Fix issue with self-referencing dataclass

[uriyyo]

Change Summary

Fix issue when dataclass field that references on self point to inner pydantic model instead of dataclass.

from pydantic.dataclasses import dataclass


@dataclass
class Foo:
    a: Foo


# before this PR
assert Foo.__pydantic_model__.__fields__['a'].type_ is Foo.__pydantic_model__

# after this PR
assert Foo.__pydantic_model__.__fields__['a'].type_ is Foo

Related issue number

fix: #3675

Checklist

• Unit tests for the changes exist
• Tests pass on CI and coverage remains at 100%
• Documentation reflects the changes where applicable
• changes/<pull request or issue id>-<github username>.md file added describing change
  (see changes/README.md for details)
• My PR is ready to review, please add a comment including the phrase "please review" to assign reviewers

[uriyyo]

please review

[samuelcolvin]

This looks great, thanks.

@PrettyWood you happy with this as you did most of the work on dataclasses?

[PrettyWood]

LGTM !

[da2ce7]

It would be great to get merged this #3713 and #3819 , as I'm trying to use the dataclasses for my data structures.

[samuelcolvin]

thanks so much for this.

[pvicente]

@uriyyo I think this fix should have been released as 2.0 as interface is changing, not as 1.9.1 and it's not backward compatible. No matter it has a default value because there are **field_definitions kwargs after it. Mypy was complaining and I had to add cls_kwargs={}, when calling it after updating from 1.9.0 to 1.9.1, so it wasn't backward compatible call and shouldn't have been a semver patch version release.

See how I was calling it

    model = pydantic.create_model(
        model_name,
        __config__=None,
        __base__=None,
        __module__=__name__,
        __validators__={},
        **{k: (v, ...) for k, v in attributes.items()},

and error from mypy was:

error: No overload variant of "create_model" matches argument types "str", "None", "None", "str", "Dict[<nothing>, <nothing>]", "Dict[str, Tuple[type, ellipsis]]"

[samuelcolvin]

please create a separate issue.

[pvicente]

@samuelcolvin #4106