mac_latin2.py 8.5 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183
  1. """ Python Character Mapping Codec generated from 'LATIN2.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 IncrementalEncoder(codecs.IncrementalEncoder):
  14. def encode(self, input, final=False):
  15. return codecs.charmap_encode(input,self.errors,encoding_map)[0]
  16. class IncrementalDecoder(codecs.IncrementalDecoder):
  17. def decode(self, input, final=False):
  18. return codecs.charmap_decode(input,self.errors,decoding_map)[0]
  19. class StreamWriter(Codec,codecs.StreamWriter):
  20. pass
  21. class StreamReader(Codec,codecs.StreamReader):
  22. pass
  23. ### encodings module API
  24. def getregentry():
  25. return codecs.CodecInfo(
  26. name='mac-latin2',
  27. encode=Codec().encode,
  28. decode=Codec().decode,
  29. incrementalencoder=IncrementalEncoder,
  30. incrementaldecoder=IncrementalDecoder,
  31. streamreader=StreamReader,
  32. streamwriter=StreamWriter,
  33. )
  34. ### Decoding Map
  35. decoding_map = codecs.make_identity_dict(range(256))
  36. decoding_map.update({
  37. 0x0080: 0x00c4, # LATIN CAPITAL LETTER A WITH DIAERESIS
  38. 0x0081: 0x0100, # LATIN CAPITAL LETTER A WITH MACRON
  39. 0x0082: 0x0101, # LATIN SMALL LETTER A WITH MACRON
  40. 0x0083: 0x00c9, # LATIN CAPITAL LETTER E WITH ACUTE
  41. 0x0084: 0x0104, # LATIN CAPITAL LETTER A WITH OGONEK
  42. 0x0085: 0x00d6, # LATIN CAPITAL LETTER O WITH DIAERESIS
  43. 0x0086: 0x00dc, # LATIN CAPITAL LETTER U WITH DIAERESIS
  44. 0x0087: 0x00e1, # LATIN SMALL LETTER A WITH ACUTE
  45. 0x0088: 0x0105, # LATIN SMALL LETTER A WITH OGONEK
  46. 0x0089: 0x010c, # LATIN CAPITAL LETTER C WITH CARON
  47. 0x008a: 0x00e4, # LATIN SMALL LETTER A WITH DIAERESIS
  48. 0x008b: 0x010d, # LATIN SMALL LETTER C WITH CARON
  49. 0x008c: 0x0106, # LATIN CAPITAL LETTER C WITH ACUTE
  50. 0x008d: 0x0107, # LATIN SMALL LETTER C WITH ACUTE
  51. 0x008e: 0x00e9, # LATIN SMALL LETTER E WITH ACUTE
  52. 0x008f: 0x0179, # LATIN CAPITAL LETTER Z WITH ACUTE
  53. 0x0090: 0x017a, # LATIN SMALL LETTER Z WITH ACUTE
  54. 0x0091: 0x010e, # LATIN CAPITAL LETTER D WITH CARON
  55. 0x0092: 0x00ed, # LATIN SMALL LETTER I WITH ACUTE
  56. 0x0093: 0x010f, # LATIN SMALL LETTER D WITH CARON
  57. 0x0094: 0x0112, # LATIN CAPITAL LETTER E WITH MACRON
  58. 0x0095: 0x0113, # LATIN SMALL LETTER E WITH MACRON
  59. 0x0096: 0x0116, # LATIN CAPITAL LETTER E WITH DOT ABOVE
  60. 0x0097: 0x00f3, # LATIN SMALL LETTER O WITH ACUTE
  61. 0x0098: 0x0117, # LATIN SMALL LETTER E WITH DOT ABOVE
  62. 0x0099: 0x00f4, # LATIN SMALL LETTER O WITH CIRCUMFLEX
  63. 0x009a: 0x00f6, # LATIN SMALL LETTER O WITH DIAERESIS
  64. 0x009b: 0x00f5, # LATIN SMALL LETTER O WITH TILDE
