Issue #3672: Reject surrogates in utf-8 codec; add surrogates error
handler.
diff --git a/Python/marshal.c b/Python/marshal.c
index bf7a26b..4ad873e 100644
--- a/Python/marshal.c
+++ b/Python/marshal.c
@@ -312,7 +312,9 @@
}
else if (PyUnicode_CheckExact(v)) {
PyObject *utf8;
- utf8 = PyUnicode_AsUTF8String(v);
+ utf8 = PyUnicode_EncodeUTF8(PyUnicode_AS_UNICODE(v),
+ PyUnicode_GET_SIZE(v),
+ "surrogates");
if (utf8 == NULL) {
p->depth--;
p->error = WFERR_UNMARSHALLABLE;
@@ -810,7 +812,7 @@
retval = NULL;
break;
}
- v = PyUnicode_DecodeUTF8(buffer, n, NULL);
+ v = PyUnicode_DecodeUTF8(buffer, n, "surrogates");
PyMem_DEL(buffer);
retval = v;
break;