ctypes — A foreign function library for Python¶

Loading dynamic link libraries¶

ctypes exports the cdll, and on Windows windll and oledll objects, for loading dynamic link libraries.

You load libraries by accessing them as attributes of these objects. cdll loads libraries which export functions using the standard cdecl calling convention, while windll libraries call functions using the stdcall calling convention. oledll also uses the stdcall calling convention, and assumes the functions return a Windows HRESULT error code. The error code is used to automatically raise an OSError exception when the function call fails.

Here are some examples for Windows. Note that msvcrt is the MS standard C library containing most standard C functions, and uses the cdecl calling convention:

>>> from ctypes import *
>>> print(windll.kernel32)
<WinDLL 'kernel32', handle ... at ...>
>>> print(cdll.msvcrt)
<CDLL 'msvcrt', handle ... at ...>
>>> libc = cdll.msvcrt
>>>

Windows appends the usual .dll file suffix automatically.

On Linux, it is required to specify the filename including the extension to load a library, so attribute access can not be used to load libraries. 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:

>>> cdll.LoadLibrary("libc.so.6")
<CDLL 'libc.so.6', handle ... at ...>
>>> libc = CDLL("libc.so.6")
>>> libc
<CDLL 'libc.so.6', handle ... at ...>
>>>

Accessing functions from loaded dlls¶

Functions are accessed as attributes of dll objects:

>>> libc.printf
<_FuncPtr object at 0x...>
>>> print(windll.kernel32.GetModuleHandleA)
<_FuncPtr object at 0x...>
>>> print(windll.kernel32.MyOwnFunction)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "ctypes.py", line 239, in __getattr__
    func = _StdcallFuncPtr(name, self)
AttributeError: function 'MyOwnFunction' not found
>>>

Note that win32 system dlls like kernel32 and user32 often export ANSI as well as UNICODE versions of a function. The UNICODE version is exported with a W appended to the name, while the ANSI version is exported with an A appended to the name. The win32 GetModuleHandle function, which returns a module handle for a given module name, has the following C prototype, and a macro is used to expose one of them as GetModuleHandle depending on whether UNICODE is defined or not:

/* ANSI version */
HMODULE GetModuleHandleA(LPCSTR lpModuleName);
/* UNICODE version */
HMODULE GetModuleHandleW(LPCWSTR lpModuleName);

windll does not try to select one of them by magic, you must access the version you need by specifying GetModuleHandleA or GetModuleHandleW explicitly, and then call it with bytes or string objects respectively.

Sometimes, dlls export functions with names which aren’t valid Python identifiers, like "??2@YAPAXI@Z". In this case you have to use getattr() to retrieve the function:

>>> getattr(cdll.msvcrt, "??2@YAPAXI@Z")
<_FuncPtr object at 0x...>
>>>

On Windows, some dlls export functions not by name but by ordinal. These functions can be accessed by indexing the dll object with the ordinal number:

>>> cdll.kernel32[1]
<_FuncPtr object at 0x...>
>>> cdll.kernel32[0]
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "ctypes.py", line 310, in __getitem__
    func = _StdcallFuncPtr(name, self)
AttributeError: function ordinal 0 not found
>>>

Calling functions¶

You can call these functions like any other Python callable. This example uses the rand() function, which takes no arguments and returns a pseudo-random integer:

>>> print(libc.rand())
1804289383

On Windows, you can call the GetModuleHandleA() function, which returns a win32 module handle (passing None as single argument to call it with a NULL pointer):

>>> print(hex(windll.kernel32.GetModuleHandleA(None)))
0x1d000000
>>>

ValueError is raised when you call an stdcall function with the cdecl calling convention, or vice versa:

>>> cdll.kernel32.GetModuleHandleA(None)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ValueError: Procedure probably called with not enough arguments (4 bytes missing)
>>>

>>> windll.msvcrt.printf(b"spam")
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ValueError: Procedure probably called with too many arguments (4 bytes in excess)
>>>

To find out the correct calling convention you have to look into the C header file or the documentation for the function you want to call.

On Windows, ctypes uses win32 structured exception handling to prevent crashes from general protection faults when functions are called with invalid argument values:

>>> windll.kernel32.GetModuleHandleA(32)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
OSError: exception: access violation reading 0x00000020
>>>

There are, however, enough ways to crash Python with ctypes, so you should be careful anyway. The faulthandler module can be helpful in debugging crashes (e.g. from segmentation faults produced by erroneous C library calls).

