test_codecencodings_jp.py 4.8 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126
  1. #
  2. # test_codecencodings_jp.py
  3. # Codec encoding tests for Japanese encodings.
  4. #
  5. from test import multibytecodec_support
  6. import unittest
  7. class Test_CP932(multibytecodec_support.TestBase, unittest.TestCase):
  8. encoding = 'cp932'
  9. tstring = multibytecodec_support.load_teststring('shift_jis')
  10. codectests = (
  11. # invalid bytes
  12. (b"abc\x81\x00\x81\x00\x82\x84", "strict", None),
  13. (b"abc\xf8", "strict", None),
  14. (b"abc\x81\x00\x82\x84", "replace", "abc\ufffd\x00\uff44"),
  15. (b"abc\x81\x00\x82\x84\x88", "replace", "abc\ufffd\x00\uff44\ufffd"),
  16. (b"abc\x81\x00\x82\x84", "ignore", "abc\x00\uff44"),
  17. (b"ab\xEBxy", "replace", "ab\uFFFDxy"),
  18. (b"ab\xF0\x39xy", "replace", "ab\uFFFD9xy"),
  19. (b"ab\xEA\xF0xy", "replace", 'ab\ufffd\ue038y'),
  20. # sjis vs cp932
  21. (b"\\\x7e", "replace", "\\\x7e"),
  22. (b"\x81\x5f\x81\x61\x81\x7c", "replace", "\uff3c\u2225\uff0d"),
  23. )
  24. euc_commontests = (
  25. # invalid bytes
  26. (b"abc\x80\x80\xc1\xc4", "strict", None),
  27. (b"abc\x80\x80\xc1\xc4", "replace", "abc\ufffd\ufffd\u7956"),
  28. (b"abc\x80\x80\xc1\xc4\xc8", "replace", "abc\ufffd\ufffd\u7956\ufffd"),
  29. (b"abc\x80\x80\xc1\xc4", "ignore", "abc\u7956"),
  30. (b"abc\xc8", "strict", None),
  31. (b"abc\x8f\x83\x83", "replace", "abc\ufffd\ufffd\ufffd"),
  32. (b"\x82\xFCxy", "replace", "\ufffd\ufffdxy"),
  33. (b"\xc1\x64", "strict", None),
  34. (b"\xa1\xc0", "strict", "\uff3c"),
  35. (b"\xa1\xc0\\", "strict", "\uff3c\\"),
  36. (b"\x8eXY", "replace", "\ufffdXY"),
  37. )
  38. class Test_EUC_JIS_2004(multibytecodec_support.TestBase,
  39. unittest.TestCase):
  40. encoding = 'euc_jis_2004'
  41. tstring = multibytecodec_support.load_teststring('euc_jisx0213')
  42. codectests = euc_commontests
  43. xmlcharnametest = (
  44. "\xab\u211c\xbb = \u2329\u1234\u232a",
  45. b"\xa9\xa8ℜ\xa9\xb2 = ⟨ሴ⟩"
  46. )
  47. class Test_EUC_JISX0213(multibytecodec_support.TestBase,
  48. unittest.TestCase):
  49. encoding = 'euc_jisx0213'
  50. tstring = multibytecodec_support.load_teststring('euc_jisx0213')
  51. codectests = euc_commontests
  52. xmlcharnametest = (
  53. "\xab\u211c\xbb = \u2329\u1234\u232a",
  54. b"\xa9\xa8ℜ\xa9\xb2 = ⟨ሴ⟩"
  55. )
  56. class Test_EUC_JP_COMPAT(multibytecodec_support.TestBase,
  57. unittest.TestCase):
  58. encoding = 'euc_jp'
  59. tstring = multibytecodec_support.load_teststring('euc_jp')
  60. codectests = euc_commontests + (
  61. ("\xa5", "strict", b"\x5c"),
  62. ("\u203e", "strict", b"\x7e"),
  63. )
  64. shiftjis_commonenctests = (
  65. (b"abc\x80\x80\x82\x84", "strict", None),
  66. (b"abc\xf8", "strict", None),
  67. (b"abc\x80\x80\x82\x84def", "ignore", "abc\uff44def"),
  68. )
  69. class Test_SJIS_COMPAT(multibytecodec_support.TestBase, unittest.TestCase):
  70. encoding = 'shift_jis'
  71. tstring = multibytecodec_support.load_teststring('shift_jis')
  72. codectests = shiftjis_commonenctests + (
  73. (b"abc\x80\x80\x82\x84", "replace", "abc\ufffd\ufffd\uff44"),
  74. (b"abc\x80\x80\x82\x84\x88", "replace", "abc\ufffd\ufffd\uff44\ufffd"),
  75. (b"\\\x7e", "strict", "\\\x7e"),
  76. (b"\x81\x5f\x81\x61\x81\x7c", "strict", "\uff3c\u2016\u2212"),
  77. (b"abc\x81\x39", "replace", "abc\ufffd9"),
  78. (b"abc\xEA\xFC", "replace", "abc\ufffd\ufffd"),
  79. (b"abc\xFF\x58", "replace", "abc\ufffdX"),
  80. )
  81. class Test_SJIS_2004(multibytecodec_support.TestBase, unittest.TestCase):
  82. encoding = 'shift_jis_2004'
  83. tstring = multibytecodec_support.load_teststring('shift_jis')
  84. codectests = shiftjis_commonenctests + (
  85. (b"\\\x7e", "strict", "\xa5\u203e"),
  86. (b"\x81\x5f\x81\x61\x81\x7c", "strict", "\\\u2016\u2212"),
  87. (b"abc\xEA\xFC", "strict", "abc\u64bf"),
  88. (b"\x81\x39xy", "replace", "\ufffd9xy"),
  89. (b"\xFF\x58xy", "replace", "\ufffdXxy"),
  90. (b"\x80\x80\x82\x84xy", "replace", "\ufffd\ufffd\uff44xy"),
  91. (b"\x80\x80\x82\x84\x88xy", "replace", "\ufffd\ufffd\uff44\u5864y"),
  92. (b"\xFC\xFBxy", "replace", '\ufffd\u95b4y'),
  93. )
  94. xmlcharnametest = (
  95. "\xab\u211c\xbb = \u2329\u1234\u232a",
  96. b"\x85Gℜ\x85Q = ⟨ሴ⟩"
  97. )
  98. class Test_SJISX0213(multibytecodec_support.TestBase, unittest.TestCase):
  99. encoding = 'shift_jisx0213'
  100. tstring = multibytecodec_support.load_teststring('shift_jisx0213')
  101. codectests = shiftjis_commonenctests + (
  102. (b"abc\x80\x80\x82\x84", "replace", "abc\ufffd\ufffd\uff44"),
  103. (b"abc\x80\x80\x82\x84\x88", "replace", "abc\ufffd\ufffd\uff44\ufffd"),
  104. # sjis vs cp932
  105. (b"\\\x7e", "replace", "\xa5\u203e"),
  106. (b"\x81\x5f\x81\x61\x81\x7c", "replace", "\x5c\u2016\u2212"),
  107. )
  108. xmlcharnametest = (
  109. "\xab\u211c\xbb = \u2329\u1234\u232a",
  110. b"\x85Gℜ\x85Q = ⟨ሴ⟩"
  111. )
  112. if __name__ == "__main__":
  113. unittest.main()