Structured Output fails with enum field in pydantic schema.

[MeerrettichMeister]

#54 enabled the expected support for pydantic schemas for structured output.

However the feature works just partially since when including an enum field in the pydantic class the request fails due to unrecognized/unsupported fields in the request body to the oci endpoint.

oci_chat = ChatOCIGenAI(
    model_id="google.gemini-2.5-flash",
    service_endpoint="https://inference.generativeai.us-chicago-1.oci.oraclecloud.com",
    compartment_id=...,
)


class Domain(Enum):
    NATURE = "NATURE"
    TECHNICAL = "TECHNICAL"
    HISTORICAL = "HISTORICAL"


class Formatted(BaseModel):
    answer: str
    user_request: str
    domain: Domain


structured = oci_chat.with_structured_output(Formatted, method="json_schema")
structured_response = structured.invoke(
    [HumanMessage("Tell me something about the statue of liberty")]
)

**print(structured_response)**

Running the provided code returns the endpoint error.

Pydantic generates a json schema where the enum is placed behind a $def field and the attributes then just reference the field with $ref
It seems that the generative ai endpoint is not able to parse that schema.

For the sake of future proving I would advise to test the pydantic parsing with more complex schemas i.e. nested schemas, enum types, literalt types etc.