None, integers, bytes objects and (unicode) strings are the only native Python objects that can directly be used as parameters in these function calls. None is passed as a C NULL pointer, bytes objects and strings are passed as pointer to the memory block that contains their data (char* or wchar_t*). Python integers are passed as the platform’s default C int type, their value is masked to fit into the C type.

Before we move on calling functions with other parameter types, we have to learn more about ctypes data types.

Fundamental data types¶

ctypes defines a number of primitive C compatible data types:

1. The constructor accepts any object with a truth value.

All these types can be created by calling them with an optional initializer of the correct type and value:

>>> c_int()
c_long(0)
>>> c_wchar_p("Hello, World")
c_wchar_p(140018365411392)
>>> c_ushort(-3)
c_ushort(65533)
>>>

Since these types are mutable, their value can also be changed afterwards:

>>> i = c_int(42)
>>> print(i)
c_long(42)
>>> print(i.value)
42
>>> i.value = -99
>>> print(i.value)
-99
>>>

Assigning a new value to instances of the pointer types c_char_p, c_wchar_p, and c_void_p changes the memory location they point to, not the contents of the memory block (of course not, because Python string objects are immutable):

>>> s = "Hello, World"
>>> c_s = c_wchar_p(s)
>>> print(c_s)
c_wchar_p(139966785747344)
>>> print(c_s.value)
Hello World
>>> c_s.value = "Hi, there"
>>> print(c_s)              # the memory location has changed
c_wchar_p(139966783348904)
>>> print(c_s.value)
Hi, there
>>> print(s)                # first object is unchanged
Hello, World
>>>

You should be careful, however, not to pass them to functions expecting pointers to mutable memory. If you need mutable memory blocks, ctypes has a create_string_buffer() function which creates these in various ways. The current memory block contents can be accessed (or changed) with the raw property; if you want to access it as NUL terminated string, use the value property:

>>> from ctypes import *
>>> p = create_string_buffer(3)            # create a 3 byte buffer, initialized to NUL bytes
>>> print(sizeof(p), repr(p.raw))
3 b'\x00\x00\x00'
>>> p = create_string_buffer(b"Hello")     # create a buffer containing a NUL terminated string
>>> print(sizeof(p), repr(p.raw))
6 b'Hello\x00'
>>> print(repr(p.value))
b'Hello'
>>> p = create_string_buffer(b"Hello", 10) # create a 10 byte buffer
>>> print(sizeof(p), repr(p.raw))
10 b'Hello\x00\x00\x00\x00\x00'
>>> p.value = b"Hi"
>>> print(sizeof(p), repr(p.raw))
10 b'Hi\x00lo\x00\x00\x00\x00\x00'
>>>

The create_string_buffer() function replaces the old c_buffer() function (which is still available as an alias). To create a mutable memory block containing unicode characters of the C type wchar_t, use the create_unicode_buffer() function.

Calling functions, continued¶

Note that printf prints to the real standard output channel, not to sys.stdout, so these examples will only work at the console prompt, not from within IDLE or PythonWin:

>>> printf = libc.printf
>>> printf(b"Hello, %s\n", b"World!")
Hello, World!
14
>>> printf(b"Hello, %S\n", "World!")
Hello, World!
14
>>> printf(b"%d bottles of beer\n", 42)
42 bottles of beer
19
>>> printf(b"%f bottles of beer\n", 42.5)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ctypes.ArgumentError: argument 2: TypeError: Don't know how to convert parameter 2
>>>

As has been mentioned before, all Python types except integers, strings, and bytes objects have to be wrapped in their corresponding ctypes type, so that they can be converted to the required C data type:

>>> printf(b"An int %d, a double %f\n", 1234, c_double(3.14))
An int 1234, a double 3.140000
31
>>>

Calling variadic functions¶

On a lot of platforms calling variadic functions through ctypes is exactly the same as calling functions with a fixed number of parameters. On some platforms, and in particular ARM64 for Apple Platforms, the calling convention for variadic functions is different than that for regular functions.

On those platforms it is required to specify the argtypes attribute for the regular, non-variadic, function arguments:

libc.printf.argtypes = [ctypes.c_char_p]

Because specifying the attribute does not inhibit portability it is advised to always specify argtypes for all variadic functions.

