The Decorator Principle
First of all, I will give you an introduction to the Decorator Principle. This principle is not limited to Python, you can use it in every other language as well. However, Python has some built-in syntax for decorators and the decorator principle fits the language design as well. Understanding this principle will make it much easier to familiarize yourself with decorators and understand their functionality.
To explain the decorator principle, I will take the menu of a restaurant as an example. We assume that every dish on the menu has a corresponding function or class in our code. Whether you use classes of functions with decorators may vary based on the programming language and architecture you are using, but the core concept of the pattern stays the same. To keep it simple, we will only use functions for this example.
At first, the menu only offers a burger and a steak. So we just have two functions handling both dishes:
But the restaurant has a problem: there are customers who want custom versions of the dishes. Maybe someone wants another sauce or extra ingredients. To cover all these cases, we need functions for every single version of the dishes. As you can imagine, this is an absolutely terrible idea and design as it is inflexible and violates a lot of code principles like KISS or DRY.
As you can see, it gets really messy. And this is where decorators come in! They provide a simple and flexible way to customize functions and classes by adding extra behavior. In our example, we create a decorator for each customization, which takes the original function of the dish and adds the special wish of the customer. By using this, we only have to code the special customer wishes once and separated!
Then we can decorate our dish function according to the individual wishes of the customer. A function call with some decorators can then look like this:
As the figure shows, every decorator calls the previous function and appends some additional content to it. Therefore, decorators depend on the function they are called on.
Implementation in Python
In Python, decorators are basically callables. A callable in Python can be a function or class, in general any object that you can call. However, the easiest and most common way to implement decorators in Python is to use functions. The special thing about them is that they take a function as an input argument and return a function as well.
Make sure to return only the function and don’t call it like modified_function() for instance instead.
Since we want to create a decorator, we want to modify the behavior of the function given as input. A common way to do so is to create a new function that wraps the input function. In this wrapper function we can call the input function and do some other stuff too:
It is important to note that the wrapper function takes *args and **kwargs as arguments. If you are not familiar with this syntax, they take the normal and the keyword arguments passed into the function and put it into a list and a dictionary. Inside the wrapper, we reverse this by unpacking the arguments in the function call. In this way, we ensure that we get all the parameters that were passed to the original function and that we pass them on correctly.
To apply our decorator, we just use the @ syntax and attach it to a function. When the Python interpreter loads the code, it will call our decorator and pass the function automatically as the argument. Then our function gets replaced by the returned one from the decorator. With this in mind, you may realize that the syntax for decorators is just syntactic sugar and simply replace the function with a new one. So both of the following examples work the same:
For this example, the output of the function without the decorator is the following:
>>> calculate_answer(7) 1631
If we append the double_result decorator to the function, we can see that the result is doubled.
>>> calculate_answer(7) 3262
Overview: Advanced techniques for Python Decorators
Decorators are very powerful, and with the basic implementation of the previous chapter, you are actually well served. Nevertheless, there are some advanced techniques that can be valuable in some cases. The following is a brief overview of what you can do. However, if you want to read more about specific topics, let me know in the comments!
Decorators replace a function with another one, but what if we want the new function to have the attributes of the old one? If we take a closer look to the decorated function of the previous example, the name in the code stays the same, but the actual function name changed!
>>> calculate_answer.__name__ 'wrapper'
In most the cases this should not matter, but if it does, we can use the wraps decorator from the functools module. This will copy the function attributes to the wrapper and prevent weird name confusion.
Not only can you decorate functions, you can also decorate classes . This is rather used in other object-oriented languages like Java, but there are use-cases in Python as well. You can implement the Singleton pattern with class-decorators for instance.
Maybe you want to customize your decorators too, so you don’t have to write multiple ones. In general, this is quite easy, but it also adds a new layer of complexity to your decorator. The solution is to simply wrap your decorator in another function! This function takes the arguments you need and returns the decorator. With the decorator syntax it’s possible to call this function with your arguments and use the returned decorator instead. Python 3.9 also introduced more syntactic flexibility with the decorator syntax, you can take a look in our corresponding post if you are interested.
A decorator can also handle state. There are multiple possible solutions to implement this. For instance, you can use instance objects of classes as decorators by implementing the __call__ method. However, since functions are also objects, you can simply add attributes to functions in order to add state.
I hope this post gave you a good introduction to decorators in Python and you learned how to create them on your own. If you have any thoughts, feedback or questions feel free to leave a comment below!
There is nothing that excites me more than solving problems and understanding complex concepts. I want to share this enthusiasm by making complex things as simple as possible, so everyone can understand them. Nothing is more frustrating than trying to understand a topic with incomprehensible explanations. That’s why I want to inspire you to become a better developer by simplifying concepts and explaining issues clearly.
Our Book Recommendations on Python (Affiliate)