iso8859_4.py 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307
  1. """ Python Character Mapping Codec iso8859_4 generated from 'MAPPINGS/ISO8859/8859-4.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='iso8859-4',
  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. '\x00' # 0x00 -> NULL
  34. '\x01' # 0x01 -> START OF HEADING
  35. '\x02' # 0x02 -> START OF TEXT
  36. '\x03' # 0x03 -> END OF TEXT
  37. '\x04' # 0x04 -> END OF TRANSMISSION
  38. '\x05' # 0x05 -> ENQUIRY
  39. '\x06' # 0x06 -> ACKNOWLEDGE
  40. '\x07' # 0x07 -> BELL
  41. '\x08' # 0x08 -> BACKSPACE
  42. '\t' # 0x09 -> HORIZONTAL TABULATION
  43. '\n' # 0x0A -> LINE FEED
  44. '\x0b' # 0x0B -> VERTICAL TABULATION
  45. '\x0c' # 0x0C -> FORM FEED
  46. '\r' # 0x0D -> CARRIAGE RETURN
  47. '\x0e' # 0x0E -> SHIFT OUT
  48. '\x0f' # 0x0F -> SHIFT IN
  49. '\x10' # 0x10 -> DATA LINK ESCAPE
  50. '\x11' # 0x11 -> DEVICE CONTROL ONE
  51. '\x12' # 0x12 -> DEVICE CONTROL TWO
  52. '\x13' # 0x13 -> DEVICE CONTROL THREE
  53. '\x14' # 0x14 -> DEVICE CONTROL FOUR
  54. '\x15' # 0x15 -> NEGATIVE ACKNOWLEDGE
  55. '\x16' # 0x16 -> SYNCHRONOUS IDLE
  56. '\x17' # 0x17 -> END OF TRANSMISSION BLOCK
  57. '\x18' # 0x18 -> CANCEL
  58. '\x19' # 0x19 -> END OF MEDIUM
  59. '\x1a' # 0x1A -> SUBSTITUTE
  60. '\x1b' # 0x1B -> ESCAPE
  61. '\x1c' # 0x1C -> FILE SEPARATOR
  62. '\x1d' # 0x1D -> GROUP SEPARATOR
  63. '\x1e' # 0x1E -> RECORD SEPARATOR
  64. '\x1f' # 0x1F -> UNIT SEPARATOR
  65. ' ' # 0x20 -> SPACE
  66. '!' # 0x21 -> EXCLAMATION MARK
  67. '"' # 0x22 -> QUOTATION MARK
  68. '#' # 0x23 -> NUMBER SIGN
  69. '$' # 0x24 -> DOLLAR SIGN
  70. '%' # 0x25 -> PERCENT SIGN
  71. '&' # 0x26 -> AMPERSAND
  72. "'" # 0x27 -> APOSTROPHE
  73. '(' # 0x28 -> LEFT PARENTHESIS
  74. ')' # 0x29 -> RIGHT PARENTHESIS
  75. '*' # 0x2A -> ASTERISK
  76. '+' # 0x2B -> PLUS SIGN
  77. ',' # 0x2C -> COMMA
  78. '-' # 0x2D -> HYPHEN-MINUS
  79. '.' # 0x2E -> FULL STOP
  80. '/' # 0x2F -> SOLIDUS
  81. '0' # 0x30 -> DIGIT ZERO
  82. '1' # 0x31 -> DIGIT ONE
  83. '2' # 0x32 -> DIGIT TWO
  84. '3' # 0x33 -> DIGIT THREE
  85. '4' # 0x34 -> DIGIT FOUR
  86. '5' # 0x35 -> DIGIT FIVE
  87. '6' # 0x36 -> DIGIT SIX
  88. '7' # 0x37 -> DIGIT SEVEN
  89. '8' # 0x38 -> DIGIT EIGHT
  90. '9' # 0x39 -> DIGIT NINE
  91. ':' # 0x3A -> COLON
  92. ';' # 0x3B -> SEMICOLON
