gh-68114: Fix handling for removed PyArg_ParseTuple 'w' formatters (GH-8204)

Co-authored-by: Joe Jevnik <joe@quantopian.com>
Co-authored-by: Petr Viktorin <encukou@gmail.com>

Lib/test/test_capi/test_getargs.py

@@ -856,20 +856,24 @@ class Bytes_TestCase(unittest.TestCase):
 
     def test_w_star(self):
         # getargs_w_star() modifies first and last byte
-        from _testcapi import getargs_w_star
+        # getargs_w_star_opt() takes additional optional args: with one
-        self.assertRaises(TypeError, getargs_w_star, 'abc\xe9')
+        #   argument it should behave the same as getargs_w_star
-        self.assertRaises(TypeError, getargs_w_star, b'bytes')
+        from _testcapi import getargs_w_star, getargs_w_star_opt
-        self.assertRaises(TypeError, getargs_w_star, b'nul:\0')
+        for func in (getargs_w_star, getargs_w_star_opt):
-        self.assertRaises(TypeError, getargs_w_star, memoryview(b'bytes'))
+            with self.subTest(func=func):
-        buf = bytearray(b'bytearray')
+                self.assertRaises(TypeError, func, 'abc\xe9')
-        self.assertEqual(getargs_w_star(buf), b'[ytearra]')
+                self.assertRaises(TypeError, func, b'bytes')
-        self.assertEqual(buf, bytearray(b'[ytearra]'))
+                self.assertRaises(TypeError, func, b'nul:\0')
-        buf = bytearray(b'memoryview')
+                self.assertRaises(TypeError, func, memoryview(b'bytes'))
-        self.assertEqual(getargs_w_star(memoryview(buf)), b'[emoryvie]')
+                buf = bytearray(b'bytearray')
-        self.assertEqual(buf, bytearray(b'[emoryvie]'))
+                self.assertEqual(func(buf), b'[ytearra]')
-        self.assertRaises(TypeError, getargs_w_star, None)
+                self.assertEqual(buf, bytearray(b'[ytearra]'))
-        self.assertRaises(TypeError, getargs_w_star, NONCONTIG_WRITABLE)
+                buf = bytearray(b'memoryview')
-        self.assertRaises(TypeError, getargs_w_star, NONCONTIG_READONLY)
+                self.assertEqual(func(memoryview(buf)), b'[emoryvie]')
+                self.assertEqual(buf, bytearray(b'[emoryvie]'))
+                self.assertRaises(TypeError, func, None)
+                self.assertRaises(TypeError, func, NONCONTIG_WRITABLE)
+                self.assertRaises(TypeError, func, NONCONTIG_READONLY)
 
     def test_getargs_empty(self):
         from _testcapi import getargs_empty
@@ -1112,9 +1116,9 @@ class SkipitemTest(unittest.TestCase):
             c = chr(i)
 
             # skip parentheses, the error reporting is inconsistent about them
-            # skip 'e', it's always a two-character code
+            # skip 'e' and 'w', they're always two-character codes
             # skip '|' and '$', they don't represent arguments anyway
-            if c in '()e|$':
+            if c in '()ew|$':
                 continue
 
             # test the format unit when not skipped
@@ -1152,7 +1156,7 @@ class SkipitemTest(unittest.TestCase):
         dict_b = {'b':1}
         keywords = ["a", "b"]
 
-        supported = ('s#', 's*', 'z#', 'z*', 'y#', 'y*', 'w#', 'w*')
+        supported = ('s#', 's*', 'z#', 'z*', 'y#', 'y*', 'w*')
         for c in string.ascii_letters:
             for c2 in '#*':
                 f = c + c2

Misc/NEWS.d/next/C API/2024-03-18-17-29-52.gh-issue-68114.W7R_lI.rst

@@ -0,0 +1,2 @@
+Fixed skipitem()'s handling of the old 'w' and 'w#' formatters.  These are
+no longer supported and now raise an exception if used.

Modules/_testcapi/getargs.c

@@ -141,6 +141,122 @@ getargs_w_star(PyObject *self, PyObject *args)
     return result;
 }
 
