利用pybind11进行C++与Python混合编程

1、pybind的安装

1.1、安装依赖

sudo apt-get install python-dev  (or python3-dev)
sudo apt-get install cmake
sudo pip install pytest
sudo pip install numpy
sudo pip install scipy

1.2、pip安装

 sudo pip install pybind11

1.3、源码安装

• 1、创建目录
• 2、获取pybind11源码包 https://github/pybind/pybind11
• 3、下载Eigen http://eigen.tuxfamily/index.php?title=Main_Page
• 4、在目录执行 mkdir build
• 5 、cd build
• 6、cmake ..
• 7、make check -j 4
• 8、 sudo make install

1.4、使用brew 安装

brew install pybind11

2、使用pybind进行C++代码编写

---

这里以ROS为列，获取Baxter机器人上方的图片，为了简洁方便看，这里只放头文件和cmake文件

---

2.1、头文件

• rosBase.hpp

#ifndef _ROS_H_
#define _ROS_H_
#include<ros/ros.h>
#include <std_msgs/Float64.h>
#include<string.h>
#include<robotstates_su/robotstates.h>
class rosBase{
    public:
    rosBase(const std::string node_name,int argc=0,char**argv=nullptr);
};
#endif

• baxterImage.hpp

#include<rosBase.hpp>
#include<image_transport/image_transport.h>
#include<cv_bridge/cv_bridge.h>
#include<sensor_msgs/image_encodings.h>
#include <opencv2/opencv.hpp>
#include "opencv2/core/core.hpp"
#include<opencv2/imgproc/imgproc.hpp>
#include<opencv2/highgui/highgui.hpp>
#include<opencv/cv.h>
#include<zeromqConnect.hpp>
#include<vector>
#include<Python.h>
#include<pybind11/embed.h> 
static const std::string OPENCV_WINDOW = "Open-CV display window";
#define headImage "/cameras/head_camera/image"
#define leftImage "/cameras/left_hand_camera/image"
#define rightImage "/cameras/right_hand_camera/image"

class baxterImage:public rosBase{
    public:
    baxterImage(const std::string nodeName, std::string IP="*",int Port=5556,int rate=1000);
    void getImage(const std::string cameraName);
    void processImageCallBack(const sensor_msgs::ImageConstPtr& msg);
    bool baxterOk();
    void baxterSleep();
    void baxterRate(int rate);
    void baxterSpinOnce();
    private:
        ros::NodeHandle nh;
        ros::Subscriber headSub;
        ros::Subscriber leftSub;
        ros::Subscriber rigthSub;
        void *publisher=NULL;
        int rate;
        const std::string Port;
        const std::string IP;
        rosBase rosbase;
};

2.2、下面是一个重要的文件 bindings.cpp

用于把C++代码和python代码进行绑定

#include<pybind11/pybind11.h>
#include<baxterImage.hpp>
#include<pybind11/stl.h>
#include<pybind11/eigen.h>
namespace py = pybind11;
PYBIND11_MODULE(baxter,m)
{
    py::class_<rosBase>rosbase(m,"rosBase");
    rosbase.def(py::init<const std::string&>());
    py::class_<baxterImage>(m,"baxterImage",rosbase)
        .def(py::init<const std::string &,const std::string&,int ,int>())
        .def("getImage",&baxterImage::getImage)
        .def("baxterOk",&baxterImage::baxterOk)
        .def("baxterSleep",&baxterImage::baxterSleep)
        .def("baxterSpinOnce",&baxterImage::baxterSpinOnce)
        .def("baxterRate",&baxterImage::baxterRate);
}

可以对着下面图像进行理解

2.3、Cmake文件

cmake_minimum_required(VERSION 2.8.3)
project(Ros_control)
## Compile as C++11, supported in ROS Kinetic and newer
add_compile_options(-std=c++11)
find_package(  OpenCV REQUIRED)
find_package(Eigen3 REQUIRED)
include_directories(
 include
  "/opt/ros/kinetic/include/"
  "/usr/include/"
  "/usr/include/boost/"
  "/usr/lib/python3.7/dist-packages/numpy/core/include/"
  "/usr/include/python3.7m/"
  "/headless/baxter_ws/devel/include/"
)
LINK_DIRECTORIES(
  "/opt/ros/kinetic/lib/"
  "/usr/lib/x86_64-linux-gnu/"
  "/opt/ros/kinetic/lib/x86_64-linux-gnu/"
  "/lib64/"
)
include_directories(SYSTEM ${EIGEN3_INCLUDE_DIRS})
add_subdirectory(
  lib/pybind11
)
pybind11_add_module(baxter "src/rosBase.cpp" "src/baxterImage.cpp"  "src/bindings.cpp")
target_link_libraries(
baxter PRIVATE   ${OpenCV_LIBS}
)

可以把pybind11 的lib库拷入到编译目录的lib库中(不是必须，能找到对应的库就行)
编译时会根据python版本生成baxter.cpython-37m-x86_64-linux-gnu.so文件

3、使用

把生成的so文件放到python文件使用的地方，或者让使用的文件能够找到此so文件

3.1、使用的案例代码

import baxter
#push = baxter.baxterImage("pushimage","*",6666,1000)
push = baxter.baxterImage("pushimage","192.168.1.105",6666,1000)
push.getImage("h")
while push.baxterOk():
    push.baxterSpinOnce()

参考学习