使用ctypes在python中调用c函数

Question

I'm trying to call c functions from python, I have the following code in c.

struct _returndata
{
    double* data;
    int row;
    int col;
};

int mlfAddmatrixW(struct _returndata* retArr)
{
    double data[] = {1,2,3,4,5,6,7,8,9} 
    retArr->row = 3;
    retArr->col = 3;        
    memcpy(retArr->data, data, 9*sizeof(double));            
    return  1;
}

This is my code in python:

class RETARRAY(Structure):
    _fields_= [("data", c_double*9),
              ("row", c_int),
            ("col", c_int)]

if __name__ == '__main__':      

    dll = CDLL("/home/robu/Documents/tmo_compile/libmatrix/distrib/libmatrixwrapper.so")    

    #Initializing the matrix 
    retArr = pointer(RETARRAY())

    for i in retArr.contents.data:
        print i;

    dll.mlfAddmatrixW(pointer(retArr))
    for i in retArr.contents.data:
        print i;

    print retArr.contents.row
    print retArr.contents.col

The content of the data has changed, but the col and row is still 0. How can I fix that? Is it possible to create a dynamic array in python , because in this case I created an array with 9 elements ("data", c_double*9), . I know the size of the array after I called mlfAddmatrixW function, the size of the array will be col*row .

Answer 1

You have a different struct in C and Python: one has a pointer to double, the other an array of doubles. Try something like:

NineDoubles = c_double * 9

class RETARRAY(Structure):
    _fields_= [("data", POINTER(c_double)),
              ("row", c_int),
              ("col", c_int)]

#Initializing the matrix 
data = NineDoubles()
retArr = RETARRAY()
retArr.data = data

dll.mlfAddmatrixW(pointer(retArr))