Calling functions with your own custom data types¶

You can also customize ctypes argument conversion to allow instances of your own classes be used as function arguments. ctypes looks for an _as_parameter_ attribute and uses this as the function argument. The attribute must be an integer, string, bytes, a ctypes instance, or an object with an _as_parameter_ attribute:

>>> class Bottles:
...     def __init__(self, number):
...         self._as_parameter_ = number
...
>>> bottles = Bottles(42)
>>> printf(b"%d bottles of beer\n", bottles)
42 bottles of beer
19
>>>

If you don’t want to store the instance’s data in the _as_parameter_ instance variable, you could define a property which makes the attribute available on request.

Specifying the required argument types (function prototypes)¶

It is possible to specify the required argument types of functions exported from DLLs by setting the argtypes attribute.

argtypes must be a sequence of C data types (the printf() function is probably not a good example here, because it takes a variable number and different types of parameters depending on the format string, on the other hand this is quite handy to experiment with this feature):

>>> printf.argtypes = [c_char_p, c_char_p, c_int, c_double]
>>> printf(b"String '%s', Int %d, Double %f\n", b"Hi", 10, 2.2)
String 'Hi', Int 10, Double 2.200000
37
>>>

Specifying a format protects against incompatible argument types (just as a prototype for a C function), and tries to convert the arguments to valid types:

>>> printf(b"%d %d %d", 1, 2, 3)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ctypes.ArgumentError: argument 2: TypeError: 'int' object cannot be interpreted as ctypes.c_char_p
>>> printf(b"%s %d %f\n", b"X", 2, 3)
X 2 3.000000
13
>>>

If you have defined your own classes which you pass to function calls, you have to implement a from_param() class method for them to be able to use them in the argtypes sequence. The from_param() class method receives the Python object passed to the function call, it should do a typecheck or whatever is needed to make sure this object is acceptable, and then return the object itself, its _as_parameter_ attribute, or whatever you want to pass as the C function argument in this case. Again, the result should be an integer, string, bytes, a ctypes instance, or an object with an _as_parameter_ attribute.

Return types¶

By default functions are assumed to return the C int type. Other return types can be specified by setting the restype attribute of the function object.

The C prototype of time() is time_t time(time_t *). Because time_t might be of a different type than the default return type int, you should specify the restype attribute:

>>> libc.time.restype = c_time_t

The argument types can be specified using argtypes:

>>> libc.time.argtypes = (POINTER(c_time_t),)

To call the function with a NULL pointer as first argument, use None:

>>> print(libc.time(None))
1150640792

Here is a more advanced example, it uses the strchr() function, which expects a string pointer and a char, and returns a pointer to a string:

>>> strchr = libc.strchr
>>> strchr(b"abcdef", ord("d"))
8059983
>>> strchr.restype = c_char_p    # c_char_p is a pointer to a string
>>> strchr(b"abcdef", ord("d"))
b'def'
>>> print(strchr(b"abcdef", ord("x")))
None
>>>

If you want to avoid the ord("x") calls above, you can set the argtypes attribute, and the second argument will be converted from a single character Python bytes object into a C char:

>>> strchr.restype = c_char_p
>>> strchr.argtypes = [c_char_p, c_char]
>>> strchr(b"abcdef", b"d")
b'def'
>>> strchr(b"abcdef", b"def")
Traceback (most recent call last):
ctypes.ArgumentError: argument 2: TypeError: one character bytes, bytearray or integer expected
>>> print(strchr(b"abcdef", b"x"))
None
>>> strchr(b"abcdef", b"d")
b'def'
>>>

You can also use a callable Python object (a function or a class for example) as the restype attribute, if the foreign function returns an integer. The callable will be called with the integer the C function returns, and the result of this call will be used as the result of your function call. This is useful to check for error return values and automatically raise an exception:

>>> GetModuleHandle = windll.kernel32.GetModuleHandleA
>>> def ValidHandle(value):
...     if value == 0:
...         raise WinError()
...     return value
...
>>>
>>> GetModuleHandle.restype = ValidHandle
>>> GetModuleHandle(None)
486539264
>>> GetModuleHandle("something silly")
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "<stdin>", line 3, in ValidHandle
OSError: [Errno 126] The specified module could not be found.
>>>

WinError is a function which will call Windows FormatMessage() api to get the string representation of an error code, and returns an exception. WinError takes an optional error code parameter, if no one is used, it calls GetLastError() to retrieve it.

