Address case where model_construct on a class which defines model_post_init fails with AttributeError: __pydantic_private__ when subsequently model_copy'd

[babygrimes]

Change Summary

See detail as reported in Issue #9122

Note: this patch specifically addresses the handling of __pydantic_private__ in __copy__, __deepcopy__ and __eq__. It does not address usage in several other places in main.py including: __{get,set,del}attr__, __{get,set}state__. It was unclear whether these were required, as eventually the model created by model_construct will in fact have __pydantic_private__ set (possibly to None).

:## Related issue number: #9122

Checklist

• The pull request title is a good summary of the changes - it will be used in the changelog
• Unit tests for the changes exist
• Tests pass on CI
• Documentation reflects the changes where applicable [N/A]
• My PR is ready to review

Please Review

Fix #9122

Selected Reviewer: @alexmojaki

[codspeed-hq]

CodSpeed Performance Report

Merging #9168 will not alter performance

_{Comparing babygrimes:9122-deepcopy-pydantic-private (d2660e5) with main (6322b24)}

Summary

✅ 13 untouched benchmarks

[sydney-runkle]

Looking good, just requested some minor changes!

[babygrimes]

Made another push which uses a local private: dict[str, Any] | None to quiet make typecheck but still uses the getattr() approach. Please review and let me know if there are any additional changes required here!

[sydney-runkle]

@babygrimes,

Ah, I see. Let's go with your initial approach then (it feels a bit cleaner to me). I'll go ahead and approve - we can merge once you revert back to that approach. Thanks a bunch - looking forward to merging soon!

[sydney-runkle]

As noted above -- approved, will merge when you revert back to your initial approach with the hasattr checks followed by the direct attribute accesses (thanks for pointing out, we already use that pattern in model_construct

[babygrimes]

Changes have been reverted to the hasattr() followed by direct attribute access, also reverting ALL changes in model_construct.

[babygrimes]

Please review and merge, or let me know if anything else is needed here. And, finally, THANKS for pydantic - a great solution to such a core problem :)

[sydney-runkle]

Looks fantastic. Thank you for your contribution!!