[Enum] add __deepcopy__ method to Enum

[ethanfurman]

Linked PRs

• gh-106602: Add __copy__ and __deepcopy__ in Enum #106666
• [3.11] gh-106602: [Enum] Add __copy__ and __deepcopy__ (GH-106666) #106694
• [3.12] gh-106602: [Enum] Add __copy__ and __deepcopy__ (GH-106666) #106695

[Agent-Hellboy]

Hi @ethanfurman, I thought of implementing this.

did you mean we should be able to make a deep copy of the whole enum like adding is check in here
https://github.com/python/cpython/blob/1f2921b72c369b19c2e32aaedb9f8c63e0cb8b48/Lib/test/test_enum.py#L803C1-L808C35

if yes, then we need to make changes in copy module because Enums are getting created using a custom metaclass
and deep copy returns the object as it is if the class is a subclass of a type
https://github.com/python/cpython/blob/1f2921b72c369b19c2e32aaedb9f8c63e0cb8b48/Lib/copy.py#L137C1-L139C42

Where did the bug get reported? Do you know if I can see the StackOverflow link?

if it's about copying a member (adding deepcopy in Enum) then I will explore more

    def __deepcopy__(self, memo):
        new_enum = self.__class__(self.value)

currently, it is taking the same object for new_enum as well as self

[ethanfurman]

[question about is check]

Yes, those should both be is checks -- enum members are singletons, so a copy of any kind should return the exact same object.

[copy module questions]

An enum class, such as Color, is processed by that branch; a member, such as Color.RED, falls through to the else branch where a __deepcopy__ is looked for.

Enum.__copy__ and Enum.__deepcopy__ just need to return self.

[bug report question]

This issue is where the bug was reported.

[ethanfurman]

The current test_copy is testing an enum class -- we need to add a test for enum members. All those copy tests should be using is.

[Agent-Hellboy]

This issue is where the bug was reported.

Ohh, I thought, this one also got reported on StackOverflow just like a calendar module feature I once picked.

I have added a PR
Btw, I learned a lot about Enum's internal design because of the wrong assumption that deep-copy(Enum or its member) will return a new Enum

[ethanfurman]

Thanks, @Agent-Hellboy, for the help!

[Agent-Hellboy]

Thanks @ethanfurman, looking forward to contributions like these.