mac_greek.py 14 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307
  1. """ Python Character Mapping Codec mac_greek generated from 'MAPPINGS/VENDORS/APPLE/GREEK.TXT' with gencodec.py.
  2. """#"
  3. import codecs
  4. ### Codec APIs
  5. class Codec(codecs.Codec):
  6. def encode(self,input,errors='strict'):
  7. return codecs.charmap_encode(input,errors,encoding_table)
  8. def decode(self,input,errors='strict'):
  9. return codecs.charmap_decode(input,errors,decoding_table)
  10. class IncrementalEncoder(codecs.IncrementalEncoder):
  11. def encode(self, input, final=False):
  12. return codecs.charmap_encode(input,self.errors,encoding_table)[0]
  13. class IncrementalDecoder(codecs.IncrementalDecoder):
  14. def decode(self, input, final=False):
  15. return codecs.charmap_decode(input,self.errors,decoding_table)[0]
  16. class StreamWriter(Codec,codecs.StreamWriter):
  17. pass
  18. class StreamReader(Codec,codecs.StreamReader):
  19. pass
  20. ### encodings module API
  21. def getregentry():
  22. return codecs.CodecInfo(
  23. name='mac-greek',
  24. encode=Codec().encode,
  25. decode=Codec().decode,
  26. incrementalencoder=IncrementalEncoder,
  27. incrementaldecoder=IncrementalDecoder,
  28. streamreader=StreamReader,
  29. streamwriter=StreamWriter,
  30. )
  31. ### Decoding Table
  32. decoding_table = (
  33. u'\x00' # 0x00 -> CONTROL CHARACTER
  34. u'\x01' # 0x01 -> CONTROL CHARACTER
  35. u'\x02' # 0x02 -> CONTROL CHARACTER
  36. u'\x03' # 0x03 -> CONTROL CHARACTER
  37. u'\x04' # 0x04 -> CONTROL CHARACTER
  38. u'\x05' # 0x05 -> CONTROL CHARACTER
  39. u'\x06' # 0x06 -> CONTROL CHARACTER
  40. u'\x07' # 0x07 -> CONTROL CHARACTER
  41. u'\x08' # 0x08 -> CONTROL CHARACTER
  42. u'\t' # 0x09 -> CONTROL CHARACTER
  43. u'\n' # 0x0A -> CONTROL CHARACTER
  44. u'\x0b' # 0x0B -> CONTROL CHARACTER
  45. u'\x0c' # 0x0C -> CONTROL CHARACTER
  46. u'\r' # 0x0D -> CONTROL CHARACTER
  47. u'\x0e' # 0x0E -> CONTROL CHARACTER
  48. u'\x0f' # 0x0F -> CONTROL CHARACTER
  49. u'\x10' # 0x10 -> CONTROL CHARACTER
  50. u'\x11' # 0x11 -> CONTROL CHARACTER
  51. u'\x12' # 0x12 -> CONTROL CHARACTER
  52. u'\x13' # 0x13 -> CONTROL CHARACTER
  53. u'\x14' # 0x14 -> CONTROL CHARACTER
  54. u'\x15' # 0x15 -> CONTROL CHARACTER
  55. u'\x16' # 0x16 -> CONTROL CHARACTER
  56. u'\x17' # 0x17 -> CONTROL CHARACTER
  57. u'\x18' # 0x18 -> CONTROL CHARACTER
  58. u'\x19' # 0x19 -> CONTROL CHARACTER
  59. u'\x1a' # 0x1A -> CONTROL CHARACTER
  60. u'\x1b' # 0x1B -> CONTROL CHARACTER
  61. u'\x1c' # 0x1C -> CONTROL CHARACTER
  62. u'\x1d' # 0x1D -> CONTROL CHARACTER
  63. u'\x1e' # 0x1E -> CONTROL CHARACTER
  64. u'\x1f' # 0x1F -> CONTROL CHARACTER
