cp500.py 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280
  1. """ Python Character Mapping Codec generated from 'CP500.TXT' with gencodec.py.
  2. Written by Marc-Andre Lemburg ([email protected]).
  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. 0x0004: 0x009c, # CONTROL
  24. 0x0005: 0x0009, # HORIZONTAL TABULATION
  25. 0x0006: 0x0086, # CONTROL
  26. 0x0007: 0x007f, # DELETE
  27. 0x0008: 0x0097, # CONTROL
  28. 0x0009: 0x008d, # CONTROL
  29. 0x000a: 0x008e, # CONTROL
  30. 0x0014: 0x009d, # CONTROL
  31. 0x0015: 0x0085, # CONTROL
  32. 0x0016: 0x0008, # BACKSPACE
  33. 0x0017: 0x0087, # CONTROL
  34. 0x001a: 0x0092, # CONTROL
  35. 0x001b: 0x008f, # CONTROL
  36. 0x0020: 0x0080, # CONTROL
  37. 0x0021: 0x0081, # CONTROL
  38. 0x0022: 0x0082, # CONTROL
  39. 0x0023: 0x0083, # CONTROL
  40. 0x0024: 0x0084, # CONTROL
  41. 0x0025: 0x000a, # LINE FEED
  42. 0x0026: 0x0017, # END OF TRANSMISSION BLOCK
  43. 0x0027: 0x001b, # ESCAPE
  44. 0x0028: 0x0088, # CONTROL
  45. 0x0029: 0x0089, # CONTROL
  46. 0x002a: 0x008a, # CONTROL
  47. 0x002b: 0x008b, # CONTROL
  48. 0x002c: 0x008c, # CONTROL
  49. 0x002d: 0x0005, # ENQUIRY
  50. 0x002e: 0x0006, # ACKNOWLEDGE
  51. 0x002f: 0x0007, # BELL
  52. 0x0030: 0x0090, # CONTROL
  53. 0x0031: 0x0091, # CONTROL
  54. 0x0032: 0x0016, # SYNCHRONOUS IDLE
  55. 0x0033: 0x0093, # CONTROL
  56. 0x0034: 0x0094, # CONTROL
  57. 0x0035: 0x0095, # CONTROL
  58. 0x0036: 0x0096, # CONTROL
  59. 0x0037: 0x0004, # END OF TRANSMISSION
  60. 0x0038: 0x0098, # CONTROL
  61. 0x0039: 0x0099, # CONTROL
  62. 0x003a: 0x009a, # CONTROL
  63. 0x003b: 0x009b, # CONTROL
  64. 0x003c: 0x0014, # DEVICE CONTROL FOUR
  65. 0x003d: 0x0015, # NEGATIVE ACKNOWLEDGE
  66. 0x003e: 0x009e, # CONTROL
  67. 0x003f: 0x001a, # SUBSTITUTE
  68. 0x0040: 0x0020, # SPACE
  69. 0x0041: 0x00a0, # NO-BREAK SPACE
  70. 0x0042: 0x00e2, # LATIN SMALL LETTER A WITH CIRCUMFLEX
  71. 0x0043: 0x00e4, # LATIN SMALL LETTER A WITH DIAERESIS
  72. 0x0044: 0x00e0, # LATIN SMALL LETTER A WITH GRAVE
  73. 0x0045: 0x00e1, # LATIN SMALL LETTER A WITH ACUTE
  74. 0x0046: 0x00e3, # LATIN SMALL LETTER A WITH TILDE
  75. 0x0047: 0x00e5, # LATIN SMALL LETTER A WITH RING ABOVE
  76. 0x0048: 0x00e7, # LATIN SMALL LETTER C WITH CEDILLA
  77. 0x0049: 0x00f1, # LATIN SMALL LETTER N WITH TILDE
  78. 0x004a: 0x005b, # LEFT SQUARE BRACKET
  79. 0x004b: 0x002e, # FULL STOP
  80. 0x004c: 0x003c, # LESS-THAN SIGN
  81. 0x004d: 0x0028, # LEFT PARENTHESIS
  82. 0x004e: 0x002b, # PLUS SIGN
  83. 0x004f: 0x0021, # EXCLAMATION MARK
  84. 0x0050: 0x0026, # AMPERSAND
  85. 0x0051: 0x00e9, # LATIN SMALL LETTER E WITH ACUTE
  86. 0x0052: 0x00ea, # LATIN SMALL LETTER E WITH CIRCUMFLEX
  87. 0x0053: 0x00eb, # LATIN SMALL LETTER E WITH DIAERESIS
