CODEWITHSUNDEEP
pythonctypes
Sergey Sosnin
Sergey Sosnin

Reputation: 1413

ctypes float rubbish return

I've faced with a problem when I tried to load float into python program using ctypes

C code:

float test_ret_float(){
  return 1.0;
}

In Python, all ways rubbish resulted:

print lib.test_ret_float()
>>1074161254
print c_float(lib.test_ret_float()).value
>>1074161280.0

In case of int all works fine.

It seems, that type conversation doesn't work as expected and really returns raw 4byte value that converts not to float but int, why?

Upvotes: 2

Views: 3393

Answers (2)

Francisco Esteves
Francisco Esteves

Reputation: 21

I know it is way too late but to whoever searches this and might be looking to the same issue I had.

Usage: if you passed a pointer to a variable as and argument and the C function changes replaces it with a floating point, this way you can repackage the python interpretation (integer) value to a signaled 32bit floating point

def ieee754(value):
    """Repackages values to 32bit floating point"""
    packed_v = struct.pack('>L', value)
    return struct.unpack('>f', packed_v)[0]

Upvotes: 2

Sergey Sosnin
Sergey Sosnin

Reputation: 1413

The problem was that I missed restype to setup:

lib.test_ret_float.restype = c_float

That solves the problem

Upvotes: 5

Related Questions