blob: 34d8a0ea5acc4b7f97cf4f943cfc4a78bb3d4247 [file] [log] [blame]
Marc-André Lemburga1129f42005-10-21 13:49:12 +00001""" Python Character Mapping Codec generated from 'VENDORS/MICSFT/PC/CP852.TXT' with gencodec.py.
Guido van Rossum0229bf62000-03-10 23:17:24 +00002
3"""#"
4
5import codecs
6
7### Codec APIs
8
9class Codec(codecs.Codec):
10
11 def encode(self,input,errors='strict'):
Guido van Rossum0229bf62000-03-10 23:17:24 +000012 return codecs.charmap_encode(input,errors,encoding_map)
Tim Peters469cdad2002-08-08 20:19:19 +000013
Guido van Rossum0229bf62000-03-10 23:17:24 +000014 def decode(self,input,errors='strict'):
Marc-André Lemburga1129f42005-10-21 13:49:12 +000015 return codecs.charmap_decode(input,errors,decoding_table)
Tim Peters536cf992005-12-25 23:18:31 +000016
Thomas Woutersa9773292006-04-21 09:43:23 +000017class IncrementalEncoder(codecs.IncrementalEncoder):
18 def encode(self, input, final=False):
19 return codecs.charmap_encode(input,self.errors,encoding_map)[0]
20
21class IncrementalDecoder(codecs.IncrementalDecoder):
22 def decode(self, input, final=False):
23 return codecs.charmap_decode(input,self.errors,decoding_table)[0]
24
Guido van Rossum0229bf62000-03-10 23:17:24 +000025class StreamWriter(Codec,codecs.StreamWriter):
26 pass
Tim Peters469cdad2002-08-08 20:19:19 +000027
Guido van Rossum0229bf62000-03-10 23:17:24 +000028class StreamReader(Codec,codecs.StreamReader):
29 pass
30
31### encodings module API
32
33def getregentry():
Thomas Woutersa9773292006-04-21 09:43:23 +000034 return codecs.CodecInfo(
35 name='cp852',
36 encode=Codec().encode,
37 decode=Codec().decode,
38 incrementalencoder=IncrementalEncoder,
39 incrementaldecoder=IncrementalDecoder,
40 streamreader=StreamReader,
41 streamwriter=StreamWriter,
42 )
Guido van Rossum0229bf62000-03-10 23:17:24 +000043
44### Decoding Map
45
Marc-André Lemburga866df82001-01-03 21:29:14 +000046decoding_map = codecs.make_identity_dict(range(256))
47decoding_map.update({
Tim Peters536cf992005-12-25 23:18:31 +000048 0x0080: 0x00c7, # LATIN CAPITAL LETTER C WITH CEDILLA
49 0x0081: 0x00fc, # LATIN SMALL LETTER U WITH DIAERESIS
50 0x0082: 0x00e9, # LATIN SMALL LETTER E WITH ACUTE
51 0x0083: 0x00e2, # LATIN SMALL LETTER A WITH CIRCUMFLEX
52 0x0084: 0x00e4, # LATIN SMALL LETTER A WITH DIAERESIS
53 0x0085: 0x016f, # LATIN SMALL LETTER U WITH RING ABOVE
54 0x0086: 0x0107, # LATIN SMALL LETTER C WITH ACUTE
55 0x0087: 0x00e7, # LATIN SMALL LETTER C WITH CEDILLA
56 0x0088: 0x0142, # LATIN SMALL LETTER L WITH STROKE
57 0x0089: 0x00eb, # LATIN SMALL LETTER E WITH DIAERESIS
58 0x008a: 0x0150, # LATIN CAPITAL LETTER O WITH DOUBLE ACUTE
59 0x008b: 0x0151, # LATIN SMALL LETTER O WITH DOUBLE ACUTE
60 0x008c: 0x00ee, # LATIN SMALL LETTER I WITH CIRCUMFLEX
61 0x008d: 0x0179, # LATIN CAPITAL LETTER Z WITH ACUTE
62 0x008e: 0x00c4, # LATIN CAPITAL LETTER A WITH DIAERESIS
63 0x008f: 0x0106, # LATIN CAPITAL LETTER C WITH ACUTE
64 0x0090: 0x00c9, # LATIN CAPITAL LETTER E WITH ACUTE
65 0x0091: 0x0139, # LATIN CAPITAL LETTER L WITH ACUTE
66 0x0092: 0x013a, # LATIN SMALL LETTER L WITH ACUTE
67 0x0093: 0x00f4, # LATIN SMALL LETTER O WITH CIRCUMFLEX
68 0x0094: 0x00f6, # LATIN SMALL LETTER O WITH DIAERESIS
69 0x0095: 0x013d, # LATIN CAPITAL LETTER L WITH CARON
70 0x0096: 0x013e, # LATIN SMALL LETTER L WITH CARON
71 0x0097: 0x015a, # LATIN CAPITAL LETTER S WITH ACUTE
72 0x0098: 0x015b, # LATIN SMALL LETTER S WITH ACUTE
73 0x0099: 0x00d6, # LATIN CAPITAL LETTER O WITH DIAERESIS
74 0x009a: 0x00dc, # LATIN CAPITAL LETTER U WITH DIAERESIS
75 0x009b: 0x0164, # LATIN CAPITAL LETTER T WITH CARON
76 0x009c: 0x0165, # LATIN SMALL LETTER T WITH CARON
77 0x009d: 0x0141, # LATIN CAPITAL LETTER L WITH STROKE
78 0x009e: 0x00d7, # MULTIPLICATION SIGN
79 0x009f: 0x010d, # LATIN SMALL LETTER C WITH CARON
80 0x00a0: 0x00e1, # LATIN SMALL LETTER A WITH ACUTE
81 0x00a1: 0x00ed, # LATIN SMALL LETTER I WITH ACUTE
82 0x00a2: 0x00f3, # LATIN SMALL LETTER O WITH ACUTE
83 0x00a3: 0x00fa, # LATIN SMALL LETTER U WITH ACUTE
84 0x00a4: 0x0104, # LATIN CAPITAL LETTER A WITH OGONEK
85 0x00a5: 0x0105, # LATIN SMALL LETTER A WITH OGONEK
