How are built-in types protected from overwriting (assigning to) their methods?

Question

I noted that

int.__str__ = lambda x: pass

yields an error.

I can see, why that is forbidden. But how? Can I use that in "normal" code?

Answer 1

For setting attributes directly on int itself and other built-in types (rather than their instances), this protection happens in type.__setattr__, which specifically prohibits setting attributes on built-in types:

static int
type_setattro(PyTypeObject *type, PyObject *name, PyObject *value)
{
    int res;
    if (!(type->tp_flags & Py_TPFLAGS_HEAPTYPE)) {
        PyErr_Format(
            PyExc_TypeError,
            "can't set attributes of built-in/extension type '%s'",
            type->tp_name);
        return -1;
    }
    ...

Py_TPFLAGS_HEAPTYPE is the flag that indicates if a type was defined in Python rather than C.

---

You cannot do the same thing with your own classes unless you implement them in C. You can sort of pretend to do it, by writing a metaclass with a custom __setattr__, but that makes using other useful metaclasses more complicated, and it still doesn't prevent someone calling type.__setattr__ on your classes directly. (Trying a similar trick with object.__setattr__(int, ...) doesn't work, because there's a specific check to catch that.)

---

You didn't ask about instances of built-in types, but they're also interesting. Instances of most built-in types can't have attributes set on them simply because there's nowhere to put such attributes - no __dict__. Rather than having a special "no setting allowed" __setattr__, or lacking a __setattr__, they usually inherit __setattr__ from object, which knows how to handle objects with no __dict__:

descr = _PyType_Lookup(tp, name);

if (descr != NULL) {
    Py_INCREF(descr);
    f = descr->ob_type->tp_descr_set;
    if (f != NULL) {
        res = f(descr, obj, value);
        goto done;
    }
}

if (dict == NULL) {
    dictptr = _PyObject_GetDictPtr(obj);
    if (dictptr == NULL) {
        if (descr == NULL) {
            PyErr_Format(PyExc_AttributeError,
                         "'%.100s' object has no attribute '%U'",
                         tp->tp_name, name);
        }
        else {
            PyErr_Format(PyExc_AttributeError,
                         "'%.50s' object attribute '%U' is read-only",
                         tp->tp_name, name);
        }
        goto done;
    }
    ...