  88. 0x0054: 0x00e8, # LATIN SMALL LETTER E WITH GRAVE
  89. 0x0055: 0x00ed, # LATIN SMALL LETTER I WITH ACUTE
  90. 0x0056: 0x00ee, # LATIN SMALL LETTER I WITH CIRCUMFLEX
  91. 0x0057: 0x00ef, # LATIN SMALL LETTER I WITH DIAERESIS
  92. 0x0058: 0x00ec, # LATIN SMALL LETTER I WITH GRAVE
  93. 0x0059: 0x00df, # LATIN SMALL LETTER SHARP S (GERMAN)
  94. 0x005a: 0x005d, # RIGHT SQUARE BRACKET
  95. 0x005b: 0x0024, # DOLLAR SIGN
  96. 0x005c: 0x002a, # ASTERISK
  97. 0x005d: 0x0029, # RIGHT PARENTHESIS
  98. 0x005e: 0x003b, # SEMICOLON
  99. 0x005f: 0x005e, # CIRCUMFLEX ACCENT
  100. 0x0060: 0x002d, # HYPHEN-MINUS
  101. 0x0061: 0x002f, # SOLIDUS
  102. 0x0062: 0x00c2, # LATIN CAPITAL LETTER A WITH CIRCUMFLEX
  103. 0x0063: 0x00c4, # LATIN CAPITAL LETTER A WITH DIAERESIS
  104. 0x0064: 0x00c0, # LATIN CAPITAL LETTER A WITH GRAVE
  105. 0x0065: 0x00c1, # LATIN CAPITAL LETTER A WITH ACUTE
  106. 0x0066: 0x00c3, # LATIN CAPITAL LETTER A WITH TILDE
  107. 0x0067: 0x00c5, # LATIN CAPITAL LETTER A WITH RING ABOVE
  108. 0x0068: 0x00c7, # LATIN CAPITAL LETTER C WITH CEDILLA
  109. 0x0069: 0x00d1, # LATIN CAPITAL LETTER N WITH TILDE
  110. 0x006a: 0x00a6, # BROKEN BAR
  111. 0x006b: 0x002c, # COMMA
  112. 0x006c: 0x0025, # PERCENT SIGN
  113. 0x006d: 0x005f, # LOW LINE
  114. 0x006e: 0x003e, # GREATER-THAN SIGN
  115. 0x006f: 0x003f, # QUESTION MARK
  116. 0x0070: 0x00f8, # LATIN SMALL LETTER O WITH STROKE
  117. 0x0071: 0x00c9, # LATIN CAPITAL LETTER E WITH ACUTE
  118. 0x0072: 0x00ca, # LATIN CAPITAL LETTER E WITH CIRCUMFLEX
  119. 0x0073: 0x00cb, # LATIN CAPITAL LETTER E WITH DIAERESIS
  120. 0x0074: 0x00c8, # LATIN CAPITAL LETTER E WITH GRAVE
  121. 0x0075: 0x00cd, # LATIN CAPITAL LETTER I WITH ACUTE
  122. 0x0076: 0x00ce, # LATIN CAPITAL LETTER I WITH CIRCUMFLEX
  123. 0x0077: 0x00cf, # LATIN CAPITAL LETTER I WITH DIAERESIS
  124. 0x0078: 0x00cc, # LATIN CAPITAL LETTER I WITH GRAVE
  125. 0x0079: 0x0060, # GRAVE ACCENT
  126. 0x007a: 0x003a, # COLON
  127. 0x007b: 0x0023, # NUMBER SIGN
  128. 0x007c: 0x0040, # COMMERCIAL AT
  129. 0x007d: 0x0027, # APOSTROPHE
  130. 0x007e: 0x003d, # EQUALS SIGN
  131. 0x007f: 0x0022, # QUOTATION MARK
  132. 0x0080: 0x00d8, # LATIN CAPITAL LETTER O WITH STROKE
  133. 0x0081: 0x0061, # LATIN SMALL LETTER A
  134. 0x0082: 0x0062, # LATIN SMALL LETTER B
  135. 0x0083: 0x0063, # LATIN SMALL LETTER C
  136. 0x0084: 0x0064, # LATIN SMALL LETTER D
  137. 0x0085: 0x0065, # LATIN SMALL LETTER E
  138. 0x0086: 0x0066, # LATIN SMALL LETTER F
  139. 0x0087: 0x0067, # LATIN SMALL LETTER G
  140. 0x0088: 0x0068, # LATIN SMALL LETTER H
  141. 0x0089: 0x0069, # LATIN SMALL LETTER I
  142. 0x008a: 0x00ab, # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
  143. 0x008b: 0x00bb, # RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
