Introduction to Python ctypes

Python ctypes export the cdll and on Windows windll or respectively oledll objects, to load the requested dynamic link libraries. A dynamically linked library is a compiled binary that is linked at runtime to the executable main process. On Windows platforms, these binaries are called Dynamic Link Libraries (DLL) and on Linux, they are called shared objects (SO). You can load these linked libraries by accessing them as the attributes of the cdll, windll or oledll objects. Now, we will demonstrate a very brief example for Windows and Linux to get the current time directly out of the time function in libc (this library defines the system calls and other basic facilities such as open, printf, or exit).

Note that in the case of Windows, msvcrt is the MS standard C library containing most of the standard C functions and uses the cdecl calling convention (on Linux systems, the similar library would be libc.so.6):

C:\Users\Admin>python

>>> from ctypes import *
>>> libc = cdll.msvcrt
>>> print libc.time(None)
1428180920

Windows appends the usual .dll file suffix automatically. On Linux, it is required to specify the filename, including the extension, to load the chosen library. Either the LoadLibrary() method of the DLL loaders should be used or you should load the library by creating an instance of CDLL by calling the constructor, as shown in the following code:

(labenv)user@lab:~$ python

>>> from ctypes import *
>>> libc = CDLL("libc.so.6")
>>> print libc.time(None)
1428180920

As shown in these two examples, it is very easy to be able to call to a dynamic library and use a function that is exported. You will be using this technique many times throughout the book, so it is important that you understand how it works.

When looking at the two examples from the earlier section in detail, you can see that we use None as one of the parameters for a dynamically linked C library. This is possible because None, integers, longs, byte strings, and unicode strings are the native Python objects that can be directly used as the parameters in these function calls. None is passed as a C, NULL pointer, byte strings, and unicode strings are passed as pointers to the memory block that contains their data (char * or wchar_t *). Python integers and Python longs are passed as the platform's default C int type, their value is masked to fit into the C type. A complete overview of the Python types and their corresponding ctype types can be seen in Table 1:

This table is very helpful because all the Python types except integers, strings, and unicode strings have to be wrapped in their corresponding ctypes type so that they can be converted to the required C data type in the linked library and not throw the TypeError exceptions, as shown in the following code:

(labenv)user@lab:~$ python

>>> from ctypes import *
>>> libc = CDLL("libc.so.6")
>>> printf = libc.printf

>>> printf("An int %d, a double %f\n", 4711, 47.11)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ctypes.ArgumentError: argument 3: <type 'exceptions.TypeError'>: Don't know how to convert parameter 3

>>> printf("An int %d, a double %f\n", 4711, c_double(47.11))
An int 4711, a double 47.110000

from ctypes import *

class case(Union):
        _fields_ = [
        ("evidence_int", c_int),
        ("evidence_long", c_long),
        ("evidence_char", c_char * 4)
        ]

value = raw_input("Enter new evidence number:")
new_evidence = case(int(value))
print "Evidence number as a int: %i" % new_evidence.evidence_int
print "Evidence number as a long: %ld" % new_evidence.evidence_long
print "Evidence number as a char: %s" % new_evidence.evidence_char

(labenv)user@lab:~$ python new_evidence.py

Enter new evidence number:42

Evidence number as a long: 42
Evidence number as a int: 42
Evidence number as a char: *

from ctypes import *

class case(Structure):
        _fields_ = [
        ("name", c_char * 16),
        ("number", c_int),
        ("investigator_name", c_char * 8)
        ]