86 0x00a6: 0x017d, # LATIN CAPITAL LETTER Z WITH CARON
87 0x00a7: 0x017e, # LATIN SMALL LETTER Z WITH CARON
88 0x00a8: 0x0118, # LATIN CAPITAL LETTER E WITH OGONEK
89 0x00a9: 0x0119, # LATIN SMALL LETTER E WITH OGONEK
90 0x00aa: 0x00ac, # NOT SIGN
91 0x00ab: 0x017a, # LATIN SMALL LETTER Z WITH ACUTE
92 0x00ac: 0x010c, # LATIN CAPITAL LETTER C WITH CARON
93 0x00ad: 0x015f, # LATIN SMALL LETTER S WITH CEDILLA
94 0x00ae: 0x00ab, # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
95 0x00af: 0x00bb, # RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
96 0x00b0: 0x2591, # LIGHT SHADE
97 0x00b1: 0x2592, # MEDIUM SHADE
98 0x00b2: 0x2593, # DARK SHADE
99 0x00b3: 0x2502, # BOX DRAWINGS LIGHT VERTICAL
100 0x00b4: 0x2524, # BOX DRAWINGS LIGHT VERTICAL AND LEFT
101 0x00b5: 0x00c1, # LATIN CAPITAL LETTER A WITH ACUTE
102 0x00b6: 0x00c2, # LATIN CAPITAL LETTER A WITH CIRCUMFLEX
103 0x00b7: 0x011a, # LATIN CAPITAL LETTER E WITH CARON
104 0x00b8: 0x015e, # LATIN CAPITAL LETTER S WITH CEDILLA
105 0x00b9: 0x2563, # BOX DRAWINGS DOUBLE VERTICAL AND LEFT
106 0x00ba: 0x2551, # BOX DRAWINGS DOUBLE VERTICAL
107 0x00bb: 0x2557, # BOX DRAWINGS DOUBLE DOWN AND LEFT
108 0x00bc: 0x255d, # BOX DRAWINGS DOUBLE UP AND LEFT
109 0x00bd: 0x017b, # LATIN CAPITAL LETTER Z WITH DOT ABOVE
110 0x00be: 0x017c, # LATIN SMALL LETTER Z WITH DOT ABOVE
111 0x00bf: 0x2510, # BOX DRAWINGS LIGHT DOWN AND LEFT
112 0x00c0: 0x2514, # BOX DRAWINGS LIGHT UP AND RIGHT
113 0x00c1: 0x2534, # BOX DRAWINGS LIGHT UP AND HORIZONTAL
114 0x00c2: 0x252c, # BOX DRAWINGS LIGHT DOWN AND HORIZONTAL
115 0x00c3: 0x251c, # BOX DRAWINGS LIGHT VERTICAL AND RIGHT
116 0x00c4: 0x2500, # BOX DRAWINGS LIGHT HORIZONTAL
117 0x00c5: 0x253c, # BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL
118 0x00c6: 0x0102, # LATIN CAPITAL LETTER A WITH BREVE
119 0x00c7: 0x0103, # LATIN SMALL LETTER A WITH BREVE
120 0x00c8: 0x255a, # BOX DRAWINGS DOUBLE UP AND RIGHT
121 0x00c9: 0x2554, # BOX DRAWINGS DOUBLE DOWN AND RIGHT
122 0x00ca: 0x2569, # BOX DRAWINGS DOUBLE UP AND HORIZONTAL
123 0x00cb: 0x2566, # BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL
124 0x00cc: 0x2560, # BOX DRAWINGS DOUBLE VERTICAL AND RIGHT
125 0x00cd: 0x2550, # BOX DRAWINGS DOUBLE HORIZONTAL
126 0x00ce: 0x256c, # BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL
127 0x00cf: 0x00a4, # CURRENCY SIGN
128 0x00d0: 0x0111, # LATIN SMALL LETTER D WITH STROKE
129 0x00d1: 0x0110, # LATIN CAPITAL LETTER D WITH STROKE
130 0x00d2: 0x010e, # LATIN CAPITAL LETTER D WITH CARON
131 0x00d3: 0x00cb, # LATIN CAPITAL LETTER E WITH DIAERESIS
132 0x00d4: 0x010f, # LATIN SMALL LETTER D WITH CARON
133 0x00d5: 0x0147, # LATIN CAPITAL LETTER N WITH CARON
134 0x00d6: 0x00cd, # LATIN CAPITAL LETTER I WITH ACUTE
135 0x00d7: 0x00ce, # LATIN CAPITAL LETTER I WITH CIRCUMFLEX
136 0x00d8: 0x011b, # LATIN SMALL LETTER E WITH CARON
137 0x00d9: 0x2518, # BOX DRAWINGS LIGHT UP AND LEFT
138 0x00da: 0x250c, # BOX DRAWINGS LIGHT DOWN AND RIGHT
139 0x00db: 0x2588, # FULL BLOCK
140 0x00dc: 0x2584, # LOWER HALF BLOCK
141 0x00dd: 0x0162, # LATIN CAPITAL LETTER T WITH CEDILLA
142 0x00de: 0x016e, # LATIN CAPITAL LETTER U WITH RING ABOVE
143 0x00df: 0x2580, # UPPER HALF BLOCK
144 0x00e0: 0x00d3, # LATIN CAPITAL LETTER O WITH ACUTE
145 0x00e1: 0x00df, # LATIN SMALL LETTER SHARP S
146 0x00e2: 0x00d4, # LATIN CAPITAL LETTER O WITH CIRCUMFLEX
147 0x00e3: 0x0143, # LATIN CAPITAL LETTER N WITH ACUTE
148 0x00e4: 0x0144, # LATIN SMALL LETTER N WITH ACUTE
149 0x00e5: 0x0148, # LATIN SMALL LETTER N WITH CARON
150 0x00e6: 0x0160, # LATIN CAPITAL LETTER S WITH CARON
151 0x00e7: 0x0161, # LATIN SMALL LETTER S WITH CARON
152 0x00e8: 0x0154, # LATIN CAPITAL LETTER R WITH ACUTE
153 0x00e9: 0x00da, # LATIN CAPITAL LETTER U WITH ACUTE
154 0x00ea: 0x0155, # LATIN SMALL LETTER R WITH ACUTE
155 0x00eb: 0x0170, # LATIN CAPITAL LETTER U WITH DOUBLE ACUTE
156 0x00ec: 0x00fd, # LATIN SMALL LETTER Y WITH ACUTE
157 0x00ed: 0x00dd, # LATIN CAPITAL LETTER Y WITH ACUTE
158 0x00ee: 0x0163, # LATIN SMALL LETTER T WITH CEDILLA
159 0x00ef: 0x00b4, # ACUTE ACCENT
160 0x00f0: 0x00ad, # SOFT HYPHEN
161 0x00f1: 0x02dd, # DOUBLE ACUTE ACCENT
