bug: `Annotated` causes fields to not show up with description

[maxschulz-COL]

Description of the bug

When using Annotated, pydantic fields do not show up as attributes anymore, have no hyperlink, and the field description is lost

To Reproduce

 class ExampleModel(BaseModel):
     """An example model."""
 
     field_with_constraints_and_description: Annotated[
         int, Field(default=5, ge=0, le=100, description="Shows constraints within doc string.")
     ]

Full traceback

Expected behavior

I would expect them to show up as attributes with their description.

[maxschulz-COL]

P.S I also think the contraints do not show up?

[krishanunayak]

In my case, although it was working earlier, after few forced library updates, suddenly the class itself is also not recognized as a pydantic model. Hence none of fields, config, validators, or even json schema are showing up. My (partial) setup is

griffe==1.12.1
griffe-pydantic==1.1.6
mkdocs==1.6.1
mkdocs-autorefs==1.4.2
mkdocs-get-deps==0.2.0
mkdocs-material==9.6.17
mkdocs-material-extensions==1.3.1
mkdocstrings==0.30.0
mkdocstrings-python==1.17.0
pydantic==2.11.7
pydantic_core==2.33.2
pymdown-extensions==10.16.1

Not sure which versions I was using earlier, as I only fixed the minor versions in pyproject.toml.

[krishanunayak]

While running mkdocs serve -s -v I get the following two warnings
griffe: Base class pydantic.BaseModel is not loaded, or not static, it cannot be resolved
griffe: Base class sqlalchemy.orm.DeclarativeBase is not loaded, or not static, it cannot be resolved

[pawamoy]

We don't actually support Annotated yet. Marking as feature 🙂

[maxschulz-COL]

We don't actually support Annotated yet. Marking as feature 🙂

I know this is probably a lot of work, but would there be any ETA possible? This is pretty much the thing that stops us from converting to mkdocsstrings over at Vizro :)

[pawamoy]

One man work, no ETA 😅 It shouldn't be that hard, we have all the necessary API to identify an ...: Annotated[..., Field(...)] instead of ...: int = Field(...), both during static and dynamic analysis.

[maxschulz-COL]

Haha I see 😄 makes totally sense. Asked slightly differently then for planning: is there an idea if this would be tackled in the next say 2-4 weeks?

[pawamoy]

I'd say probably not 🤷 Happy to review PRs though!