  65. 0x009c: 0x00fa, # LATIN SMALL LETTER U WITH ACUTE
  66. 0x009d: 0x011a, # LATIN CAPITAL LETTER E WITH CARON
  67. 0x009e: 0x011b, # LATIN SMALL LETTER E WITH CARON
  68. 0x009f: 0x00fc, # LATIN SMALL LETTER U WITH DIAERESIS
  69. 0x00a0: 0x2020, # DAGGER
  70. 0x00a1: 0x00b0, # DEGREE SIGN
  71. 0x00a2: 0x0118, # LATIN CAPITAL LETTER E WITH OGONEK
  72. 0x00a4: 0x00a7, # SECTION SIGN
  73. 0x00a5: 0x2022, # BULLET
  74. 0x00a6: 0x00b6, # PILCROW SIGN
  75. 0x00a7: 0x00df, # LATIN SMALL LETTER SHARP S
  76. 0x00a8: 0x00ae, # REGISTERED SIGN
  77. 0x00aa: 0x2122, # TRADE MARK SIGN
  78. 0x00ab: 0x0119, # LATIN SMALL LETTER E WITH OGONEK
  79. 0x00ac: 0x00a8, # DIAERESIS
  80. 0x00ad: 0x2260, # NOT EQUAL TO
  81. 0x00ae: 0x0123, # LATIN SMALL LETTER G WITH CEDILLA
  82. 0x00af: 0x012e, # LATIN CAPITAL LETTER I WITH OGONEK
  83. 0x00b0: 0x012f, # LATIN SMALL LETTER I WITH OGONEK
  84. 0x00b1: 0x012a, # LATIN CAPITAL LETTER I WITH MACRON
  85. 0x00b2: 0x2264, # LESS-THAN OR EQUAL TO
  86. 0x00b3: 0x2265, # GREATER-THAN OR EQUAL TO
  87. 0x00b4: 0x012b, # LATIN SMALL LETTER I WITH MACRON
  88. 0x00b5: 0x0136, # LATIN CAPITAL LETTER K WITH CEDILLA
  89. 0x00b6: 0x2202, # PARTIAL DIFFERENTIAL
  90. 0x00b7: 0x2211, # N-ARY SUMMATION
  91. 0x00b8: 0x0142, # LATIN SMALL LETTER L WITH STROKE
  92. 0x00b9: 0x013b, # LATIN CAPITAL LETTER L WITH CEDILLA
  93. 0x00ba: 0x013c, # LATIN SMALL LETTER L WITH CEDILLA
  94. 0x00bb: 0x013d, # LATIN CAPITAL LETTER L WITH CARON
  95. 0x00bc: 0x013e, # LATIN SMALL LETTER L WITH CARON
  96. 0x00bd: 0x0139, # LATIN CAPITAL LETTER L WITH ACUTE
  97. 0x00be: 0x013a, # LATIN SMALL LETTER L WITH ACUTE
  98. 0x00bf: 0x0145, # LATIN CAPITAL LETTER N WITH CEDILLA
  99. 0x00c0: 0x0146, # LATIN SMALL LETTER N WITH CEDILLA
  100. 0x00c1: 0x0143, # LATIN CAPITAL LETTER N WITH ACUTE
  101. 0x00c2: 0x00ac, # NOT SIGN
  102. 0x00c3: 0x221a, # SQUARE ROOT
  103. 0x00c4: 0x0144, # LATIN SMALL LETTER N WITH ACUTE
  104. 0x00c5: 0x0147, # LATIN CAPITAL LETTER N WITH CARON
  105. 0x00c6: 0x2206, # INCREMENT
  106. 0x00c7: 0x00ab, # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
  107. 0x00c8: 0x00bb, # RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
  108. 0x00c9: 0x2026, # HORIZONTAL ELLIPSIS
  109. 0x00ca: 0x00a0, # NO-BREAK SPACE
  110. 0x00cb: 0x0148, # LATIN SMALL LETTER N WITH CARON
  111. 0x00cc: 0x0150, # LATIN CAPITAL LETTER O WITH DOUBLE ACUTE
  112. 0x00cd: 0x00d5, # LATIN CAPITAL LETTER O WITH TILDE
  113. 0x00ce: 0x0151, # LATIN SMALL LETTER O WITH DOUBLE ACUTE
  114. 0x00cf: 0x014c, # LATIN CAPITAL LETTER O WITH MACRON
  115. 0x00d0: 0x2013, # EN DASH
  116. 0x00d1: 0x2014, # EM DASH
  117. 0x00d2: 0x201c, # LEFT DOUBLE QUOTATION MARK
