
Function 'safe_encode ' in 'encodeutils' module treats 'UTF-8' and 'utf-8' encodings as different. But it should understand different aliases, that have different text cases. It allows us avoid redundant coding/decoding. Also added unittests. Change-Id: I4c446952fc904c1231cccbda1cd4d2a4cce5c55f Closes-Bug: #1342050
106 lines
4.0 KiB
Python
106 lines
4.0 KiB
Python
# -*- coding: utf-8 -*-
|
|
|
|
# Copyright 2014 Red Hat, Inc.
|
|
# All Rights Reserved.
|
|
#
|
|
# Licensed under the Apache License, Version 2.0 (the "License"); you may
|
|
# not use this file except in compliance with the License. You may obtain
|
|
# a copy of the License at
|
|
#
|
|
# http://www.apache.org/licenses/LICENSE-2.0
|
|
#
|
|
# Unless required by applicable law or agreed to in writing, software
|
|
# distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
|
|
# WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
|
|
# License for the specific language governing permissions and limitations
|
|
# under the License.
|
|
|
|
import mock
|
|
from oslotest import base as test_base
|
|
import six
|
|
|
|
from oslo.utils import encodeutils
|
|
|
|
|
|
class EncodeUtilsTest(test_base.BaseTestCase):
|
|
|
|
def test_safe_decode(self):
|
|
safe_decode = encodeutils.safe_decode
|
|
self.assertRaises(TypeError, safe_decode, True)
|
|
self.assertEqual(six.u('ni\xf1o'), safe_decode(six.b("ni\xc3\xb1o"),
|
|
incoming="utf-8"))
|
|
if six.PY2:
|
|
# In Python 3, bytes.decode() doesn't support anymore
|
|
# bytes => bytes encodings like base64
|
|
self.assertEqual(six.u("test"), safe_decode("dGVzdA==",
|
|
incoming='base64'))
|
|
|
|
self.assertEqual(six.u("strange"), safe_decode(six.b('\x80strange'),
|
|
errors='ignore'))
|
|
|
|
self.assertEqual(six.u('\xc0'), safe_decode(six.b('\xc0'),
|
|
incoming='iso-8859-1'))
|
|
|
|
# Forcing incoming to ascii so it falls back to utf-8
|
|
self.assertEqual(six.u('ni\xf1o'), safe_decode(six.b('ni\xc3\xb1o'),
|
|
incoming='ascii'))
|
|
|
|
self.assertEqual(six.u('foo'), safe_decode(b'foo'))
|
|
|
|
def test_safe_encode_none_instead_of_text(self):
|
|
self.assertRaises(TypeError, encodeutils.safe_encode, None)
|
|
|
|
def test_safe_encode_bool_instead_of_text(self):
|
|
self.assertRaises(TypeError, encodeutils.safe_encode, True)
|
|
|
|
def test_safe_encode_int_instead_of_text(self):
|
|
self.assertRaises(TypeError, encodeutils.safe_encode, 1)
|
|
|
|
def test_safe_encode_list_instead_of_text(self):
|
|
self.assertRaises(TypeError, encodeutils.safe_encode, [])
|
|
|
|
def test_safe_encode_dict_instead_of_text(self):
|
|
self.assertRaises(TypeError, encodeutils.safe_encode, {})
|
|
|
|
def test_safe_encode_tuple_instead_of_text(self):
|
|
self.assertRaises(TypeError, encodeutils.safe_encode, ('foo', 'bar', ))
|
|
|
|
def test_safe_encode_py2(self):
|
|
if six.PY2:
|
|
# In Python 3, str.encode() doesn't support anymore
|
|
# text => text encodings like base64
|
|
self.assertEqual(
|
|
six.b("dGVzdA==\n"),
|
|
encodeutils.safe_encode("test", encoding='base64'),
|
|
)
|
|
else:
|
|
self.skipTest("Requires py2.x")
|
|
|
|
def test_safe_encode_force_incoming_utf8_to_ascii(self):
|
|
# Forcing incoming to ascii so it falls back to utf-8
|
|
self.assertEqual(
|
|
six.b('ni\xc3\xb1o'),
|
|
encodeutils.safe_encode(six.b('ni\xc3\xb1o'), incoming='ascii'),
|
|
)
|
|
|
|
def test_safe_encode_same_encoding_different_cases(self):
|
|
with mock.patch.object(encodeutils, 'safe_decode', mock.Mock()):
|
|
utf8 = encodeutils.safe_encode(
|
|
six.u('foo\xf1bar'), encoding='utf-8')
|
|
self.assertEqual(
|
|
encodeutils.safe_encode(utf8, 'UTF-8', 'utf-8'),
|
|
encodeutils.safe_encode(utf8, 'utf-8', 'UTF-8'),
|
|
)
|
|
self.assertEqual(
|
|
encodeutils.safe_encode(utf8, 'UTF-8', 'utf-8'),
|
|
encodeutils.safe_encode(utf8, 'utf-8', 'utf-8'),
|
|
)
|
|
encodeutils.safe_decode.assert_has_calls([])
|
|
|
|
def test_safe_encode_different_encodings(self):
|
|
text = six.u('foo\xc3\xb1bar')
|
|
result = encodeutils.safe_encode(
|
|
text=text, incoming='utf-8', encoding='iso-8859-1')
|
|
self.assertNotEqual(text, result)
|
|
self.assertNotEqual(six.b("foo\xf1bar"), result)
|