162 0x00f2: 0x02db, # OGONEK
163 0x00f3: 0x02c7, # CARON
164 0x00f4: 0x02d8, # BREVE
165 0x00f5: 0x00a7, # SECTION SIGN
166 0x00f6: 0x00f7, # DIVISION SIGN
167 0x00f7: 0x00b8, # CEDILLA
168 0x00f8: 0x00b0, # DEGREE SIGN
169 0x00f9: 0x00a8, # DIAERESIS
170 0x00fa: 0x02d9, # DOT ABOVE
171 0x00fb: 0x0171, # LATIN SMALL LETTER U WITH DOUBLE ACUTE
172 0x00fc: 0x0158, # LATIN CAPITAL LETTER R WITH CARON
173 0x00fd: 0x0159, # LATIN SMALL LETTER R WITH CARON
174 0x00fe: 0x25a0, # BLACK SQUARE
175 0x00ff: 0x00a0, # NO-BREAK SPACE
Marc-André Lemburga866df82001-01-03 21:29:14 +0000176})
Guido van Rossum0229bf62000-03-10 23:17:24 +0000177
Marc-André Lemburga1129f42005-10-21 13:49:12 +0000178### Decoding Table
179
180decoding_table = (
Guido van Rossumef87d6e2007-05-02 19:09:54 +0000181 '\x00' # 0x0000 -> NULL
182 '\x01' # 0x0001 -> START OF HEADING
183 '\x02' # 0x0002 -> START OF TEXT
184 '\x03' # 0x0003 -> END OF TEXT
185 '\x04' # 0x0004 -> END OF TRANSMISSION
186 '\x05' # 0x0005 -> ENQUIRY
187 '\x06' # 0x0006 -> ACKNOWLEDGE
188 '\x07' # 0x0007 -> BELL
189 '\x08' # 0x0008 -> BACKSPACE
190 '\t' # 0x0009 -> HORIZONTAL TABULATION
191 '\n' # 0x000a -> LINE FEED
192 '\x0b' # 0x000b -> VERTICAL TABULATION
193 '\x0c' # 0x000c -> FORM FEED
194 '\r' # 0x000d -> CARRIAGE RETURN
195 '\x0e' # 0x000e -> SHIFT OUT
196 '\x0f' # 0x000f -> SHIFT IN
197 '\x10' # 0x0010 -> DATA LINK ESCAPE
198 '\x11' # 0x0011 -> DEVICE CONTROL ONE
199 '\x12' # 0x0012 -> DEVICE CONTROL TWO
200 '\x13' # 0x0013 -> DEVICE CONTROL THREE
201 '\x14' # 0x0014 -> DEVICE CONTROL FOUR
202 '\x15' # 0x0015 -> NEGATIVE ACKNOWLEDGE
203 '\x16' # 0x0016 -> SYNCHRONOUS IDLE
204 '\x17' # 0x0017 -> END OF TRANSMISSION BLOCK
205 '\x18' # 0x0018 -> CANCEL
206 '\x19' # 0x0019 -> END OF MEDIUM
207 '\x1a' # 0x001a -> SUBSTITUTE
208 '\x1b' # 0x001b -> ESCAPE
209 '\x1c' # 0x001c -> FILE SEPARATOR
210 '\x1d' # 0x001d -> GROUP SEPARATOR
211 '\x1e' # 0x001e -> RECORD SEPARATOR
212 '\x1f' # 0x001f -> UNIT SEPARATOR
213 ' ' # 0x0020 -> SPACE
214 '!' # 0x0021 -> EXCLAMATION MARK
215 '"' # 0x0022 -> QUOTATION MARK
216 '#' # 0x0023 -> NUMBER SIGN
217 '$' # 0x0024 -> DOLLAR SIGN
218 '%' # 0x0025 -> PERCENT SIGN
219 '&' # 0x0026 -> AMPERSAND
220 "'" # 0x0027 -> APOSTROPHE
221 '(' # 0x0028 -> LEFT PARENTHESIS
222 ')' # 0x0029 -> RIGHT PARENTHESIS
223 '*' # 0x002a -> ASTERISK
224 '+' # 0x002b -> PLUS SIGN
225 ',' # 0x002c -> COMMA
226 '-' # 0x002d -> HYPHEN-MINUS
227 '.' # 0x002e -> FULL STOP
228 '/' # 0x002f -> SOLIDUS
229 '0' # 0x0030 -> DIGIT ZERO
230 '1' # 0x0031 -> DIGIT ONE
231 '2' # 0x0032 -> DIGIT TWO
232 '3' # 0x0033 -> DIGIT THREE
233 '4' # 0x0034 -> DIGIT FOUR
234 '5' # 0x0035 -> DIGIT FIVE
235 '6' # 0x0036 -> DIGIT SIX
236 '7' # 0x0037 -> DIGIT SEVEN
237 '8' # 0x0038 -> DIGIT EIGHT
238 '9' # 0x0039 -> DIGIT NINE
239 ':' # 0x003a -> COLON
240 ';' # 0x003b -> SEMICOLON
241 '<' # 0x003c -> LESS-THAN SIGN
242 '=' # 0x003d -> EQUALS SIGN
243 '>' # 0x003e -> GREATER-THAN SIGN
244 '?' # 0x003f -> QUESTION MARK
245 '@' # 0x0040 -> COMMERCIAL AT
246 'A' # 0x0041 -> LATIN CAPITAL LETTER A
247 'B' # 0x0042 -> LATIN CAPITAL LETTER B
248 'C' # 0x0043 -> LATIN CAPITAL LETTER C
249 'D' # 0x0044 -> LATIN CAPITAL LETTER D
250 'E' # 0x0045 -> LATIN CAPITAL LETTER E
251 'F' # 0x0046 -> LATIN CAPITAL LETTER F
252 'G' # 0x0047 -> LATIN CAPITAL LETTER G
253 'H' # 0x0048 -> LATIN CAPITAL LETTER H
254 'I' # 0x0049 -> LATIN CAPITAL LETTER I
255 'J' # 0x004a -> LATIN CAPITAL LETTER J
256 'K' # 0x004b -> LATIN CAPITAL LETTER K
257 'L' # 0x004c -> LATIN CAPITAL LETTER L
258 'M' # 0x004d -> LATIN CAPITAL LETTER M
259 'N' # 0x004e -> LATIN CAPITAL LETTER N
260 'O' # 0x004f -> LATIN CAPITAL LETTER O
261 'P' # 0x0050 -> LATIN CAPITAL LETTER P
262 'Q' # 0x0051 -> LATIN CAPITAL LETTER Q
263 'R' # 0x0052 -> LATIN CAPITAL LETTER R
264 'S' # 0x0053 -> LATIN CAPITAL LETTER S
265 'T' # 0x0054 -> LATIN CAPITAL LETTER T
266 'U' # 0x0055 -> LATIN CAPITAL LETTER U
267 'V' # 0x0056 -> LATIN CAPITAL LETTER V
268 'W' # 0x0057 -> LATIN CAPITAL LETTER W
269 'X' # 0x0058 -> LATIN CAPITAL LETTER X
270 'Y' # 0x0059 -> LATIN CAPITAL LETTER Y
271 'Z' # 0x005a -> LATIN CAPITAL LETTER Z
