[3.12] pythongh-111942: Fix SystemError in the TextIOWrapper constructor (pythonGH-112061) (pythonGH-112089)

In non-debug more the check for the "errors" argument is skipped,
and then PyUnicode_AsUTF8() can fail, but its result was not checked.

Co-authored-by: Victor Stinner <[email protected]>
(cherry picked from commit 9302f05)

Co-authored-by: Serhiy Storchaka <[email protected]>

Author: vstinner

Lib/test/test_io.py

@@ -2629,9 +2629,7 @@ def test_constructor(self):
         if support.Py_DEBUG or sys.flags.dev_mode or self.is_C:
             with self.assertRaises(UnicodeEncodeError):
                 t.__init__(b, encoding="utf-8", errors='\udcfe')
-        if support.Py_DEBUG or sys.flags.dev_mode:
-            # TODO: If encoded to UTF-8, should also be checked for
-            # embedded null characters.
+        if support.Py_DEBUG or sys.flags.dev_mode or self.is_C:
             with self.assertRaises(ValueError):
                 t.__init__(b, encoding="utf-8", errors='replace\0')
         with self.assertRaises(TypeError):

Misc/NEWS.d/next/Library/2023-11-14-18-43-55.gh-issue-111942.x1pnrj.rst

@@ -0,0 +1,2 @@
+Fix SystemError in the TextIOWrapper constructor with non-encodable "errors"
+argument in non-debug mode.

Modules/_io/textio.c

@@ -1099,6 +1099,15 @@ _io_TextIOWrapper___init___impl(textio *self, PyObject *buffer,
     else if (io_check_errors(errors)) {
         return -1;
     }
+    Py_ssize_t errors_len;
+    const char *errors_str = PyUnicode_AsUTF8AndSize(errors, &errors_len);
+    if (errors_str == NULL) {
+        return -1;
+    }
+    if (strlen(errors_str) != (size_t)errors_len) {
+        PyErr_SetString(PyExc_ValueError, "embedded null character");
+        return -1;
+    }
 
     if (validate_newline(newline) < 0) {
         return -1;
@@ -1171,11 +1180,11 @@ _io_TextIOWrapper___init___impl(textio *self, PyObject *buffer,
     Py_INCREF(buffer);
 
     /* Build the decoder object */
-    if (_textiowrapper_set_decoder(self, codec_info, PyUnicode_AsUTF8(errors)) != 0)
+    if (_textiowrapper_set_decoder(self, codec_info, errors_str) != 0)
         goto error;
 
     /* Build the encoder object */
-    if (_textiowrapper_set_encoder(self, codec_info, PyUnicode_AsUTF8(errors)) != 0)
+    if (_textiowrapper_set_encoder(self, codec_info, errors_str) != 0)
         goto error;
 
     /* Finished sorting out the codec details */