HDF5 (TRIQS/h5 Integration)

clair/c2py integrates with the TRIQS/h5 library, enabling serialization and deserialization of C++ classes to HDF5 files from Python.

If your C++ class implements HDF5 support via TRIQS/h5 (the h5_write / h5_read friend functions), clair automatically makes the wrapped objects usable with the TRIQS/h5 HDFArchive.

C++ Example:

// HDF5 serialization
#include <c2py/c2py.hpp>
#include <h5/h5.hpp>
#include <vector>

namespace mylib {

  class matrix {
    std::vector<std::vector<double>> data_;

    public:
    matrix(size_t rows, size_t cols) : data_(rows, std::vector<double>(cols, 0.0)) {}

    // HDF5 write
    friend void h5_write(h5::group g, std::string const &name, matrix const &m) { h5::write(g, name, m.data_); }

    // HDF5 read
    friend void h5_read(h5::group g, std::string const &name, matrix &m) { h5::read(g, name, m.data_); }
  };

} // namespace mylib

Python Usage:

The wrapped type can be stored in and retrieved from a TRIQS/h5 HDFArchive:

from h5 import HDFArchive
from hdf5 import Matrix          # module 'hdf5', class mylib::matrix -> Matrix

m = Matrix(3, 3)

# Write to an HDF5 archive
with HDFArchive("data.h5", "w") as ar:
    ar["matrix"] = m

# Read it back
with HDFArchive("data.h5", "r") as ar:
    m2 = ar["matrix"]

Notes:

• The HDF5 interface relies on the TRIQS/h5 C++ library for the serialization logic (the h5_write / h5_read friends shown above).

• For each class with HDF5 support, clair generates a __write_hdf5__ method and registers the type with the h5.formats registry, so it works transparently with the TRIQS/h5 HDFArchive.