272 '[' # 0x005b -> LEFT SQUARE BRACKET
273 '\\' # 0x005c -> REVERSE SOLIDUS
274 ']' # 0x005d -> RIGHT SQUARE BRACKET
275 '^' # 0x005e -> CIRCUMFLEX ACCENT
276 '_' # 0x005f -> LOW LINE
277 '`' # 0x0060 -> GRAVE ACCENT
278 'a' # 0x0061 -> LATIN SMALL LETTER A
279 'b' # 0x0062 -> LATIN SMALL LETTER B
280 'c' # 0x0063 -> LATIN SMALL LETTER C
281 'd' # 0x0064 -> LATIN SMALL LETTER D
282 'e' # 0x0065 -> LATIN SMALL LETTER E
283 'f' # 0x0066 -> LATIN SMALL LETTER F
284 'g' # 0x0067 -> LATIN SMALL LETTER G
285 'h' # 0x0068 -> LATIN SMALL LETTER H
286 'i' # 0x0069 -> LATIN SMALL LETTER I
287 'j' # 0x006a -> LATIN SMALL LETTER J
288 'k' # 0x006b -> LATIN SMALL LETTER K
289 'l' # 0x006c -> LATIN SMALL LETTER L
290 'm' # 0x006d -> LATIN SMALL LETTER M
291 'n' # 0x006e -> LATIN SMALL LETTER N
292 'o' # 0x006f -> LATIN SMALL LETTER O
293 'p' # 0x0070 -> LATIN SMALL LETTER P
294 'q' # 0x0071 -> LATIN SMALL LETTER Q
295 'r' # 0x0072 -> LATIN SMALL LETTER R
296 's' # 0x0073 -> LATIN SMALL LETTER S
297 't' # 0x0074 -> LATIN SMALL LETTER T
298 'u' # 0x0075 -> LATIN SMALL LETTER U
299 'v' # 0x0076 -> LATIN SMALL LETTER V
300 'w' # 0x0077 -> LATIN SMALL LETTER W
301 'x' # 0x0078 -> LATIN SMALL LETTER X
302 'y' # 0x0079 -> LATIN SMALL LETTER Y
303 'z' # 0x007a -> LATIN SMALL LETTER Z
304 '{' # 0x007b -> LEFT CURLY BRACKET
305 '|' # 0x007c -> VERTICAL LINE
306 '}' # 0x007d -> RIGHT CURLY BRACKET
307 '~' # 0x007e -> TILDE
308 '\x7f' # 0x007f -> DELETE
309 '\xc7' # 0x0080 -> LATIN CAPITAL LETTER C WITH CEDILLA
310 '\xfc' # 0x0081 -> LATIN SMALL LETTER U WITH DIAERESIS
311 '\xe9' # 0x0082 -> LATIN SMALL LETTER E WITH ACUTE
312 '\xe2' # 0x0083 -> LATIN SMALL LETTER A WITH CIRCUMFLEX
313 '\xe4' # 0x0084 -> LATIN SMALL LETTER A WITH DIAERESIS
314 '\u016f' # 0x0085 -> LATIN SMALL LETTER U WITH RING ABOVE
315 '\u0107' # 0x0086 -> LATIN SMALL LETTER C WITH ACUTE
316 '\xe7' # 0x0087 -> LATIN SMALL LETTER C WITH CEDILLA
317 '\u0142' # 0x0088 -> LATIN SMALL LETTER L WITH STROKE
318 '\xeb' # 0x0089 -> LATIN SMALL LETTER E WITH DIAERESIS
319 '\u0150' # 0x008a -> LATIN CAPITAL LETTER O WITH DOUBLE ACUTE
320 '\u0151' # 0x008b -> LATIN SMALL LETTER O WITH DOUBLE ACUTE
321 '\xee' # 0x008c -> LATIN SMALL LETTER I WITH CIRCUMFLEX
322 '\u0179' # 0x008d -> LATIN CAPITAL LETTER Z WITH ACUTE
323 '\xc4' # 0x008e -> LATIN CAPITAL LETTER A WITH DIAERESIS
324 '\u0106' # 0x008f -> LATIN CAPITAL LETTER C WITH ACUTE
325 '\xc9' # 0x0090 -> LATIN CAPITAL LETTER E WITH ACUTE
326 '\u0139' # 0x0091 -> LATIN CAPITAL LETTER L WITH ACUTE
327 '\u013a' # 0x0092 -> LATIN SMALL LETTER L WITH ACUTE
328 '\xf4' # 0x0093 -> LATIN SMALL LETTER O WITH CIRCUMFLEX
329 '\xf6' # 0x0094 -> LATIN SMALL LETTER O WITH DIAERESIS
330 '\u013d' # 0x0095 -> LATIN CAPITAL LETTER L WITH CARON
331 '\u013e' # 0x0096 -> LATIN SMALL LETTER L WITH CARON
332 '\u015a' # 0x0097 -> LATIN CAPITAL LETTER S WITH ACUTE
333 '\u015b' # 0x0098 -> LATIN SMALL LETTER S WITH ACUTE
334 '\xd6' # 0x0099 -> LATIN CAPITAL LETTER O WITH DIAERESIS
335 '\xdc' # 0x009a -> LATIN CAPITAL LETTER U WITH DIAERESIS
336 '\u0164' # 0x009b -> LATIN CAPITAL LETTER T WITH CARON
337 '\u0165' # 0x009c -> LATIN SMALL LETTER T WITH CARON
338 '\u0141' # 0x009d -> LATIN CAPITAL LETTER L WITH STROKE
339 '\xd7' # 0x009e -> MULTIPLICATION SIGN
340 '\u010d' # 0x009f -> LATIN SMALL LETTER C WITH CARON
341 '\xe1' # 0x00a0 -> LATIN SMALL LETTER A WITH ACUTE
342 '\xed' # 0x00a1 -> LATIN SMALL LETTER I WITH ACUTE
343 '\xf3' # 0x00a2 -> LATIN SMALL LETTER O WITH ACUTE
344 '\xfa' # 0x00a3 -> LATIN SMALL LETTER U WITH ACUTE
345 '\u0104' # 0x00a4 -> LATIN CAPITAL LETTER A WITH OGONEK
346 '\u0105' # 0x00a5 -> LATIN SMALL LETTER A WITH OGONEK
347 '\u017d' # 0x00a6 -> LATIN CAPITAL LETTER Z WITH CARON
348 '\u017e' # 0x00a7 -> LATIN SMALL LETTER Z WITH CARON
349 '\u0118' # 0x00a8 -> LATIN CAPITAL LETTER E WITH OGONEK
350 '\u0119' # 0x00a9 -> LATIN SMALL LETTER E WITH OGONEK
351 '\xac' # 0x00aa -> NOT SIGN
352 '\u017a' # 0x00ab -> LATIN SMALL LETTER Z WITH ACUTE
353 '\u010c' # 0x00ac -> LATIN CAPITAL LETTER C WITH CARON
354 '\u015f' # 0x00ad -> LATIN SMALL LETTER S WITH CEDILLA
