Special case normalization of empty strings. Fixes #924361.
Backported to 2.3.
diff --git a/Lib/test/test_unicodedata.py b/Lib/test/test_unicodedata.py
index 61b4ffb..8157fb3 100644
--- a/Lib/test/test_unicodedata.py
+++ b/Lib/test/test_unicodedata.py
@@ -170,6 +170,7 @@
def test_normalize(self):
self.assertRaises(TypeError, self.db.normalize)
self.assertRaises(ValueError, self.db.normalize, 'unknown', u'xx')
+ self.assertEqual(self.db.normalize('NFKC', u''), u'')
# The rest can be found in test_normalization.py
# which requires an external file.
diff --git a/Modules/unicodedata.c b/Modules/unicodedata.c
index 311db29..ba218a3 100644
--- a/Modules/unicodedata.c
+++ b/Modules/unicodedata.c
@@ -515,6 +515,13 @@
&form, &PyUnicode_Type, &input))
return NULL;
+ if (PyUnicode_GetSize(input) == 0) {
+ /* Special case empty input strings, since resizing
+ them later would cause internal errors. */
+ Py_INCREF(input);
+ return input;
+ }
+
if (strcmp(form, "NFC") == 0)
return nfc_nfkc(input, 0);
if (strcmp(form, "NFKC") == 0)