  118. 0x00d3: 0x201d, # RIGHT DOUBLE QUOTATION MARK
  119. 0x00d4: 0x2018, # LEFT SINGLE QUOTATION MARK
  120. 0x00d5: 0x2019, # RIGHT SINGLE QUOTATION MARK
  121. 0x00d6: 0x00f7, # DIVISION SIGN
  122. 0x00d7: 0x25ca, # LOZENGE
  123. 0x00d8: 0x014d, # LATIN SMALL LETTER O WITH MACRON
  124. 0x00d9: 0x0154, # LATIN CAPITAL LETTER R WITH ACUTE
  125. 0x00da: 0x0155, # LATIN SMALL LETTER R WITH ACUTE
  126. 0x00db: 0x0158, # LATIN CAPITAL LETTER R WITH CARON
  127. 0x00dc: 0x2039, # SINGLE LEFT-POINTING ANGLE QUOTATION MARK
  128. 0x00dd: 0x203a, # SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
  129. 0x00de: 0x0159, # LATIN SMALL LETTER R WITH CARON
  130. 0x00df: 0x0156, # LATIN CAPITAL LETTER R WITH CEDILLA
  131. 0x00e0: 0x0157, # LATIN SMALL LETTER R WITH CEDILLA
  132. 0x00e1: 0x0160, # LATIN CAPITAL LETTER S WITH CARON
  133. 0x00e2: 0x201a, # SINGLE LOW-9 QUOTATION MARK
  134. 0x00e3: 0x201e, # DOUBLE LOW-9 QUOTATION MARK
  135. 0x00e4: 0x0161, # LATIN SMALL LETTER S WITH CARON
  136. 0x00e5: 0x015a, # LATIN CAPITAL LETTER S WITH ACUTE
  137. 0x00e6: 0x015b, # LATIN SMALL LETTER S WITH ACUTE
  138. 0x00e7: 0x00c1, # LATIN CAPITAL LETTER A WITH ACUTE
  139. 0x00e8: 0x0164, # LATIN CAPITAL LETTER T WITH CARON
  140. 0x00e9: 0x0165, # LATIN SMALL LETTER T WITH CARON
  141. 0x00ea: 0x00cd, # LATIN CAPITAL LETTER I WITH ACUTE
  142. 0x00eb: 0x017d, # LATIN CAPITAL LETTER Z WITH CARON
  143. 0x00ec: 0x017e, # LATIN SMALL LETTER Z WITH CARON
  144. 0x00ed: 0x016a, # LATIN CAPITAL LETTER U WITH MACRON
  145. 0x00ee: 0x00d3, # LATIN CAPITAL LETTER O WITH ACUTE
  146. 0x00ef: 0x00d4, # LATIN CAPITAL LETTER O WITH CIRCUMFLEX
  147. 0x00f0: 0x016b, # LATIN SMALL LETTER U WITH MACRON
  148. 0x00f1: 0x016e, # LATIN CAPITAL LETTER U WITH RING ABOVE
  149. 0x00f2: 0x00da, # LATIN CAPITAL LETTER U WITH ACUTE
  150. 0x00f3: 0x016f, # LATIN SMALL LETTER U WITH RING ABOVE
  151. 0x00f4: 0x0170, # LATIN CAPITAL LETTER U WITH DOUBLE ACUTE
  152. 0x00f5: 0x0171, # LATIN SMALL LETTER U WITH DOUBLE ACUTE
  153. 0x00f6: 0x0172, # LATIN CAPITAL LETTER U WITH OGONEK
  154. 0x00f7: 0x0173, # LATIN SMALL LETTER U WITH OGONEK
  155. 0x00f8: 0x00dd, # LATIN CAPITAL LETTER Y WITH ACUTE
  156. 0x00f9: 0x00fd, # LATIN SMALL LETTER Y WITH ACUTE
  157. 0x00fa: 0x0137, # LATIN SMALL LETTER K WITH CEDILLA
  158. 0x00fb: 0x017b, # LATIN CAPITAL LETTER Z WITH DOT ABOVE
  159. 0x00fc: 0x0141, # LATIN CAPITAL LETTER L WITH STROKE
  160. 0x00fd: 0x017c, # LATIN SMALL LETTER Z WITH DOT ABOVE
  161. 0x00fe: 0x0122, # LATIN CAPITAL LETTER G WITH CEDILLA
  162. 0x00ff: 0x02c7, # CARON
  163. })
  164. ### Encoding Map
  165. encoding_map = codecs.make_encoding_map(decoding_map)