How to test if PyObject has an iterator

Question

I am implementing a C function as an extension for Python. Inside abstract.h, I found the following:

/* ==== Iterators ================================================ */

/* Takes an object and returns an iterator for it.
   This is typically a new iterator but if the argument is an iterator, this
   returns itself. */
PyAPI_FUNC(PyObject *) PyObject_GetIter(PyObject *);

/* Returns 1 if the object 'obj' provides iterator protocols, and 0 otherwise.

   This function always succeeds. */
PyAPI_FUNC(int) PyIter_Check(PyObject *);

When I try to get iterators using PyObject_GetIter on obviously non-iterable objects like a number, I get the error SystemError: <built-in function xxx> returned a result with an error set.

static PyObject *my_method(PyObject *self, PyObject *args) 
{
    PyObject *obj;
    PyArg_ParseTuple(args, "O", &obj)
    //  printf("\ncheck %d",PyIter_Check(obj)); // always 0
    PyObject *iter = PyObject_GetIter(obj); // throws error
    return PyLong_FromLong(0);
}

I would like to handle errors on my own. So I tried to use the PyIter_Check to test if the object has an iterator. However, this function returned 0 for all objects I provided, including the iterable ones.

I thought that it might be caused by the PyAPI_FUNC() macro but I found it in pyport.h and it seems to be only adding __declspec.

• Why is the function PyIter_Check returning zero for all objects?

Answer 1

PyIter_Check is for checking whether an object is an iterator, not whether it can provide one. There doesn’t appear to be a PyIterable_Check.

Moreover, Python pretty much enforces EAFP: since anything could provide an __iter__ that raises an exception, you have to check for an error from PyObject_GetIter anyway, so the only point of the *_Check functions is to provide early sanity checks (sometimes with better error messages).