First, some background:

What Computers Do

Generally, computers take data as input, follow algorithms, and produce data as output.

Data

Generally, data must be held for a while during execution of the algorithm. Python uses variables to store data.

>>> 123
>>> foo = 7

Data Types

Data comes in a number of types:

Atomic (can’t be split into smaller parts)

>>> foo = 7    # integer
>>> goo = 3.14 # float (floating point integer)
>>> hoo = "z"  # string (Python has no separate type for 'character')

Structured

>>> koo = "My hovercraft is full of eels." # string of characters
>>> moo = [2, 13, 474]                     # list (of numbers)
>>> noo = [5, "bob", 6.02, "doggy"]        # list with more than one type of element
>>> qoo = (1,2,45)                         # tuple (as in 'double', 'quadruple', '7-tuple')
>>> roo = ("frog", 14)                     # tuple with elements of different types
>>> soo = {8, 2048}                        # set
>>> too = {"dynomite", 7.62}               # set with elements of different types
>>> voo = {

Deeper structure

>>> woo = [5.56, (2, 4, "Burrito"), ["a", 1723], "klingon"] # list with structures as elements

Putting data into complex structures, especially if they (as often) are uniform in shape, can get boring, should be automated where possible, and computers can be used for this.

Objects and Object-Oriented Programming (OOP)

Predefined structures for data can be defined and used. Most computer languages refer to

these standardized structures as objects
the structure definitions (not the concrete objects, more like the blueprints) as classes
the process of creating an object of a class as instantiation
one specific object as an instance of a class
pieces of data held in Objects as attributes
ways of doing things to the attributes held in objects, or ways of doing things using attributes as inputs, as methods
Programming making use of this way of organizing and using data is Object-Oriented Programming.

Object-Oriented Programming in Python

In Python, to define a class: (this one is as simple as possible, and is useless other than counting, since it does not have any attributes or methods)

Define a class

We can define a class very simply.

class ScaryAnimal:
    pass

class is a keyword. Capitalization is exact. Followed by a semicolon.

ScaryAnimal is a name, and, by convention, is CamelCase (LikeTheseWordsHere)

pass is Python placeholder code. We really should add methods and attributes.

Make an object of that class

We can make (‘instantiate’) one (at the REPL, or in stored code)

my_first_monster = ScaryAnimal() # make me a scary animal, and store that in the variable 'my_first_monster'
my_first_monster # let's see what Python thinks my_first_monster is

__main__. means that ScaryAnimal is a ‘top-level’ class

object means my_first_monster is a Python object

The hex number like (but this can vary every time we run code) 0x7fef14a46410 is the address, in memory, where my_first_monster starts.

Make the object so that it can store some useful information

__init__ This is a method, which is a function specialized on a class.

__init__ is a special class, and that is what the underscores are there to indicate.

When Python instantiates a class (makes an object of that class), Python looks for a method named __init__, and, if Python finds it, executes the code therein.

__init__ is also called the constructor method, but I think this is a little off.

We can create objects without __init__ being defined (so without being found by Python).

If __init__ is found (and we should create one, at least for top-level classes) it

creates attributes (I think of them as named slots), and
gives these attributes initial values.

To be clear, we did, in fact, create my_first_monster without an __init__ method, so it is possible. Just not particularly useful, except if we just want to count monsters.

Attributes

Attributes are variables specialized on each object. If each monster has an attribute like number_of_eyes, we can have different monsters with different numbers of eyes.

Attributes (named slots that belong to a specific object) are created, and given an initial value by, __init__.

If we want to specify, at the time we create a new object, the value of some attribute of the new object (e.g., eye_count = 8, or has_fangs = True) we can identify the parameters we must pass when instantiating this class.

A Second Class

class ScarierAnimal:
 
    def __init__(self, number_of_eyes, has_fangs): 
        self.number_of_eyes = number_of_eyes
        self.has_fangs = has_fangs

self is used to refer to the new object we are building, so we can use Python code to manipulate that new object.

Even though we have defined an __init__ for ScarierAnimal, We can still instantiate an object of type ScarierAnimal without having it assigned any attributes.

(I do not think this is good.)

>>> my_second_monster = ScarierAnimal
>>> my_second_monster.number_of_eyes
ERROR

But, now that we have defined __init__, taking two parameters, we can instantiate a ScarierAnimal by passing values, like so:

>>> my_third_monster = ScarierAnimal(16, False)
>>> my_third_monster.number_of_eyes
16

Note that, in our definition of __init__, there are three parameters. The first one, self, however, is special. It is automatically assigned the value of the object itself, so we do not (and better not try to) pass a value via self.

Note the names of the parameters on the right side of assignments like

        self.eye_count = eye_count

do not need to match the names of the attributes on the left side.

ALSO NOTE that they conventionally do.

A Silly Class

# This would work, but would be silly:

class ScaryAnimalOdd:
  def __init__(self, darth, leia):
    self.eye_count = darth
    self.has_fangs = leia
    

>>> critter_2 = ScaryAnimalOdd(3, False)
>>> critter_2.eye_count
3

More about Methods

Methods are functions for changing or using values held in attributes. They are specified as part of the class definition, and so apply to every instance of that class.

Perhaps we know that monsters sometimes lose eyes, and sometimes must be defanged.

We can define a couple of methods as part of a class.

    def defang(self):
        self.has_fangs = False
 
    def poke_eye_out(self):
        self.eye_count = self.eye_count - 1

Like in a class definition

class EvenMoreScaryAnimal:
 
    def __init__(self, number_of_eyes, has_fangs): 
        self.number_of_eyes = number_of_eyes
        self.has_fangs = has_fangs
        
    def defang(self):
        self.has_fangs = False
 
    def poke_eye_out(self):

>>> my_fourth_monster = EvenMoreScaryAnimal(6, True)
>>> my_fourth_monster.number_of_eyes # no parens, so we want to learn about the **attribute** 'eye_count'
6

>>>my_fourth_monster.has_fangs 
True

>>> my_fourth_monster.poke_eye_out()
>>> my_fourth_monster.number_of_eyes
5

Notes:

the parentheses indicate that we are invoking a function/method.
even though self is a parameter in the definition of methods, no value is ever expected to be passed via self, and one (the special one of the object being called, like a leash) is what self evaluates to within the method definition.

Sub-Classes

Often making sub-classifications is useful, because, although objects in a sub-class (like SportsCar) might share attributes (engine_model) and methods with (add_miles()) with a parent class (like MotorVehicle), they might also have different attributes or methods (racing_league).

We can efficiently capture this by creating subclasses.

Python (3) OOP