Just discovered this when I was glancing through the pystream (a CUDA library for Python) source codes.
In Python2.5+, it is possible to directly load simple C shared libraries from Python without the need for writing C wrappers for it. (SWIG or manually writing one). This can be achieved using the Python ctypes module.
Lets take a simple example:
/* test.c */
int multiply(int a,int b){
return a * b;
}
Compile it as a SO library (ref: Writing and using shared libraries):
gcc -c -fPIC test.c
gcc -shared -fPIC -o libtest.so test.o
Now we got an SO file with a test function, lets load it in Python
In [1]: import ctypes
In [2]: libtest = ctypes.cdll.LoadLibrary('/path/to/libtest.so')
In [3]: libtest.multiply(30,99)
Out[3]: 2970
Hope this would be useful to someone.
More details : http://docs.python.org/library/ctypes.html
Happy Hacking :D
In Python2.5+, it is possible to directly load simple C shared libraries from Python without the need for writing C wrappers for it. (SWIG or manually writing one). This can be achieved using the Python ctypes module.
Lets take a simple example:
/* test.c */
int multiply(int a,int b){
return a * b;
}
Compile it as a SO library (ref: Writing and using shared libraries):
gcc -c -fPIC test.c
gcc -shared -fPIC -o libtest.so test.o
Now we got an SO file with a test function, lets load it in Python
In [1]: import ctypes
In [2]: libtest = ctypes.cdll.LoadLibrary('/path/to/libtest.so')
In [3]: libtest.multiply(30,99)
Out[3]: 2970
Hope this would be useful to someone.
More details : http://docs.python.org/library/ctypes.html
Happy Hacking :D