Question

I am making my first venture into integrating C and Python 2.7.3. For starters, I’m just trying to write a C module for Python that can do basic addition. (It is called npfind because once I figure this out, I want to write a find method for numpy)

npfind.h:

#include <math.h>

extern int add(int a, int b);

npfind.c:

#include "npfind.h"

int add(int a, int b)
{
    return a + b;
}

pynpfind.c:

#include "Python.h"
#include "npfind.h"

static char* py_add_doc = "Adds two numbers.";
static PyObject* py_add(PyObject* self, PyObject* args)
{
    int a, b, r;

    if (!PyArg_ParseTuple(args, "ii", &a, &b))
    {
        return NULL;
    }

    r = a + b;

    return Py_BuildValue("i", r);
}

static PyMethodDef* _npfindmethods = {
    {"add", py_add, METH_VARARGS, py_add_doc},
    {NULL, NULL, 0, NULL}
};

void init_npfind(void)
{
    PyObject* mod;
    mod = Py_InitModule("_npfind", _npfindmethods);
}

npfind.py:

from _npfind import *

#Do stuff with the methods

npfindsetup.py

from distutils.core import setup, Extension

setup(name="npfind", version="1.0", py_modules = ['npfind.py'],
      ext_modules=[Extension("_npfind", ["pynpfind.c", "npfind.c"])])

After all that, on Windows 7, I type

python npfindsetup.py build_ext --inplace --compiler=mingw32

Which seems to work. When I then try to find npfind.py, I get this error:

Traceback (most recent call last):
  File "npfind.py", line 1, in <module>
    from _npfind import *
ValueError: module functions cannot set METH_CLASS or METH_STATIC

I cannot figure out what it is talking about. What are METH_CLASS and METH_STATIC, and why I am I trying to set them?

Answer 1

You are declaring the _npfindmethods as a pointer, and trying to initialize it as an array. When I build a code copied from your snippets, I get a lot of warnings like:

a.c:24:5: warning: braces around scalar initializer [enabled by default]
a.c:24:5: warning: (near initialization for '_npfindmethods') [enabled by default]
a.c:24:5: warning: initialization from incompatible pointer type [enabled by default]
a.c:24:5: warning: (near initialization for '_npfindmethods') [enabled by default]
(...)

The variable is initialized with an incorrect value, and thus Python finds random data inside.

---

You should declare _npfindmethods as an array instead:

static PyMethodDef _npfindmethods[] = {
    {"add", py_add, METH_VARARGS, py_add_doc},
    {NULL, NULL, 0, NULL}
};

Now it will be initialized as you expect it to. Also, because now py_add_doc needs to have constant address, you have to make it an array as well:

static char py_add_doc[] = "Adds two numbers.";

So, your final pynpfind.c would look like:

#include "Python.h"
#include "npfind.h"

static char py_add_doc[] = "Adds two numbers.";
static PyObject* py_add(PyObject* self, PyObject* args)
{
    int a, b, r;

    if (!PyArg_ParseTuple(args, "ii", &a, &b))
    {
        return NULL;
    }

    r = a + b;

    return Py_BuildValue("i", r);
}

static PyMethodDef _npfindmethods[] = {
    {"add", py_add, METH_VARARGS, py_add_doc},
    {NULL, NULL, 0, NULL}
};

void init_npfind(void)
{
    PyObject* mod;
    mod = Py_InitModule("_npfind", _npfindmethods);
}