Skip to content

feat: optimize runtime performance by reducing copy input attachment. #298

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Oct 30, 2025
Merged

feat: optimize runtime performance by reducing copy input attachment. #298

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
53 changes: 32 additions & 21 deletions xllm/api_service/api_service.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -95,9 +95,11 @@ void APIService::CompletionsHttp(::google::protobuf::RpcController* controller,
google::protobuf::Arena::CreateMessage<proto::CompletionResponse>(arena);

auto ctrl = reinterpret_cast<brpc::Controller*>(controller);
std::string attachment = std::move(ctrl->request_attachment().to_string());
std::string error;
auto st = json2pb::JsonToProtoMessage(attachment, req_pb, &error);
json2pb::Json2PbOptions options;
butil::IOBuf& buf = ctrl->request_attachment();
butil::IOBufAsZeroCopyInputStream iobuf_stream(buf);
auto st = json2pb::JsonToProtoMessage(&iobuf_stream, req_pb, options, &error);
if (!st) {
ctrl->SetFailed(error);
LOG(ERROR) << "parse json to proto failed: " << error;
Expand Down Expand Up @@ -127,16 +129,14 @@ void ChatCompletionsImpl(std::unique_ptr<Service>& service,
auto resp_pb =
google::protobuf::Arena::CreateMessage<typename ChatCall::ResType>(arena);

std::string attachment = std::move(ctrl->request_attachment().to_string());
std::string error;

google::protobuf::util::JsonParseOptions options;
options.ignore_unknown_fields = true;
auto json_status =
google::protobuf::util::JsonStringToMessage(attachment, req_pb, options);
if (!json_status.ok()) {
ctrl->SetFailed(json_status.ToString());
LOG(ERROR) << "parse json to proto failed: " << json_status.ToString();
json2pb::Json2PbOptions options;
butil::IOBuf& buf = ctrl->request_attachment();
butil::IOBufAsZeroCopyInputStream iobuf_stream(buf);
auto st = json2pb::JsonToProtoMessage(&iobuf_stream, req_pb, options, &error);
if (!st) {
ctrl->SetFailed(error);
LOG(ERROR) << "parse json to proto failed: " << buf.to_string();
return;
}

Expand Down Expand Up @@ -201,9 +201,11 @@ void APIService::EmbeddingsHttp(::google::protobuf::RpcController* controller,
google::protobuf::Arena::CreateMessage<proto::EmbeddingResponse>(arena);

auto ctrl = reinterpret_cast<brpc::Controller*>(controller);
std::string attachment = std::move(ctrl->request_attachment().to_string());
std::string error;
auto st = json2pb::JsonToProtoMessage(attachment, req_pb, &error);
json2pb::Json2PbOptions options;
butil::IOBuf& buf = ctrl->request_attachment();
butil::IOBufAsZeroCopyInputStream iobuf_stream(buf);
auto st = json2pb::JsonToProtoMessage(&iobuf_stream, req_pb, options, &error);
if (!st) {
ctrl->SetFailed(error);
LOG(ERROR) << "parse json to proto failed: " << error;
Expand Down Expand Up @@ -248,10 +250,13 @@ void APIService::ImageGenerationHttp(
auto resp_pb =
google::protobuf::Arena::CreateMessage<proto::ImageGenerationResponse>(
arena);

auto ctrl = reinterpret_cast<brpc::Controller*>(controller);
std::string attachment = std::move(ctrl->request_attachment().to_string());
std::string error;
auto st = json2pb::JsonToProtoMessage(attachment, req_pb, &error);
json2pb::Json2PbOptions options;
butil::IOBuf& buf = ctrl->request_attachment();
butil::IOBufAsZeroCopyInputStream iobuf_stream(buf);
auto st = json2pb::JsonToProtoMessage(&iobuf_stream, req_pb, options, &error);
if (!st) {
ctrl->SetFailed(error);
LOG(ERROR) << "parse json to proto failed: " << error;
Expand Down Expand Up @@ -290,9 +295,11 @@ void APIService::RerankHttp(::google::protobuf::RpcController* controller,
google::protobuf::Arena::CreateMessage<proto::RerankResponse>(arena);

auto ctrl = reinterpret_cast<brpc::Controller*>(controller);
std::string attachment = std::move(ctrl->request_attachment().to_string());
std::string error;
auto st = json2pb::JsonToProtoMessage(attachment, req_pb, &error);
json2pb::Json2PbOptions options;
butil::IOBuf& buf = ctrl->request_attachment();
butil::IOBufAsZeroCopyInputStream iobuf_stream(buf);
auto st = json2pb::JsonToProtoMessage(&iobuf_stream, req_pb, options, &error);
if (!st) {
ctrl->SetFailed(error);
LOG(ERROR) << "parse json to proto failed: " << error;
Expand Down Expand Up @@ -398,9 +405,11 @@ void APIService::LinkCluster(::google::protobuf::RpcController* controller,
google::protobuf::Arena::CreateMessage<proto::RpcStatus>(arena);

auto ctrl = reinterpret_cast<brpc::Controller*>(controller);
std::string attachment = std::move(ctrl->request_attachment().to_string());
std::string error;
auto st = json2pb::JsonToProtoMessage(attachment, req_pb, &error);
json2pb::Json2PbOptions options;
butil::IOBuf& buf = ctrl->request_attachment();
butil::IOBufAsZeroCopyInputStream iobuf_stream(buf);
auto st = json2pb::JsonToProtoMessage(&iobuf_stream, req_pb, options, &error);
if (!st) {
ctrl->SetFailed(error);
LOG(ERROR) << "parse json to proto failed: " << error;
Expand Down Expand Up @@ -452,9 +461,11 @@ void APIService::UnlinkCluster(::google::protobuf::RpcController* controller,
google::protobuf::Arena::CreateMessage<proto::RpcStatus>(arena);

auto ctrl = reinterpret_cast<brpc::Controller*>(controller);
std::string attachment = std::move(ctrl->request_attachment().to_string());
std::string error;
auto st = json2pb::JsonToProtoMessage(attachment, req_pb, &error);
json2pb::Json2PbOptions options;
butil::IOBuf& buf = ctrl->request_attachment();
butil::IOBufAsZeroCopyInputStream iobuf_stream(buf);
auto st = json2pb::JsonToProtoMessage(&iobuf_stream, req_pb, options, &error);
if (!st) {
ctrl->SetFailed(error);
LOG(ERROR) << "parse json to proto failed: " << error;
Expand Down