Please note that a much more powerful error checking mechanism is available through the errcheck attribute; see the reference manual for details.

Passing pointers (or: passing parameters by reference)¶

Sometimes a C api function expects a pointer to a data type as parameter, probably to write into the corresponding location, or if the data is too large to be passed by value. This is also known as passing parameters by reference.

ctypes exports the byref() function which is used to pass parameters by reference. The same effect can be achieved with the pointer() function, although pointer() does a lot more work since it constructs a real pointer object, so it is faster to use byref() if you don’t need the pointer object in Python itself:

>>> i = c_int()
>>> f = c_float()
>>> s = create_string_buffer(b'\000' * 32)
>>> print(i.value, f.value, repr(s.value))
0 0.0 b''
>>> libc.sscanf(b"1 3.14 Hello", b"%d %f %s",
...             byref(i), byref(f), s)
3
>>> print(i.value, f.value, repr(s.value))
1 3.1400001049 b'Hello'
>>>

Structures and unions¶

Structures and unions must derive from the Structure and Union base classes which are defined in the ctypes module. Each subclass must define a _fields_ attribute. _fields_ must be a list of 2-tuples, containing a field name and a field type.

The field type must be a ctypes type like c_int, or any other derived ctypes type: structure, union, array, pointer.

Here is a simple example of a POINT structure, which contains two integers named x and y, and also shows how to initialize a structure in the constructor:

>>> from ctypes import *
>>> class POINT(Structure):
...     _fields_ = [("x", c_int),
...                 ("y", c_int)]
...
>>> point = POINT(10, 20)
>>> print(point.x, point.y)
10 20
>>> point = POINT(y=5)
>>> print(point.x, point.y)
0 5
>>> POINT(1, 2, 3)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: too many initializers
>>>

You can, however, build much more complicated structures. A structure can itself contain other structures by using a structure as a field type.

Here is a RECT structure which contains two POINTs named upperleft and lowerright:

>>> class RECT(Structure):
...     _fields_ = [("upperleft", POINT),
...                 ("lowerright", POINT)]
...
>>> rc = RECT(point)
>>> print(rc.upperleft.x, rc.upperleft.y)
0 5
>>> print(rc.lowerright.x, rc.lowerright.y)
0 0
>>>

Nested structures can also be initialized in the constructor in several ways:

>>> r = RECT(POINT(1, 2), POINT(3, 4))
>>> r = RECT((1, 2), (3, 4))

Field descriptors can be retrieved from the class, they are useful for debugging because they can provide useful information:

>>> print(POINT.x)
<Field type=c_long, ofs=0, size=4>
>>> print(POINT.y)
<Field type=c_long, ofs=4, size=4>
>>>

Structure/union alignment and byte order¶

By default, Structure and Union fields are aligned in the same way the C compiler does it. It is possible to override this behavior by specifying a _pack_ class attribute in the subclass definition. This must be set to a positive integer and specifies the maximum alignment for the fields. This is what #pragma pack(n) also does in MSVC. It is also possible to set a minimum alignment for how the subclass itself is packed in the same way #pragma align(n) works in MSVC. This can be achieved by specifying a _align_ class attribute in the subclass definition.

ctypes uses the native byte order for Structures and Unions. To build structures with non-native byte order, you can use one of the BigEndianStructure, LittleEndianStructure, BigEndianUnion, and LittleEndianUnion base classes. These classes cannot contain pointer fields.

Bit fields in structures and unions¶

It is possible to create structures and unions containing bit fields. Bit fields are only possible for integer fields, the bit width is specified as the third item in the _fields_ tuples:

>>> class Int(Structure):
...     _fields_ = [("first_16", c_int, 16),
...                 ("second_16", c_int, 16)]
...
>>> print(Int.first_16)
<Field type=c_long, ofs=0:0, bits=16>
>>> print(Int.second_16)
<Field type=c_long, ofs=0:16, bits=16>
>>>

Arrays¶

Arrays are sequences, containing a fixed number of instances of the same type.

