You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

iso8859_5.py 5.4KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140
  1. """ Python Character Mapping Codec generated from '8859-5.TXT' with gencodec.py.
  2. Written by Marc-Andre Lemburg (mal@lemburg.com).
  3. (c) Copyright CNRI, All Rights Reserved. NO WARRANTY.
  4. (c) Copyright 2000 Guido van Rossum.
  5. """#"
  6. import codecs
  7. ### Codec APIs
  8. class Codec(codecs.Codec):
  9. def encode(self,input,errors='strict'):
  10. return codecs.charmap_encode(input,errors,encoding_map)
  11. def decode(self,input,errors='strict'):
  12. return codecs.charmap_decode(input,errors,decoding_map)
  13. class StreamWriter(Codec,codecs.StreamWriter):
  14. pass
  15. class StreamReader(Codec,codecs.StreamReader):
  16. pass
  17. ### encodings module API
  18. def getregentry():
  19. return (Codec().encode,Codec().decode,StreamReader,StreamWriter)
  20. ### Decoding Map
  21. decoding_map = codecs.make_identity_dict(range(256))
  22. decoding_map.update({
  23. 0x00a1: 0x0401, # CYRILLIC CAPITAL LETTER IO
  24. 0x00a2: 0x0402, # CYRILLIC CAPITAL LETTER DJE
  25. 0x00a3: 0x0403, # CYRILLIC CAPITAL LETTER GJE
  26. 0x00a4: 0x0404, # CYRILLIC CAPITAL LETTER UKRAINIAN IE
  27. 0x00a5: 0x0405, # CYRILLIC CAPITAL LETTER DZE
  28. 0x00a6: 0x0406, # CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I
  29. 0x00a7: 0x0407, # CYRILLIC CAPITAL LETTER YI
  30. 0x00a8: 0x0408, # CYRILLIC CAPITAL LETTER JE
  31. 0x00a9: 0x0409, # CYRILLIC CAPITAL LETTER LJE
  32. 0x00aa: 0x040a, # CYRILLIC CAPITAL LETTER NJE
  33. 0x00ab: 0x040b, # CYRILLIC CAPITAL LETTER TSHE
  34. 0x00ac: 0x040c, # CYRILLIC CAPITAL LETTER KJE
  35. 0x00ae: 0x040e, # CYRILLIC CAPITAL LETTER SHORT U
  36. 0x00af: 0x040f, # CYRILLIC CAPITAL LETTER DZHE
  37. 0x00b0: 0x0410, # CYRILLIC CAPITAL LETTER A
  38. 0x00b1: 0x0411, # CYRILLIC CAPITAL LETTER BE
  39. 0x00b2: 0x0412, # CYRILLIC CAPITAL LETTER VE
  40. 0x00b3: 0x0413, # CYRILLIC CAPITAL LETTER GHE
  41. 0x00b4: 0x0414, # CYRILLIC CAPITAL LETTER DE
  42. 0x00b5: 0x0415, # CYRILLIC CAPITAL LETTER IE
  43. 0x00b6: 0x0416, # CYRILLIC CAPITAL LETTER ZHE
  44. 0x00b7: 0x0417, # CYRILLIC CAPITAL LETTER ZE
  45. 0x00b8: 0x0418, # CYRILLIC CAPITAL LETTER I
  46. 0x00b9: 0x0419, # CYRILLIC CAPITAL LETTER SHORT I
  47. 0x00ba: 0x041a, # CYRILLIC CAPITAL LETTER KA
  48. 0x00bb: 0x041b, # CYRILLIC CAPITAL LETTER EL
  49. 0x00bc: 0x041c, # CYRILLIC CAPITAL LETTER EM
  50. 0x00bd: 0x041d, # CYRILLIC CAPITAL LETTER EN
  51. 0x00be: 0x041e, # CYRILLIC CAPITAL LETTER O
  52. 0x00bf: 0x041f, # CYRILLIC CAPITAL LETTER PE
  53. 0x00c0: 0x0420, # CYRILLIC CAPITAL LETTER ER
  54. 0x00c1: 0x0421, # CYRILLIC CAPITAL LETTER ES
  55. 0x00c2: 0x0422, # CYRILLIC CAPITAL LETTER TE
  56. 0x00c3: 0x0423, # CYRILLIC CAPITAL LETTER U
  57. 0x00c4: 0x0424, # CYRILLIC CAPITAL LETTER EF
  58. 0x00c5: 0x0425, # CYRILLIC CAPITAL LETTER HA
  59. 0x00c6: 0x0426, # CYRILLIC CAPITAL LETTER TSE
