fix: do not ignore Config fields information in dataclass

[PrettyWood]

Change Summary

We used to keep only the default value and the type when creating a dataclass, which would then follow the default workflow. But now we can use metadata in dataclasses.field or directly a pydantic.Field instance. We hence ignore the information set in the Config

Related issue number

Closes #2426

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)

[codecov]

Codecov Report

Merging #2437 (ac21529) into master (a8d50ae) will not change coverage.
The diff coverage is 100.00%.

@@            Coverage Diff            @@
##            master     #2437   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files           25        25           
  Lines         5080      5085    +5     
  Branches      1041      1042    +1     
=========================================
+ Hits          5080      5085    +5     

Impacted Files	Coverage Δ
pydantic/dataclasses.py	100.00% <100.00%> (ø)
pydantic/main.py	100.00% <100.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update a8d50ae...ac21529. Read the comment docs.

[samuelcolvin]

Would this also fix #2424?

[PrettyWood]

Unfortunetaly no. #2424 seems trickier.

[samuelcolvin]

Humm, I think this could lead to unexpected behaviour - get_field_info is not designed to return kwargs for Field, I think it's not used for this anywhere else.

Perhaps we could add some tests of using get_field_info to create Fields and update the docstring?

[samuelcolvin]

But in theory this should all be done when creating the field as it is normally, maybe we should work out why it's not?

[PrettyWood]

This is really an issue of merging infos from config to an instance of Field. As a matter of fact it's the same with create_model

from pydantic import create_model, Field

class Config:
    fields = {'a': {'description': 'descr'}}

Ad = create_model('Ad', __config__=Config, a=(str, ...))
assert Ad.schema()['properties'] == {'a': {'title': 'A', 'description': 'descr', 'type': 'string'}}

Ad = create_model('Ad', __config__=Config, a=(str, Field(...)))
assert Ad.schema()['properties'] == {'a': {'title': 'A', 'type': 'string'}}

@samuelcolvin I would suggest we don't include this in 1.8.1 and consider this as a classic "bug".
We can then take time to refactor calmly some stuff if needed as you suggested

[samuelcolvin]

humm, I understand what you mean, but I think given that it's new since 1.8, we should try to fix it if possible.

I will look into it now.

[samuelcolvin]

Turns out the same thing even happens with a vanilla pydantic model:

from pydantic import Field, BaseModel

class Foo(BaseModel):
    a: str = Field(...)

    class Config:
        fields = {'a': {'description': 'descr'}}

debug(Foo.schema()['properties'])

I'll look into it now.

This is not new in 1.8, it was the same in v1.7.3. Still, should be fixed.

[samuelcolvin]

superseded by #2461