Skip to content

2017-09-10 Python

Object Oriented Python

Object Oriented Programming with Python#

In python, everything is an object. Even functions are function objects.

The keyword class is used to define eventual objects

Classes have attributes, which are similar to variables but are defined inside a class and belong to that class.

Classes have methods to give them abilities

When we use a class, the resulting object is an instance.

Example#

    class NewClass:
        name_attribute = "Kenneth"

        def name_method(self):
            return self.name

    new_instance = NewClass()
    new_instance.name_method()
  • By convention classes always start with a capital letter
  • If they have multiple words in, then each first letter is capitalised

Most basic class#

class Thief:
    pass

>>> from characters import Thief
>>> surfer190 = Thief()
>>> surfer190
<characters.Thief object at 0x101cbe278>

It is like running a function but classes don’t run. It creates an instance of the class.

Attributes#

class Thief:
    sneaky = True

Access the class member with . syntax

>>> surfer190.sneaky
True

You can also get the classes member with

>> Thief.sneaky
True

But instances are responsible for their own attribute values

>>> surfer190.sneaky = False
>>> surfer190.sneaky
False

Delete an instance of a class (object)

>>> del surfer190

Methods#

Methods are functions that belong to a class

Self#

Whenever they are used, they are used by an instance. Not the actual class. That is why methods always take at least one parameter that is the instance using the method By convention that parameter is always called self

If you call the class method diretly you get an error unless you give it the instance:

    >>> from characters import Thief
    >>> surfer = Thief()
    >>> surfer.pickpocket()
    Called by <characters.Thief object at 0x101bbe2e8>
    False
    >>> Thief.pickpocket()
    Traceback (most recent call last):
    File "<stdin>", line 1, in <module>
    TypeError: pickpocket() missing 1 required positional argument: 'self'
    >>> Thief.pickpocket(surfer)
    Called by <characters.Thief object at 0x101bbe2e8>
    True

Another thing is you can always refer to self in the method:

    import random

    class Thief:
        sneaky = True

        def pickpocket(self):
            print("Called by {}".format(self))
            return bool(random.randint(0, 1))

Init#

When you create a new method python looks for a method called def __init__

Dunders indicate that it is a method python will run on it’s own. We need not run it ourselves.

    def __init__(self, name, sneaky=True, **kwargs):
        self.name = name
        self.sneaky = sneaky

        for key, value in kwargs.items():
            # Very useful function
            setattr(self, key, value)

when you initialise you will be required to provide a name parameter

**kwargs is a dictionary of key-value pairs

Class attributes#

Attributes set on the class are universal and can be changed just by specifying in . notation

    redcar.laps = 0

To prevent this you have to set the number of laps attribute in __init__

Inheritance#

You can inherit attributes and methods from a parent class by adding that class in the definition

class Thief(Character):
    ...

Note: Every class in python inherits from the built-in class called object

In python 2 you had to do this manually but python3 improved this to automatically inherit

Using the Super Class#

We can use methods on demand from the super class from the child class with super()

Usually done of overridden classes. So redefine the method and call super()

When using super() Must include the method name and required arguments too.

Sub classes can take different arguments from their parent classes

    class Character:
        def __init__(self, name, **kwargs):
            self.name = name

            for key, value in kwargs.items():
                setattr(self, key, value)


    class Thief(Character):
        sneaky = True

        def __init__(self, name, sneaky=True, **kwargs):
            '''
            Super must be called first as sneaky could be defined in key word arguments
            '''
            super().__init__(name, **kwargs)
            self.sneaky = sneaky

Refactoring#

Tech jargon for rearranging code nto a more logical state, by deleting, renaming, comibing and breaking code up.

Inferiting from multiple classes (Multiple Inheritance)#

You can inherit from multiple classes with an inheritance chain. The order that you inherit from classes is important because of super()

The MRO - Method Resolution Order

You can use class.__mro__ or inspect.getmro() to look at your class’s method resolution order (MRO)

Tightly coupled code / design means that classes have to know a lot about other classes. Makes it harder to add in new functionality.

Loosely coupled code is the way to go, so it becomes easier to mix and match.

