I've been working on a project where I want to control how elements are added to a list. For example, I need to make sure that every element added is greater than 0. I thought about using the descriptor pattern to manage this, but I found out that the usual dunder methods like `__set__` or `__setitem__` don't apply. This made me wonder why Python lists don't have dunder methods for append and extend, especially since it would make it easier to implement controlled access similar to what we have with other Python objects. I know I could subclass the list and override those methods, but I'm curious why descriptors can't be applied to lists in the way I hoped.
4 Answers
You can also subclass `UserList`. This gives you a bit more flexibility if you're looking to control list behavior specifically.
Just to clarify, descriptors might not work on lists the way you think they should. Dunder methods are used because they’re called indirectly via features like list item assignments, whereas append and extend are regular methods. Therefore, you can add your logic directly in those methods.
You might want to check out `collections.abc.MutableSequence`. It provides a way to create a controlled list and might give you the behavior you’re looking for. It handles some of the mutator methods for you.
The reason you won't find dunder methods for append and extend is that these methods are considered normal methods, not built-in functions or operators. If you want controlled access when adding elements, overriding these mutator methods is the way to go. The `MutableSequence` already implements some of those for you, so it simplifies the process.
I see your point! It would have been cool if append was a dunder method to enhance polymorphism, but I get that it’s a design choice.
I was suggesting that it would be beneficial to have the flexibility of descriptors for any attribute type, not just the limited subset that Python uses.