alibaba/yalantinglibs

A collection of modern C++ libraries, include coro_rpc, struct_pack, struct_json, struct_xml, struct_pb, easylog, async_simple

yaLanTingLibs

A Collection of C++20 libraries, include struct_pack, struct_json, struct_xml, struct_yaml, struct_pb, easylog, coro_rpc, coro_http and async_simple

中文版

yaLanTingLibs is a collection of modern c++ util libraries, now it contains struct_pack, struct_json, struct_xml, struct_yaml, struct_pb, easylog, coro_rpc, coro_io, coro_http and async_simple, more and more cool libraries will be added into yaLanTingLibs in the future.

The target of yaLanTingLibs: provide very easy and high performance modern C++ libraries for developers, it can help to quickly build high performance applications.

OS (Compiler Version)	Status
Ubuntu 22.04 (clang 14.0.0)
Ubuntu 22.04 (gcc 11.2.0)
macOS Monterey 12 (AppleClang 14.0.0.14000029)
Windows Server 2022 (MSVC 19.33.31630.0)

Quick Start

compiler requirements

If your compiler don't support C++20, yalantinglibs will only compile the serialization libraries (struct_pack, struct_json, struct_xml, struct_yaml, easylog support C++17). Make sure you have such compilers:

• g++9 above;
• clang++6 above
• msvc 14.20 above;

Otherwise, yalantinglibs will compile all the libraries. Make sure you have such compilers:

• g++10 above;
• clang++13 above
• msvc 14.29 above;

You can also use cmake option -DENABLE_CPP_20=ON or -DENABLE_CPP_20=OFF to control it.

Install & Compile

By Vcpkg

1. Install vcpkg
2. run ./vcpkg install yalantinglibs
3. If you use cmake, add those codes:

find_package(yalantinglibs CONFIG REQUIRED)
target_link_libraries(main PRIVATE yalantinglibs::yalantinglibs)

Manually Install

Yalantinglibs is a head-only library. You can just copy ./include/ylt directory into your project. But we suggest you use cmake to install it.

1. clone repo

git clone https://github.com/alibaba/yalantinglibs.git
cd yalantinglibs
mkdir build
cd build

2. build & test

• We suggest you compile the example and test the code first:

cmake ..
cmake --build . --config debug # add -j, if you have enough memory to parallel compile
ctest . # run tests

• Build in bazel:

bazel build ylt # Please make sure bazel in you bin path.
bazel build coro_http_example # Or replace in anyone you want to build and test.
# Actually you might take it in other project in prefix @com_alibaba_yalangtinglibs, like
bazel build @com_alibaba_yalangtinglibs://ylt

bazel version > 7
bazel build ylt --enable_bzlmod

You can see the test/example/benchmark executable file in ./build/output/.

• Or you can just skip build example/test/benchmark:

# You can use those option to skip build unit-test & benchmark & example: 
cmake .. -DBUILD_EXAMPLES=OFF -DBUILD_BENCHMARK=OFF -DBUILD_UNIT_TESTS=OFF

3. install

cmake --install . # --prefix ./user_defined_install_path 

4. start develop

• Use Cmake:

After install ylt, copy then open the directory src/*/examples, then:

mkdir build
cd build
cmake ..
cmake --build .

Cmake FetchContent

You can also import ylt by cmake FetchContent

cmake_minimum_required(VERSION 3.15)
project(ylt_test)

include(FetchContent)

FetchContent_Declare(
    yalantinglibs
    GIT_REPOSITORY https://github.com/alibaba/yalantinglibs.git
    GIT_TAG 0766d839fe52eb12ac7ecd34bc39a76399cfde41 # optional ( default master / main )
    GIT_SHALLOW 1 # optional ( --depth=1 )
)

FetchContent_MakeAvailable(yalantinglibs)
add_executable(main main.cpp)

target_link_libraries(main yalantinglibs::yalantinglibs)
target_compile_features(main PRIVATE cxx_std_20)

Compile Manually:

1. Add include/ directory to include path(skip it if you have install ylt into default include path).
2. Add include/ylt/thirdparty to include path(skip it if you have install ylt by cmake).
3. Add include/ylt/standalone to include path(skip it if you have install ylt by cmake).
4. Enable c++20 standard by option -std=c++20(g++/clang++) or /std:c++20(msvc)
5. If you use any header with coro_ prefix, add link option -pthread in linux, add option -fcoroutines when you use g++10.

