c2py

C++ to Python interface

Using the following instructions you can call your existing C/C++ modules from Python. There are multiple ways you can do this, but here I'm using SWIG for the truly lazy following instructions that can be found here.

Installing SWIG

You need to install SWIG:

• you can download it from here;
• or install it using a package manager:

sudo apt install -y swig3.0   # Ubuntu, Elementary, ...
sudo pacman -S --noconfirm swig  # Arch, Manjaro, ...
sudo yum install swig  # Fedora, CentOS, Redhat

TL;DR

The scripts run_c.sh and run_cpp.sh do everything explained here for C and C++ languages, respectively. You can just run them after installing SWIG and they should do everything, hopefully without a problem.

Step by step example

The codes in this repo represent a (hopefully) working example. The instructions assume you are using C, but using C++ is not much different and the differences are explained at each step. Also refer to run_c.sh and run_cpp.sh if you get confused. I also try to specify which files each command produces at each step:

0. Have your C/C++ code ready: swig_example.c
1. Create a simple header file that exposes the functionality you want to use (like header.h)
2. Create a simple interface file using the previous header file (like swig_example.i)
3. Build the C/C++ wrapper:

   swig3.0 -python swig_example.i  # outputs `swig_example_wrap.c`

   in C++ you have to add the -c++ option as well.

   swig3.0 -python -c++ swig_example.i  # outputs `swig_example_wrap.cxx`

4. Compile the code to get the .so file:

   gcc -fPIC -c swig_example.c swig_example_wrap.c -I /usr/include/python3.5/  # outputs `swig_example.o` and `swig_example_wrap.o`

   in C++ you need to use g++ instead:

   g++ -fPIC -c swig_example.cpp swig_example_wrap.cxx -I /usr/include/python3.5/  # outputs `swig_example.o` and `swig_example_wrap.o`

   • If you have a C++ code, just use g++ instaed of gcc
   • These commands might not work out of the box in your system. These are some hints that might help you in that case:
     • the include paths and the python module name might be different
     • you can use locate include/python to locate the include path
     • system-wide installations of python usually use /usr/include/ or /usr/local/include/ but Anaconda normally uses the home directory, e.g. /home/<user>/miniconda3/include/
5. build the .so file:

   ld -shared swig_example.o swig_example_wrap.o -o _swig_example.so  # outputs `_swig_example.so`

   in C++ you have to use g++ here as well:

   g++ -shared swig_example.o swig_example_wrap.o -o _swig_example.so  # outputs `_swig_example.so`

6. Use it in python!

   python3 -c "import swig_example; print('Result of computing factorial of 5 (implemented in C++) from Python:', swig_example.fact(5))"

   or simply open the Python interpreter and run the following:

   import swig_example
   swig_example.fact(10)