[C++/Python] 如何在C++中使用一個Python類? (Use Python-defined class in C++)

最近在做基於OpenCV的車牌識別, 其中需要用到深度學習的一些代碼(Python), 所以一開始的時候開發語言選擇了Python(禍患之源).

固然現在Python的速度不算太慢, 但你一定要用Python來操作圖像, 實現某些算法的時候, 效率就變得非常重要. 可惜的是, Python在大多數算法實現中, 由於其循環操作實在是太慢, 導致實現的算法效率非常之低.

所以現在我要把深度學習中的一個類(分類器)轉換到C++中, 在這個過程之前, 需要做一些test projects, 我主要參照的文章是: C++調用Python(3).

開發環境為 VS2012, WIN7 64.

所有代碼都在最后放出, 這里先逐步講解.

首先我們需要導入合適的頭文件: Python.h, 它位於你的Python安裝目錄的include目錄下. (2.6, 2.7版本實測通過)

使用VS的話, 可以按如下方法設置一下項目屬性:

1. 調試 -> xxx屬性(最后一欄) -> VC++目錄 -> 右側庫目錄
2. 增加C:\Python27\libs, 具體路徑個人可能不同.

這樣就完成了最基本的配置, 下面我們開始導入.py文件並使用它.

我們使用的.py文件非常簡單, 代碼如下:

#!/usr/bin/python
# Filename: testpy.py
class Person:  4     def sayHi(self):  5         print 'hi'
class Second:  7     def invoke(self,obj):  8  obj.sayHi()  9 def sayhi(name): 10     print 'hi',name;

注: 下述所有導入方法在導入失敗時不會報錯, 只會返回空指針.

第一步是導入.py文件:

1. 使用PyObject* pModule來存儲導入的.py文件模塊, 調用的方法是PyImport_ImportModule(path):  PyObject* pModule = PyImport_ImportModule("testpy"); 
2. 使用PyObject* pDict來存儲導入模塊中的方法字典, 調用的方法是PyModule_GetDict(module):  PyObject* pDict = PyModule_GetDict(pModule); 

這樣就完成了.py文件的導入.

第二步是導入已導入模塊中的方法或類:

1. 獲取方法, 調用的方法是PyDict_GetItemString(dict, methodName): PyObject* pFunHi = PyDict_GetItemString(pDict, "sayhi"); 
2. 獲取類, 調用的方法同上, 注意紅體部分的字符串對應於.py文件中的類/方法名:  PyObject* pClassSecond = PyDict_GetItemString(pDict, "Second"); 

第三步是使用導入的方法或類:

1. 使用方法, 調用PyObject_CallFunction(pFunc, "s", args)即可:  PyObject_CallFunction(pFunHi, "s", "lhb"); 
2. 使用類構造對象, 調用PyInstance_New(pClass, NULL, NULL)即可:  PyObject* pInstanceSecond = PyInstance_New(pClassSecond, NULL, NULL); , 注意其中的pClassSecond為第二步.2中獲取的類指針
3. 使用類對象的方法, 調用PyObject_CallMethod(pInstance, methodname, "O", args)即可:  PyObject_CallMethod(pInstanceSecond, "invoke", "O", pInstancePerson); 
4. 上述調用中的"s"和"O"代表的是參數列表的類型, 我們可以在 Py_BuildValue 找到所有的類型, 本文最后也附了此表.

最后不要忘記銷毀這些對象:  Py_DECREF(pointer); 

下面是C++的實現代碼, 代碼來自於我參考的博客, 略有修改.

 

/*
 * test.cpp  3  * Created on: 2010-8-12  4  * Author: lihaibo  5  */
#include <C:/Python27/include/Python.h>
#include <iostream>
#include <string>

