C++ Wrapper for Python Matplotlib Lib

I was looking for a light-weighted C++ library for visualization to integrate into my programs when I realized that there are no officialy small ones existing. One attempt that I noticed was a C++ wrapper for the famous visualization python library: Matplotlib (the original one can be found HERE; and the modified program with CMake build can be found HERE). However, looking at the source codes to find a way to expand the library, I found out that the current approach directly utilizing Python/C API; makes it quite difficult for an expansion. That is the start of this library, written with pybind11 in hope of providing a more convenient way to connect with Matplotlib library.

Dependencies

• OS: Ubuntu (tested on 16.04, 19.04)
• compiler: tested with gcc-6
• C++11 and above
• python2/3
• pybind11 : C++ library for seamless conversion between C++ and python
• matplotlib
• cmake: >= 3.8

How to Build

Commands

cd /path/to/this/root/directory
mkdir build && cd build && cmake ../
make -j`nproc`

Build Options

• BUILD_EXAMPLES=ON/OFF (default OFF): whether to build examples or not.
• WITH_DEBUG=ON/OFF (default OFF): whether to enable debug macro or not.
• WITH_TEST=ON/OFF (default OFF): whether to build test or not.
• WITH_DOC=ON/OFF (default ON): whether to build doxygen documentations or not.

Examples

All the examples can be found in examples directory.

Basic Operations

Source Code

Important parts of the code:

    // import modules of matplotlib library
    pe::vis::Matplotlib mpllib;

    // check if the modules are imported successully or not
    if (!mpllib.imported()) {
        std::cout << "Failed to import matplotlib library\n";
        exit(EXIT_FAILURE);
    }

    ...

    // create a figure of desired size
    mpllib.figure(1200, 780);

    // plot with default parameters
    mpllib.plot(x, y);

    // plot with user-input parameters
    mpllib.plot(x, w,
                {
                    {"color", pe::vis::Matplotlib::createAnyBaseMapData<std::string>("r")},
                    {"linestyle", pe::vis::Matplotlib::createAnyBaseMapData<std::string>("--")},
                });

    // plot with user-input parameters
    mpllib.plot(x, z,
                {
                    {"label", pe::vis::Matplotlib::createAnyBaseMapData<std::string>("log(x)")},
                    {"lw", pe::vis::Matplotlib::createAnyBaseMapData<int>(2)},
                });

    // limit for x axis
    mpllib.xlim(0, 1000 * 1000);

    // add grid
    mpllib.grid();

    // add title
    mpllib.title("Sample plot");

    // add legend with default location (best)
    mpllib.legend();

    // save figure
    mpllib.savefig("Basic.png");

    // show figure
    mpllib.show();

It can be seen here that user can input flexible parameters of "arbitary" types (currently support bool, int, float, double, std::string - but the list of supported types can be easily extended) into the function; which make it easier to modify parameters just like in python code. To pass new parameters, you need to create a map like the following:

{
    {"label", pe::vis::Matplotlib::createAnyBaseMapData<std::string>("log(x)")},
    {"lw", pe::vis::Matplotlib::createAnyBaseMapData<int>(2)},
}

Each pair has the following structure:

{[string of name of the python parameter] : pe::vis::Matplotlib::createAnyBaseMapData<[type of the parameter]>([value of the parameter])}

Fancy Figure

Source Code

3D Figure

Source Code

Path

Source Code

Bezier Path

Source Code

Hyperplane

Source Code

More examples can be easily created by using current wrapped functions or adding more modules, functions from python side.