Detailed explanation of how Python calls C/C++ underlying libraries and transfers values ​​to each other

As a script interpretation language, Python itself is well integrated with C++, so using Python to develop and call the C/C++ underlying library where performance requirements are required is simply an artifact. This article introduces in detail the problem of Python calling C/C++ underlying libraries and passing values ​​to each other. Let’s take a look.

Preface

Development environment:

Centos 7 + Python 3.5.1 + Qt Creator (just compiled with Qt Creator, and does not use any libraries of QT)

Python calls C/C++ libraries, there are two ways I can do it now

1.extern "C" export (It is troublesome to pass values ​​to each other, this method is not recommended):

Make the C/C++ library the same DLL or .so as usual, for example:

//.h文件
#include <Python.h>
//.cpp文件
//C/C++ my.so 或者my.dll
enter "C" void printHello()
{
  std::cout<<"Hello World"<<std::endl;
}

#Python
import ctypes 
from ctypes import *
loadso = ctypes.cdll.LoadLibrary 
mylib = loadso("./my.so")
mylib.printHello()
>>>Hello world

Code explanation:

my.so There is a C export functionprintHello()

import ctypes : Import an official library, as the name suggests, it is related to C

loadso = ctypes.cdll.LoadLibrary : loadso represents the function used to load the library

mylib = loadso(“./my.so”)　 //Or loadso(“my.dll”) Load my.so library

mylib.printHello(): Call Library function

The above code can output normally: Hello World, but they do not pass values ​​to each other

Python and C++ pass values ​​to each other

//.h文件
#include <Python.h>
//.cpp文件
enter "C" int printHello(const char* str)
{
  std::cout<<str<<std::endl;
  return 1;  
}

Then the problem with Python comes

str = create_string_buffer(b"Hello World")
#mylib.printHello("Hello World") 这里死活就是显示：H，*(str+4)才是'e',*(str+8) 是'l' 依次类推
print (mylib.printHello(str))
>>>Hello World
>>>1
#由于对Python不是特别的熟悉 怎么也做不到显示C++返回的字符串， Python只能显示C++返回的字符串子能看到一个地址而已

2.Python extension C/C++

Not much to say, just go to the code

//.h文件 本来这是C++连接Mysql 我只摘抄部分代#include <Python.h>
//.cpp文件
//传递多个参数 Python传过来的参数在args里面
PyObject* printfHello(PyObject* self,PyObject* args)
{
  int i=0
   const char* str;
  if (!PyArg_ParseTuple(args, "i|s", &i,&str))   //i 表示整形 s 表示字符串
    return PyLong_FromLong(0);
  print("%d,%s",i,str);
  return Py_BuildValue("s","OK");  //向Python返回OK字符串
}
//映射 知道MFC的一看就懂
static PyMethodDef MyMethods[] = {
{"printfHello", printfHello, METH_VARARGS,  //"printHello" 中可调用的函数 METH_VARARGS ：带有参数   METH_NOARGS：无参数
"print"},   //说明
{"connect", connect, METH_VARARGS,
"connect mysql"},
{NULL, NULL, 0, NULL}
};
static PyObject* UtilError;
// 向Python中注册模块
static struct PyModuleDef spammodule = { 
PyModuleDef_HEAD_INIT,
"libMysqlUtil", //模块名字 import libMysqlUtil
"C++ Connect Mysql",
-1,
MyMethods
};//PyInit_libMysqlUtil 注意名字 一定要PyInit_ 加上你的模块名字 不然Python import 会提示没有定义 PyInit_你的模块名字 PyMODINIT_FUNC PyInit_libMysqlUtil(void) { PyObject* m = nullptr; m = PyModule_Create(&spammodule);
//m= Py_InitModule(....) Python 2.7 if(!m) { return m; } UtilError = PyErr_NewException("Util.error",NULL,NULL); Py_INCREF(UtilError); PyModule_AddObject(m,"error",UtilError); return m; }

#python
import libMysqlUtil
libMysqlUtil.printHello(1,"hello World")
>>>1,hello World
>>>OK

Summary

So far, Python and C/C++ communicate with each other and can meet most needs. Structure value transfer has not been studied yet. For classes, just use pointers. , there is a pointer in C++, and the pointer is converted into an integer in Python. When Python passes this integer to C++, it uses PyArg_ParseTuple to turn the integer into a class pointer.

For more detailed explanations of how Python calls C/C++ underlying libraries and transfers values ​​to each other, please pay attention to the PHP Chinese website!