+static PyObject *
+getargs_w_star_opt(PyObject *self, PyObject *args)
+{
+    Py_buffer buffer;
+    Py_buffer buf2;
+    int number = 1;
+
+    if (!PyArg_ParseTuple(args, "w*|w*i:getargs_w_star",
+                          &buffer, &buf2, &number)) {
+        return NULL;
+    }
+
+    if (2 <= buffer.len) {
+        char *str = buffer.buf;
+        str[0] = '[';
+        str[buffer.len-1] = ']';
+    }
+
+    PyObject *result = PyBytes_FromStringAndSize(buffer.buf, buffer.len);
+    PyBuffer_Release(&buffer);
+    return result;
+}
+
+/* Test the old w and w# codes that no longer work */
+static PyObject *
+test_w_code_invalid(PyObject *self, PyObject *arg)
+{
+    static const char * const keywords[] = {"a", "b", "c", "d", NULL};
+    char *formats_3[] = {"O|w#$O",
+                         "O|w$O",
+                         "O|w#O",
+                         "O|wO",
+                         NULL};
+    char *formats_4[] = {"O|w#O$O",
+                         "O|wO$O",
+                         "O|Ow#O",
+                         "O|OwO",
+                         "O|Ow#$O",
+                         "O|Ow$O",
+                         NULL};
+    size_t n;
+    PyObject *args;
+    PyObject *kwargs;
+    PyObject *tmp;
+
+    if (!(args = PyTuple_Pack(1, Py_None))) {
+        return NULL;
+    }
+
+    kwargs = PyDict_New();
+    if (!kwargs) {
+        Py_DECREF(args);
+        return NULL;
+    }
+
+    if (PyDict_SetItemString(kwargs, "c", Py_None)) {
+        Py_DECREF(args);
+        Py_XDECREF(kwargs);
+        return NULL;
+    }
+
+    for (n = 0; formats_3[n]; ++n) {
+        if (PyArg_ParseTupleAndKeywords(args, kwargs, formats_3[n],
+                                        (char**) keywords,
+                                        &tmp, &tmp, &tmp)) {
+            Py_DECREF(args);
+            Py_DECREF(kwargs);
+            PyErr_Format(PyExc_AssertionError,
+                         "test_w_code_invalid_suffix: %s",
+                         formats_3[n]);
+            return NULL;
+        }
+        else {
+            if (!PyErr_ExceptionMatches(PyExc_SystemError)) {
+                Py_DECREF(args);
+                Py_DECREF(kwargs);
+                return NULL;
+            }
+            PyErr_Clear();
+        }
+    }
+
+    if (PyDict_DelItemString(kwargs, "c") ||
+        PyDict_SetItemString(kwargs, "d", Py_None)) {
+
+        Py_DECREF(kwargs);
+        Py_DECREF(args);
+        return NULL;
+    }
+
+    for (n = 0; formats_4[n]; ++n) {
+        if (PyArg_ParseTupleAndKeywords(args, kwargs, formats_4[n],
+                                        (char**) keywords,
+                                        &tmp, &tmp, &tmp, &tmp)) {
+            Py_DECREF(args);
+            Py_DECREF(kwargs);
+            PyErr_Format(PyExc_AssertionError,
+                         "test_w_code_invalid_suffix: %s",
+                         formats_4[n]);
+            return NULL;
+        }
+        else {
+            if (!PyErr_ExceptionMatches(PyExc_SystemError)) {
+                Py_DECREF(args);
+                Py_DECREF(kwargs);
+                return NULL;
+            }
+            PyErr_Clear();
+        }
+    }
+
+    Py_DECREF(args);
+    Py_DECREF(kwargs);
+    Py_RETURN_NONE;
+}
+
 static PyObject *
 getargs_empty(PyObject *self, PyObject *args, PyObject *kwargs)
 {
@@ -684,6 +800,7 @@ static PyMethodDef test_methods[] = {
     {"getargs_s_star",          getargs_s_star,                  METH_VARARGS},
     {"getargs_tuple",           getargs_tuple,                   METH_VARARGS},
     {"getargs_w_star",          getargs_w_star,                  METH_VARARGS},
+    {"getargs_w_star_opt",      getargs_w_star_opt,              METH_VARARGS},
     {"getargs_empty",           _PyCFunction_CAST(getargs_empty), METH_VARARGS|METH_KEYWORDS},
     {"getargs_y",               getargs_y,                       METH_VARARGS},
     {"getargs_y_hash",          getargs_y_hash,                  METH_VARARGS},
@@ -693,6 +810,7 @@ static PyMethodDef test_methods[] = {
     {"getargs_z_star",          getargs_z_star,                  METH_VARARGS},
     {"parse_tuple_and_keywords", parse_tuple_and_keywords,       METH_VARARGS},
     {"gh_99240_clear_args",     gh_99240_clear_args,             METH_VARARGS},
+    {"test_w_code_invalid",     test_w_code_invalid,             METH_NOARGS},
     {NULL},
 };
 

Python/getargs.c

@@ -2641,6 +2641,11 @@ skipitem(const char **p_format, va_list *p_va, int flags)
             if (p_va != NULL) {
                 (void) va_arg(*p_va, char **);
             }
+            if (c == 'w' && *format != '*')
+            {
+                /* after 'w', only '*' is allowed */
+                goto err;
+            }
             if (*format == '#') {
                 if (p_va != NULL) {
                     (void) va_arg(*p_va, Py_ssize_t *);