VertexModel incompatibility with FastAPI's endpoint function definitions

[eHorn96]

Hi there,

Thanks for the library, first and foremost. It makes developing the business logic with ArangoDB easy af.

An issue I encountered is the incompatibility with FastAPI's endpoint function definitions.

async def create_process(
    current_user: Annotated[User, Depends(get_current_user)],
    process_data: Annotated[Process, Body(...)]): # This fails

Serializing to json just makes it look super ugly in swagger.

I forked your repo to help you on this tho. Surely there's a way to fix this.

Cheers!

Erik

[nadobando]

Hi Erik, thanks for your appreciation,
can you share the part of the swagger/openapi so I can reproduce it?

I'm creating another library FuriousAPI which is an extension for FastAPI
and there are submodules for working with:

1. SQLModel FuriousAPI-SQLModel - ✅ done
2. Beanie/mongo FuriousAPI-Beanie - ✅ done
3. PydangORM - FuriousAPI-pydango - 🔁 need to create the implementation

Regards, @nadobando

[eHorn96]

Hi @nadobando , I am currently a little busy right now, so I won't have time for the next few weeks I'm afraid. Sorry for the unclear description of the issue. Usually in FastAPI you can declare the class as-is in the signature and its model will be displayed in the Swagger docs. The Process class above has been declared as a subclass according to your documentation.

class Process(VertexModel):
    name: str | None = None
    process_type: str | None = None
    status: Literal["Waiting", "Running", "Done", "Failed"] | None = None
    uses: Annotated[list[NodeAttribute], Relation[Uses]] = "test"

    class Collection(VertexCollectionConfig):
        name = "process"
        indexes = [
            PersistentIndex(fields=["name"], unique=True),
        ] 

However I tried to make it work, it just didn't. Process.model_dump() won't work, and (sorry I worded it wrongly) Process.model_schema_json() resulted in the whole serialized JSON to be displayed in Swagger. Hope it helps.

Cheers!

[nadobando-vinesight]

@eHorn96 I will try to look at this, I want to start working on the upgrade of pydantic, which will also add circular graphs feature