vllm.entrypoints.openai.engine.protocol ¶

GenerationError ¶

Bases: Exception

raised when finish_reason indicates internal server error (500)

Source code in vllm/entrypoints/openai/engine/protocol.py

class GenerationError(Exception):
    """raised when finish_reason indicates internal server error (500)"""

    def __init__(self, message: str = "Internal server error"):
        super().__init__(message)
        self.status_code = HTTPStatus.INTERNAL_SERVER_ERROR

validate_structural_tag_response_format ¶

validate_structural_tag_response_format(
    response_format: AnyStructuralTagResponseFormat
    | dict[str, Any],
) -> None

Validate structural tags before they are sent to the engine.

Engine-side validation reports malformed structural tags as generation failures. OpenAI request parsing should classify them as bad requests.

Source code in vllm/entrypoints/openai/engine/protocol.py

def validate_structural_tag_response_format(
    response_format: AnyStructuralTagResponseFormat | dict[str, Any],
) -> None:
    """Validate structural tags before they are sent to the engine.

    Engine-side validation reports malformed structural tags as generation
    failures. OpenAI request parsing should classify them as bad requests.
    """
    import json

    from pydantic import TypeAdapter, ValidationError

    if isinstance(response_format, dict):
        try:
            response_format = TypeAdapter(
                AnyStructuralTagResponseFormat
            ).validate_python(response_format)
        except ValidationError as exc:
            raise VLLMValidationError(
                "Invalid response_format structural_tag specification.",
                parameter="response_format",
            ) from exc

    try:
        payload = json.dumps(response_format.model_dump(by_alias=True))
        validate_structural_tag_payload(payload, parameter="response_format")
    except (TypeError, ValueError) as exc:
        raise VLLMValidationError(
            "Invalid response_format structural_tag specification.",
            parameter="response_format",
        ) from exc