Function create_response

examples/server/server.py:16087–16171

(  # pyright: ignore[reportUnusedFunction]
        http_request: Request,
        body: CreateResponseRequest,
    )

    @app.post("/v1/responses")
    async def create_response(  # pyright: ignore[reportUnusedFunction]
        http_request: Request,
        body: CreateResponseRequest,
    ):
        service: CompletionService = app.state.service
        formatter = service.formatter
        try:
            chat_parts = formatter.response_request_to_chat_parts(body)
            parts = formatter.completion_request_from_response_chat_parts(chat_parts)
        except CompletionRequestValidationError as exc:
            raise bad_request(exc) from exc
        response_tools = chat_parts.tools
        try:
            request = service.request_from_prepared(
                payload=parts.payload,
                prompt_text=parts.prompt_text,
                prompt_plan=parts.prompt_plan,
                grammar_text=parts.grammar_text,
                chat_tool_name=parts.tool_name,
            )
            stream, cancel = service.submit_request(request)
        except CompletionRequestValidationError as exc:
            raise bad_request(exc) from exc
        if body.stream:
            started_indices: set[int] = set()
            parsed_states: Dict[int, ResponseParser] = {}
            stream_state = OpenAIFormatter.ResponsesStream(
                body=body,
                response_id="resp_" + request.id,
                created_at=float(request.created),
                model=service.scheduler.model.model_path,
            )

            def response_chunk_payloads(
                completion_chunk: CompletionChunk,
            ) -> Iterable[BaseModel | Dict[str, Any]]:
                chat_chunks = formatter.convert_completion_chunk_to_chat_chunks(
                    completion_chunk,
                    started_indices,
                    parts.tool_name,
                    tools=response_tools,
                    parsed_states=parsed_states,
                    generation_prompt=parts.generation_prompt,
                )
                payloads: List[BaseModel | Dict[str, Any]] = []
                for chat_chunk in chat_chunks:
                    payloads.extend(
                        formatter.convert_chat_chunk_to_response_events(
                            chat_chunk,
                            stream_state,
                        )
                    )
                return payloads

            return StreamingResponse(
                stream_sse_outputs(
                    formatter,
                    http_request,