The recommended way to create array types is by multiplying a data type with a positive integer:

TenPointsArrayType = POINT * 10

Here is an example of a somewhat artificial data type, a structure containing 4 POINTs among other stuff:

>>> from ctypes import *
>>> class POINT(Structure):
...     _fields_ = ("x", c_int), ("y", c_int)
...
>>> class MyStruct(Structure):
...     _fields_ = [("a", c_int),
...                 ("b", c_float),
...                 ("point_array", POINT * 4)]
>>>
>>> print(len(MyStruct().point_array))
4
>>>

Instances are created in the usual way, by calling the class:

arr = TenPointsArrayType()
for pt in arr:
    print(pt.x, pt.y)

The above code print a series of 0 0 lines, because the array contents is initialized to zeros.

Initializers of the correct type can also be specified:

>>> from ctypes import *
>>> TenIntegers = c_int * 10
>>> ii = TenIntegers(1, 2, 3, 4, 5, 6, 7, 8, 9, 10)
>>> print(ii)
<c_long_Array_10 object at 0x...>
>>> for i in ii: print(i, end=" ")
...
1 2 3 4 5 6 7 8 9 10
>>>

Pointers¶

Pointer instances are created by calling the pointer() function on a ctypes type:

>>> from ctypes import *
>>> i = c_int(42)
>>> pi = pointer(i)
>>>

Pointer instances have a contents attribute which returns the object to which the pointer points, the i object above:

>>> pi.contents
c_long(42)
>>>

Note that ctypes does not have OOR (original object return), it constructs a new, equivalent object each time you retrieve an attribute:

>>> pi.contents is i
False
>>> pi.contents is pi.contents
False
>>>

Assigning another c_int instance to the pointer’s contents attribute would cause the pointer to point to the memory location where this is stored:

>>> i = c_int(99)
>>> pi.contents = i
>>> pi.contents
c_long(99)
>>>

Pointer instances can also be indexed with integers:

>>> pi[0]
99
>>>

Assigning to an integer index changes the pointed to value:

>>> print(i)
c_long(99)
>>> pi[0] = 22
>>> print(i)
c_long(22)
>>>

It is also possible to use indexes different from 0, but you must know what you’re doing, just as in C: You can access or change arbitrary memory locations. Generally you only use this feature if you receive a pointer from a C function, and you know that the pointer actually points to an array instead of a single item.

Behind the scenes, the pointer() function does more than simply create pointer instances, it has to create pointer types first. This is done with the POINTER() function, which accepts any ctypes type, and returns a new type:

>>> PI = POINTER(c_int)
>>> PI
<class 'ctypes.LP_c_long'>
>>> PI(42)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: expected c_long instead of int
>>> PI(c_int(42))
<ctypes.LP_c_long object at 0x...>
>>>

Calling the pointer type without an argument creates a NULL pointer. NULL pointers have a False boolean value:

>>> null_ptr = POINTER(c_int)()
>>> print(bool(null_ptr))
False
>>>

ctypes checks for NULL when dereferencing pointers (but dereferencing invalid non-NULL pointers would crash Python):

>>> null_ptr[0]
Traceback (most recent call last):
    ....
ValueError: NULL pointer access
>>>

>>> null_ptr[0] = 1234
Traceback (most recent call last):
    ....
ValueError: NULL pointer access
>>>

Type conversions¶

Usually, ctypes does strict type checking. This means, if you have POINTER(c_int) in the argtypes list of a function or as the type of a member field in a structure definition, only instances of exactly the same type are accepted. There are some exceptions to this rule, where ctypes accepts other objects. For example, you can pass compatible array instances instead of pointer types. So, for POINTER(c_int), ctypes accepts an array of c_int:

>>> class Bar(Structure):
...     _fields_ = [("count", c_int), ("values", POINTER(c_int))]
...
>>> bar = Bar()
>>> bar.values = (c_int * 3)(1, 2, 3)
>>> bar.count = 3
>>> for i in range(bar.count):
...     print(bar.values[i])
...
1
2
3
>>>

In addition, if a function argument is explicitly declared to be a pointer type (such as POINTER(c_int)) in argtypes, an object of the pointed type (c_int in this case) can be passed to the function. ctypes will apply the required byref() conversion in this case automatically.