  93. '<' # 0x3C -> LESS-THAN SIGN
  94. '=' # 0x3D -> EQUALS SIGN
  95. '>' # 0x3E -> GREATER-THAN SIGN
  96. '?' # 0x3F -> QUESTION MARK
  97. '@' # 0x40 -> COMMERCIAL AT
  98. 'A' # 0x41 -> LATIN CAPITAL LETTER A
  99. 'B' # 0x42 -> LATIN CAPITAL LETTER B
  100. 'C' # 0x43 -> LATIN CAPITAL LETTER C
  101. 'D' # 0x44 -> LATIN CAPITAL LETTER D
  102. 'E' # 0x45 -> LATIN CAPITAL LETTER E
  103. 'F' # 0x46 -> LATIN CAPITAL LETTER F
  104. 'G' # 0x47 -> LATIN CAPITAL LETTER G
  105. 'H' # 0x48 -> LATIN CAPITAL LETTER H
  106. 'I' # 0x49 -> LATIN CAPITAL LETTER I
  107. 'J' # 0x4A -> LATIN CAPITAL LETTER J
  108. 'K' # 0x4B -> LATIN CAPITAL LETTER K
  109. 'L' # 0x4C -> LATIN CAPITAL LETTER L
  110. 'M' # 0x4D -> LATIN CAPITAL LETTER M
  111. 'N' # 0x4E -> LATIN CAPITAL LETTER N
  112. 'O' # 0x4F -> LATIN CAPITAL LETTER O
  113. 'P' # 0x50 -> LATIN CAPITAL LETTER P
  114. 'Q' # 0x51 -> LATIN CAPITAL LETTER Q
  115. 'R' # 0x52 -> LATIN CAPITAL LETTER R
  116. 'S' # 0x53 -> LATIN CAPITAL LETTER S
  117. 'T' # 0x54 -> LATIN CAPITAL LETTER T
  118. 'U' # 0x55 -> LATIN CAPITAL LETTER U
  119. 'V' # 0x56 -> LATIN CAPITAL LETTER V
  120. 'W' # 0x57 -> LATIN CAPITAL LETTER W
  121. 'X' # 0x58 -> LATIN CAPITAL LETTER X
  122. 'Y' # 0x59 -> LATIN CAPITAL LETTER Y
  123. 'Z' # 0x5A -> LATIN CAPITAL LETTER Z
  124. '[' # 0x5B -> LEFT SQUARE BRACKET
  125. '\\' # 0x5C -> REVERSE SOLIDUS
  126. ']' # 0x5D -> RIGHT SQUARE BRACKET
  127. '^' # 0x5E -> CIRCUMFLEX ACCENT
  128. '_' # 0x5F -> LOW LINE
  129. '`' # 0x60 -> GRAVE ACCENT
  130. 'a' # 0x61 -> LATIN SMALL LETTER A
  131. 'b' # 0x62 -> LATIN SMALL LETTER B
  132. 'c' # 0x63 -> LATIN SMALL LETTER C
  133. 'd' # 0x64 -> LATIN SMALL LETTER D
  134. 'e' # 0x65 -> LATIN SMALL LETTER E
  135. 'f' # 0x66 -> LATIN SMALL LETTER F
  136. 'g' # 0x67 -> LATIN SMALL LETTER G
  137. 'h' # 0x68 -> LATIN SMALL LETTER H
  138. 'i' # 0x69 -> LATIN SMALL LETTER I
  139. 'j' # 0x6A -> LATIN SMALL LETTER J
  140. 'k' # 0x6B -> LATIN SMALL LETTER K
  141. 'l' # 0x6C -> LATIN SMALL LETTER L
  142. 'm' # 0x6D -> LATIN SMALL LETTER M
  143. 'n' # 0x6E -> LATIN SMALL LETTER N
  144. 'o' # 0x6F -> LATIN SMALL LETTER O
  145. 'p' # 0x70 -> LATIN SMALL LETTER P
  146. 'q' # 0x71 -> LATIN SMALL LETTER Q
  147. 'r' # 0x72 -> LATIN SMALL LETTER R
  148. 's' # 0x73 -> LATIN SMALL LETTER S