  60. 0x00c7: 0x0427, # CYRILLIC CAPITAL LETTER CHE
  61. 0x00c8: 0x0428, # CYRILLIC CAPITAL LETTER SHA
  62. 0x00c9: 0x0429, # CYRILLIC CAPITAL LETTER SHCHA
  63. 0x00ca: 0x042a, # CYRILLIC CAPITAL LETTER HARD SIGN
  64. 0x00cb: 0x042b, # CYRILLIC CAPITAL LETTER YERU
  65. 0x00cc: 0x042c, # CYRILLIC CAPITAL LETTER SOFT SIGN
  66. 0x00cd: 0x042d, # CYRILLIC CAPITAL LETTER E
  67. 0x00ce: 0x042e, # CYRILLIC CAPITAL LETTER YU
  68. 0x00cf: 0x042f, # CYRILLIC CAPITAL LETTER YA
  69. 0x00d0: 0x0430, # CYRILLIC SMALL LETTER A
  70. 0x00d1: 0x0431, # CYRILLIC SMALL LETTER BE
  71. 0x00d2: 0x0432, # CYRILLIC SMALL LETTER VE
  72. 0x00d3: 0x0433, # CYRILLIC SMALL LETTER GHE
  73. 0x00d4: 0x0434, # CYRILLIC SMALL LETTER DE
  74. 0x00d5: 0x0435, # CYRILLIC SMALL LETTER IE
  75. 0x00d6: 0x0436, # CYRILLIC SMALL LETTER ZHE
  76. 0x00d7: 0x0437, # CYRILLIC SMALL LETTER ZE
  77. 0x00d8: 0x0438, # CYRILLIC SMALL LETTER I
  78. 0x00d9: 0x0439, # CYRILLIC SMALL LETTER SHORT I
  79. 0x00da: 0x043a, # CYRILLIC SMALL LETTER KA
  80. 0x00db: 0x043b, # CYRILLIC SMALL LETTER EL
  81. 0x00dc: 0x043c, # CYRILLIC SMALL LETTER EM
  82. 0x00dd: 0x043d, # CYRILLIC SMALL LETTER EN
  83. 0x00de: 0x043e, # CYRILLIC SMALL LETTER O
  84. 0x00df: 0x043f, # CYRILLIC SMALL LETTER PE
  85. 0x00e0: 0x0440, # CYRILLIC SMALL LETTER ER
  86. 0x00e1: 0x0441, # CYRILLIC SMALL LETTER ES
  87. 0x00e2: 0x0442, # CYRILLIC SMALL LETTER TE
  88. 0x00e3: 0x0443, # CYRILLIC SMALL LETTER U
  89. 0x00e4: 0x0444, # CYRILLIC SMALL LETTER EF
  90. 0x00e5: 0x0445, # CYRILLIC SMALL LETTER HA
  91. 0x00e6: 0x0446, # CYRILLIC SMALL LETTER TSE
  92. 0x00e7: 0x0447, # CYRILLIC SMALL LETTER CHE
  93. 0x00e8: 0x0448, # CYRILLIC SMALL LETTER SHA
  94. 0x00e9: 0x0449, # CYRILLIC SMALL LETTER SHCHA
  95. 0x00ea: 0x044a, # CYRILLIC SMALL LETTER HARD SIGN
  96. 0x00eb: 0x044b, # CYRILLIC SMALL LETTER YERU
  97. 0x00ec: 0x044c, # CYRILLIC SMALL LETTER SOFT SIGN
  98. 0x00ed: 0x044d, # CYRILLIC SMALL LETTER E
  99. 0x00ee: 0x044e, # CYRILLIC SMALL LETTER YU
  100. 0x00ef: 0x044f, # CYRILLIC SMALL LETTER YA
  101. 0x00f0: 0x2116, # NUMERO SIGN
  102. 0x00f1: 0x0451, # CYRILLIC SMALL LETTER IO
  103. 0x00f2: 0x0452, # CYRILLIC SMALL LETTER DJE
  104. 0x00f3: 0x0453, # CYRILLIC SMALL LETTER GJE
  105. 0x00f4: 0x0454, # CYRILLIC SMALL LETTER UKRAINIAN IE
  106. 0x00f5: 0x0455, # CYRILLIC SMALL LETTER DZE
  107. 0x00f6: 0x0456, # CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I
  108. 0x00f7: 0x0457, # CYRILLIC SMALL LETTER YI
  109. 0x00f8: 0x0458, # CYRILLIC SMALL LETTER JE
  110. 0x00f9: 0x0459, # CYRILLIC SMALL LETTER LJE
  111. 0x00fa: 0x045a, # CYRILLIC SMALL LETTER NJE
  112. 0x00fb: 0x045b, # CYRILLIC SMALL LETTER TSHE
  113. 0x00fc: 0x045c, # CYRILLIC SMALL LETTER KJE
  114. 0x00fd: 0x00a7, # SECTION SIGN
  115. 0x00fe: 0x045e, # CYRILLIC SMALL LETTER SHORT U
  116. 0x00ff: 0x045f, # CYRILLIC SMALL LETTER DZHE
  117. })
  118. ### Encoding Map
  119. encoding_map = {}
  120. for k,v in decoding_map.items():
  121. encoding_map[v] = k