Advanced use of Python decorators and metaclasses


While introducing people to Python metaclasses I realized that sometimes the big problem of the most powerful Python features is that programmers do not perceive how they may simplify their usual tasks. Therefore, features like metaclasses are considered a fancy but rather unuseful addition to a standard OOP language, instead of a real game changer.

This post wants to show how to use metaclasses and decorators to create a powerful class that can be inherited and customized by easily adding decorated methods.

Metaclasses and decorators: a match made in space

Metaclasses are a complex topic, and most of the times even advanced programmers do not see a wide range of practical uses for them. Chances are that this is the part of Python (or other languages that support metaclasses, like Smalltalk and Ruby) that fits the least the "standard" object-oriented patterns or solutions found in C++ and Java, just to mention two big players.

Indeed metaclasess usually come ... more

Python 3 OOP Part 6 - Abstract Base Classes

This post is available as an IPython Notebook here

The Inspection Club

As you know, Python leverages polymorphism at its maximum by dealing only with generic references to objects. This makes OOP not an addition to the language but part of its structure from the ground up. Moreover, Python pushes the EAFP appoach, which tries to avoid direct inspection of objects as much as possible.

It is however very interesting to read what Guido van Rossum says in PEP 3119: Invocation means interacting with an object by invoking its methods. Usually this is combined with polymorphism, so that invoking a given method may run different code depending on the type of an object. Inspection means the ability for external code (outside of the object's methods) to examine the type or properties of that object, and make decisions on how to treat that ... more

Python 3 OOP Part 5 - Metaclasses

This post is available as an IPython Notebook here

The Type Brothers

The first step into the most intimate secrets of Python objects comes from two components we already met in the first post: class and object. These two things are the very fundamental elements of Python OOP system, so it is worth spending some time to understand how they work and relate each other.

First of all recall that in Python everything is an object, that is everything inherits from object. Thus, object seems to be the deepest thing you can find digging into Python variables. Let's check this

>>> a = 5
>>> type(a... more

Python 3 OOP Part 4 - Polymorphism

This post is available as an IPython Notebook here

Good Morning, Polymorphism

The term polymorphism, in the OOP lingo, refers to the ability of an object to adapt the code to the type of the data it is processing.

Polymorphism has two major applications in an OOP language. The first is that an object may provide different implementations of one of its methods depending on the type of the input parameters. The second is that code written for a given type of data may be used on data with a derived type, i.e. methods understand the class hierarchy of a type.

In Python polymorphism is one of the key concepts, and we can say that it is a built-in feature. Let us deal with it step by step.

First of all, you know that in Python the type of a variable is not explicitly declared. Beware that this does not mean that Python variables are untyped. On the contrary, everything in ... more

Python 3 OOP Part 3 - Delegation: composition and inheritance

This post is available as an IPython Notebook here

The Delegation Run

If classes are objects what is the difference between types and instances?

When I talk about "my cat" I am referring to a concrete instance of the "cat" concept, which is a subtype of "animal". So, despite being both objects, while types can be specialized, instances cannot.

Usually an object B is said to be a specialization of an object A when:

  • B has all the features of A
  • B can provide new features
  • B can perform some or all the tasks performed by A in a different way

Those targets are very general and valid for any system and the key to achieve them with the maximum reuse of already existing components is delegation. Delegation means that an object shall perform only what it knows best, and leave the rest to other ... more

Python 3 OOP Part 2 - Classes and members

This post is available as an IPython Notebook here

Python Classes Strike Again

The Python implementation of classes has some peculiarities. The bare truth is that in Python the class of an object is an object itself. You can check this by issuing type() on the class

>>> a = 1
>>> type(a)
<class 'int'>
>>> type(int)
<class 'type'>

This shows that the int class is an object, an instance of the ... more

Python 3 OOP Part 1 - Objects and types

This post is available as an IPython Notebook here

About this series

Object-oriented programming (OOP) has been the leading programming paradigm for several decades now, starting from the initial attempts back in the 60s to some of the most important languages used nowadays. Being a set of programming concepts and design methodologies, OOP can never be said to be "correctly" or "fully" implemented by a language: indeed there are as many implementations as languages.

So one of the most interesting aspects of OOP languages is to understand how they implement those concepts. In this post I am going to try and start analyzing the OOP implementation of the Python language. Due to the richness of the topic, however, I consider this attempt just like a set of thoughts for Python beginners trying to find their way into this beautiful (and sometimes peculiar) language.

This series of posts wants to ... more