  149. 't' # 0x74 -> LATIN SMALL LETTER T
  150. 'u' # 0x75 -> LATIN SMALL LETTER U
  151. 'v' # 0x76 -> LATIN SMALL LETTER V
  152. 'w' # 0x77 -> LATIN SMALL LETTER W
  153. 'x' # 0x78 -> LATIN SMALL LETTER X
  154. 'y' # 0x79 -> LATIN SMALL LETTER Y
  155. 'z' # 0x7A -> LATIN SMALL LETTER Z
  156. '{' # 0x7B -> LEFT CURLY BRACKET
  157. '|' # 0x7C -> VERTICAL LINE
  158. '}' # 0x7D -> RIGHT CURLY BRACKET
  159. '~' # 0x7E -> TILDE
  160. '\x7f' # 0x7F -> DELETE
  161. '\x80' # 0x80 -> <control>
  162. '\x81' # 0x81 -> <control>
  163. '\x82' # 0x82 -> <control>
  164. '\x83' # 0x83 -> <control>
  165. '\x84' # 0x84 -> <control>
  166. '\x85' # 0x85 -> <control>
  167. '\x86' # 0x86 -> <control>
  168. '\x87' # 0x87 -> <control>
  169. '\x88' # 0x88 -> <control>
  170. '\x89' # 0x89 -> <control>
  171. '\x8a' # 0x8A -> <control>
  172. '\x8b' # 0x8B -> <control>
  173. '\x8c' # 0x8C -> <control>
  174. '\x8d' # 0x8D -> <control>
  175. '\x8e' # 0x8E -> <control>
  176. '\x8f' # 0x8F -> <control>
  177. '\x90' # 0x90 -> <control>
  178. '\x91' # 0x91 -> <control>
  179. '\x92' # 0x92 -> <control>
  180. '\x93' # 0x93 -> <control>
  181. '\x94' # 0x94 -> <control>
  182. '\x95' # 0x95 -> <control>
  183. '\x96' # 0x96 -> <control>
  184. '\x97' # 0x97 -> <control>
  185. '\x98' # 0x98 -> <control>
  186. '\x99' # 0x99 -> <control>
  187. '\x9a' # 0x9A -> <control>
  188. '\x9b' # 0x9B -> <control>
  189. '\x9c' # 0x9C -> <control>
  190. '\x9d' # 0x9D -> <control>
  191. '\x9e' # 0x9E -> <control>
  192. '\x9f' # 0x9F -> <control>
  193. '\xa0' # 0xA0 -> NO-BREAK SPACE
  194. '\u0104' # 0xA1 -> LATIN CAPITAL LETTER A WITH OGONEK
  195. '\u0138' # 0xA2 -> LATIN SMALL LETTER KRA
  196. '\u0156' # 0xA3 -> LATIN CAPITAL LETTER R WITH CEDILLA
  197. '\xa4' # 0xA4 -> CURRENCY SIGN
  198. '\u0128' # 0xA5 -> LATIN CAPITAL LETTER I WITH TILDE
  199. '\u013b' # 0xA6 -> LATIN CAPITAL LETTER L WITH CEDILLA
  200. '\xa7' # 0xA7 -> SECTION SIGN
  201. '\xa8' # 0xA8 -> DIAERESIS
  202. '\u0160' # 0xA9 -> LATIN CAPITAL LETTER S WITH CARON
  203. '\u0112' # 0xAA -> LATIN CAPITAL LETTER E WITH MACRON
  204. '\u0122' # 0xAB -> LATIN CAPITAL LETTER G WITH CEDILLA
  205. '\u0166' # 0xAC -> LATIN CAPITAL LETTER T WITH STROKE
  206. '\xad' # 0xAD -> SOFT HYPHEN
  207. '\u017d' # 0xAE -> LATIN CAPITAL LETTER Z WITH CARON
