a b/medacy/tools/unicode_to_ascii.py
1
"""
2
Extracted from the UTF-8 to ASCII conversion program found at
3
https://metamap.nlm.nih.gov/ReplaceUTF8.shtml
4
"""
5
6
7
UNICODE_TO_ASCII = {
8
    '\u221E': " ",
9
    '\u02C2': " ",
10
    '\u2218': " ",
11
    '\u0020': " ",
12
    '\u0021': "!",
13
    '\u0023': "#",
14
    '\u0024': "$",
15
    '\u0025': "%",
16
    '\u0026': "&",
17
    '\u0028': "(",
18
    '\u0029': ")",
19
    '\u002a': "*",
20
    '\u002b': "+",
21
    '\u002c': ",",
22
    '\u002d': "-",
23
    '\u002e': ".",
24
    '\u002f': "/",
25
    '\u0030': "0",
26
    '\u0031': "1",
27
    '\u0032': "2",
28
    '\u0033': "3",
29
    '\u0034': "4",
30
    '\u0035': "5",
31
    '\u0036': "6",
32
    '\u0037': "7",
33
    '\u0038': "8",
34
    '\u0039': "9",
35
    '\u003a': ":",
36
    '\u003b': ";",
37
    '\u003c': "<",
38
    '\u003d': "=",
39
    '\u003e': ">",
40
    '\u003f': "?",
41
    '\u0040': "@",
42
    '\u0041': "A",
43
    '\u0042': "B",
44
    '\u0043': "C",
45
    '\u0044': "D",
46
    '\u0045': "E",
47
    '\u0046': "F",
48
    '\u0047': "G",
49
    '\u0048': "H",
50
    '\u0049': "I",
51
    '\u004a': "J",
52
    '\u004b': "K",
53
    '\u004c': "L",
54
    '\u004d': "M",
55
    '\u004e': "N",
56
    '\u004f': "O",
57
    '\u0050': "P",
58
    '\u0051': "Q",
59
    '\u0052': "R",
60
    '\u0053': "S",
61
    '\u0054': "T",
62
    '\u0055': "U",
63
    '\u0056': "V",
64
    '\u0057': "W",
65
    '\u0058': "X",
66
    '\u0059': "Y",
67
    '\u005a': "Z",
68
    '\u005b': "[",
69
    '\u005d': "]",
70
    '\u005e': "^",
71
    '\u005f': "_",
72
    '\u0060': "`",
73
    '\u0061': "a",
74
    '\u0062': "b",
75
    '\u0063': "c",
76
    '\u0064': "d",
77
    '\u0065': "e",
78
    '\u0066': "f",
79
    '\u0067': "g",
80
    '\u0068': "h",
81
    '\u0069': "i",
82
    '\u006a': "j",
83
    '\u006b': "k",
84
    '\u006c': "l",
85
    '\u006d': "m",
86
    '\u006e': "n",
87
    '\u006f': "o",
88
    '\u0070': "p",
89
    '\u0071': "q",
90
    '\u0072': "r",
91
    '\u0073': "s",
92
    '\u0074': "t",
93
    '\u0075': "u",
94
    '\u0076': "v",
95
    '\u0077': "w",
96
    '\u0078': "x",
97
    '\u0079': "y",
98
    '\u007a': "z",
99
    '\u007b': "{",
100
    '\u007c': "|",
101
    '\u007d': "}",
102
    '\u007e': "~",
103
    '\u00a0': " ",
104
    '\u00a1': "!",
105
    '\u00a2': "c", # Cent sign
106
    '\u00a3': "L", # Pound sign
107
    '\u00a4': " ", # Currency sign
108
    '\u00a5': "Y", # Yen sign
109
    '\u00a6': "|",
110
    '\u00a7': "S", # Section Sign
111
    '\u00a8': "",
112
    '\u00a9': "(c)",
113
    '\u00aa': "a",
114
    '\u00ac': "!", # Not sign
115
    '\u00ad': "-",
116
    '\u00ae': "(r)",
117
    '\u00af': "-", # Macron
118
    '\u00b0': " degrees ", # Degree
119
    '\u00b1': "+/-",
120
    '\u00b2': "2",
121
    '\u00b3': "3",
122
    '\u00b4': "'",
123
    '\u00b5': "u",
124
    '\u00b6': " ", # Paragraph
125
    '\u00b7': ".", # Middle dot
126
    '\u00b8': " ", # Cedilla
127
    '\u00b9': "1",
128
    '\u00ba': "o",
129
    '\u00bc': "1/4",
130
    '\u00bd': "1/2",
131
    '\u00be': "3/4",
132
    '\u00bf': "?", # Inverted Question Mark
133
    '\u00c0': "A",
134
    '\u00c1': "A",
135
    '\u00c2': "A",
136
    '\u00c3': "A",
137