To set a POINTER type field to NULL, you can assign None:

>>> bar.values = None
>>>

Sometimes you have instances of incompatible types. In C, you can cast one type into another type. ctypes provides a cast() function which can be used in the same way. The Bar structure defined above accepts POINTER(c_int) pointers or c_int arrays for its values field, but not instances of other types:

>>> bar.values = (c_byte * 4)()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: incompatible types, c_byte_Array_4 instance instead of LP_c_long instance
>>>

For these cases, the cast() function is handy.

The cast() function can be used to cast a ctypes instance into a pointer to a different ctypes data type. cast() takes two parameters, a ctypes object that is or can be converted to a pointer of some kind, and a ctypes pointer type. It returns an instance of the second argument, which references the same memory block as the first argument:

>>> a = (c_byte * 4)()
>>> cast(a, POINTER(c_int))
<ctypes.LP_c_long object at ...>
>>>

So, cast() can be used to assign to the values field of Bar the structure:

>>> bar = Bar()
>>> bar.values = cast((c_byte * 4)(), POINTER(c_int))
>>> print(bar.values[0])
0
>>>

Incomplete Types¶

Incomplete Types are structures, unions or arrays whose members are not yet specified. In C, they are specified by forward declarations, which are defined later:

struct cell; /* forward declaration */

struct cell {
    char *name;
    struct cell *next;
};

The straightforward translation into ctypes code would be this, but it does not work:

>>> class cell(Structure):
...     _fields_ = [("name", c_char_p),
...                 ("next", POINTER(cell))]
...
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "<stdin>", line 2, in cell
NameError: name 'cell' is not defined
>>>

because the new class cell is not available in the class statement itself. In ctypes, we can define the cell class and set the _fields_ attribute later, after the class statement:

>>> from ctypes import *
>>> class cell(Structure):
...     pass
...
>>> cell._fields_ = [("name", c_char_p),
...                  ("next", POINTER(cell))]
>>>

Let’s try it. We create two instances of cell, and let them point to each other, and finally follow the pointer chain a few times:

>>> c1 = cell()
>>> c1.name = b"foo"
>>> c2 = cell()
>>> c2.name = b"bar"
>>> c1.next = pointer(c2)
>>> c2.next = pointer(c1)
>>> p = c1
>>> for i in range(8):
...     print(p.name, end=" ")
...     p = p.next[0]
...
foo bar foo bar foo bar foo bar
>>>

Callback functions¶

ctypes allows creating C callable function pointers from Python callables. These are sometimes called callback functions.

First, you must create a class for the callback function. The class knows the calling convention, the return type, and the number and types of arguments this function will receive.

The CFUNCTYPE() factory function creates types for callback functions using the cdecl calling convention. On Windows, the WINFUNCTYPE() factory function creates types for callback functions using the stdcall calling convention.

Both of these factory functions are called with the result type as first argument, and the callback functions expected argument types as the remaining arguments.

I will present an example here which uses the standard C library’s qsort() function, that is used to sort items with the help of a callback function. qsort() will be used to sort an array of integers:

>>> IntArray5 = c_int * 5
>>> ia = IntArray5(5, 1, 7, 33, 99)
>>> qsort = libc.qsort
>>> qsort.restype = None
>>>

qsort() must be called with a pointer to the data to sort, the number of items in the data array, the size of one item, and a pointer to the comparison function, the callback. The callback will then be called with two pointers to items, and it must return a negative integer if the first item is smaller than the second, a zero if they are equal, and a positive integer otherwise.

So our callback function receives pointers to integers, and must return an integer. First we create the type for the callback function:

>>> CMPFUNC = CFUNCTYPE(c_int, POINTER(c_int), POINTER(c_int))
>>>

To get started, here is a simple callback that shows the values it gets passed:

>>> def py_cmp_func(a, b):
...     print("py_cmp_func", a[0], b[0])
...     return 0
...
>>> cmp_func = CMPFUNC(py_cmp_func)
>>>

The result:

>>> qsort(ia, len(ia), sizeof(c_int), cmp_func)
py_cmp_func 5 1
py_cmp_func 33 99
py_cmp_func 7 33
py_cmp_func 5 7
py_cmp_func 1 7
>>>

Now we can actually compare the two items and return a useful result:

>>> def py_cmp_func(a, b):
...     print("py_cmp_func", a[0], b[0])
...     return a[0] - b[0]
...
>>>
>>> qsort(ia, len(ia), sizeof(c_int), CMPFUNC(py_cmp_func))
py_cmp_func 5 1
py_cmp_func 33 99
py_cmp_func 7 33
py_cmp_func 1 7
py_cmp_func 5 7
>>>

