Reputation: 11
I face the following problem. In C I wrote the following:
#include <stdio.h>
typedef struct {
double *arr;
int length;
} str;
void f(str*);
int main (void){
double x[3] = {0.1,0.2,0.3};
str aa;
aa.length = 3;
aa.arr = x;
f(&aa);
return 0;
}
void f(str *ss){
int i;
printf("%d\n",ss->length);
for (i=0; i<ss->length; i++) {
printf("%e\n",ss->arr[i]);
}
}
If I compile it to an executable it works correctly. I receive:
3
0.1
0.2
0.3
as it should be. After building the shared library 'pointertostrucCtypes.so' from the C-code above I call the function f in python as follows:
ptrToDouble = ctypes.POINTER(ctypes.c_double)
class pystruc (ctypes.Structure):
_fields_=[
("length",ctypes.c_int),
("arr",ptrToDouble)
]
aa = pystruc()
aa.length = ctypes.c_int(4)
xx = numpy.arange(4,dtype=ctypes.c_double)
aa.arr = xx.ctypes.data_as(ptrToDouble)
myfunc = ctypes.CDLL('pointertostrucCtypes.so')
myfunc.f.argtypes = [ctypes.POINTER(pystruc)]
myfunc.f(ctypes.byref(aa))
It results always in printing out an arbitrary integer and afterwards it gives me a segmentation fault. because length does not fit. Has somebody an idea what I do wrong here ?
Upvotes: 1
Views: 1000
Reputation: 4285
Your fields are reversed. Try:
class pystruc (ctypes.Structure):
_fields_=[
("arr",ptrToDouble)
("length",ctypes.c_int),
]
Upvotes: 3