More Details:

For more details, see the cmake file here and there.

Introduction

coro_rpc

Very easy-to-use, coroutine-based, high performance rpc framework with C++20, more than 0.4M QPS per thread in pipeline mode. coro_rpc is a header only library.

You can finish a rpc server and rpc client in 5 minutes!

English Introduction

[English API] (TODO)

Talk (Chinese) of coro_rpc on purecpp conference.

Video (Chinese) on purecpp conference, start from 04:55:08 of the video record.

quick example

1.define a rpc function as a local normal function.

// rpc_service.hpp
inline std::string_view echo(std::string_view str) { return str; }

2.register rpc function and start a server

#include "rpc_service.hpp"
#include <ylt/coro_rpc/coro_rpc_server.hpp>

int main() {
  coro_rpc_server server(/*thread_num =*/10, /*port =*/9000);
  server.register_handler<echo>(); // register function echo
  server.start(); // start the server & block
}

3.rpc client call rpc service

#include "rpc_service.hpp"
#include <ylt/coro_rpc/coro_rpc_client.hpp>

Lazy<void> test_client() {
  coro_rpc_client client;
  co_await client.connect("localhost", /*port =*/"9000"); // connect to the server

  auto r = co_await client.call<echo>("hello coro_rpc"); // call remote function echo
  std::cout << r.result.value() << "\n"; //will print "hello coro_rpc"
}

int main() {
  syncAwait(test_client());
}

More examples here.

struct_pack

Based on compile-time reflection, very easy to use, high performance serialization library, struct_pack is a header only library, it is used by coro_rpc now.

Only one line code to finish serialization and deserialization, 2-20x faster than protobuf.

quick example

struct person {
  int64_t id;
  std::string name;
  int age;
  double salary;
};

person person1{.id = 1, .name = "hello struct pack", .age = 20, .salary = 1024.42};

// one line code serialize
std::vector<char> buffer = struct_pack::serialize(person1);

// one line code deserialization
auto person2 = deserialize<person>(buffer);

struct_pack is very fast.

English Introduction

[English API] (TODO)

(Slides) A Faster Serialization Library Based on Compile-time Reflection and C++ 20 of struct_pack on CppCon2022

(Video) A Faster Serialization Library Based on Compile-time Reflection and C++ 20 on cppcon2022

(Slides)(Chinese) of struct_pack on purecpp conference.

(Video)(Chinese) on purecpp conference, start from 01:32:20 of the video record.

See more examples here.

struct_json

reflection-based json lib, very easy to do struct to json and json to struct.

quick example

#include "ylt/struct_json/json_reader.h"
#include "ylt/struct_json/json_writer.h"

struct person {
  std::string name;
  int age;
};
YLT_REFL(person, name, age);

int main() {
  person p{.name = "tom", .age = 20};
  std::string str;
  struct_json::to_json(p, str); // {"name":"tom","age":20}

  person p1;
  struct_json::from_json(p1, str);
}

struct_xml

reflection-based xml lib, very easy to do struct to xml and xml to struct.

