Problem
-------

Actual behaviour:

>>> getattr('foobar', 123)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: getattr(): attribute name must be string
>>> hasattr('foobar', 123)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: hasattr(): attribute name must be string

Expected behaviour:

>>> getattr('foobar', 123)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: attribute name must be string, not 'int'
>>> hasattr('foobar', 123)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: attribute name must be string, not 'int'

Motivation:

>>> setattr('foobar', 123, 'baz')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: attribute name must be string, not 'int'
>>> delattr('foobar', 123)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: attribute name must be string, not 'int'

Solution
--------

In the function builtin_getattr defined in Python/bltinmodule.c, we remove the following lines:

if (!PyUnicode_Check(name)) {
        PyErr_SetString(PyExc_TypeError,
                        "getattr(): attribute name must be string");
        return NULL;
    }

because the expected TypeError message is already implemented in the subsequent call to the functions _PyObject_LookupAttr and PyObject_GetAttr defined in Objects/object.c:

PyObject *
PyObject_GetAttr(PyObject *v, PyObject *name)
{
    PyTypeObject *tp = Py_TYPE(v);
    if (!PyUnicode_Check(name)) {
        PyErr_Format(PyExc_TypeError,
                     "attribute name must be string, not '%.200s'",
                     Py_TYPE(name)->tp_name);
        return NULL;
    }

[…]

int
_PyObject_LookupAttr(PyObject *v, PyObject *name, PyObject **result)
{
    PyTypeObject *tp = Py_TYPE(v);

    if (!PyUnicode_Check(name)) {
        PyErr_Format(PyExc_TypeError,
                     "attribute name must be string, not '%.200s'",
                     Py_TYPE(name)->tp_name);
        *result = NULL;
        return -1;
    }

[…]

Likewise, in the function builtin_hasattr_impl defined in Python/bltinmodule.c, we remove the following lines:

if (!PyUnicode_Check(name)) {
        PyErr_SetString(PyExc_TypeError,
                        "hasattr(): attribute name must be string");
        return NULL;
    }

because the expected TypeError message is already implemented in the subsequent call to the function _PyObject_LookupAttr defined in Objects/object.c.

Good point. That code was originally added in bpo-420304 because every exception raised in PyObject_GetAttr() (including a TypeError for non-string name) was silenced in hasattr() and 3-argument getattr(). It was changed in Python 3, so this code duplication is no longer needed.

The name of the function was added in error messages in a9b9c9f. Now it will gone. It is a minor regression, but I think that it is fine. Not all error messages contain a function name. In this case the context is clear from the traceback, and the error is not specific to these two functions, setattr() and delattr() raise the same error.

New changeset 16bf9bd by Géry Ogam in branch 'main':
bpo-44024: Improve the TypeError message in getattr and hasattr (GH-25863)
16bf9bd

Thank you for your contribution Géry.

Thanks for the review Serhiy.