gh-146563: add exception note for invalid Expat handler return values (#146565)

Author: picnixz

Include/internal/pycore_pyerrors.h

@@ -29,7 +29,8 @@ PyAPI_FUNC(PyObject*) _PyErr_FormatFromCause(
     ...
     );
 
-extern int _PyException_AddNote(
+// Export for 'pyexpat' shared extension.
+PyAPI_FUNC(int) _PyException_AddNote(
      PyObject *exc,
      PyObject *note);
 

Lib/test/test_pyexpat.py

@@ -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):

Misc/NEWS.d/next/Library/2026-03-28-11-31-32.gh-issue-146563.cXtSym.rst

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

Modules/pyexpat.c

@@ -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;
+    }
+    int rc = _PyException_AddNote(exc, note);
+    Py_DECREF(note);
+    if (rc < 0) {
+        goto error;
+    };
+    goto done;
+
+error:
+    PyErr_Clear();
+done:
+    PyErr_SetRaisedException(exc);
+}
+
 #define RC_HANDLER(RETURN_TYPE, NAME, PARAMS,       \
                    INIT, PARSE_FORMAT, CONVERSION,  \
                    RETURN_VARIABLE, GETUSERDATA)    \
@@ -536,6 +558,9 @@ my_ ## NAME ## Handler PARAMS {                     \
     }                                               \
     CONVERSION                                      \
     Py_DECREF(rv);                                  \
+    if (PyErr_Occurred()) {                         \
+        invalid_expat_handler_rv(#NAME);            \
+    }                                               \
     return RETURN_VARIABLE;                         \
 }