355 '\xab' # 0x00ae -> LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
356 '\xbb' # 0x00af -> RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
357 '\u2591' # 0x00b0 -> LIGHT SHADE
358 '\u2592' # 0x00b1 -> MEDIUM SHADE
359 '\u2593' # 0x00b2 -> DARK SHADE
360 '\u2502' # 0x00b3 -> BOX DRAWINGS LIGHT VERTICAL
361 '\u2524' # 0x00b4 -> BOX DRAWINGS LIGHT VERTICAL AND LEFT
362 '\xc1' # 0x00b5 -> LATIN CAPITAL LETTER A WITH ACUTE
363 '\xc2' # 0x00b6 -> LATIN CAPITAL LETTER A WITH CIRCUMFLEX
364 '\u011a' # 0x00b7 -> LATIN CAPITAL LETTER E WITH CARON
365 '\u015e' # 0x00b8 -> LATIN CAPITAL LETTER S WITH CEDILLA
366 '\u2563' # 0x00b9 -> BOX DRAWINGS DOUBLE VERTICAL AND LEFT
367 '\u2551' # 0x00ba -> BOX DRAWINGS DOUBLE VERTICAL
368 '\u2557' # 0x00bb -> BOX DRAWINGS DOUBLE DOWN AND LEFT
369 '\u255d' # 0x00bc -> BOX DRAWINGS DOUBLE UP AND LEFT
370 '\u017b' # 0x00bd -> LATIN CAPITAL LETTER Z WITH DOT ABOVE
371 '\u017c' # 0x00be -> LATIN SMALL LETTER Z WITH DOT ABOVE
372 '\u2510' # 0x00bf -> BOX DRAWINGS LIGHT DOWN AND LEFT
373 '\u2514' # 0x00c0 -> BOX DRAWINGS LIGHT UP AND RIGHT
374 '\u2534' # 0x00c1 -> BOX DRAWINGS LIGHT UP AND HORIZONTAL
375 '\u252c' # 0x00c2 -> BOX DRAWINGS LIGHT DOWN AND HORIZONTAL
376 '\u251c' # 0x00c3 -> BOX DRAWINGS LIGHT VERTICAL AND RIGHT
377 '\u2500' # 0x00c4 -> BOX DRAWINGS LIGHT HORIZONTAL
378 '\u253c' # 0x00c5 -> BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL
379 '\u0102' # 0x00c6 -> LATIN CAPITAL LETTER A WITH BREVE
380 '\u0103' # 0x00c7 -> LATIN SMALL LETTER A WITH BREVE
381 '\u255a' # 0x00c8 -> BOX DRAWINGS DOUBLE UP AND RIGHT
382 '\u2554' # 0x00c9 -> BOX DRAWINGS DOUBLE DOWN AND RIGHT
383 '\u2569' # 0x00ca -> BOX DRAWINGS DOUBLE UP AND HORIZONTAL
384 '\u2566' # 0x00cb -> BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL
385 '\u2560' # 0x00cc -> BOX DRAWINGS DOUBLE VERTICAL AND RIGHT
386 '\u2550' # 0x00cd -> BOX DRAWINGS DOUBLE HORIZONTAL
387 '\u256c' # 0x00ce -> BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL
388 '\xa4' # 0x00cf -> CURRENCY SIGN
389 '\u0111' # 0x00d0 -> LATIN SMALL LETTER D WITH STROKE
390 '\u0110' # 0x00d1 -> LATIN CAPITAL LETTER D WITH STROKE
391 '\u010e' # 0x00d2 -> LATIN CAPITAL LETTER D WITH CARON
392 '\xcb' # 0x00d3 -> LATIN CAPITAL LETTER E WITH DIAERESIS
393 '\u010f' # 0x00d4 -> LATIN SMALL LETTER D WITH CARON
394 '\u0147' # 0x00d5 -> LATIN CAPITAL LETTER N WITH CARON
395 '\xcd' # 0x00d6 -> LATIN CAPITAL LETTER I WITH ACUTE
396 '\xce' # 0x00d7 -> LATIN CAPITAL LETTER I WITH CIRCUMFLEX
397 '\u011b' # 0x00d8 -> LATIN SMALL LETTER E WITH CARON
398 '\u2518' # 0x00d9 -> BOX DRAWINGS LIGHT UP AND LEFT
399 '\u250c' # 0x00da -> BOX DRAWINGS LIGHT DOWN AND RIGHT
400 '\u2588' # 0x00db -> FULL BLOCK
401 '\u2584' # 0x00dc -> LOWER HALF BLOCK
402 '\u0162' # 0x00dd -> LATIN CAPITAL LETTER T WITH CEDILLA
403 '\u016e' # 0x00de -> LATIN CAPITAL LETTER U WITH RING ABOVE
404 '\u2580' # 0x00df -> UPPER HALF BLOCK
405 '\xd3' # 0x00e0 -> LATIN CAPITAL LETTER O WITH ACUTE
406 '\xdf' # 0x00e1 -> LATIN SMALL LETTER SHARP S
407 '\xd4' # 0x00e2 -> LATIN CAPITAL LETTER O WITH CIRCUMFLEX
408 '\u0143' # 0x00e3 -> LATIN CAPITAL LETTER N WITH ACUTE
409 '\u0144' # 0x00e4 -> LATIN SMALL LETTER N WITH ACUTE
410 '\u0148' # 0x00e5 -> LATIN SMALL LETTER N WITH CARON
411 '\u0160' # 0x00e6 -> LATIN CAPITAL LETTER S WITH CARON
412 '\u0161' # 0x00e7 -> LATIN SMALL LETTER S WITH CARON
413 '\u0154' # 0x00e8 -> LATIN CAPITAL LETTER R WITH ACUTE
414 '\xda' # 0x00e9 -> LATIN CAPITAL LETTER U WITH ACUTE
415 '\u0155' # 0x00ea -> LATIN SMALL LETTER R WITH ACUTE
416 '\u0170' # 0x00eb -> LATIN CAPITAL LETTER U WITH DOUBLE ACUTE
417 '\xfd' # 0x00ec -> LATIN SMALL LETTER Y WITH ACUTE
418 '\xdd' # 0x00ed -> LATIN CAPITAL LETTER Y WITH ACUTE
419 '\u0163' # 0x00ee -> LATIN SMALL LETTER T WITH CEDILLA
420 '\xb4' # 0x00ef -> ACUTE ACCENT
421 '\xad' # 0x00f0 -> SOFT HYPHEN
422 '\u02dd' # 0x00f1 -> DOUBLE ACUTE ACCENT
423 '\u02db' # 0x00f2 -> OGONEK
424 '\u02c7' # 0x00f3 -> CARON
425 '\u02d8' # 0x00f4 -> BREVE
426 '\xa7' # 0x00f5 -> SECTION SIGN
427 '\xf7' # 0x00f6 -> DIVISION SIGN
428 '\xb8' # 0x00f7 -> CEDILLA
429 '\xb0' # 0x00f8 -> DEGREE SIGN