As we can easily check, our array is sorted now:

>>> for i in ia: print(i, end=" ")
...
1 5 7 33 99
>>>

The function factories can be used as decorator factories, so we may as well write:

>>> @CFUNCTYPE(c_int, POINTER(c_int), POINTER(c_int))
... def py_cmp_func(a, b):
...     print("py_cmp_func", a[0], b[0])
...     return a[0] - b[0]
...
>>> qsort(ia, len(ia), sizeof(c_int), py_cmp_func)
py_cmp_func 5 1
py_cmp_func 33 99
py_cmp_func 7 33
py_cmp_func 1 7
py_cmp_func 5 7
>>>

Accessing values exported from dlls¶

Some shared libraries not only export functions, they also export variables. An example in the Python library itself is the Py_Version, Python runtime version number encoded in a single constant integer.

ctypes can access values like this with the in_dll() class methods of the type. pythonapi is a predefined symbol giving access to the Python C api:

>>> version = ctypes.c_int.in_dll(ctypes.pythonapi, "Py_Version")
>>> print(hex(version.value))
0x30c00a0

An extended example which also demonstrates the use of pointers accesses the PyImport_FrozenModules pointer exported by Python.

Quoting the docs for that value:

This pointer is initialized to point to an array of _frozen records, terminated by one whose members are all NULL or zero. When a frozen module is imported, it is searched in this table. Third-party code could play tricks with this to provide a dynamically created collection of frozen modules.

So manipulating this pointer could even prove useful. To restrict the example size, we show only how this table can be read with ctypes:

>>> from ctypes import *
>>>
>>> class struct_frozen(Structure):
...     _fields_ = [("name", c_char_p),
...                 ("code", POINTER(c_ubyte)),
...                 ("size", c_int),
...                 ("get_code", POINTER(c_ubyte)),  # Function pointer
...                ]
...
>>>

We have defined the _frozen data type, so we can get the pointer to the table:

>>> FrozenTable = POINTER(struct_frozen)
>>> table = FrozenTable.in_dll(pythonapi, "_PyImport_FrozenBootstrap")
>>>

Since table is a pointer to the array of struct_frozen records, we can iterate over it, but we just have to make sure that our loop terminates, because pointers have no size. Sooner or later it would probably crash with an access violation or whatever, so it’s better to break out of the loop when we hit the NULL entry:

>>> for item in table:
...     if item.name is None:
...         break
...     print(item.name.decode("ascii"), item.size)
...
_frozen_importlib 31764
_frozen_importlib_external 41499
zipimport 12345
>>>

The fact that standard Python has a frozen module and a frozen package (indicated by the negative size member) is not well known, it is only used for testing. Try it out with import __hello__ for example.

Surprises¶

There are some edges in ctypes where you might expect something other than what actually happens.

Consider the following example:

>>> from ctypes import *
>>> class POINT(Structure):
...     _fields_ = ("x", c_int), ("y", c_int)
...
>>> class RECT(Structure):
...     _fields_ = ("a", POINT), ("b", POINT)
...
>>> p1 = POINT(1, 2)
>>> p2 = POINT(3, 4)
>>> rc = RECT(p1, p2)
>>> print(rc.a.x, rc.a.y, rc.b.x, rc.b.y)
1 2 3 4
>>> # now swap the two points
>>> rc.a, rc.b = rc.b, rc.a
>>> print(rc.a.x, rc.a.y, rc.b.x, rc.b.y)
3 4 3 4
>>>

Hm. We certainly expected the last statement to print 3 4 1 2. What happened? Here are the steps of the rc.a, rc.b = rc.b, rc.a line above:

>>> temp0, temp1 = rc.b, rc.a
>>> rc.a = temp0
>>> rc.b = temp1
>>>

Note that temp0 and temp1 are objects still using the internal buffer of the rc object above. So executing rc.a = temp0 copies the buffer contents of temp0 into rc ‘s buffer. This, in turn, changes the contents of temp1. So, the last assignment rc.b = temp1, doesn’t have the expected effect.

Keep in mind that retrieving sub-objects from Structure, Unions, and Arrays doesn’t copy the sub-object, instead it retrieves a wrapper object accessing the root-object’s underlying buffer.

