert True == sphere_origin_0.overlaps(Sphere(0, 0, 0, 0))
12assert True == (sphere_origin_0 == Sphere(0, 0, 0, 0))
13assert False == (sphere_origin_0 == Sphere(0, 0, 0, 1))
14assert "Sphere(x=0, y=0, z=0, radius=0)" == str(sphere_origin_0)
15
16sphere = Sphere(1, 2, 3, 4)
17assert 1 == sphere.x
18assert 2 == sphere.y
19assert 3 == sphere.z
20assert 4 == sphere.radius
21assert 8 == sphere.diameter()
22assert 0.01 > abs(sphere.surface_area() - 201.06)
23assert 0.01 > abs(sphere.volume() - 268.08)
24assert 0.01 > abs(sphere.distance_between_centres(Sphere(0, 0, 0, 0)) - 3.74)
25assert 0.01 > abs(sphere.distance_between_edges(Sphere(0, 0, 0, 0)) - (-0.26))
26assert True == sphere.overlaps(Sphere(0, 0, 0, 0))
27assert False == (sphere == Sphere(0, 0, 0, 0))
28assert True == (sphere == Sphere(1, 2, 3, 4))
29assert "Sphere(x=1, y=2, z=3, radius=4)" == str(sphere)

20.2. For Next Class