    '\u00c4': "A",
138
    '\u00c5': "A",
139
    '\u00c6': "AE",
140
    '\u00c7': "C",
141
    '\u00c8': "E",
142
    '\u00c9': "E",
143
    '\u00ca': "E",
144
    '\u00cb': "E",
145
    '\u00cc': "I",
146
    '\u00cd': "I",
147
    '\u00ce': "I",
148
    '\u00cf': "I",
149
    '\u00d0': "D", # Cap Eth
150
    '\u00d1': "N",
151
    '\u00d2': "O",
152
    '\u00d3': "O",
153
    '\u00d4': "O",
154
    '\u00d5': "O",
155
    '\u00d6': "O",
156
    '\u00d7': "*",
157
    '\u00d8': "O",
158
    '\u00d9': "U",
159
    '\u00da': "U",
160
    '\u00db': "U",
161
    '\u00dc': "U",
162
    '\u00dd': "Y",
163
    '\u00de': "P", # Cap Thorn
164
    '\u00df': "beta",
165
    '\u00e0': "a",
166
    '\u00e1': "a",
167
    '\u00e2': "a",
168
    '\u00e3': "a",
169
    '\u00e4': "a",
170
    '\u00e5': "a",
171
    '\u00e6': "ae",
172
    '\u00e7': "c",
173
    '\u00e8': "e",
174
    '\u00e9': "e",
175
    '\u00ea': "e",
176
    '\u00eb': "e",
177
    '\u00ec': "i",
178
    '\u00ed': "i",
179
    '\u00ee': "i",
180
    '\u00ef': "i",
181
    '\u00f1': "n",
182
    '\u00f2': "o",
183
    '\u00f3': "o",
184
    '\u00f4': "o",
185
    '\u00f5': "o",
186
    '\u00f6': "o",
187
    '\u00f7': "/",
188
    '\u00f8': "o",
189
    '\u00f9': "u",
190
    '\u00fa': "u",
191
    '\u00fb': "u",
192
    '\u00fc': "u",
193
    '\u00fd': "y",
194
    '\u00fe': "p", # Thorn
195
    '\u00ff': "y",
196
    '\u0100': "A",
197
    '\u0101': "a",
198
    '\u0102': "A",
199
    '\u0103': "a",
200
    '\u0104': "A",
201
    '\u0105': "a",
202
    '\u0106': "C",
203
    '\u0107': "c",
204
    '\u0108': "C",
205
    '\u0109': "c",
206
    '\u010a': "C",
207
    '\u010b': "c",
208
    '\u010c': "C",
209
    '\u010d': "c",
210
    '\u010e': "D",
211
    '\u010f': "d",
212
    '\u0110': "D",
213
    '\u0111': "d",
214
    '\u0112': "E",
215
    '\u0113': "e",
216
    '\u0114': "E",
217
    '\u0115': "e",
218
    '\u0116': "E",
219
    '\u0117': "e",
220
    '\u0118': "E",
221
    '\u0119': "e",
222
    '\u011a': "E",
223
    '\u011b': "e",
224
    '\u011c': "G",
225
    '\u011d': "g",
226
    '\u011e': "G",
227
    '\u011f': "g",
228
    '\u0120': "G",
229
    '\u0121': "g",
230
    '\u0122': "G",
231
    '\u0123': "g",
232
    '\u0124': "H",
233
    '\u0125': "h",
234
    '\u0126': "H",
235
    '\u0127': "h",
236
    '\u0128': "I",
237
    '\u0129': "i",
238
    '\u012a': "I",
239
    '\u012b': "i",
240
    '\u012c': "I",
241
    '\u012d': "i",
242
    '\u012e': "I",
243
    '\u012f': "i",
244
    '\u0130': "I",
245
    '\u0131': "i",
246
    '\u0132': "IJ",
247
    '\u0133': "ij",
248
    '\u0134': "J",
249
    '\u0135': "j",
250
    '\u0136': "K",
251
    '\u0137': "k",
252
    '\u0139': "L",
253
    '\u013a': "l",
254
    '\u013b': "L",
255
    '\u013c': "l",
256
    '\u013d': "L",
257
    '\u013e': "l",
258
    '\u013f': "L",
259
    '\u0140': "l",
260
    '\u0141': "L",
261
    '\u0142': "l",
262
    '\u0143': "N",
263
    '\u0144': "n",
264
    '\u0145': "N",
265
    '\u0146': "n",
266
    '\u0147': "N",
267
    '\u0148': "n",
268
    '\u0149': "n",
269
    '\u014c': "O",
270
    '\u014d': "o",
271
    '\u014e': "O",
272
    '\u014f': "o",
273
    '\u0150': "O",
274
    '\u0151': "o",
275
    '\u0152': "OE",
276
    '\u0153': "oe",