430 '\xa8' # 0x00f9 -> DIAERESIS
431 '\u02d9' # 0x00fa -> DOT ABOVE
432 '\u0171' # 0x00fb -> LATIN SMALL LETTER U WITH DOUBLE ACUTE
433 '\u0158' # 0x00fc -> LATIN CAPITAL LETTER R WITH CARON
434 '\u0159' # 0x00fd -> LATIN SMALL LETTER R WITH CARON
435 '\u25a0' # 0x00fe -> BLACK SQUARE
436 '\xa0' # 0x00ff -> NO-BREAK SPACE
Marc-André Lemburga1129f42005-10-21 13:49:12 +0000437)
438
Guido van Rossum0229bf62000-03-10 23:17:24 +0000439### Encoding Map
440
Marc-André Lemburga1129f42005-10-21 13:49:12 +0000441encoding_map = {
Tim Peters536cf992005-12-25 23:18:31 +0000442 0x0000: 0x0000, # NULL
443 0x0001: 0x0001, # START OF HEADING
444 0x0002: 0x0002, # START OF TEXT
445 0x0003: 0x0003, # END OF TEXT
446 0x0004: 0x0004, # END OF TRANSMISSION
447 0x0005: 0x0005, # ENQUIRY
448 0x0006: 0x0006, # ACKNOWLEDGE
449 0x0007: 0x0007, # BELL
450 0x0008: 0x0008, # BACKSPACE
451 0x0009: 0x0009, # HORIZONTAL TABULATION
452 0x000a: 0x000a, # LINE FEED
453 0x000b: 0x000b, # VERTICAL TABULATION
454 0x000c: 0x000c, # FORM FEED
455 0x000d: 0x000d, # CARRIAGE RETURN
456 0x000e: 0x000e, # SHIFT OUT
457 0x000f: 0x000f, # SHIFT IN
458 0x0010: 0x0010, # DATA LINK ESCAPE
459 0x0011: 0x0011, # DEVICE CONTROL ONE
460 0x0012: 0x0012, # DEVICE CONTROL TWO
461 0x0013: 0x0013, # DEVICE CONTROL THREE
462 0x0014: 0x0014, # DEVICE CONTROL FOUR
463 0x0015: 0x0015, # NEGATIVE ACKNOWLEDGE
464 0x0016: 0x0016, # SYNCHRONOUS IDLE
465 0x0017: 0x0017, # END OF TRANSMISSION BLOCK
466 0x0018: 0x0018, # CANCEL
467 0x0019: 0x0019, # END OF MEDIUM
468 0x001a: 0x001a, # SUBSTITUTE
469 0x001b: 0x001b, # ESCAPE
470 0x001c: 0x001c, # FILE SEPARATOR
471 0x001d: 0x001d, # GROUP SEPARATOR
472 0x001e: 0x001e, # RECORD SEPARATOR
473 0x001f: 0x001f, # UNIT SEPARATOR
474 0x0020: 0x0020, # SPACE
475 0x0021: 0x0021, # EXCLAMATION MARK
476 0x0022: 0x0022, # QUOTATION MARK
477 0x0023: 0x0023, # NUMBER SIGN
478 0x0024: 0x0024, # DOLLAR SIGN
479 0x0025: 0x0025, # PERCENT SIGN
480 0x0026: 0x0026, # AMPERSAND
481 0x0027: 0x0027, # APOSTROPHE
482 0x0028: 0x0028, # LEFT PARENTHESIS
483 0x0029: 0x0029, # RIGHT PARENTHESIS
484 0x002a: 0x002a, # ASTERISK
485 0x002b: 0x002b, # PLUS SIGN
486 0x002c: 0x002c, # COMMA
487 0x002d: 0x002d, # HYPHEN-MINUS
488 0x002e: 0x002e, # FULL STOP
489 0x002f: 0x002f, # SOLIDUS
490 0x0030: 0x0030, # DIGIT ZERO
491 0x0031: 0x0031, # DIGIT ONE
492 0x0032: 0x0032, # DIGIT TWO
493 0x0033: 0x0033, # DIGIT THREE
494 0x0034: 0x0034, # DIGIT FOUR
495 0x0035: 0x0035, # DIGIT FIVE
496 0x0036: 0x0036, # DIGIT SIX
497 0x0037: 0x0037, # DIGIT SEVEN
498 0x0038: 0x0038, # DIGIT EIGHT
499 0x0039: 0x0039, # DIGIT NINE
500 0x003a: 0x003a, # COLON
501 0x003b: 0x003b, # SEMICOLON
502 0x003c: 0x003c, # LESS-THAN SIGN
503 0x003d: 0x003d, # EQUALS SIGN
504 0x003e: 0x003e, # GREATER-THAN SIGN
505 0x003f: 0x003f, # QUESTION MARK
506 0x0040: 0x0040, # COMMERCIAL AT
507 0x0041: 0x0041, # LATIN CAPITAL LETTER A
508 0x0042: 0x0042, # LATIN CAPITAL LETTER B
509 0x0043: 0x0043, # LATIN CAPITAL LETTER C
510 0x0044: 0x0044, # LATIN CAPITAL LETTER D
511 0x0045: 0x0045, # LATIN CAPITAL LETTER E
512 0x0046: 0x0046, # LATIN CAPITAL LETTER F
513 0x0047: 0x0047, # LATIN CAPITAL LETTER G
514 0x0048: 0x0048, # LATIN CAPITAL LETTER H
515 0x0049: 0x0049, # LATIN CAPITAL LETTER I
516 0x004a: 0x004a, # LATIN CAPITAL LETTER J
517 0x004b: 0x004b, # LATIN CAPITAL LETTER K
518 0x004c: 0x004c, # LATIN CAPITAL LETTER L
519 0x004d: 0x004d, # LATIN CAPITAL LETTER M
520 0x004e: 0x004e, # LATIN CAPITAL LETTER N
521 0x004f: 0x004f, # LATIN CAPITAL LETTER O
522 0x0050: 0x0050, # LATIN CAPITAL LETTER P
523 0x0051: 0x0051, # LATIN CAPITAL LETTER Q
524 0x0052: 0x0052, # LATIN CAPITAL LETTER R
525 0x0053: 0x0053, # LATIN CAPITAL LETTER S
526 0x0054: 0x0054, # LATIN CAPITAL LETTER T
527 0x0055: 0x0055, # LATIN CAPITAL LETTER U
528 0x0056: 0x0056, # LATIN CAPITAL LETTER V
529 0x0057: 0x0057, # LATIN CAPITAL LETTER W
530 0x0058: 0x0058, # LATIN CAPITAL LETTER X
531 0x0059: 0x0059, # LATIN CAPITAL LETTER Y
532 0x005a: 0x005a, # LATIN CAPITAL LETTER Z
533 0x005b: 0x005b, # LEFT SQUARE BRACKET
534 0x005c: 0x005c, # REVERSE SOLIDUS
