Use clair-c2py manually

The following two commands generate the Python bindings and compile the corresponding Python module for our Getting Started example:

clair-c2py getting_started.cpp -- -std=c++20 `c2py_flags -i`
clang++ getting_started.cpp -std=c++20 -shared -o getting_started.so `c2py_flags`

That’s it. The Python module getting_started is ready to be used.

Let us break down the commands in more detail.

Generate Python Bindings

clair-c2py getting_started.cpp -- -std=c++20 `c2py_flags -i`

This generates the C++/Python binding file getting_started.wrap.cxx and updates the original source file getting_started.cpp to include the generated bindings. A companion header getting_started.wrap.hxx is also generated when the module wraps at least one class (see Across multiple modules).

See Generate Python Bindings and Compile the Module for some more background information.

Note

  • Here, we assume that c2py is installed and that c2py_flags is available in the system path. The command c2py_flags -i provides all necessary include paths for Python.

  • In a CMake project, we typically rely on compile_commands.json in conjunction with automatic detection of Python and c2py targets (see Use CMake integration [Recommended] for more details).

Compile the Module

clang++ getting_started.cpp -std=c++20 -shared -o getting_started.so `c2py_flags`

This compiles the Python C++ extension.

Note

  • The command c2py_flags provides all necessary include and linker paths for Python.

  • Any C++20 compiler can be used to compile the bindings (clang, gcc, etc.). It is independent of the clair_c2py tool and of the LLVM/clang version it is based on.