quick example

#include "ylt/struct_xml/xml_reader.h"
#include "ylt/struct_xml/xml_writer.h"

struct person {
  std::string name;
  int age;
};
YLT_REFL(person, name, age);

void basic_usage() {
  std::string xml = R"(
<person>
    <name>tom</name>
    <age>20</age>
</person>
)";

  person p;
  bool r = struct_xml::from_xml(p, xml.data());
  assert(r);
  assert(p.name == "tom" && p.age == 20);

  std::string str;
  r = struct_xml::to_xml_pretty(p, str);
  assert(r);
  std::cout << str;
}

struct_yaml

reflection-based yaml lib, very easy to do struct to yaml and yaml to struct.

quick example

#include "ylt/struct_yaml/yaml_reader.h"
#include "ylt/struct_yaml/yaml_writer.h"

struct person {
  std::string name;
  int age;
};
YLT_REFL(person, name, age);

void basic_usage() {
    // serialization the structure to the string
    person p = {"admin", 20};
    std::string ss;
    struct_yaml::to_yaml(ss, p);
    std::cout << ss << std::endl;
    
    std::string yaml = R"(
    name : tom
    age : 30
    )";
    
    // deserialization the structure from the string
    struct_yaml::from_yaml(p, yaml);
}

coro_http

coro_http is a C++20 coroutine http(https) library, include server and client, functions: get/post, websocket, multipart file upload, chunked and ranges download etc. more examples

get/post

#include "ylt/coro_http/coro_http_server.hpp"
#include "ylt/coro_http/coro_http_client.hpp"
using namespace coro_http;

async_simple::coro::Lazy<void> basic_usage() {
  coro_http_server server(1, 9001);
  server.set_http_handler<GET>(
      "/get", [](coro_http_request &req, coro_http_response &resp) {
        resp.set_status_and_content(status_type::ok, "ok");
      });

  server.set_http_handler<GET>(
      "/coro",
      [](coro_http_request &req,
         coro_http_response &resp) -> async_simple::coro::Lazy<void> {
        resp.set_status_and_content(status_type::ok, "ok");
        co_return;
      });
  server.aync_start(); // aync_start() don't block, sync_start() will block.
  std::this_thread::sleep_for(300ms);  // wait for server start

  coro_http_client client{};
  auto result = co_await client.async_get("http://127.0.0.1:9001/get");
  assert(result.status == 200);
  assert(result.resp_body == "ok");
  for (auto [key, val] : result.resp_headers) {
    std::cout << key << ": " << val << "\n";
  }
}

async_simple::coro::Lazy<void> get_post(coro_http_client &client) {
  std::string uri = "http://www.example.com";
  auto result = co_await client.async_get(uri);
  std::cout << result.status << "\n";
  
  result = co_await client.async_post(uri, "hello", req_content_type::string);
  std::cout << result.status << "\n";
}

int main() {
  coro_http_client client{};
  async_simple::coro::syncAwait(get_post(client));
}

websocket

async_simple::coro::Lazy<void> websocket(coro_http_client &client) {
  // connect to your websocket server.
  bool r = co_await client.async_connect("ws://example.com/ws");
  if (!r) {
    co_return;
  }

  co_await client.write_websocket("hello websocket");
  auto data = co_await client.read_websocket();
  CHECK(data.resp_body == "hello websocket");
  co_await client.write_websocket("test again");
  data = co_await client.read_websocket();
  CHECK(data.resp_body == "test again");
  co_await client.write_websocket("ws close");
  data = co_await client.read_websocket();
  CHECK(data.net_err == asio::error::eof);
  CHECK(data.resp_body == "ws close");
}

upload/download

async_simple::coro::Lazy<void> upload_files(coro_http_client &client) {
  std::string uri = "http://example.com";
  
  client.add_str_part("hello", "world");
  client.add_str_part("key", "value");
  client.add_file_part("test", "test.jpg");
  auto result = co_await client.async_upload(uri);
  std::cout << result.status << "\n";
  
  result = co_await client.async_upload(uri, "test", "test.jpg");
}

async_simple::coro::Lazy<void> download_files(coro_http_client &client) {
  // chunked download
  auto result = co_await client.async_download("http://example.com/test.jpg",
                                               "myfile.jpg");
  std::cout << result.status << "\n";
  
