Viewing file:  test_codecencodings_jp.py (4.22 KB) -rw-r--r--
Select action/file-type:
 (+) |  (+) |  (+) | Code (+) | Session (+) |  (+) | SDB (+) |  (+) |  (+) |  (+) |  (+) |  (+) |
#! /usr/bin/env python2.4
#
# test_codecencodings_jp.py
# Codec encoding tests for Japanese encodings.
#
# $CJKCodecs: test_codecencodings_jp.py,v 1.3 2004/06/19 06:09:55 perky Exp $
from test import test_support
from test import test_multibytecodec_support
import unittest
class Test_CP932(test_multibytecodec_support.TestBase, unittest.TestCase):
encoding = 'cp932'
tstring = test_multibytecodec_support.load_teststring('shift_jis')
codectests = (
# invalid bytes
("abc\x81\x00\x81\x00\x82\x84", "strict", None),
("abc\xf8", "strict", None),
("abc\x81\x00\x82\x84", "replace", u"abc\ufffd\uff44"),
("abc\x81\x00\x82\x84\x88", "replace", u"abc\ufffd\uff44\ufffd"),
("abc\x81\x00\x82\x84", "ignore", u"abc\uff44"),
# sjis vs cp932
("\\\x7e", "replace", u"\\\x7e"),
("\x81\x5f\x81\x61\x81\x7c", "replace", u"\uff3c\u2225\uff0d"),
)
class Test_EUC_JISX0213(test_multibytecodec_support.TestBase,
unittest.TestCase):
encoding = 'euc_jisx0213'
tstring = test_multibytecodec_support.load_teststring('euc_jisx0213')
codectests = (
# invalid bytes
("abc\x80\x80\xc1\xc4", "strict", None),
("abc\xc8", "strict", None),
("abc\x80\x80\xc1\xc4", "replace", u"abc\ufffd\u7956"),
("abc\x80\x80\xc1\xc4\xc8", "replace", u"abc\ufffd\u7956\ufffd"),
("abc\x80\x80\xc1\xc4", "ignore", u"abc\u7956"),
("abc\x8f\x83\x83", "replace", u"abc\ufffd"),
("\xc1\x64", "strict", None),
("\xa1\xc0", "strict", u"\uff3c"),
)
xmlcharnametest = (
u"\xab\u211c\xbb = \u2329\u1234\u232a",
"\xa9\xa8ℜ\xa9\xb2 = ⟨ሴ⟩"
)
eucjp_commontests = (
("abc\x80\x80\xc1\xc4", "strict", None),
("abc\xc8", "strict", None),
("abc\x80\x80\xc1\xc4", "replace", u"abc\ufffd\u7956"),
("abc\x80\x80\xc1\xc4\xc8", "replace", u"abc\ufffd\u7956\ufffd"),
("abc\x80\x80\xc1\xc4", "ignore", u"abc\u7956"),
("abc\x8f\x83\x83", "replace", u"abc\ufffd"),
("\xc1\x64", "strict", None),
)
class Test_EUC_JP_COMPAT(test_multibytecodec_support.TestBase,
unittest.TestCase):
encoding = 'euc_jp'
tstring = test_multibytecodec_support.load_teststring('euc_jp')
codectests = eucjp_commontests + (
("\xa1\xc0\\", "strict", u"\uff3c\\"),
(u"\xa5", "strict", "\x5c"),
(u"\u203e", "strict", "\x7e"),
)
shiftjis_commonenctests = (
("abc\x80\x80\x82\x84", "strict", None),
("abc\xf8", "strict", None),
("abc\x80\x80\x82\x84", "replace", u"abc\ufffd\uff44"),
("abc\x80\x80\x82\x84\x88", "replace", u"abc\ufffd\uff44\ufffd"),
("abc\x80\x80\x82\x84def", "ignore", u"abc\uff44def"),
)
class Test_SJIS_COMPAT(test_multibytecodec_support.TestBase, unittest.TestCase):
encoding = 'shift_jis'
tstring = test_multibytecodec_support.load_teststring('shift_jis')
codectests = shiftjis_commonenctests + (
("\\\x7e", "strict", u"\\\x7e"),
("\x81\x5f\x81\x61\x81\x7c", "strict", u"\uff3c\u2016\u2212"),
)
class Test_SJISX0213(test_multibytecodec_support.TestBase, unittest.TestCase):
encoding = 'shift_jisx0213'
tstring = test_multibytecodec_support.load_teststring('shift_jisx0213')
codectests = (
# invalid bytes
("abc\x80\x80\x82\x84", "strict", None),
("abc\xf8", "strict", None),
("abc\x80\x80\x82\x84", "replace", u"abc\ufffd\uff44"),
("abc\x80\x80\x82\x84\x88", "replace", u"abc\ufffd\uff44\ufffd"),
("abc\x80\x80\x82\x84def", "ignore", u"abc\uff44def"),
# sjis vs cp932
("\\\x7e", "replace", u"\xa5\u203e"),
("\x81\x5f\x81\x61\x81\x7c", "replace", u"\x5c\u2016\u2212"),
)
xmlcharnametest = (
u"\xab\u211c\xbb = \u2329\u1234\u232a",
"\x85Gℜ\x85Q = ⟨ሴ⟩"
)
def test_main():
suite = unittest.TestSuite()
suite.addTest(unittest.makeSuite(Test_CP932))
suite.addTest(unittest.makeSuite(Test_EUC_JISX0213))
suite.addTest(unittest.makeSuite(Test_EUC_JP_COMPAT))
suite.addTest(unittest.makeSuite(Test_SJIS_COMPAT))
suite.addTest(unittest.makeSuite(Test_SJISX0213))
test_support.run_suite(suite)
if __name__ == "__main__":
test_main()
|