Decorator is a structural design pattern that lets you attach new behaviors to objects by placing these objects inside special wrapper objects that contain the behaviors.
Imagine that you're working on a notification library which lets other programs notify their users about important events.
The initial version of the library was based on the
Notifier class that had only a few fields, a constructor and a single
send method. The method could accept a message argument from a client and send the message to a list of emails that were passed to the notifier via its constructor. A third-party app which acted as a client was supposed to create and configure the notifier object once, and then use it each time something important happened.
At some point, you realize that users of the library expect more than just email notifications. Many of them would like to receive an SMS about critical issues. Others would like to be notified on Facebook and, of course, the corporate users would love to get Slack notifications.
How hard can that be? You extended the
Notifier class and put the additional notification methods into new subclasses. Now the client was supposed to instantiate the desired notification class and use it for all further notifications.
But then someone reasonably asked you, "Why can't you use several notification types at once? If your house is on fire, you'd probably want to be informed through every channel."
You tried to address that problem by creating special subclasses which combined several notification methods within one class. However, it quickly became apparent that this approach would bloat the code immensely, not only the library code but the client code as well.
You have to find some other way to structure notifications classes so that their number won't accidentally break some Guinness record.
Extending a class is the first thing that comes to mind when you need to alter an object's behavior. However, inheritance has several serious caveats that you need to be aware of.
- Inheritance is static. You can't alter the behavior of an existing object at runtime. You can only replace the whole object with another one that's created from a different subclass.
- Subclasses can have just one parent class. In most languages, inheritance doesn't let a class inherit behaviors of multiple classes at the same time.
One of the ways to overcome these caveats is by using Aggregation or Composition instead of Inheritance. Both of the alternatives work almost the same way: one object has a reference to another and delegates it some work, whereas with inheritance, the object itself is able to do that work, inheriting the behavior from its superclass.
With this new approach you can easily substitute the linked "helper" object with another, changing the behavior of the container at runtime. An object can use the behavior of various classes, having references to multiple objects and delegating them all kinds of work. Aggregation/composition is the key principle behind many design patterns, including Decorator. On that note, let's return to the pattern discussion.
"Wrapper" is the alternative nickname for the Decorator pattern that clearly expresses the main idea of the pattern. A wrapper is an object that can be linked with some target object. The wrapper contains the same set of methods as the target and delegates to it all requests it receives. However, the wrapper may alter the result by doing something either before or after it passes the request to the target.
When does a simple wrapper become the real decorator? As I mentioned, the wrapper implements the same interface as the wrapped object. That's why from the client's perspective these objects are identical. Make the wrapper's reference field accept any object that follows that interface. This will let you cover an object in multiple wrappers, adding the combined behavior of all the wrappers to it.
In our notifications example, let's leave the simple email notification behavior inside the base
Notifier class, but turn all other notification methods into decorators.
The client code would need to wrap a basic notifier object into a set of decorators that match the client's preferences. The resulting objects will be structured as a stack.
The last decorator in the stack would be the object that the client actually works with. Since all decorators implement the same interface as the base notifier, the rest of the client code won't care whether it works with the "pure" notifier object or the decorated one.
We could apply the same approach to other behaviors such as formatting messages or composing the recipient list. The client can decorate the object with any custom decorators, as long as they follow the same interface as the others.
Wearing clothes is an example of using decorators. When you're cold, you wrap yourself in a sweater. If you're still cold with a sweater, you can wear a jacket on top. If it's raining, you can put on a raincoat. All of these garments "extend" your basic behavior but aren't part of you, and you can easily take off any piece of clothing whenever you don't need it.
The Component declares the common interface for both wrappers and wrapped objects.
Concrete Component is a class of objects being wrapped. It defines the basic behavior, which can be altered by decorators.
The Base Decorator class has a field for referencing a wrapped object. The field's type should be declared as the component interface so it can contain both concrete components and decorators. The base decorator delegates all operations to the wrapped object.
Concrete Decorators define extra behaviors that can be added to components dynamically. Concrete decorators override methods of the base decorator and execute their behavior either before or after calling the parent method.
The Client can wrap components in multiple layers of decorators, as long as it works with all objects via the component interface.
In this example, the Decorator pattern lets you compress and encrypt sensitive data independently from the code that actually uses this data.
The application wraps the data source object with a pair of decorators. Both wrappers change the way the data is written to and read from the disk:
Just before the data is written to disk, the decorators encrypt and compress it. The original class writes the encrypted and protected data to the file without knowing about the change.
Right after the data is read from disk, it goes through the same decorators, which decompress and decode it.
The decorators and the data source class implement the same interface, which makes them all interchangeable in the client code.
Use the Decorator pattern when you need to be able to assign extra behaviors to objects at runtime without breaking the code that uses these objects.
The Decorator lets you structure your business logic into layers, create a decorator for each layer and compose objects with various combinations of this logic at runtime. The client code can treat all these objects in the same way, since they all follow a common interface.
Use the pattern when it's awkward or not possible to extend an object's behavior using inheritance.
Many programming languages have the
final keyword that can be used to prevent further extension of a class. For a final class, the only way to reuse the existing behavior would be to wrap the class with your own wrapper, using the Decorator pattern.
How to Implement
Make sure your business domain can be represented as a primary component with multiple optional layers over it.
Figure out what methods are common to both the primary component and the optional layers. Create a component interface and declare those methods there.
Create a concrete component class and define the base behavior in it.
Create a base decorator class. It should have a field for storing a reference to a wrapped object. The field should be declared with the component interface type to allow linking to concrete components as well as decorators. The base decorator must delegate all work to the wrapped object.
Make sure all classes implement the component interface.
Create concrete decorators by extending them from the base decorator. A concrete decorator must execute its behavior before or after the call to the parent method (which always delegates to the wrapped object).
The client code must be responsible for creating decorators and composing them in the way the client needs.
Pros and Cons
- You can extend an object's behavior without making a new subclass.
- You can add or remove responsibilities from an object at runtime.
- You can combine several behaviors by wrapping an object into multiple decorators.
- Single Responsibility Principle. You can divide a monolithic class that implements many possible variants of behavior into several smaller classes.
- It's hard to remove a specific wrapper from the wrappers stack.
- It's hard to implement a decorator in such a way that its behavior doesn't depend on the order in the decorators stack.
- The initial configuration code of layers might look pretty ugly.
Relations with Other Patterns
Adapter changes the interface of an existing object, while Decorator enhances an object without changing its interface. In addition, Decorator supports recursive composition, which isn't possible when you use Adapter.
Chain of Responsibility and Decorator have very similar class structures. Both patterns rely on recursive composition to pass the execution through a series of objects. However, there are several crucial differences.
The CoR handlers can execute arbitrary operations independently of each other. They can also stop passing the request further at any point. On the other hand, various Decorators can extend the object's behavior while keeping it consistent with the base interface. In addition, decorators aren't allowed to break the flow of the request.
A Decorator is like a Composite but only has one child component. There's another significant difference: Decorator adds additional responsibilities to the wrapped object, while Composite just "sums up" its children's results.
However, the patterns can also cooperate: you can use Decorator to extend the behavior of a specific object in the Composite tree.
Decorator and Proxy have similar structures, but very different intents. Both patterns are built on the composition principle, where one object is supposed to delegate some of the work to another. The difference is that a Proxy usually manages the life cycle of its service object on its own, whereas the composition of Decorators is always controlled by the client.