AttributeError with pydantic 2.10.0 & 2.10.1 (fixed in 2.10.2)

[twhalm]

With pydantic v2.10.0, importing pyairtable causes the following error:

Exception has occurred: AttributeError
partially initialized module 'pyairtable' has no attribute 'api' (most likely due to a circular import)
  File "file.py", line 2, in <module>
    from pyairtable import Api
AttributeError: partially initialized module 'pyairtable' has no attribute 'api' (most likely due to a circular import)

This error does not occur with earlier versions of pydantic.

[shaunagm]

We're getting the same error.

Edit: it looks like this is an issue with Pydantic 2.10 which introduced a number of regressions being actively worked on that have bubbled up to pyairtable

[JasonCrowe]

I backed off to 2.9 and it's working.

[Eagles0607]

Backing off to 2.9 worked for me as well.

[sungmincs]

Pydantic 2.10.1 still doesn't fix this issue.

[Viicos]

This should be fixed in 2.10.2, coming next week. The reason you are getting this issue is because the following _api private field annotation:

pyairtable/pyairtable/models/_base.py

Lines 124 to 135 in 34eb78e

	class RestfulModel(AirtableModel):
	"""
	Base model for any data structures that wrap around a REST API endpoint.
	Subclasses can pass a number of keyword arguments to control serialization behavior:
	* ``url=``: format string for building the URL to be used when saving changes to this model.
	"""
	__url_pattern: ClassVar[str] = ""
	_api: "pyairtable.api.api.Api" = pydantic.PrivateAttr()

failed to be evaluated in Pydantic <= 2.9. In 2.10, we refactored our forward annotation evaluation logic to be more correct, meaning it does evaluate the forward annotation as expected. However, this leads to a circular import because the pyairtable.api.api module is already in the process of being imported when this forward annotation is evaluated.

2.10.2 will include a fix to not evaluate forward annotations for private fields, because we don't make use of the annotation any way.

[mesozoic]

Thanks @Viicos for keeping us updated, and for working toward a patch to fix the issue. 🙌

In the meantime I've opened #412 in case anyone out there has ideas for how to remove the circular dependencies caused by type annotations in this library.

[Viicos]

In the meantime I've opened #412 in case anyone out there has ideas for how to remove the circular dependencies caused by type annotations in this library.

Note that as long as these circular issues happens in private attributes, it shouldn't be an issue anymore in the next patch release. But yeah, if they are used for "normal" Pydantic fields, this is going to be an issue.

[coronarita1991]

I got a same error as above while importing pyairtable, and solved with downgrading pydantic with 2.9.2. After solving issue I could reach this post. Anyway, thanks.

[LouisMonierSpark]

Super painful issue. This should be prioritized asap.

[Viicos]

2.10.2 is now released, and fixes the issue.

[mesozoic]

Thank you for the fix, @Viicos! I'll resolve this issue now.

Note that as long as these circular issues happens in private attributes, it shouldn't be an issue anymore in the next patch release. But yeah, if they are used for "normal" Pydantic fields, this is going to be an issue.

Heard! I don't foresee us adding any more things like this, but it'll still be an interesting mental exercise to rethink the dependency graph within the library. Thanks again for your help.