  144. 0x008c: 0x00f0, # LATIN SMALL LETTER ETH (ICELANDIC)
  145. 0x008d: 0x00fd, # LATIN SMALL LETTER Y WITH ACUTE
  146. 0x008e: 0x00fe, # LATIN SMALL LETTER THORN (ICELANDIC)
  147. 0x008f: 0x00b1, # PLUS-MINUS SIGN
  148. 0x0090: 0x00b0, # DEGREE SIGN
  149. 0x0091: 0x006a, # LATIN SMALL LETTER J
  150. 0x0092: 0x006b, # LATIN SMALL LETTER K
  151. 0x0093: 0x006c, # LATIN SMALL LETTER L
  152. 0x0094: 0x006d, # LATIN SMALL LETTER M
  153. 0x0095: 0x006e, # LATIN SMALL LETTER N
  154. 0x0096: 0x006f, # LATIN SMALL LETTER O
  155. 0x0097: 0x0070, # LATIN SMALL LETTER P
  156. 0x0098: 0x0071, # LATIN SMALL LETTER Q
  157. 0x0099: 0x0072, # LATIN SMALL LETTER R
  158. 0x009a: 0x00aa, # FEMININE ORDINAL INDICATOR
  159. 0x009b: 0x00ba, # MASCULINE ORDINAL INDICATOR
  160. 0x009c: 0x00e6, # LATIN SMALL LIGATURE AE
  161. 0x009d: 0x00b8, # CEDILLA
  162. 0x009e: 0x00c6, # LATIN CAPITAL LIGATURE AE
  163. 0x009f: 0x00a4, # CURRENCY SIGN
  164. 0x00a0: 0x00b5, # MICRO SIGN
  165. 0x00a1: 0x007e, # TILDE
  166. 0x00a2: 0x0073, # LATIN SMALL LETTER S
  167. 0x00a3: 0x0074, # LATIN SMALL LETTER T
  168. 0x00a4: 0x0075, # LATIN SMALL LETTER U
  169. 0x00a5: 0x0076, # LATIN SMALL LETTER V
  170. 0x00a6: 0x0077, # LATIN SMALL LETTER W
  171. 0x00a7: 0x0078, # LATIN SMALL LETTER X
  172. 0x00a8: 0x0079, # LATIN SMALL LETTER Y
  173. 0x00a9: 0x007a, # LATIN SMALL LETTER Z
  174. 0x00aa: 0x00a1, # INVERTED EXCLAMATION MARK
  175. 0x00ab: 0x00bf, # INVERTED QUESTION MARK
  176. 0x00ac: 0x00d0, # LATIN CAPITAL LETTER ETH (ICELANDIC)
  177. 0x00ad: 0x00dd, # LATIN CAPITAL LETTER Y WITH ACUTE
  178. 0x00ae: 0x00de, # LATIN CAPITAL LETTER THORN (ICELANDIC)
  179. 0x00af: 0x00ae, # REGISTERED SIGN
  180. 0x00b0: 0x00a2, # CENT SIGN
  181. 0x00b1: 0x00a3, # POUND SIGN
  182. 0x00b2: 0x00a5, # YEN SIGN
  183. 0x00b3: 0x00b7, # MIDDLE DOT
  184. 0x00b4: 0x00a9, # COPYRIGHT SIGN
  185. 0x00b5: 0x00a7, # SECTION SIGN
  186. 0x00b7: 0x00bc, # VULGAR FRACTION ONE QUARTER
  187. 0x00b8: 0x00bd, # VULGAR FRACTION ONE HALF
  188. 0x00b9: 0x00be, # VULGAR FRACTION THREE QUARTERS
  189. 0x00ba: 0x00ac, # NOT SIGN
  190. 0x00bb: 0x007c, # VERTICAL LINE
  191. 0x00bc: 0x00af, # MACRON
  192. 0x00bd: 0x00a8, # DIAERESIS
  193. 0x00be: 0x00b4, # ACUTE ACCENT
  194. 0x00bf: 0x00d7, # MULTIPLICATION SIGN
  195. 0x00c0: 0x007b, # LEFT CURLY BRACKET
  196. 0x00c1: 0x0041, # LATIN CAPITAL LETTER A
  197. 0x00c2: 0x0042, # LATIN CAPITAL LETTER B
  198. 0x00c3: 0x0043, # LATIN CAPITAL LETTER C
  199. 0x00c4: 0x0044, # LATIN CAPITAL LETTER D
  200. 0x00c5: 0x0045, # LATIN CAPITAL LETTER E
  201. 0x00c6: 0x0046, # LATIN CAPITAL LETTER F
  202. 0x00c7: 0x0047, # LATIN CAPITAL LETTER G
  203. 0x00c8: 0x0048, # LATIN CAPITAL LETTER H