Useful functions#

isinstance('abc', str) - Tells you whether something is an instance of a particular class

    >>> isinstance(5.22, (int, str))
    False
    >>> isinstance(5.22, (int, float))
    True

You can use a tuple to check against

Easter Egg#

    >>> isinstance(True, int)
    True
    >>> isinstance(True, bool)
    True

issubclass(bool, int) - Tells you whether a class is an instance of another class

    >>> issubclass(Thief, Character)
    True
    >>> issubclass(bool ,int)
    True

type('stephen') - Tells you the type of object an instance is

Better to use isinstance as it will check the full inheritance tree

There are a lot of dunder attributes __attribute__

  • myvar.__class__ - tells you the class of an instance

Can even get the class name:

    >>> stephen = Thief('stephen')
    >>> stephen.__class__
    <class 'characters.Thief'>
    >>> stephen.__class__.__name__
    'Thief'

Python leans towards ducktyping where if a class looks and quacks like a duck it should be considered a duck. It is still handy to tell if instance of class is expected.

Another good one is dir(my_var) which shows you available methods

Magic Methods#

Methods that python calls for you. We have seen __init__ already.

__str__: Returns a string to identify object whenever it is turned into a string

__repr__: Return official string representation, used for debugging, as much info as possible

__int__ / __float__: Return integer and float representations

    >>> five = NumString(5)
    >>> five
    <numstring.NumString object at 0x101bbe278>
    >>> str(five)
    '5'
    >>> int(five)
    5
    >>> float(five)
    5.0

__add__: Addition Math operation, from the left hand side

__radd__: Reflective addition, addition from the right hand side

__iadd__: which is += (i stands for in place )

You can check the full list in Emulating numeric types docs

    class NumString:
        def __init__(self, value):
            self.value = str(value)

        def __str__(self):
            return self.value

        def __int__(self):
            return int(self.value)

        def __float__(self):
            return float(self.value)

        def __add__(self, other):
            return self.value + str(other)

        def __radd__(self, other):
            return self + other

        def __iadd__(self, other):
            self.value = self + other
            return self.value

The len(my_obj) is implemented with __len__

To check if item in my_obj you use __contains__

If __contains__ does not exist, python uses __iter__ or __getitem__

    def __iter__(self):
        for item in self.slots:
            yield item

Making an iterable#

What is yield? It is very similar to return but it does not immediately stop execution like return does

yield lets you send items out of the method as they are available and keep on working

This construct is called a generator

We can simplify above to:

        def __iter__(self):
            yield from self.slots

Because self.slots is a list, python knows how to iterate through it.

When you use these default built-in methods with your custom class, it makes them easier to use and more what you are used to.

Custom verisions of built-ins#

When building custom classes you will usually make use of built-in classes of your language. The two most important of which are __init__ and __new__

__new__ - is for customising an immutable data type

immutable means the only time you should change them is at creation time

Unlike init, new does a return

New does not take self, as it is a special method that operates on a class, not an instance

With immutable types, it can be unsafe to use super() inside of __new__, it is better to use parent method directly eg:

    self = str.__new__(*args, **kwargs)

Take this code:

    for _ in range(count): 
        ....

The _ implies that you don’t care what that value is

And if you use copy.copy() you are not using references, but copy by reference

Mimicking dot notation#

    class javascriptObject(dict):
        def __getattribute__(self, item):
            try:
                return self[item]
            except KeyError:
                return super().__getattribute__(item)

Another immutable example#

    class Double(int):
        def __new__(value, *args, **kwargs):
            self = int.__new__(value, *args, **kwargs)
            self *= 2
            return self

Class Methods#

  • Construct an instance of your class without having to have a class instance already
  • They often work as a factory for an object
  • Don’t take self as their first argument, they take the class but that is a reserved keyword so cls is used (sometimes klass is used)
  • cls(books) is basically calling the constructor __init__
  • @classmethod is a decorator - a function that takes another function, does something with it then returns it
    class Book:
    def __init__(self, title, author):
        self.title = title
        self.author = author

    def __str__(self):
        return '{} by {}'.format(self.title, self.author)

class BookCase:
    def __init__(self, books=None):
        self.books = books

    @classmethod
    def create_bookcase(cls, book_list):
        books = []
        for title, author in book_list:
            books.append(Book(title, author))
        return cls(books)

