Using array.array in Python ctypes

Question

I have to interface a Python program with a C library. The particular function I need to call takes an array and returns a double. The following function has the same signature and is easier to understand than my own:

double sum(double * array, const int length) {
 double total = 0;
 int i;
 for (i=0; i<length; i++) {
     total += array[i];
 }
 return total;
}

My current solution is:

import ctypes
lib = ctypes.CDLL(library_name) 

l = 10
arr = tuple(range(l))

lib.sum.restype = ctypes.c_double
values = (ctypes.c_double * l)(*arr)
ret = lib.sum(values, l)

But I use the array module a lot in my code and it seemed to me that using them with the C code should be more straight forward as it is a typed array. So I tried to feed the C function directly with an array but it did not work. To make it work, I wrapped array like this:

class Array(array):

   @property
   def _as_parameter_(self):
      return (TYPES[self.typecode] * len(self))(*self)

where TYPES map the typecode from the array to ctypes types:

   TYPES = {'c': ctypes.c_char,
            'b': ctypes.c_byte,
            'B': ctypes.c_ubyte,
            '?': ctypes.c_bool,
            'h': ctypes.c_short,
            'H': ctypes.c_ushort,
            'i': ctypes.c_int,
            'I': ctypes.c_uint,
            'l': ctypes.c_long,
            'L': ctypes.c_ulong,
            'q': ctypes.c_longlong,
            'Q': ctypes.c_ulonglong,
            'f': ctypes.c_float,
            'd': ctypes.c_double}

Is there a way to replace _as_parameter_ by something that does not create another array?

Thanks

HYRY · Accepted Answer · 2012-05-17 00:56:00Z

4

use array.buffer_info() to get the address and length, and cast() the address to POINTER(c_double):

from array import array
buf = array("d", range(101))
addr, count = buf.buffer_info()
print lib.sum(cast(addr, POINTER(c_double)), count)

answered May 17, 2012 at 0:56

HYRY

97.8k28 gold badges197 silver badges192 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

user1767754 Over a year ago

where is cast defined?

Brian Larsen · Accepted Answer · 2012-06-20 21:37:45Z

0

Not that this is a better answer but for completeness if you are using numpy it is just a bit different as numpy can handle the cast part for you.

import numpy as np
data = np.arange(101, dtype=ctypes.c_double) # making this match seems important sometimes
print lib.sum(data.ctypes.data_as(ctypes.POINTER(ctypes.c_double)), len(data))

edited Jun 20, 2012 at 21:37

answered Jun 5, 2012 at 1:53

Brian Larsen

1,76617 silver badges28 bronze badges

Collectives™ on Stack Overflow

Using array.array in Python ctypes

2 Answers 2

1 Comment

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

1 Comment

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related