cpp_python_socket

• Simple TCP/IP socket comunication wrapper between c++ and Python for IPC.
• Summarized as a small tool to boost working efficiency.
• Support multi-channel and 16-bit images, especially RGB-D pairs.
• Efficient for Real-time prediction based on PyTorch/Python with streaming device input built on C++ API.

General Information

• Branch master covers simple string communication, relies on standard libraries.
• Branch image_transferring adds image transferring capabilities, relies on:

1. OpenCV for c++: https://github.com/opencv/opencvBuild
2. OpenCV for Python: pip install opencv-python

Only tested in Ubuntu 16.04, write an issue should you find any error.

Quick Start

If intending to use C++ code, add this 3 things to your CMakeLists.txt:

• add_subdirectory(cpp_python_socket)
• Append cpp_python_socket/cpp/include to include_directories(...
• Append cpp_sockets to target_link_libraries(... of your library/executable.

Test the code

1. Change directory cd cpp_python_socket/

2. Build it: ./build.sh

3. Run unit test:

• Terminal 1: python python/server.py
• Terminal 2: cd cpp; ./run_cpp_client_test.sh

Usage examples

Python Server:

from cpp_python_socket.python.server import Server

if __name__ == "__main__":
  server = Server("127.0.0.1", 5001)
  message = server.receive()
  print("[CLIENT]:" + message)
  server.send("Shut up")

C++ client:

#include <iostream>
#include "client.hpp"

int main() {
    socket_communication::Client client("127.0.0.1", 5001);  // ip, port
    client.Send("Hello hello!");  // Send string
    std::string answer = client.Receive();  // Receive string
    std::cout << "[SERVER]: " << answer << std::endl;
}