  65. u' ' # 0x20 -> SPACE
  66. u'!' # 0x21 -> EXCLAMATION MARK
  67. u'"' # 0x22 -> QUOTATION MARK
  68. u'#' # 0x23 -> NUMBER SIGN
  69. u'$' # 0x24 -> DOLLAR SIGN
  70. u'%' # 0x25 -> PERCENT SIGN
  71. u'&' # 0x26 -> AMPERSAND
  72. u"'" # 0x27 -> APOSTROPHE
  73. u'(' # 0x28 -> LEFT PARENTHESIS
  74. u')' # 0x29 -> RIGHT PARENTHESIS
  75. u'*' # 0x2A -> ASTERISK
  76. u'+' # 0x2B -> PLUS SIGN
  77. u',' # 0x2C -> COMMA
  78. u'-' # 0x2D -> HYPHEN-MINUS
  79. u'.' # 0x2E -> FULL STOP
  80. u'/' # 0x2F -> SOLIDUS
  81. u'0' # 0x30 -> DIGIT ZERO
  82. u'1' # 0x31 -> DIGIT ONE
  83. u'2' # 0x32 -> DIGIT TWO
  84. u'3' # 0x33 -> DIGIT THREE
  85. u'4' # 0x34 -> DIGIT FOUR
  86. u'5' # 0x35 -> DIGIT FIVE
  87. u'6' # 0x36 -> DIGIT SIX
  88. u'7' # 0x37 -> DIGIT SEVEN
  89. u'8' # 0x38 -> DIGIT EIGHT
  90. u'9' # 0x39 -> DIGIT NINE
  91. u':' # 0x3A -> COLON
  92. u';' # 0x3B -> SEMICOLON
  93. u'<' # 0x3C -> LESS-THAN SIGN
  94. u'=' # 0x3D -> EQUALS SIGN
  95. u'>' # 0x3E -> GREATER-THAN SIGN
  96. u'?' # 0x3F -> QUESTION MARK
  97. u'@' # 0x40 -> COMMERCIAL AT
  98. u'A' # 0x41 -> LATIN CAPITAL LETTER A
  99. u'B' # 0x42 -> LATIN CAPITAL LETTER B
  100. u'C' # 0x43 -> LATIN CAPITAL LETTER C
  101. u'D' # 0x44 -> LATIN CAPITAL LETTER D
  102. u'E' # 0x45 -> LATIN CAPITAL LETTER E
  103. u'F' # 0x46 -> LATIN CAPITAL LETTER F
  104. u'G' # 0x47 -> LATIN CAPITAL LETTER G
  105. u'H' # 0x48 -> LATIN CAPITAL LETTER H
  106. u'I' # 0x49 -> LATIN CAPITAL LETTER I
  107. u'J' # 0x4A -> LATIN CAPITAL LETTER J
  108. u'K' # 0x4B -> LATIN CAPITAL LETTER K
  109. u'L' # 0x4C -> LATIN CAPITAL LETTER L
  110. u'M' # 0x4D -> LATIN CAPITAL LETTER M
  111. u'N' # 0x4E -> LATIN CAPITAL LETTER N
  112. u'O' # 0x4F -> LATIN CAPITAL LETTER O
  113. u'P' # 0x50 -> LATIN CAPITAL LETTER P
  114. u'Q' # 0x51 -> LATIN CAPITAL LETTER Q
  115. u'R' # 0x52 -> LATIN CAPITAL LETTER R
  116. u'S' # 0x53 -> LATIN CAPITAL LETTER S
  117. u'T' # 0x54 -> LATIN CAPITAL LETTER T
  118. u'U' # 0x55 -> LATIN CAPITAL LETTER U
  119. u'V' # 0x56 -> LATIN CAPITAL LETTER V
  120. u'W' # 0x57 -> LATIN CAPITAL LETTER W
  121. u'X' # 0x58 -> LATIN CAPITAL LETTER X
  122. u'Y' # 0x59 -> LATIN CAPITAL LETTER Y
  123. u'Z' # 0x5A -> LATIN CAPITAL LETTER Z
  124. u'[' # 0x5B -> LEFT SQUARE BRACKET
  125. u'\\' # 0x5C -> REVERSE SOLIDUS
  126. u']' # 0x5D -> RIGHT SQUARE BRACKET
  127. u'^' # 0x5E -> CIRCUMFLEX ACCENT
  128. u'_' # 0x5F -> LOW LINE
  129. u'`' # 0x60 -> GRAVE ACCENT
  130. u'a' # 0x61 -> LATIN SMALL LETTER A
  131. u'b' # 0x62 -> LATIN SMALL LETTER B
  132. u'c' # 0x63 -> LATIN SMALL LETTER C
  133. u'd' # 0x64 -> LATIN SMALL LETTER D
  134. u'e' # 0x65 -> LATIN SMALL LETTER E
