gh-146563: add exception note for invalid Expat handler return values

Author: picnixz

Include/internal/pycore_global_objects_fini_generated.h

@@ -299,6 +299,7 @@ struct _Py_global_strings {
         STRUCT_FOR_ID(aclose)
         STRUCT_FOR_ID(add)
         STRUCT_FOR_ID(add_done_callback)
+        STRUCT_FOR_ID(add_note)
         STRUCT_FOR_ID(adobe)
         STRUCT_FOR_ID(after_in_child)
         STRUCT_FOR_ID(after_in_parent)

Include/internal/pycore_global_strings.h

@@ -510,6 +510,34 @@ def _test_exception(self, have_source):
             self.assertIn('call_with_frame("StartElement"',
                           entries[1].line)
 
+    def test_invalid_NotStandalone(self):
+        parser = expat.ParserCreate()
+        parser.NotStandaloneHandler = mock.Mock(return_value="bad value")
+        parser.ElementDeclHandler = lambda _1, _2: None
+
+        payload = b"""\
+<!DOCTYPE quotations SYSTEM "quotations.dtd" [<!ELEMENT root ANY>]><root/>
+"""
+        with self.assertRaises(TypeError) as cm:
+            parser.Parse(payload, True)
+        parser.NotStandaloneHandler.assert_called_once()
+
+        notes = ["invalid 'NotStandalone' event handler return value"]
+        self.assertEqual(cm.exception.__notes__, notes)
+
+    def test_invalid_ExternalEntityRefHandler(self):
+        parser = expat.ParserCreate()
+        parser.UseForeignDTD()
+        parser.SetParamEntityParsing(expat.XML_PARAM_ENTITY_PARSING_ALWAYS)
+        parser.ExternalEntityRefHandler = mock.Mock(return_value=None)
+
+        with self.assertRaises(TypeError) as cm:
+            parser.Parse(b"<?xml version='1.0'?><element/>", True)
+        parser.ExternalEntityRefHandler.assert_called_once()
+
+        notes = ["invalid 'ExternalEntityRef' event handler return value"]
+        self.assertEqual(cm.exception.__notes__, notes)
+
 
 # Test Current* members:
 class PositionTest(unittest.TestCase):

Include/internal/pycore_runtime_init_generated.h

@@ -0,0 +1,2 @@
+:mod:`xml.parser.expat`: add an exception note when a custom Expat handler
+return value cannot be properly interpreted. Patch by Bénédikt Tran.

Include/internal/pycore_unicodeobject_generated.h

@@ -503,6 +503,28 @@ my_StartElementHandler(void *userData,
     }
 }
 
+static inline void
+invalid_expat_handler_rv(const char *name)
+{
+    PyObject *exc = PyErr_GetRaisedException();
+    assert(exc != NULL);
+    PyObject *note = PyUnicode_FromFormat("invalid '%s' event handler return value", name);
+    if (note == NULL) {
+        goto error;
+    }
+    PyObject *res = PyObject_CallMethodOneArg(exc, &_Py_ID(add_note), note);
+    Py_DECREF(note);
+    if (res == NULL) {
+        goto error;
+    }
+    goto done;
+
+error:
+    PyErr_Clear();
+done:
+    _PyErr_ChainExceptions1(exc);
+}
+
 #define RC_HANDLER(RETURN_TYPE, NAME, PARAMS,       \
                    INIT, PARSE_FORMAT, CONVERSION,  \
                    RETURN_VARIABLE, GETUSERDATA)    \
@@ -536,6 +558,10 @@ my_ ## NAME ## Handler PARAMS {                     \
     }                                               \
     CONVERSION                                      \
     Py_DECREF(rv);                                  \
+    if (PyErr_Occurred()) {                         \
+        invalid_expat_handler_rv(#NAME);            \
+        return RETURN_VARIABLE;                     \
+    }                                               \
     return RETURN_VARIABLE;                         \
 }