535 0x005d: 0x005d, # RIGHT SQUARE BRACKET
536 0x005e: 0x005e, # CIRCUMFLEX ACCENT
537 0x005f: 0x005f, # LOW LINE
538 0x0060: 0x0060, # GRAVE ACCENT
539 0x0061: 0x0061, # LATIN SMALL LETTER A
540 0x0062: 0x0062, # LATIN SMALL LETTER B
541 0x0063: 0x0063, # LATIN SMALL LETTER C
542 0x0064: 0x0064, # LATIN SMALL LETTER D
543 0x0065: 0x0065, # LATIN SMALL LETTER E
544 0x0066: 0x0066, # LATIN SMALL LETTER F
545 0x0067: 0x0067, # LATIN SMALL LETTER G
546 0x0068: 0x0068, # LATIN SMALL LETTER H
547 0x0069: 0x0069, # LATIN SMALL LETTER I
548 0x006a: 0x006a, # LATIN SMALL LETTER J
549 0x006b: 0x006b, # LATIN SMALL LETTER K
550 0x006c: 0x006c, # LATIN SMALL LETTER L
551 0x006d: 0x006d, # LATIN SMALL LETTER M
552 0x006e: 0x006e, # LATIN SMALL LETTER N
553 0x006f: 0x006f, # LATIN SMALL LETTER O
554 0x0070: 0x0070, # LATIN SMALL LETTER P
555 0x0071: 0x0071, # LATIN SMALL LETTER Q
556 0x0072: 0x0072, # LATIN SMALL LETTER R
557 0x0073: 0x0073, # LATIN SMALL LETTER S
558 0x0074: 0x0074, # LATIN SMALL LETTER T
559 0x0075: 0x0075, # LATIN SMALL LETTER U
560 0x0076: 0x0076, # LATIN SMALL LETTER V
561 0x0077: 0x0077, # LATIN SMALL LETTER W
562 0x0078: 0x0078, # LATIN SMALL LETTER X
563 0x0079: 0x0079, # LATIN SMALL LETTER Y
564 0x007a: 0x007a, # LATIN SMALL LETTER Z
565 0x007b: 0x007b, # LEFT CURLY BRACKET
566 0x007c: 0x007c, # VERTICAL LINE
567 0x007d: 0x007d, # RIGHT CURLY BRACKET
568 0x007e: 0x007e, # TILDE
569 0x007f: 0x007f, # DELETE
570 0x00a0: 0x00ff, # NO-BREAK SPACE
571 0x00a4: 0x00cf, # CURRENCY SIGN
572 0x00a7: 0x00f5, # SECTION SIGN
573 0x00a8: 0x00f9, # DIAERESIS
574 0x00ab: 0x00ae, # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
575 0x00ac: 0x00aa, # NOT SIGN
576 0x00ad: 0x00f0, # SOFT HYPHEN
577 0x00b0: 0x00f8, # DEGREE SIGN
578 0x00b4: 0x00ef, # ACUTE ACCENT
579 0x00b8: 0x00f7, # CEDILLA
580 0x00bb: 0x00af, # RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
581 0x00c1: 0x00b5, # LATIN CAPITAL LETTER A WITH ACUTE
582 0x00c2: 0x00b6, # LATIN CAPITAL LETTER A WITH CIRCUMFLEX
583 0x00c4: 0x008e, # LATIN CAPITAL LETTER A WITH DIAERESIS
584 0x00c7: 0x0080, # LATIN CAPITAL LETTER C WITH CEDILLA
585 0x00c9: 0x0090, # LATIN CAPITAL LETTER E WITH ACUTE
586 0x00cb: 0x00d3, # LATIN CAPITAL LETTER E WITH DIAERESIS
587 0x00cd: 0x00d6, # LATIN CAPITAL LETTER I WITH ACUTE
588 0x00ce: 0x00d7, # LATIN CAPITAL LETTER I WITH CIRCUMFLEX
589 0x00d3: 0x00e0, # LATIN CAPITAL LETTER O WITH ACUTE
590 0x00d4: 0x00e2, # LATIN CAPITAL LETTER O WITH CIRCUMFLEX
591 0x00d6: 0x0099, # LATIN CAPITAL LETTER O WITH DIAERESIS
592 0x00d7: 0x009e, # MULTIPLICATION SIGN
593 0x00da: 0x00e9, # LATIN CAPITAL LETTER U WITH ACUTE
594 0x00dc: 0x009a, # LATIN CAPITAL LETTER U WITH DIAERESIS
595 0x00dd: 0x00ed, # LATIN CAPITAL LETTER Y WITH ACUTE
596 0x00df: 0x00e1, # LATIN SMALL LETTER SHARP S
597 0x00e1: 0x00a0, # LATIN SMALL LETTER A WITH ACUTE
598 0x00e2: 0x0083, # LATIN SMALL LETTER A WITH CIRCUMFLEX
599 0x00e4: 0x0084, # LATIN SMALL LETTER A WITH DIAERESIS
600 0x00e7: 0x0087, # LATIN SMALL LETTER C WITH CEDILLA
601 0x00e9: 0x0082, # LATIN SMALL LETTER E WITH ACUTE
602 0x00eb: 0x0089, # LATIN SMALL LETTER E WITH DIAERESIS
603 0x00ed: 0x00a1, # LATIN SMALL LETTER I WITH ACUTE
604 0x00ee: 0x008c, # LATIN SMALL LETTER I WITH CIRCUMFLEX
605 0x00f3: 0x00a2, # LATIN SMALL LETTER O WITH ACUTE
606 0x00f4: 0x0093, # LATIN SMALL LETTER O WITH CIRCUMFLEX
607 0x00f6: 0x0094, # LATIN SMALL LETTER O WITH DIAERESIS
608 0x00f7: 0x00f6, # DIVISION SIGN
609 0x00fa: 0x00a3, # LATIN SMALL LETTER U WITH ACUTE
610 0x00fc: 0x0081, # LATIN SMALL LETTER U WITH DIAERESIS
611 0x00fd: 0x00ec, # LATIN SMALL LETTER Y WITH ACUTE
612 0x0102: 0x00c6, # LATIN CAPITAL LETTER A WITH BREVE
613 0x0103: 0x00c7, # LATIN SMALL LETTER A WITH BREVE
614 0x0104: 0x00a4, # LATIN CAPITAL LETTER A WITH OGONEK
615 0x0105: 0x00a5, # LATIN SMALL LETTER A WITH OGONEK
616 0x0106: 0x008f, # LATIN CAPITAL LETTER C WITH ACUTE
617 0x0107: 0x0086, # LATIN SMALL LETTER C WITH ACUTE
618 0x010c: 0x00ac, # LATIN CAPITAL LETTER C WITH CARON
619 0x010d: 0x009f, # LATIN SMALL LETTER C WITH CARON
620 0x010e: 0x00d2, # LATIN CAPITAL LETTER D WITH CARON