  135. u'f' # 0x66 -> LATIN SMALL LETTER F
  136. u'g' # 0x67 -> LATIN SMALL LETTER G
  137. u'h' # 0x68 -> LATIN SMALL LETTER H
  138. u'i' # 0x69 -> LATIN SMALL LETTER I
  139. u'j' # 0x6A -> LATIN SMALL LETTER J
  140. u'k' # 0x6B -> LATIN SMALL LETTER K
  141. u'l' # 0x6C -> LATIN SMALL LETTER L
  142. u'm' # 0x6D -> LATIN SMALL LETTER M
  143. u'n' # 0x6E -> LATIN SMALL LETTER N
  144. u'o' # 0x6F -> LATIN SMALL LETTER O
  145. u'p' # 0x70 -> LATIN SMALL LETTER P
  146. u'q' # 0x71 -> LATIN SMALL LETTER Q
  147. u'r' # 0x72 -> LATIN SMALL LETTER R
  148. u's' # 0x73 -> LATIN SMALL LETTER S
  149. u't' # 0x74 -> LATIN SMALL LETTER T
  150. u'u' # 0x75 -> LATIN SMALL LETTER U
  151. u'v' # 0x76 -> LATIN SMALL LETTER V
  152. u'w' # 0x77 -> LATIN SMALL LETTER W
  153. u'x' # 0x78 -> LATIN SMALL LETTER X
  154. u'y' # 0x79 -> LATIN SMALL LETTER Y
  155. u'z' # 0x7A -> LATIN SMALL LETTER Z
  156. u'{' # 0x7B -> LEFT CURLY BRACKET
  157. u'|' # 0x7C -> VERTICAL LINE
  158. u'}' # 0x7D -> RIGHT CURLY BRACKET
  159. u'~' # 0x7E -> TILDE
  160. u'\x7f' # 0x7F -> CONTROL CHARACTER
  161. u'\xc4' # 0x80 -> LATIN CAPITAL LETTER A WITH DIAERESIS
  162. u'\xb9' # 0x81 -> SUPERSCRIPT ONE
  163. u'\xb2' # 0x82 -> SUPERSCRIPT TWO
  164. u'\xc9' # 0x83 -> LATIN CAPITAL LETTER E WITH ACUTE
  165. u'\xb3' # 0x84 -> SUPERSCRIPT THREE
  166. u'\xd6' # 0x85 -> LATIN CAPITAL LETTER O WITH DIAERESIS
  167. u'\xdc' # 0x86 -> LATIN CAPITAL LETTER U WITH DIAERESIS
  168. u'\u0385' # 0x87 -> GREEK DIALYTIKA TONOS
  169. u'\xe0' # 0x88 -> LATIN SMALL LETTER A WITH GRAVE
  170. u'\xe2' # 0x89 -> LATIN SMALL LETTER A WITH CIRCUMFLEX
  171. u'\xe4' # 0x8A -> LATIN SMALL LETTER A WITH DIAERESIS
  172. u'\u0384' # 0x8B -> GREEK TONOS
  173. u'\xa8' # 0x8C -> DIAERESIS
  174. u'\xe7' # 0x8D -> LATIN SMALL LETTER C WITH CEDILLA
  175. u'\xe9' # 0x8E -> LATIN SMALL LETTER E WITH ACUTE
  176. u'\xe8' # 0x8F -> LATIN SMALL LETTER E WITH GRAVE
  177. u'\xea' # 0x90 -> LATIN SMALL LETTER E WITH CIRCUMFLEX
  178. u'\xeb' # 0x91 -> LATIN SMALL LETTER E WITH DIAERESIS
  179. u'\xa3' # 0x92 -> POUND SIGN
  180. u'\u2122' # 0x93 -> TRADE MARK SIGN
  181. u'\xee' # 0x94 -> LATIN SMALL LETTER I WITH CIRCUMFLEX
  182. u'\xef' # 0x95 -> LATIN SMALL LETTER I WITH DIAERESIS
  183. u'\u2022' # 0x96 -> BULLET
  184. u'\xbd' # 0x97 -> VULGAR FRACTION ONE HALF
  185. u'\u2030' # 0x98 -> PER MILLE SIGN
  186. u'\xf4' # 0x99 -> LATIN SMALL LETTER O WITH CIRCUMFLEX
  187. u'\xf6' # 0x9A -> LATIN SMALL LETTER O WITH DIAERESIS
  188. u'\xa6' # 0x9B -> BROKEN BAR
  189. u'\u20ac' # 0x9C -> EURO SIGN # before Mac OS 9.2.2, was SOFT HYPHEN
  190. u'\xf9' # 0x9D -> LATIN SMALL LETTER U WITH GRAVE
