[Python-checkins] bpo-34482: test datetime classes' handling of non-UTF-8-encodable strings (GH-8878)

Tue Oct 23 02:36:11 EDT 2018

https://github.com/python/cpython/commit/3b0047d8e982b10b34ab05fd207b7d513cc1188a
commit: 3b0047d8e982b10b34ab05fd207b7d513cc1188a
branch: master
author: Alexey Izbyshev <izbyshev at ispras.ru>
committer: Tal Einat <taleinat+github at gmail.com>
date: 2018年10月23日T09:36:08+03:00
summary:
bpo-34482: test datetime classes' handling of non-UTF-8-encodable strings (GH-8878)
files:
M Lib/test/datetimetester.py

diff --git a/Lib/test/datetimetester.py b/Lib/test/datetimetester.py
index 122f6b55ba33..06fe647fb1a4 100644
--- a/Lib/test/datetimetester.py
+++ b/Lib/test/datetimetester.py
@@ -303,6 +303,8 @@ def test_tzname(self):
 self.assertEqual('UTC+09:30', timezone(9.5 * HOUR).tzname(None))
 self.assertEqual('UTC-00:01', timezone(timedelta(minutes=-1)).tzname(None))
 self.assertEqual('XYZ', timezone(-5 * HOUR, 'XYZ').tzname(None))
+ # bpo-34482: Check that surrogates are handled properly.
+ self.assertEqual('\ud800', timezone(ZERO, '\ud800').tzname(None))
 
 # Sub-minute offsets:
 self.assertEqual('UTC+01:06:40', timezone(timedelta(0, 4000)).tzname(None))
@@ -1308,6 +1310,12 @@ def test_strftime(self):
 except ValueError:
 pass
 
+ # bpo-34482: Check that surrogates don't cause a crash.
+ try:
+ t.strftime('%y\ud800%m')
+ except UnicodeEncodeError:
+ pass
+
 #check that this standard extension works
 t.strftime("%f")
 
@@ -1747,6 +1755,9 @@ def test_isoformat(self):
 self.assertEqual(t.isoformat('T'), "0001年02月03日T04:05:01.000123")
 self.assertEqual(t.isoformat(' '), "0001年02月03日 04:05:01.000123")
 self.assertEqual(t.isoformat('\x00'), "0001年02月03日\x0004:05:01.000123")
+ # bpo-34482: Check that surrogates are handled properly.
+ self.assertEqual(t.isoformat('\ud800'),
+ "0001年02月03日\ud80004:05:01.000123")
 self.assertEqual(t.isoformat(timespec='hours'), "0001年02月03日T04")
 self.assertEqual(t.isoformat(timespec='minutes'), "0001年02月03日T04:05")
 self.assertEqual(t.isoformat(timespec='seconds'), "0001年02月03日T04:05:01")
@@ -1755,6 +1766,8 @@ def test_isoformat(self):
 self.assertEqual(t.isoformat(timespec='auto'), "0001年02月03日T04:05:01.000123")
 self.assertEqual(t.isoformat(sep=' ', timespec='minutes'), "0001年02月03日 04:05")
 self.assertRaises(ValueError, t.isoformat, timespec='foo')
+ # bpo-34482: Check that surrogates are handled properly.
+ self.assertRaises(ValueError, t.isoformat, timespec='\ud800')
 # str is ISO format with the separator forced to a blank.
 self.assertEqual(str(t), "0001年02月03日 04:05:01.000123")
 
@@ -2286,6 +2299,19 @@ def test_strptime(self):
 self.assertIs(type(expected), self.theclass)
 self.assertIs(type(got), self.theclass)
 
+ # bpo-34482: Check that surrogates are handled properly.
+ inputs = [
+ ('2004-12-01\ud80013:02:47.197', '%Y-%m-%d\ud800%H:%M:%S.%f'),
+ ('2004\ud80012-01 13:02:47.197', '%Y\ud800%m-%d %H:%M:%S.%f'),
+ ('2004-12-01 13:02\ud80047.197', '%Y-%m-%d %H:%M\ud800%S.%f'),
+ ]
+ for string, format in inputs:
+ with self.subTest(string=string, format=format):
+ expected = _strptime._strptime_datetime(self.theclass, string,
+ format)
+ got = self.theclass.strptime(string, format)
+ self.assertEqual(expected, got)
+
 strptime = self.theclass.strptime
 self.assertEqual(strptime("+0002", "%z").utcoffset(), 2 * MINUTE)
 self.assertEqual(strptime("-0002", "%z").utcoffset(), -2 * MINUTE)
@@ -2353,6 +2379,12 @@ def test_more_strftime(self):
 t = t.replace(tzinfo=tz)
 self.assertEqual(t.strftime("%z"), "-0200" + z)
 
+ # bpo-34482: Check that surrogates don't cause a crash.
+ try:
+ t.strftime('%y\ud800%m %H\ud800%M')
+ except UnicodeEncodeError:
+ pass
+
 def test_extract(self):
 dt = self.theclass(2002, 3, 4, 18, 45, 3, 1234)
 self.assertEqual(dt.date(), date(2002, 3, 4))
@@ -2878,6 +2910,8 @@ def test_isoformat(self):
 self.assertEqual(t.isoformat(timespec='microseconds'), "12:34:56.123456")
 self.assertEqual(t.isoformat(timespec='auto'), "12:34:56.123456")
 self.assertRaises(ValueError, t.isoformat, timespec='monkey')
+ # bpo-34482: Check that surrogates are handled properly.
+ self.assertRaises(ValueError, t.isoformat, timespec='\ud800')
 
 t = self.theclass(hour=12, minute=34, second=56, microsecond=999500)
 self.assertEqual(t.isoformat(timespec='milliseconds'), "12:34:56.999")
@@ -2928,6 +2962,12 @@ def test_strftime(self):
 # A naive object replaces %z and %Z with empty strings.
 self.assertEqual(t.strftime("'%z' '%Z'"), "'' ''")
 
+ # bpo-34482: Check that surrogates don't cause a crash.
+ try:
+ t.strftime('%H\ud800%M')
+ except UnicodeEncodeError:
+ pass
+
 def test_format(self):
 t = self.theclass(1, 2, 3, 4)
 self.assertEqual(t.__format__(''), str(t))