  204. 0x00c9: 0x0049, # LATIN CAPITAL LETTER I
  205. 0x00ca: 0x00ad, # SOFT HYPHEN
  206. 0x00cb: 0x00f4, # LATIN SMALL LETTER O WITH CIRCUMFLEX
  207. 0x00cc: 0x00f6, # LATIN SMALL LETTER O WITH DIAERESIS
  208. 0x00cd: 0x00f2, # LATIN SMALL LETTER O WITH GRAVE
  209. 0x00ce: 0x00f3, # LATIN SMALL LETTER O WITH ACUTE
  210. 0x00cf: 0x00f5, # LATIN SMALL LETTER O WITH TILDE
  211. 0x00d0: 0x007d, # RIGHT CURLY BRACKET
  212. 0x00d1: 0x004a, # LATIN CAPITAL LETTER J
  213. 0x00d2: 0x004b, # LATIN CAPITAL LETTER K
  214. 0x00d3: 0x004c, # LATIN CAPITAL LETTER L
  215. 0x00d4: 0x004d, # LATIN CAPITAL LETTER M
  216. 0x00d5: 0x004e, # LATIN CAPITAL LETTER N
  217. 0x00d6: 0x004f, # LATIN CAPITAL LETTER O
  218. 0x00d7: 0x0050, # LATIN CAPITAL LETTER P
  219. 0x00d8: 0x0051, # LATIN CAPITAL LETTER Q
  220. 0x00d9: 0x0052, # LATIN CAPITAL LETTER R
  221. 0x00da: 0x00b9, # SUPERSCRIPT ONE
  222. 0x00db: 0x00fb, # LATIN SMALL LETTER U WITH CIRCUMFLEX
  223. 0x00dc: 0x00fc, # LATIN SMALL LETTER U WITH DIAERESIS
  224. 0x00dd: 0x00f9, # LATIN SMALL LETTER U WITH GRAVE
  225. 0x00de: 0x00fa, # LATIN SMALL LETTER U WITH ACUTE
  226. 0x00df: 0x00ff, # LATIN SMALL LETTER Y WITH DIAERESIS
  227. 0x00e0: 0x005c, # REVERSE SOLIDUS
  228. 0x00e1: 0x00f7, # DIVISION SIGN
  229. 0x00e2: 0x0053, # LATIN CAPITAL LETTER S
  230. 0x00e3: 0x0054, # LATIN CAPITAL LETTER T
  231. 0x00e4: 0x0055, # LATIN CAPITAL LETTER U
  232. 0x00e5: 0x0056, # LATIN CAPITAL LETTER V
  233. 0x00e6: 0x0057, # LATIN CAPITAL LETTER W
  234. 0x00e7: 0x0058, # LATIN CAPITAL LETTER X
  235. 0x00e8: 0x0059, # LATIN CAPITAL LETTER Y
  236. 0x00e9: 0x005a, # LATIN CAPITAL LETTER Z
  237. 0x00ea: 0x00b2, # SUPERSCRIPT TWO
  238. 0x00eb: 0x00d4, # LATIN CAPITAL LETTER O WITH CIRCUMFLEX
  239. 0x00ec: 0x00d6, # LATIN CAPITAL LETTER O WITH DIAERESIS
  240. 0x00ed: 0x00d2, # LATIN CAPITAL LETTER O WITH GRAVE
  241. 0x00ee: 0x00d3, # LATIN CAPITAL LETTER O WITH ACUTE
  242. 0x00ef: 0x00d5, # LATIN CAPITAL LETTER O WITH TILDE
  243. 0x00f0: 0x0030, # DIGIT ZERO
  244. 0x00f1: 0x0031, # DIGIT ONE
  245. 0x00f2: 0x0032, # DIGIT TWO
  246. 0x00f3: 0x0033, # DIGIT THREE
  247. 0x00f4: 0x0034, # DIGIT FOUR
  248. 0x00f5: 0x0035, # DIGIT FIVE
  249. 0x00f6: 0x0036, # DIGIT SIX
  250. 0x00f7: 0x0037, # DIGIT SEVEN
  251. 0x00f8: 0x0038, # DIGIT EIGHT
  252. 0x00f9: 0x0039, # DIGIT NINE
  253. 0x00fa: 0x00b3, # SUPERSCRIPT THREE
  254. 0x00fb: 0x00db, # LATIN CAPITAL LETTER U WITH CIRCUMFLEX
  255. 0x00fc: 0x00dc, # LATIN CAPITAL LETTER U WITH DIAERESIS
  256. 0x00fd: 0x00d9, # LATIN CAPITAL LETTER U WITH GRAVE
  257. 0x00fe: 0x00da, # LATIN CAPITAL LETTER U WITH ACUTE
  258. 0x00ff: 0x009f, # CONTROL
  259. })
  260. ### Encoding Map
  261. encoding_map = codecs.make_encoding_map(decoding_map)