（原）python使用ctypes調用C/C++接口

轉載請注明出處：

http://www.cnblogs.com/darkknightzh/p/6135514.html

參考網址：

https://docs.python.org/2/library/ctypes.html——ctypes的官方文檔

http://eli.thegreenplace.net/2008/08/31/ctypes-calling-cc-code-from-python/——提供了一個不涉及類的例子

http://stackoverflow.com/questions/145270/calling-c-c-from-python——建議使用ctypes，並提供了一個簡單的例子

http://stackoverflow.com/questions/7142169/pils-image-frombuffer-expected-data-length-when-using-ctypes-array——提供了HYRY直接使用c_ubyte進行處理的例子

import Image
from ctypes import c_ubyte, cast, POINTER

buf = (c_ubyte * 400)()
pbuf = cast(buf, POINTER(c_ubyte))
pbuf2 = cast(pbuf, POINTER(c_ubyte*400))

buf is an ubyte array, pbuf is a pointer to ubyte, pbuf2 is a pointer to ubyte[400]. img1 is created from buf directly, img2 is created from pubf2.contents.

http://www.linuxidc.com/Linux/2011-10/44838.htm——傳結構體的簡單例子

一 傳簡單的指針：

具體步驟：

1. 新建mathBuf.cpp：

#include <iostream>
#include "subBuf.h"

extern "C" 
{
    int addBuf(char* data, int num, char* outData);

    subBuf* subBuf_new(){ return new subBuf(); }
    int subBuf_sub(subBuf* subfuf, char* data, int num, char* outData){ subfuf->cursubBuf(data, num, outData); }
}

int addBuf(char* data, int num, char* outData)
{
    for (int i = 0; i < num; ++i)
    {
        outData[i] = data[i] + 3;
    }
    return num;
}

2. 新建subBuf.h：

#include <iostream>

class subBuf{
    public:
        subBuf(){}
        int cursubBuf(char* data, int num, char* outData)
        {
            for (int i = 0; i < num; ++i)
            {
                outData[i] = data[i] - 5;
            }
            return num;
        }
};

3. 終端中輸入如下命令，生成libmathBuf.so：

g++ -std=c++11 -shared -fPIC -o libmathBuf.so mathBuf.cpp

4. 新建test.py：

from ctypes import * # cdll, c_int
lib = cdll.LoadLibrary('libmathBuf.so')

class callsubBuf(object):
    def __init__(self):
        self.obj = lib.subBuf_new()

    def callcursubBuf(self, data, num, outData):
        lib.subBuf_sub(self.obj, data, num, outData)

callAddBuf = lib.addBuf

num = 4
numbytes = c_int(num)

data_in = (c_byte * num)()
for i in range(num):
    data_in[i] = i

print("initial input data buf:")
for i in range(num):
    print(data_in[i])

#pdata_in = cast(data_in, POINTER(c_ubyte))
#pdata_in2 = cast(pdata_in, POINTER(c_ubyte*num))

data_out = (c_byte * num)()

ret = callAddBuf(data_in, numbytes, data_out)

print("after call addBuf with C, output buf:")
for i in range(num):
    print(data_out[i])

f = callsubBuf()
f.callcursubBuf(data_in, numbytes, data_out)

print("after call cursubBuf with  C++ class, output buf:")
for i in range(num):
    print(data_out[i])

5. 運行test.py，輸出如下：

說明：

1) test.py如果使用c_byte，則對應C中的unsigned char。

2) 程序使用了2個文件，subBuf.h和mathBuf.cpp。實際上可以使用一個文件，但是class要在extern "C"的上面，否則即便聲明了class subBuf，也會提示invalid use of incomplete type ‘class subBuf’：

3) addSub函數的實現要在extern "C"的下面，否則會提示error: conflicting declaration of XXX with ‘C’ linkage：

4) ctypes中對應的c和python類型如下（具體參見ctypes的官方文檔）：

ctypes type	C type	Python type
c_bool	_Bool	bool (*)
c_char	char	1-character string
c_wchar	wchar_t	1-character unicode string
c_byte	char	int/long
c_ubyte	unsigned char	int/long
c_short	short	int/long
c_ushort	unsigned short	int/long
c_int	int	int/long
c_uint	unsigned int	int/long
c_long	long	int/long
c_ulong	unsigned long	int/long
c_longlong	__int64 or long long	int/long
c_ulonglong	unsigned __int64 or unsigned long long	int/long
c_float	float	float
c_double	double	float
c_longdouble	long double	float
c_char_p	char * (NUL terminated)	string or None
c_wchar_p	wchar_t * (NUL terminated)	unicode or None
c_void_p	void *	int/long or None

*. The constructor accepts any object with a truth value.

5) 除此之外：

http://blog.csdn.net/tobacco5648/article/details/41083369

提供了使用void*傳遞緩沖區的簡單說明。

 

二 傳結構體

1. 新建structPoint.cpp：

#include <stdio.h>
#include <stdlib.h>

struct structImg
{  
    int width;  
    int height;
    int channels; 
    char* buf;
};

extern "C" 
{
    void showStructureInfo(structImg p);
}

void showStructureInfo(structImg p) 
{  
    printf("%d %d %d\n", p.width, p.height, p.channels);
    for(int i=0;i< p.width*p.height*p.channels; ++i)
        printf("%d: %d\n", i, p.buf[i]);
}

2. 終端中將該文件編譯成.so庫：

g++ -std=c++11 -shared -fPIC -o libstructPoint.so structPoint.cpp

3. 新建test.py：

from ctypes import *
lib = cdll.LoadLibrary('libstructPoint.so')

class structImgTest(Structure):   
    _fields_ =[('width', c_int),  
               ('height', c_int),
               ('channels', c_int),
               ('buf', POINTER(c_ubyte))] 


def getstructImg(width, height, channels):
    #cwidth = c_int(width)
    #cheight = c_int(height)
    #cchannels = c_int(channels)
    num = width * height * channels
    data_buf = (c_byte * num)()
    for i in range(num):
        data_buf[i] = i

    pbuf = cast(data_buf, POINTER(c_ubyte))

    st = structImgTest(width, height, channels, pbuf)
    return st


width = 4
height = 3
channels = 2
st = getstructImg(width, height, channels)

callTest = lib.showStructureInfo
callTest(st)

print st.width   
print st.height 

# if declaration of test is test(structImg* p), then use the following line
pst = pointer(st)  # not sure if "POINTER" points to ctypes type, while "pointer" points to a variable

print pst.contents.width  
print pst.contents.height 

4. 終端中運行test.py，結果如下：

可見輸出的結果正確。

=========================================================================

190130更新：

下面兩個網址有使用ctypes的比較詳細的說明，也可以參考一下。

https://cvstuff.wordpress.com/2014/11/20/wraping-c-code-with-python-ctypes-the-python-side/

https://cvstuff.wordpress.com/2014/11/27/wraping-c-code-with-python-ctypes-memory-and-pointers/

190130更新結束

=========================================================================