Apply Victor's suggestions, and actually fix the bug

Co-authored-by: vstinner@python.org

Author: StanFromIreland

Modules/_remote_debugging/object_reading.c

@@ -6,6 +6,7 @@
  ******************************************************************************/
 
 #include "_remote_debugging.h"
+#include "internal/pycore_unicodeobject.h"  // _Py_MAX_UNICODE
 #include <limits.h>
 
 /* ============================================================================
@@ -153,6 +154,22 @@ read_py_str(
         }
     }
 
+    if (kind == PyUnicode_4BYTE_KIND) {
+        Py_UCS4 *ucs4 = (Py_UCS4 *)data;
+        for (Py_ssize_t i = 0; i < len; i++) {
+            Py_UCS4 ch = ucs4[i];
+            if (ch > _Py_MAX_UNICODE) {
+                PyMem_Free(data);
+                PyErr_Format(PyExc_ValueError,
+                             "character U+%x is not in range [U+0000; U+%x]",
+                             ch, _Py_MAX_UNICODE);
+                set_exception_cause(unwinder, PyExc_ValueError,
+                                    "Unicode string contains invalid character");
+                return NULL;
+            }
+        }
+    }
+
     PyObject *result = PyUnicode_FromKindAndData(kind, data, len);
     PyMem_Free(data);
     if (result == NULL) {

Modules/_testinternalcapi.c

@@ -2056,15 +2056,16 @@ unicode_transformdecimalandspacetoascii(PyObject *self, PyObject *arg)
     return _PyUnicode_TransformDecimalAndSpaceToASCII(arg);
 }
 
-/* Return a one-character str holding a code point above U+10FFFF. */
+/* Return a str holding code points above U+10FFFF. */
 static PyObject *
 new_invalid_str(PyObject *self, PyObject *Py_UNUSED(ignored))
 {
-    PyObject *str = PyUnicode_New(1, 0x10FFFF);
+    PyObject *str = PyUnicode_New(2, 0x10FFFF);
     if (str == NULL) {
         return NULL;
     }
     PyUnicode_4BYTE_DATA(str)[0] = 0x110000;
+    PyUnicode_4BYTE_DATA(str)[1] = 0x110000;
     return str;
 }