From 35804e4c63ae0a61adb71ced8ea6ddcf68908d41 Mon Sep 17 00:00:00 2001 From: Serhiy Storchaka Date: Sat, 19 Oct 2013 20:38:19 +0300 Subject: Issue #19279: UTF-7 decoder no more produces illegal strings. --- Lib/test/test_codecs.py | 30 ++++++++++++++++++++++++++++++ 1 file changed, 30 insertions(+) (limited to 'Lib/test/test_codecs.py') diff --git a/Lib/test/test_codecs.py b/Lib/test/test_codecs.py index c68088ed75..80ec541abe 100644 --- a/Lib/test/test_codecs.py +++ b/Lib/test/test_codecs.py @@ -820,6 +820,36 @@ class UTF7Test(ReadTest, unittest.TestCase): ] ) + def test_errors(self): + tests = [ + (b'a\xffb', 'a\ufffdb'), + (b'a+IK', 'a\ufffd'), + (b'a+IK-b', 'a\ufffdb'), + (b'a+IK,b', 'a\ufffdb'), + (b'a+IKx', 'a\u20ac\ufffd'), + (b'a+IKx-b', 'a\u20ac\ufffdb'), + (b'a+IKwgr', 'a\u20ac\ufffd'), + (b'a+IKwgr-b', 'a\u20ac\ufffdb'), + (b'a+IKwgr,', 'a\u20ac\ufffd'), + (b'a+IKwgr,-b', 'a\u20ac\ufffd-b'), + (b'a+IKwgrB', 'a\u20ac\u20ac\ufffd'), + (b'a+IKwgrB-b', 'a\u20ac\u20ac\ufffdb'), + (b'a+/,+IKw-b', 'a\ufffd\u20acb'), + (b'a+//,+IKw-b', 'a\ufffd\u20acb'), + (b'a+///,+IKw-b', 'a\uffff\ufffd\u20acb'), + (b'a+////,+IKw-b', 'a\uffff\ufffd\u20acb'), + ] + for raw, expected in tests: + with self.subTest(raw=raw): + self.assertRaises(UnicodeDecodeError, codecs.utf_7_decode, + raw, 'strict', True) + self.assertEqual(raw.decode('utf-7', 'replace'), expected) + + def test_nonbmp(self): + self.assertEqual('\U000104A0'.encode(self.encoding), b'+2AHcoA-') + self.assertEqual('\ud801\udca0'.encode(self.encoding), b'+2AHcoA-') + self.assertEqual(b'+2AHcoA-'.decode(self.encoding), '\U000104A0') + class UTF16ExTest(unittest.TestCase): def test_errors(self): -- cgit v1.2.1