Another example that may behave differently from what one would expect is this:

>>> s = c_char_p()
>>> s.value = b"abc def ghi"
>>> s.value
b'abc def ghi'
>>> s.value is s.value
False
>>>

Why is it printing False? ctypes instances are objects containing a memory block plus some descriptors accessing the contents of the memory. Storing a Python object in the memory block does not store the object itself, instead the contents of the object is stored. Accessing the contents again constructs a new Python object each time!

Variable-sized data types¶

ctypes provides some support for variable-sized arrays and structures.

The resize() function can be used to resize the memory buffer of an existing ctypes object. The function takes the object as first argument, and the requested size in bytes as the second argument. The memory block cannot be made smaller than the natural memory block specified by the objects type, a ValueError is raised if this is tried:

>>> short_array = (c_short * 4)()
>>> print(sizeof(short_array))
8
>>> resize(short_array, 4)
Traceback (most recent call last):
    ...
ValueError: minimum size is 8
>>> resize(short_array, 32)
>>> sizeof(short_array)
32
>>> sizeof(type(short_array))
8
>>>

This is nice and fine, but how would one access the additional elements contained in this array? Since the type still only knows about 4 elements, we get errors accessing other elements:

>>> short_array[:]
[0, 0, 0, 0]
>>> short_array[7]
Traceback (most recent call last):
    ...
IndexError: invalid index
>>>

Another way to use variable-sized data types with ctypes is to use the dynamic nature of Python, and (re-)define the data type after the required size is already known, on a case by case basis.

Finding shared libraries¶

When programming in a compiled language, shared libraries are accessed when compiling/linking a program, and when the program is run.

The purpose of the find_library() function is to locate a library in a way similar to what the compiler or runtime loader does (on platforms with several versions of a shared library the most recent should be loaded), while the ctypes library loaders act like when a program is run, and call the runtime loader directly.

The ctypes.util module provides a function which can help to determine the library to load.

ctypes.util.find_library(name)

Try to find a library and return a pathname. name is the library name without any prefix like lib, suffix like .so, .dylib or version number (this is the form used for the posix linker option -l). If no library can be found, returns None.

The exact functionality is system dependent.

On Linux, find_library() tries to run external programs (/sbin/ldconfig, gcc, objdump and ld) to find the library file. It returns the filename of the library file.

Here are some examples:

>>> from ctypes.util import find_library
>>> find_library("m")
'libm.so.6'
>>> find_library("c")
'libc.so.6'
>>> find_library("bz2")
'libbz2.so.1.0'
>>>

On macOS and Android, find_library() uses the system’s standard naming schemes and paths to locate the library, and returns a full pathname if successful:

>>> from ctypes.util import find_library
>>> find_library("c")
'/usr/lib/libc.dylib'
>>> find_library("m")
'/usr/lib/libm.dylib'
>>> find_library("bz2")
'/usr/lib/libbz2.dylib'
>>> find_library("AGL")
'/System/Library/Frameworks/AGL.framework/AGL'
>>>

On Windows, find_library() searches along the system search path, and returns the full pathname, but since there is no predefined naming scheme a call like find_library("c") will fail and return None.

If wrapping a shared library with ctypes, it may be better to determine the shared library name at development time, and hardcode that into the wrapper module instead of using find_library() to locate the library at runtime.

Loading shared libraries¶

There are several ways to load shared libraries into the Python process. One way is to instantiate one of the following classes:

class ctypes.CDLL(name, mode=DEFAULT_MODE, handle=None, use_errno=False, use_last_error=False, winmode=None)¶

Instances of this class represent loaded shared libraries. Functions in these libraries use the standard C calling convention, and are assumed to return int.

On Windows creating a CDLL instance may fail even if the DLL name exists. When a dependent DLL of the loaded DLL is not found, a OSError error is raised with the message “[WinError 126] The specified module could not be found”. This error message does not contain the name of the missing DLL because the Windows API does not return this information making this error hard to diagnose. To resolve this error and determine which DLL is not found, you need to find the list of dependent DLLs and determine which one is not found using Windows debugging and tracing tools.

Changed in version 3.12: The name parameter can now be a path-like object.

class ctypes.OleDLL(name, mode=DEFAULT_MODE, handle=None, use_errno=False, use_last_error=False, winmode=None)