621 0x010f: 0x00d4, # LATIN SMALL LETTER D WITH CARON
622 0x0110: 0x00d1, # LATIN CAPITAL LETTER D WITH STROKE
623 0x0111: 0x00d0, # LATIN SMALL LETTER D WITH STROKE
624 0x0118: 0x00a8, # LATIN CAPITAL LETTER E WITH OGONEK
625 0x0119: 0x00a9, # LATIN SMALL LETTER E WITH OGONEK
626 0x011a: 0x00b7, # LATIN CAPITAL LETTER E WITH CARON
627 0x011b: 0x00d8, # LATIN SMALL LETTER E WITH CARON
628 0x0139: 0x0091, # LATIN CAPITAL LETTER L WITH ACUTE
629 0x013a: 0x0092, # LATIN SMALL LETTER L WITH ACUTE
630 0x013d: 0x0095, # LATIN CAPITAL LETTER L WITH CARON
631 0x013e: 0x0096, # LATIN SMALL LETTER L WITH CARON
632 0x0141: 0x009d, # LATIN CAPITAL LETTER L WITH STROKE
633 0x0142: 0x0088, # LATIN SMALL LETTER L WITH STROKE
634 0x0143: 0x00e3, # LATIN CAPITAL LETTER N WITH ACUTE
635 0x0144: 0x00e4, # LATIN SMALL LETTER N WITH ACUTE
636 0x0147: 0x00d5, # LATIN CAPITAL LETTER N WITH CARON
637 0x0148: 0x00e5, # LATIN SMALL LETTER N WITH CARON
638 0x0150: 0x008a, # LATIN CAPITAL LETTER O WITH DOUBLE ACUTE
639 0x0151: 0x008b, # LATIN SMALL LETTER O WITH DOUBLE ACUTE
640 0x0154: 0x00e8, # LATIN CAPITAL LETTER R WITH ACUTE
641 0x0155: 0x00ea, # LATIN SMALL LETTER R WITH ACUTE
642 0x0158: 0x00fc, # LATIN CAPITAL LETTER R WITH CARON
643 0x0159: 0x00fd, # LATIN SMALL LETTER R WITH CARON
644 0x015a: 0x0097, # LATIN CAPITAL LETTER S WITH ACUTE
645 0x015b: 0x0098, # LATIN SMALL LETTER S WITH ACUTE
646 0x015e: 0x00b8, # LATIN CAPITAL LETTER S WITH CEDILLA
647 0x015f: 0x00ad, # LATIN SMALL LETTER S WITH CEDILLA
648 0x0160: 0x00e6, # LATIN CAPITAL LETTER S WITH CARON
649 0x0161: 0x00e7, # LATIN SMALL LETTER S WITH CARON
650 0x0162: 0x00dd, # LATIN CAPITAL LETTER T WITH CEDILLA
651 0x0163: 0x00ee, # LATIN SMALL LETTER T WITH CEDILLA
652 0x0164: 0x009b, # LATIN CAPITAL LETTER T WITH CARON
653 0x0165: 0x009c, # LATIN SMALL LETTER T WITH CARON
654 0x016e: 0x00de, # LATIN CAPITAL LETTER U WITH RING ABOVE
655 0x016f: 0x0085, # LATIN SMALL LETTER U WITH RING ABOVE
656 0x0170: 0x00eb, # LATIN CAPITAL LETTER U WITH DOUBLE ACUTE
657 0x0171: 0x00fb, # LATIN SMALL LETTER U WITH DOUBLE ACUTE
658 0x0179: 0x008d, # LATIN CAPITAL LETTER Z WITH ACUTE
659 0x017a: 0x00ab, # LATIN SMALL LETTER Z WITH ACUTE
660 0x017b: 0x00bd, # LATIN CAPITAL LETTER Z WITH DOT ABOVE
661 0x017c: 0x00be, # LATIN SMALL LETTER Z WITH DOT ABOVE
662 0x017d: 0x00a6, # LATIN CAPITAL LETTER Z WITH CARON
663 0x017e: 0x00a7, # LATIN SMALL LETTER Z WITH CARON
664 0x02c7: 0x00f3, # CARON
665 0x02d8: 0x00f4, # BREVE
666 0x02d9: 0x00fa, # DOT ABOVE
667 0x02db: 0x00f2, # OGONEK
668 0x02dd: 0x00f1, # DOUBLE ACUTE ACCENT
669 0x2500: 0x00c4, # BOX DRAWINGS LIGHT HORIZONTAL
670 0x2502: 0x00b3, # BOX DRAWINGS LIGHT VERTICAL
671 0x250c: 0x00da, # BOX DRAWINGS LIGHT DOWN AND RIGHT
672 0x2510: 0x00bf, # BOX DRAWINGS LIGHT DOWN AND LEFT
673 0x2514: 0x00c0, # BOX DRAWINGS LIGHT UP AND RIGHT
674 0x2518: 0x00d9, # BOX DRAWINGS LIGHT UP AND LEFT
675 0x251c: 0x00c3, # BOX DRAWINGS LIGHT VERTICAL AND RIGHT
676 0x2524: 0x00b4, # BOX DRAWINGS LIGHT VERTICAL AND LEFT
677 0x252c: 0x00c2, # BOX DRAWINGS LIGHT DOWN AND HORIZONTAL
678 0x2534: 0x00c1, # BOX DRAWINGS LIGHT UP AND HORIZONTAL
679 0x253c: 0x00c5, # BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL
680 0x2550: 0x00cd, # BOX DRAWINGS DOUBLE HORIZONTAL
681 0x2551: 0x00ba, # BOX DRAWINGS DOUBLE VERTICAL
682 0x2554: 0x00c9, # BOX DRAWINGS DOUBLE DOWN AND RIGHT
683 0x2557: 0x00bb, # BOX DRAWINGS DOUBLE DOWN AND LEFT
684 0x255a: 0x00c8, # BOX DRAWINGS DOUBLE UP AND RIGHT
685 0x255d: 0x00bc, # BOX DRAWINGS DOUBLE UP AND LEFT
686 0x2560: 0x00cc, # BOX DRAWINGS DOUBLE VERTICAL AND RIGHT
687 0x2563: 0x00b9, # BOX DRAWINGS DOUBLE VERTICAL AND LEFT
688 0x2566: 0x00cb, # BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL
689 0x2569: 0x00ca, # BOX DRAWINGS DOUBLE UP AND HORIZONTAL
690 0x256c: 0x00ce, # BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL
691 0x2580: 0x00df, # UPPER HALF BLOCK
692 0x2584: 0x00dc, # LOWER HALF BLOCK
693 0x2588: 0x00db, # FULL BLOCK
694 0x2591: 0x00b0, # LIGHT SHADE
695 0x2592: 0x00b1, # MEDIUM SHADE
696 0x2593: 0x00b2, # DARK SHADE
697 0x25a0: 0x00fe, # BLACK SQUARE
698}