277
    '\u0154': "R",
278
    '\u0155': "r",
279
    '\u0156': "R",
280
    '\u0157': "r",
281
    '\u0158': "R",
282
    '\u0159': "r",
283
    '\u015a': "S",
284
    '\u015b': "s",
285
    '\u015c': "S",
286
    '\u015d': "s",
287
    '\u015e': "S",
288
    '\u015f': "s",
289
    '\u0160': "S",
290
    '\u0161': "s",
291
    '\u0162': "T",
292
    '\u0163': "t",
293
    '\u0164': "T",
294
    '\u0165': "t",
295
    '\u0166': "T",
296
    '\u0167': "t",
297
    '\u0168': "U",
298
    '\u0169': "u",
299
    '\u016a': "U",
300
    '\u016b': "u",
301
    '\u016c': "U",
302
    '\u016d': "u",
303
    '\u016e': "U",
304
    '\u016f': "u",
305
    '\u0170': "U",
306
    '\u0171': "u",
307
    '\u0172': "U",
308
    '\u0173': "u",
309
    '\u0174': "W",
310
    '\u0175': "w",
311
    '\u0176': "Y",
312
    '\u0177': "y",
313
    '\u0178': "Y",
314
    '\u0179': "Z",
315
    '\u017a': "z",
316
    '\u017b': "Z",
317
    '\u017c': "z",
318
    '\u017d': "Z",
319
    '\u017e': "z",
320
    '\u017f': "s",
321
    '\u0180': "b",
322
    '\u0181': "B",
323
    '\u0182': "B",
324
    '\u0183': "b",
325
    '\u0184': "6",
326
    '\u0185': "6",
327
    '\u0186': "O",
328
    '\u0187': "C",
329
    '\u0188': "c",
330
    '\u0189': "D",
331
    '\u018a': "D",
332
    '\u018b': "D",
333
    '\u018c': "d",
334
    '\u018e': "E",
335
    '\u0190': "E",
336
    '\u0191': "F",
337
    '\u0192': "f",
338
    '\u0193': "G",
339
    '\u0194': "Gamma",
340
    '\u0195': "hv",
341
    '\u0196': "Iota",
342
    '\u0197': "I",
343
    '\u0198': "K",
344
    '\u0199': "k",
345
    '\u019a': "l",
346
    '\u019b': "lambda",
347
    '\u019c': "M",
348
    '\u019d': "N",
349
    '\u019e': "n",
350
    '\u019f': "O",
351
    '\u01a0': "O",
352
    '\u01a1': "o",
353
    '\u01a2': "OI",
354
    '\u01a3': "oi",
355
    '\u01a4': "P",
356
    '\u01a5': "p",
357
    '\u01a6': "YR",
358
    '\u01a7': "2",
359
    '\u01a8': "2",
360
    '\u01ab': "t",
361
    '\u01ac': "T",
362
    '\u01ad': "t",
363
    '\u01ae': "T",
364
    '\u01af': "U",
365
    '\u01b0': "u",
366
    '\u01b1': "Upsilon",
367
    '\u01b2': "V",
368
    '\u01b3': "Y",
369
    '\u01b4': "y",
370
    '\u01b5': "Z",
371
    '\u01b6': "z",
372
    '\u01bb': "2",
373
    '\u01bc': "5",
374
    '\u01bd': "5",
375
    '\u01c0': "|",
376
    '\u01c3': "!",
377
    '\u01c4': "DZ",
378
    '\u01c5': "Dz",
379
    '\u01c6': "dz",
380
    '\u01c7': "LJ",
381
    '\u01c8': "Lj",
382
    '\u01c9': "lj",
383
    '\u01ca': "NJ",
384
    '\u01cb': "Nj",
385
    '\u01cc': "nj",
386
    '\u01cd': "A",
387
    '\u01ce': "a",
388
    '\u01cf': "I",
389
    '\u01d0': "i",
390
    '\u01d1': "O",
391
    '\u01d2': "o",
392
    '\u01d3': "U",
393
    '\u01d4': "u",
394
    '\u01d5': "U",
395
    '\u01d6': "u",
396
    '\u01d7': "U",
397
    '\u01d8': "u",
398
    '\u01d9': "U",
399
    '\u01da': "u",
400
    '\u01db': "U",
401
    '\u01dc': "u",
402
    '\u01dd': "e",
403
    '\u01de': "A",
404
    '\u01df': "a",
405
    '\u01e0': "A",
406
    '\u01e1': "a",
407
    '\u01e2': "AE",
408
    '\u01e3': "ae",
409
    '\u01e4': "G",
410
    '\u01e5': "g",
411
    '\u01e6': "G",
412
    '\u01e7': "g",
413
    '\u01e8': "K",
414
    '\u01e9': "k",
415
    '\u01ea': "O",
416
    '\u01eb': "o",
