Getting ready

We will try a slightly more involved embedding example in a C++ program. The example is again taken from the Python online documentation (https://docs.python.org/3.5/extending/embedding.html#pure-embedding) and shows how to execute functions from a user-defined Python module by calling the compiled C++ executable.

The Python 3 example code (Py3-pure-embedding.cpp) contains the following source code (see https://docs.python.org/2/extending/embedding.html#pure-embedding for the corresponding Python 2 equivalent):

#include <Python.h>

int main(int argc, char *argv[]) {
  PyObject *pName, *pModule, *pDict, *pFunc;
  PyObject *pArgs, *pValue;
  int i;

  if (argc < 3) {
    fprintf(stderr, "Usage: pure-embedding pythonfile funcname [args]\n");
    return 1;
  }

  Py_Initialize();

  PyRun_SimpleString("import sys");
  PyRun_SimpleString("sys.path.append(\".\")");

  pName = PyUnicode_DecodeFSDefault(argv[1]);
  /* Error checking of pName left out */

  pModule = PyImport_Import(pName);
  Py_DECREF(pName);

  if (pModule != NULL) {
    pFunc = PyObject_GetAttrString(pModule, argv[2]);
    /* pFunc is a new reference */

    if (pFunc && PyCallable_Check(pFunc)) {
      pArgs = PyTuple_New(argc - 3);
      for (i = 0; i < argc - 3; ++i) {
        pValue = PyLong_FromLong(atoi(argv[i + 3]));
        if (!pValue) {
          Py_DECREF(pArgs);
          Py_DECREF(pModule);
          fprintf(stderr, "Cannot convert argument\n");
          return 1;
        }
        /* pValue reference stolen here: */
        PyTuple_SetItem(pArgs, i, pValue);
      }
      pValue = PyObject_CallObject(pFunc, pArgs);
      Py_DECREF(pArgs);
      if (pValue != NULL) {
        printf("Result of call: %ld\n", PyLong_AsLong(pValue));
        Py_DECREF(pValue);
      } else {
        Py_DECREF(pFunc);
        Py_DECREF(pModule);
        PyErr_Print();
        fprintf(stderr, "Call failed\n");
        return 1;
      }
    } else {
      if (PyErr_Occurred())
        PyErr_Print();
      fprintf(stderr, "Cannot find function \"%s\"\n", argv[2]);
    }
    Py_XDECREF(pFunc);
    Py_DECREF(pModule);
  } else {
    PyErr_Print();
    fprintf(stderr, "Failed to load \"%s\"\n", argv[1]);
    return 1;
  }
  Py_Finalize();
  return 0;
}

The Python code that we wish to embed (use_numpy.py) uses NumPy to set up a matrix with all matrix elements set to 1.0:

import numpy as np


def print_ones(rows, cols):

    A = np.ones(shape=(rows, cols), dtype=float)
    print(A)

    # we return the number of elements to verify
    # that the C++ code is able to receive return values
    num_elements = rows*cols
    return(num_elements)