  208. '\xaf' # 0xAF -> MACRON
  209. '\xb0' # 0xB0 -> DEGREE SIGN
  210. '\u0105' # 0xB1 -> LATIN SMALL LETTER A WITH OGONEK
  211. '\u02db' # 0xB2 -> OGONEK
  212. '\u0157' # 0xB3 -> LATIN SMALL LETTER R WITH CEDILLA
  213. '\xb4' # 0xB4 -> ACUTE ACCENT
  214. '\u0129' # 0xB5 -> LATIN SMALL LETTER I WITH TILDE
  215. '\u013c' # 0xB6 -> LATIN SMALL LETTER L WITH CEDILLA
  216. '\u02c7' # 0xB7 -> CARON
  217. '\xb8' # 0xB8 -> CEDILLA
  218. '\u0161' # 0xB9 -> LATIN SMALL LETTER S WITH CARON
  219. '\u0113' # 0xBA -> LATIN SMALL LETTER E WITH MACRON
  220. '\u0123' # 0xBB -> LATIN SMALL LETTER G WITH CEDILLA
  221. '\u0167' # 0xBC -> LATIN SMALL LETTER T WITH STROKE
  222. '\u014a' # 0xBD -> LATIN CAPITAL LETTER ENG
  223. '\u017e' # 0xBE -> LATIN SMALL LETTER Z WITH CARON
  224. '\u014b' # 0xBF -> LATIN SMALL LETTER ENG
  225. '\u0100' # 0xC0 -> LATIN CAPITAL LETTER A WITH MACRON
  226. '\xc1' # 0xC1 -> LATIN CAPITAL LETTER A WITH ACUTE
  227. '\xc2' # 0xC2 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX
  228. '\xc3' # 0xC3 -> LATIN CAPITAL LETTER A WITH TILDE
  229. '\xc4' # 0xC4 -> LATIN CAPITAL LETTER A WITH DIAERESIS
  230. '\xc5' # 0xC5 -> LATIN CAPITAL LETTER A WITH RING ABOVE
  231. '\xc6' # 0xC6 -> LATIN CAPITAL LETTER AE
  232. '\u012e' # 0xC7 -> LATIN CAPITAL LETTER I WITH OGONEK
  233. '\u010c' # 0xC8 -> LATIN CAPITAL LETTER C WITH CARON
  234. '\xc9' # 0xC9 -> LATIN CAPITAL LETTER E WITH ACUTE
  235. '\u0118' # 0xCA -> LATIN CAPITAL LETTER E WITH OGONEK
  236. '\xcb' # 0xCB -> LATIN CAPITAL LETTER E WITH DIAERESIS
  237. '\u0116' # 0xCC -> LATIN CAPITAL LETTER E WITH DOT ABOVE
  238. '\xcd' # 0xCD -> LATIN CAPITAL LETTER I WITH ACUTE
  239. '\xce' # 0xCE -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX
  240. '\u012a' # 0xCF -> LATIN CAPITAL LETTER I WITH MACRON
  241. '\u0110' # 0xD0 -> LATIN CAPITAL LETTER D WITH STROKE
  242. '\u0145' # 0xD1 -> LATIN CAPITAL LETTER N WITH CEDILLA
  243. '\u014c' # 0xD2 -> LATIN CAPITAL LETTER O WITH MACRON
  244. '\u0136' # 0xD3 -> LATIN CAPITAL LETTER K WITH CEDILLA
  245. '\xd4' # 0xD4 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX
  246. '\xd5' # 0xD5 -> LATIN CAPITAL LETTER O WITH TILDE
  247. '\xd6' # 0xD6 -> LATIN CAPITAL LETTER O WITH DIAERESIS
  248. '\xd7' # 0xD7 -> MULTIPLICATION SIGN
  249. '\xd8' # 0xD8 -> LATIN CAPITAL LETTER O WITH STROKE