int main(void) { 11     Py_Initialize(); // 啟動虛擬機
    if (!Py_IsInitialized()) 13         return -1; 14     // 導入模塊
    PyObject* pModule = PyImport_ImportModule("testpy"); 16     if (!pModule) { 17         printf("Cant open python file!/n"); 18         return -1; 19  } 20     // 模塊的字典列表
    PyObject* pDict = PyModule_GetDict(pModule); 22     if (!pDict) { 23         printf("Cant find dictionary./n"); 24         return -1; 25  } 26     // 演示函數調用
    PyObject* pFunHi = PyDict_GetItemString(pDict, "sayhi"); 28     PyObject_CallFunction(pFunHi, "s", "lhb"); 29  Py_DECREF(pFunHi); 30     // 演示構造一個Python對象，並調用Class的方法 31     // 獲取Second類
    PyObject* pClassSecond = PyDict_GetItemString(pDict, "Second"); 33     if (!pClassSecond) { 34         printf("Cant find second class./n"); 35         return -1; 36  } 37     //獲取Person類
    PyObject* pClassPerson = PyDict_GetItemString(pDict, "Person"); 39     if (!pClassPerson) { 40         printf("Cant find person class./n"); 41         return -1; 42  } 43     //構造Second的實例
    PyObject* pInstanceSecond = PyInstance_New(pClassSecond, NULL, NULL); 45     if (!pInstanceSecond) { 46         printf("Cant create second instance./n"); 47         return -1; 48  } 49     //構造Person的實例
    PyObject* pInstancePerson = PyInstance_New(pClassPerson, NULL, NULL); 51     if (!pInstancePerson) { 52         printf("Cant find person instance./n"); 53         return -1; 54  } 55     //把person實例傳入second的invoke方法
    PyObject_CallMethod(pInstanceSecond, "invoke", "O", pInstancePerson); 57     //釋放
 Py_DECREF(pInstanceSecond); 59  Py_DECREF(pInstancePerson); 60  Py_DECREF(pClassSecond); 61  Py_DECREF(pClassPerson); 62  Py_DECREF(pModule); 63     Py_Finalize(); // 關閉虛擬機
    return 0; 65 }

 類型參照:

s (string) [char *]

Convert a null-terminated C string to a Python object. If the C string pointer is NULL, None is used.

s# (string) [char *, int]

Convert a C string and its length to a Python object. If the C string pointer is NULL, the length is ignored and None is returned.

z (string or None) [char *]

Same as s.

z# (string or None) [char *, int]

Same as s#.

u (Unicode string) [Py_UNICODE *]

Convert a null-terminated buffer of Unicode (UCS-2 or UCS-4) data to a Python Unicode object. If the Unicode buffer pointer is NULL, Noneis returned.

u# (Unicode string) [Py_UNICODE *, int]

Convert a Unicode (UCS-2 or UCS-4) data buffer and its length to a Python Unicode object. If the Unicode buffer pointer is NULL, the length is ignored and None is returned.

i (integer) [int]

Convert a plain C int to a Python integer object.

b (integer) [char]

Convert a plain C char to a Python integer object.

h (integer) [short int]

Convert a plain C short int to a Python integer object.

l (integer) [long int]

Convert a C long int to a Python integer object.

B (integer) [unsigned char]

Convert a C unsigned char to a Python integer object.

H (integer) [unsigned short int]

Convert a C unsigned short int to a Python integer object.

I (integer/long) [unsigned int]

Convert a C unsigned int to a Python integer object or a Python long integer object, if it is larger than sys.maxint.

k (integer/long) [unsigned long]

Convert a C unsigned long to a Python integer object or a Python long integer object, if it is larger than sys.maxint.

L (long) [PY_LONG_LONG]

Convert a C long long to a Python long integer object. Only available on platforms that support long long.

K (long) [unsigned PY_LONG_LONG]

Convert a C unsigned long long to a Python long integer object. Only available on platforms that support unsigned long long.

n (int) [Py_ssize_t]

Convert a C Py_ssize_t to a Python integer or long integer.

New in version 2.5.

c (string of length 1) [char]

Convert a C int representing a character to a Python string of length 1.

d (float) [double]

Convert a C double to a Python floating point number.

f (float) [float]

Same as d.

D (complex) [Py_complex *]

Convert a C Py_complex structure to a Python complex number.

O (object) [PyObject *]

Pass a Python object untouched (except for its reference count, which is incremented by one). If the object passed in is a NULL pointer, it is assumed that this was caused because the call producing the argument found an error and set an exception. Therefore, Py_BuildValue()will return NULL but won’t raise an exception. If no exception has been raised yet, SystemError is set.

S (object) [PyObject *]

Same as O.

N (object) [PyObject *]

Same as O, except it doesn’t increment the reference count on the object. Useful when the object is created by a call to an object constructor in the argument list.

O& (object) [converter, anything]

Convert anything to a Python object through a converter function. The function is called with anything (which should be compatible withvoid *) as its argument and should return a “new” Python object, or NULL if an error occurred.

(items) (tuple) [matching-items]

Convert a sequence of C values to a Python tuple with the same number of items.

[items] (list) [matching-items]

Convert a sequence of C values to a Python list with the same number of items.

{items} (dictionary) [matching-items]

Convert a sequence of C values to a Python dictionary. Each pair of consecutive C values adds one item to the dictionary, serving as key and value, respectively.

If there is an error in the format string, the SystemError exception is set and NULL returned.