  191. u'\xfb' # 0x9E -> LATIN SMALL LETTER U WITH CIRCUMFLEX
  192. u'\xfc' # 0x9F -> LATIN SMALL LETTER U WITH DIAERESIS
  193. u'\u2020' # 0xA0 -> DAGGER
  194. u'\u0393' # 0xA1 -> GREEK CAPITAL LETTER GAMMA
  195. u'\u0394' # 0xA2 -> GREEK CAPITAL LETTER DELTA
  196. u'\u0398' # 0xA3 -> GREEK CAPITAL LETTER THETA
  197. u'\u039b' # 0xA4 -> GREEK CAPITAL LETTER LAMDA
  198. u'\u039e' # 0xA5 -> GREEK CAPITAL LETTER XI
  199. u'\u03a0' # 0xA6 -> GREEK CAPITAL LETTER PI
  200. u'\xdf' # 0xA7 -> LATIN SMALL LETTER SHARP S
  201. u'\xae' # 0xA8 -> REGISTERED SIGN
  202. u'\xa9' # 0xA9 -> COPYRIGHT SIGN
  203. u'\u03a3' # 0xAA -> GREEK CAPITAL LETTER SIGMA
  204. u'\u03aa' # 0xAB -> GREEK CAPITAL LETTER IOTA WITH DIALYTIKA
  205. u'\xa7' # 0xAC -> SECTION SIGN
  206. u'\u2260' # 0xAD -> NOT EQUAL TO
  207. u'\xb0' # 0xAE -> DEGREE SIGN
  208. u'\xb7' # 0xAF -> MIDDLE DOT
  209. u'\u0391' # 0xB0 -> GREEK CAPITAL LETTER ALPHA
  210. u'\xb1' # 0xB1 -> PLUS-MINUS SIGN
  211. u'\u2264' # 0xB2 -> LESS-THAN OR EQUAL TO
  212. u'\u2265' # 0xB3 -> GREATER-THAN OR EQUAL TO
  213. u'\xa5' # 0xB4 -> YEN SIGN
  214. u'\u0392' # 0xB5 -> GREEK CAPITAL LETTER BETA
  215. u'\u0395' # 0xB6 -> GREEK CAPITAL LETTER EPSILON
  216. u'\u0396' # 0xB7 -> GREEK CAPITAL LETTER ZETA
  217. u'\u0397' # 0xB8 -> GREEK CAPITAL LETTER ETA
  218. u'\u0399' # 0xB9 -> GREEK CAPITAL LETTER IOTA
  219. u'\u039a' # 0xBA -> GREEK CAPITAL LETTER KAPPA
  220. u'\u039c' # 0xBB -> GREEK CAPITAL LETTER MU
  221. u'\u03a6' # 0xBC -> GREEK CAPITAL LETTER PHI
  222. u'\u03ab' # 0xBD -> GREEK CAPITAL LETTER UPSILON WITH DIALYTIKA
  223. u'\u03a8' # 0xBE -> GREEK CAPITAL LETTER PSI
  224. u'\u03a9' # 0xBF -> GREEK CAPITAL LETTER OMEGA
  225. u'\u03ac' # 0xC0 -> GREEK SMALL LETTER ALPHA WITH TONOS
  226. u'\u039d' # 0xC1 -> GREEK CAPITAL LETTER NU
  227. u'\xac' # 0xC2 -> NOT SIGN
  228. u'\u039f' # 0xC3 -> GREEK CAPITAL LETTER OMICRON
  229. u'\u03a1' # 0xC4 -> GREEK CAPITAL LETTER RHO
  230. u'\u2248' # 0xC5 -> ALMOST EQUAL TO
  231. u'\u03a4' # 0xC6 -> GREEK CAPITAL LETTER TAU
  232. u'\xab' # 0xC7 -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
  233. u'\xbb' # 0xC8 -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
  234. u'\u2026' # 0xC9 -> HORIZONTAL ELLIPSIS
  235. u'\xa0' # 0xCA -> NO-BREAK SPACE
  236. u'\u03a5' # 0xCB -> GREEK CAPITAL LETTER UPSILON
  237. u'\u03a7' # 0xCC -> GREEK CAPITAL LETTER CHI
  238. u'\u0386' # 0xCD -> GREEK CAPITAL LETTER ALPHA WITH TONOS
  239. u'\u0388' # 0xCE -> GREEK CAPITAL LETTER EPSILON WITH TONOS
  240. u'\u0153' # 0xCF -> LATIN SMALL LIGATURE OE
  241. u'\u2013' # 0xD0 -> EN DASH
  242. u'\u2015' # 0xD1 -> HORIZONTAL BAR
