#6373: SystemError in str.encode('latin1', 'surrogateescape')
if the string contains unpaired surrogates.
(In debug build, crash in assert())
This can happen with normal processing, if python starts with utf-8,
then calls sys.setfilesystemencoding('latin-1')
diff --git a/Lib/test/test_codecs.py b/Lib/test/test_codecs.py
index 4ec7b58..e060471 100644
--- a/Lib/test/test_codecs.py
+++ b/Lib/test/test_codecs.py
@@ -1549,6 +1549,11 @@
self.assertEqual("foo\udca5bar".encode("iso-8859-3", "surrogateescape"),
b"foo\xa5bar")
+ def test_latin1(self):
+ # Issue6373
+ self.assertEqual("\udce4\udceb\udcef\udcf6\udcfc".encode("latin1", "surrogateescape"),
+ b"\xe4\xeb\xef\xf6\xfc")
+
def test_main():
support.run_unittest(
diff --git a/Misc/NEWS b/Misc/NEWS
index 2381844..98539e5 100644
--- a/Misc/NEWS
+++ b/Misc/NEWS
@@ -12,6 +12,10 @@
Core and Builtins
-----------------
+- Issue #6373: Fixed a RuntimeError when encoding with the latin-1 codec and
+ the 'surrogateescape' error handler, a string which contains unpaired
+ surrogates.
+
- Issue #4856: Remove checks for win NT.
Library
diff --git a/Objects/unicodeobject.c b/Objects/unicodeobject.c
index 0d4a3dd..305289b 100644
--- a/Objects/unicodeobject.c
+++ b/Objects/unicodeobject.c
@@ -4201,10 +4201,12 @@
repsize = PyBytes_Size(repunicode);
if (repsize > 1) {
/* Make room for all additional bytes. */
+ respos = str - PyBytes_AS_STRING(res);
if (_PyBytes_Resize(&res, ressize+repsize-1)) {
Py_DECREF(repunicode);
goto onError;
}
+ str = PyBytes_AS_STRING(res) + respos;
ressize += repsize-1;
}
memcpy(str, PyBytes_AsString(repunicode), repsize);