  250. '\u0172' # 0xD9 -> LATIN CAPITAL LETTER U WITH OGONEK
  251. '\xda' # 0xDA -> LATIN CAPITAL LETTER U WITH ACUTE
  252. '\xdb' # 0xDB -> LATIN CAPITAL LETTER U WITH CIRCUMFLEX
  253. '\xdc' # 0xDC -> LATIN CAPITAL LETTER U WITH DIAERESIS
  254. '\u0168' # 0xDD -> LATIN CAPITAL LETTER U WITH TILDE
  255. '\u016a' # 0xDE -> LATIN CAPITAL LETTER U WITH MACRON
  256. '\xdf' # 0xDF -> LATIN SMALL LETTER SHARP S
  257. '\u0101' # 0xE0 -> LATIN SMALL LETTER A WITH MACRON
  258. '\xe1' # 0xE1 -> LATIN SMALL LETTER A WITH ACUTE
  259. '\xe2' # 0xE2 -> LATIN SMALL LETTER A WITH CIRCUMFLEX
  260. '\xe3' # 0xE3 -> LATIN SMALL LETTER A WITH TILDE
  261. '\xe4' # 0xE4 -> LATIN SMALL LETTER A WITH DIAERESIS
  262. '\xe5' # 0xE5 -> LATIN SMALL LETTER A WITH RING ABOVE
  263. '\xe6' # 0xE6 -> LATIN SMALL LETTER AE
  264. '\u012f' # 0xE7 -> LATIN SMALL LETTER I WITH OGONEK
  265. '\u010d' # 0xE8 -> LATIN SMALL LETTER C WITH CARON
  266. '\xe9' # 0xE9 -> LATIN SMALL LETTER E WITH ACUTE
  267. '\u0119' # 0xEA -> LATIN SMALL LETTER E WITH OGONEK
  268. '\xeb' # 0xEB -> LATIN SMALL LETTER E WITH DIAERESIS
  269. '\u0117' # 0xEC -> LATIN SMALL LETTER E WITH DOT ABOVE
  270. '\xed' # 0xED -> LATIN SMALL LETTER I WITH ACUTE
  271. '\xee' # 0xEE -> LATIN SMALL LETTER I WITH CIRCUMFLEX
  272. '\u012b' # 0xEF -> LATIN SMALL LETTER I WITH MACRON
  273. '\u0111' # 0xF0 -> LATIN SMALL LETTER D WITH STROKE
  274. '\u0146' # 0xF1 -> LATIN SMALL LETTER N WITH CEDILLA
  275. '\u014d' # 0xF2 -> LATIN SMALL LETTER O WITH MACRON
  276. '\u0137' # 0xF3 -> LATIN SMALL LETTER K WITH CEDILLA
  277. '\xf4' # 0xF4 -> LATIN SMALL LETTER O WITH CIRCUMFLEX
  278. '\xf5' # 0xF5 -> LATIN SMALL LETTER O WITH TILDE
  279. '\xf6' # 0xF6 -> LATIN SMALL LETTER O WITH DIAERESIS
  280. '\xf7' # 0xF7 -> DIVISION SIGN
  281. '\xf8' # 0xF8 -> LATIN SMALL LETTER O WITH STROKE
  282. '\u0173' # 0xF9 -> LATIN SMALL LETTER U WITH OGONEK
  283. '\xfa' # 0xFA -> LATIN SMALL LETTER U WITH ACUTE
  284. '\xfb' # 0xFB -> LATIN SMALL LETTER U WITH CIRCUMFLEX
  285. '\xfc' # 0xFC -> LATIN SMALL LETTER U WITH DIAERESIS
  286. '\u0169' # 0xFD -> LATIN SMALL LETTER U WITH TILDE
  287. '\u016b' # 0xFE -> LATIN SMALL LETTER U WITH MACRON
  288. '\u02d9' # 0xFF -> DOT ABOVE
  289. )
  290. ### Encoding table
  291. encoding_table=codecs.charmap_build(decoding_table)