Remember with classmethod you don’t call init, you just call cls()

Usage:

    >>> bc = BookCase.create_bookcase([('Moby-Dick', 'Melville'), ('Jungle Book', 'Rudyard Kipling')])
    >>> bc
    <books.BookCase object at 0x1024bd470>
    >>> bc.books
    [<books.Book object at 0x1024bd5f8>, <books.Book object at 0x1024bd630>]
    >>> str(bc.books[0])
    'Moby-Dick by Melville'

Static Methods#

Static methods don’t require an instance or a class at all

Encapsulation (Hising elements of class away)#

Python motto: “We’re all adults here”

Just prepend the name of method or attribute with an underscore _

Double underscore __ makes method or underscore inaccessible outside of the class (private)

So python does name mangling on the attribute and methods

But you can find them by checking dir(obj)

Example#

    class Protected:
        __name = "Security"

        def __method(self):
            return self.__name

    >>> from protected import Protected
    >>> prot = Protected()
    >>> prot.__name
    Traceback (most recent call last):
    File "<stdin>", line 1, in <module>
    AttributeError: 'Protected' object has no attribute '__name'
    >>> prot.__method
    Traceback (most recent call last):
    File "<stdin>", line 1, in <module>
    AttributeError: 'Protected' object has no attribute '__method'
    >>> dir(prot)
    ['_Protected__method', '_Protected__name', '__class__', '__delattr__', '__dict__', '__dir__', '__doc__', '__eq__', '__format__', '__ge__', '__getattribute__', '__gt__', '__hash__', '__init__', '__init_subclass__', '__le__', '__lt__', '__module__', '__ne__', '__new__', '__reduce__', '__reduce_ex__', '__repr__', '__setattr__', '__sizeof__', '__str__', '__subclasshook__', '__weakref__']
    >>> prot._Protected__method
    <bound method Protected.__method of <protected.Protected object at 0x1024bd668>>
    >>> prot._Protected__method()
    'Security'
    >>> prot._Protected__name
    'Security'

So in python nothing is truly ever locked away

Property decorator#

If you don’t want people to know it is a method, you can use the @property decorator

    class Circle:
        def __init__(self, diameter):
            self.diameter = diameter

        @property
        def radius(self):
            return self.diameter / 2

So properties act as attributes

    small = Circle(10)
    print(small.radius)

But they cannot be set, as python does not know how to set it

    small.radius = 15

But there is a way to create a setter Use a decorator of the form @<property_name>.setter. Then use the same property method but take in another parameter

    class Circle:
        def __init__(self, diameter):
            self.diameter = diameter

        @property
        def radius(self):
            return self.diameter / 2

        @radius.setter
        def radius(self, radius):
            self.diameter = radius * 2

Implement Equality and Add#

    class Die:
        def __init__(self, sides=2, value=0):
            if not sides >= 2:
                raise ValueError("Must have at least 2 sides")

            if not isinstance(sides, int):
                raise ValueError("Sides must be a whole number")

            self.value = value or random.randint(1, sides)

        def __int__(self):
            return self.value

        def __eq__(self, other):
            return int(self) == other

        def __ne__(self, other):
            return not int(self) == other

        def __gt__(self, other):
            return int(self) > other

        def __lt__(self, other):
            return int(self) < other

        def __ge__(self, other):
            return int(self) > other or int(self) == other

        def __le__(self, other):
            return int(self) < other or int(self) == other

        def __add__(self, other):
            return int(self) + other

        def __radd__(self, other):
            return int(self) + other
Usage#
    >>> d6 = D6()
    >>> d6.value
    1
    >>> d6 < 3
    True
    >>> d6 <= 1
    True
    >>> d6 <= 2
    True
    >>> d6 >= 2
    False
    >>> d6 >= 1
    True
    >>> d6 != 4
    True
    >>> d6 == 6
    False
    >>> d6 == 1
    True

If you need to implement all of the methods try using the attrs library

Or if you just need equality and math ones then use the built-in functools.total_ordering

Interesting#

You can pass around a class Just like you would an integer or string

Just don’t call or initialise the class with () when passing it, unless you want result to be sent in