  243. u'\u201c' # 0xD2 -> LEFT DOUBLE QUOTATION MARK
  244. u'\u201d' # 0xD3 -> RIGHT DOUBLE QUOTATION MARK
  245. u'\u2018' # 0xD4 -> LEFT SINGLE QUOTATION MARK
  246. u'\u2019' # 0xD5 -> RIGHT SINGLE QUOTATION MARK
  247. u'\xf7' # 0xD6 -> DIVISION SIGN
  248. u'\u0389' # 0xD7 -> GREEK CAPITAL LETTER ETA WITH TONOS
  249. u'\u038a' # 0xD8 -> GREEK CAPITAL LETTER IOTA WITH TONOS
  250. u'\u038c' # 0xD9 -> GREEK CAPITAL LETTER OMICRON WITH TONOS
  251. u'\u038e' # 0xDA -> GREEK CAPITAL LETTER UPSILON WITH TONOS
  252. u'\u03ad' # 0xDB -> GREEK SMALL LETTER EPSILON WITH TONOS
  253. u'\u03ae' # 0xDC -> GREEK SMALL LETTER ETA WITH TONOS
  254. u'\u03af' # 0xDD -> GREEK SMALL LETTER IOTA WITH TONOS
  255. u'\u03cc' # 0xDE -> GREEK SMALL LETTER OMICRON WITH TONOS
  256. u'\u038f' # 0xDF -> GREEK CAPITAL LETTER OMEGA WITH TONOS
  257. u'\u03cd' # 0xE0 -> GREEK SMALL LETTER UPSILON WITH TONOS
  258. u'\u03b1' # 0xE1 -> GREEK SMALL LETTER ALPHA
  259. u'\u03b2' # 0xE2 -> GREEK SMALL LETTER BETA
  260. u'\u03c8' # 0xE3 -> GREEK SMALL LETTER PSI
  261. u'\u03b4' # 0xE4 -> GREEK SMALL LETTER DELTA
  262. u'\u03b5' # 0xE5 -> GREEK SMALL LETTER EPSILON
  263. u'\u03c6' # 0xE6 -> GREEK SMALL LETTER PHI
  264. u'\u03b3' # 0xE7 -> GREEK SMALL LETTER GAMMA
  265. u'\u03b7' # 0xE8 -> GREEK SMALL LETTER ETA
  266. u'\u03b9' # 0xE9 -> GREEK SMALL LETTER IOTA
  267. u'\u03be' # 0xEA -> GREEK SMALL LETTER XI
  268. u'\u03ba' # 0xEB -> GREEK SMALL LETTER KAPPA
  269. u'\u03bb' # 0xEC -> GREEK SMALL LETTER LAMDA
  270. u'\u03bc' # 0xED -> GREEK SMALL LETTER MU
  271. u'\u03bd' # 0xEE -> GREEK SMALL LETTER NU
  272. u'\u03bf' # 0xEF -> GREEK SMALL LETTER OMICRON
  273. u'\u03c0' # 0xF0 -> GREEK SMALL LETTER PI
  274. u'\u03ce' # 0xF1 -> GREEK SMALL LETTER OMEGA WITH TONOS
  275. u'\u03c1' # 0xF2 -> GREEK SMALL LETTER RHO
  276. u'\u03c3' # 0xF3 -> GREEK SMALL LETTER SIGMA
  277. u'\u03c4' # 0xF4 -> GREEK SMALL LETTER TAU
  278. u'\u03b8' # 0xF5 -> GREEK SMALL LETTER THETA
  279. u'\u03c9' # 0xF6 -> GREEK SMALL LETTER OMEGA
  280. u'\u03c2' # 0xF7 -> GREEK SMALL LETTER FINAL SIGMA
  281. u'\u03c7' # 0xF8 -> GREEK SMALL LETTER CHI
  282. u'\u03c5' # 0xF9 -> GREEK SMALL LETTER UPSILON
  283. u'\u03b6' # 0xFA -> GREEK SMALL LETTER ZETA
  284. u'\u03ca' # 0xFB -> GREEK SMALL LETTER IOTA WITH DIALYTIKA
  285. u'\u03cb' # 0xFC -> GREEK SMALL LETTER UPSILON WITH DIALYTIKA
  286. u'\u0390' # 0xFD -> GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS
  287. u'\u03b0' # 0xFE -> GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS
  288. u'\xad' # 0xFF -> SOFT HYPHEN # before Mac OS 9.2.2, was undefined
  289. )
  290. ### Encoding table
  291. encoding_table=codecs.charmap_build(decoding_table)