Method custom_route_handler

llama_cpp/server/errors.py:170–210

(request: Request)

        original_route_handler = super().get_route_handler()

        async def custom_route_handler(request: Request) -> Response:
            try:
                start_sec = time.perf_counter()
                response = await original_route_handler(request)
                elapsed_time_ms = int((time.perf_counter() - start_sec) * 1000)
                response.headers["openai-processing-ms"] = f"{elapsed_time_ms}"
                return response
            except HTTPException as unauthorized:
                # api key check failed
                raise unauthorized
            except Exception as exc:
                json_body = await request.json()
                try:
                    if "messages" in json_body:
                        # Chat completion
                        body: Optional[
                            Union[
                                CreateChatCompletionRequest,
                                CreateCompletionRequest,
                                CreateEmbeddingRequest,
                            ]
                        ] = CreateChatCompletionRequest(**json_body)
                    elif "prompt" in json_body:
                        # Text completion
                        body = CreateCompletionRequest(**json_body)
                    else:
                        # Embedding
                        body = CreateEmbeddingRequest(**json_body)
                except Exception:
                    # Invalid request body
                    body = None

                # Get proper error message from the exception
                (
                    status_code,
                    error_message,
                ) = self.error_message_wrapper(error=exc, body=body)
                return JSONResponse(
                    {"error": error_message},
                    status_code=status_code,
                )

        return custom_route_handler