  // ranges download
  result = co_await client.async_download("http://example.com/test.txt",
                                               "myfile.txt", "1-10,11-16");
  std::cout << result.status << "\n";
}

async_simple

A C++ 20 coroutine library offering simple, light-weight and easy-to-use components to write asynchronous codes. See async_simple

Details

CMAKE OPTION

config option

These option maybe useful for your project. You can enable it in your project if you import ylt by cmake fetchContent or find_package.

option	default value	description
YLT_ENABLE_SSL	OFF	enable optional ssl support for rpc/http
YLT_ENABLE_PMR	OFF	enable pmr optimize
YLT_ENABLE_IO_URING	OFF	enable io_uring in linux
YLT_ENABLE_FILE_IO_URING	OFF	enable file io_uring as backend in linux
YLT_ENABLE_STRUCT_PACK_UNPORTABLE_TYPE	OFF	enable unportable type(like wstring, int128_t) for struct_pack
YLT_ENABLE_STRUCT_PACK_OPTIMIZE	OFF	optimize struct_pack by radical template unwinding(will cost more compile time)

installation option

In default, yalantinglibs will install thirdparty librarys and standalone sublibrarires in your install path independently.

If you don't want to install the thirdparty librarys(you need install it manually), you can turn off cmake option -DINSTALL_THIRDPARTY=OFF.

If you want to install the thirdparty dependently. (install thirdparty librarys and standalone sublibrarires in ylt/thirdparty and ylt/standalone ), you can use turn off cmake option -DINSTALL_INDEPENDENT_THIRDPARTY=OFF and -DINSTALL_INDEPENDENT_STANDALONE=OFF.

option	default value
INSTALL_THIRDPARTY	ON
INSTALL_STANDALONE	ON
INSTALL_INDEPENDENT_THIRDPARTY	ON
INSTALL_INDEPENDENT_STANDALONE	ON

Those options only work in installation.

develop option

These CMake options is used for yalantinglibs developing/installing itself. They are not effected for your project, because ylt is a head-only.

option	default value
BUILD_EXAMPLES	ON
BUILD_BENCHMARK	ON
BUILD_UNIT_TESTS	ON
BUILD_*(BUILD_CORO_RPC, BUILD_STRUCT_PACK etc)	ON
COVERAGE_TEST	OFF
GENERATE_BENCHMARK_DATA	ON
CORO_RPC_USE_OTHER_RPC	ON

Thirdparty Dependency List

Here are the thirdparty libraries we used(Although async_simple is a part of ylt, it open source first, so we import it as a independence thirdparty library).

coro_io/coro_rpc/coro_http

Those dependency will by install by default. you can control it by cmake option.

• asio
• async_simple
• openssl (optional)

easylog

No dependency.

struct_pack, struct_json, struct_xml, struct_yaml

No dependency.

struct_pb

No dependency.

Standalone sublibraries

coro_http is implemented by a standalone sublibrary cinatra

struct_json、struct_xml、struct_yaml are implemented by a standalone sublibrary iguana

Benchmark

coro_rpc

options:

./benchmark_client # [threads = hardware counts] [client_pre_thread = 20] [pipeline_size = 1] [host = 127.0.0.1] [port = 9000] [test_data_path = ./test_data/echo_test] [test_seconds = 30] [warm_up_seconds = 5]

How to generate document

see Build Website

How to Contribute

1. Create an issue in the issue template.
2. Run tests and git-clang-format HEAD^ locally for the change.
3. Create a PR, fill in the PR template.
4. Choose one or more reviewers from contributors: (e.g., qicosmos, poor-circle, PikachuHyA).
5. Get approved and merged.

Discussion group

DingTalk group id: 645010455

License

yaLanTingLibs is distributed under the Apache License (Version 2.0) This product contains various third-party components under other open-source licenses. See the NOTICE file for more information.