Using a C function in Python

Question

I've tried all the solutions mentioned on the internet so far nothing worked for me.

I have a python code, to speed it up, I want that my code runs the heavy calculations in a C function. I already wrote this C function.

Then, to share the library, I did this in the terminal :

gcc -shared -Wl,-install_name,testlib.so -o testlib.so -fPIC myModule.c

which returns no error. The problem; comes when i try to launch the C function in python. Let's consider the following simple function in C :

int multiplier(int a, int b)
{

int lol = 0;

lol = a*b;

return lol;
}

I launch python3 (3.5.2), and then :

import ctypes
zelib = ctypes.CDLL("/Users/longeard/Desktop/Codes/DraII/testlib.so",ctypes.RTLD_GLOBAL)

The library should be ready to use in python by doing :

res = zelib.multiplier(2,3)

When doing that, it works and python returns

Problem is, that the function i want to use ( the multiplier function I use is just for the example ) is supposed to take floats as input and return a float. But if I now consider the same multiplier function as before but with float :

float multiplier(float a, float b)
{

float lol = 0.0;

lol = a*b;

return lol;
}

I recompile using gcc, I reimport ctypes and re-do ctypes.CDLL, and I do in python3 :

zelib.multiplier(ctypes.c_float(2),ctypes.c_float(3))

(the types.c_float are here to convert the 2 in python into a float in C ), python will return :

This is weird because if I add a printf within the function to print lol, python will print :

6.0

but still return 2, or 18 sometimes. Even though I printf and return the same variable "lol".

I tried a lot of things, and none of it worked. Do somebody have a idea please ? Thank You.

falsetru · Accepted Answer · 2016-10-18 07:52:45Z

7

You need to specify restype, argtypes of the function:

zelib = ctypes.CDLL('...')
zelib.multiplier.restype = ctypes.c_float   # return type
zelib.multiplier.argtypes = [ctypes.c_float, ctypes.c_float]  # argument types

According to 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.

and Return types in ctypes module documentation:

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.

# without specifying types
>>> import ctypes
>>> zelib = ctypes.CDLL('testlib.so')
>>> zelib.multiplier(2, 3)
0

# specifying types
>>> zelib.multiplier.restype = ctypes.c_float
>>> zelib.multiplier.argtypes = [ctypes.c_float, ctypes.c_float]
>>> zelib.multiplier(2, 3)
6.0

edited Oct 18, 2016 at 7:52

answered Oct 18, 2016 at 7:47

falsetru

371k69 gold badges769 silver badges659 bronze badges

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

3 Comments

Nicano Over a year ago

Thank you, works like a charm ! I actually tried that several times yesterday, for some reason it didn't work, but now it does.

Nicano Over a year ago

Sorry for the double comment .... DO you happen to know how to pass numpy float array as an input in this case ? And return an array in the C function as well ? Thank you.

falsetru Over a year ago

@Nicano, Check this out: docs.scipy.org/doc/numpy/reference/generated/… If you have more question, please post it as another question.

Lelanthran · Accepted Answer · 2016-10-18 08:30:52Z

2

While @falsetru's answer is the better way of doing it an alternative is to simply write your C function to use doubles.

Floats are automatically promoted to double when calling a function without a parameter list.

answered Oct 18, 2016 at 8:30

Lelanthran

1,54912 silver badges19 bronze badges

Collectives™ on Stack Overflow

Using a C function in Python

2 Answers 2

3 Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

3 Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related