417
    '\u01ec': "O",
418
    '\u01ed': "o",
419
    '\u01f0': "j",
420
    '\u01f1': "DZ",
421
    '\u01f2': "Dz",
422
    '\u01f3': "dz",
423
    '\u01f4': "G",
424
    '\u01f5': "g",
425
    '\u01f8': "N",
426
    '\u01f9': "n",
427
    '\u01fa': "A",
428
    '\u01fb': "a",
429
    '\u01fc': "AE",
430
    '\u01fd': "ae",
431
    '\u01fe': "O",
432
    '\u01ff': "o",
433
    '\u0200': "A",
434
    '\u0201': "a",
435
    '\u0202': "A",
436
    '\u0203': "a",
437
    '\u0204': "E",
438
    '\u0205': "e",
439
    '\u0206': "E",
440
    '\u0207': "e",
441
    '\u0208': "I",
442
    '\u0209': "i",
443
    '\u020a': "I",
444
    '\u020b': "i",
445
    '\u020c': "O",
446
    '\u020d': "o",
447
    '\u020e': "O",
448
    '\u020f': "o",
449
    '\u0210': "R",
450
    '\u0211': "r",
451
    '\u0212': "R",
452
    '\u0213': "r",
453
    '\u0214': "U",
454
    '\u0215': "u",
455
    '\u0216': "U",
456
    '\u0217': "u",
457
    '\u0218': "S",
458
    '\u0219': "s",
459
    '\u021a': "T",
460
    '\u021b': "t",
461
    '\u021e': "H",
462
    '\u021f': "h",
463
    '\u0220': "N",
464
    '\u0221': "d",
465
    '\u0222': "OU",
466
    '\u0223': "ou",
467
    '\u0224': "Z",
468
    '\u0225': "z",
469
    '\u0226': "A",
470
    '\u0227': "a",
471
    '\u0228': "E",
472
    '\u0229': "e",
473
    '\u022a': "O",
474
    '\u022b': "o",
475
    '\u022c': "O",
476
    '\u022d': "o",
477
    '\u022e': "O",
478
    '\u022f': "o",
479
    '\u0230': "O",
480
    '\u0231': "o",
481
    '\u0232': "Y",
482
    '\u0233': "y",
483
    '\u0234': "l",
484
    '\u0235': "n",
485
    '\u0236': "t",
486
    '\u0237': "j",
487
    '\u0238': "db",
488
    '\u0239': "qp",
489
    '\u023a': "A",
490
    '\u023b': "C",
491
    '\u023c': "c",
492
    '\u023d': "L",
493
    '\u023e': "T",
494
    '\u023f': "s",
495
    '\u0240': "z",
496
    '\u0243': "B",
497
    '\u0244': "U",
498
    '\u0245': "V",
499
    '\u0246': "E",
500
    '\u0247': "e",
501
    '\u0248': "J",
502
    '\u0249': "j",
503
    '\u024a': "Q",
504
    '\u024b': "q",
505
    '\u024c': "R",
506
    '\u024d': "r",
507
    '\u024e': "Y",
508
    '\u024f': "y",
509
    '\u0250': "a",
510
    '\u0251': "alpha",
511
    '\u0252': "alpha",
512
    '\u0253': "b",
513
    '\u0254': "o",
514
    '\u0255': "c",
515
    '\u0256': "d",
516
    '\u0257': "d",
517
    '\u025b': "E",
518
    '\u025f': "j",
519
    '\u0260': "g",
520
    '\u0261': "g",
521
    '\u0262': "G",
522
    '\u0263': "gamma",
523
    '\u0265': "h",
524
    '\u0266': "h",
525
    '\u0268': "i",
526
    '\u0269': "iota",
527
    '\u026a': "i",
528
    '\u026b': "l",
529
    '\u026c': "l",
530
    '\u026d': "l",
531
    '\u026f': "m",
532
    '\u0270': "m",
533
    '\u0271': "m",
534
    '\u0272': "n",
535
    '\u0273': "n",
536
    '\u0274': "N",
537
    '\u0275': "o",
538
    '\u0276': "OE",
539
    '\u0277': "omega",
540
    '\u0278': "phi",
541
    '\u0279': "r",
542
    '\u027a': "r",
543
    '\u027b': "r",
544
    '\u027c': "r",
545
    '\u027d': "r",
546
    '\u027e': "r",
547
    '\u0280': "R",
548
    '\u0282': "s",
549
    '\u0284': "j",
550
    '\u0287': "t",
551
    '\u0288': "t",
552
    '\u0289': "u",
553
    '\u028a': "upsilon",
554
    '\u028b': "v",
555
    '\u028c': "v",
556
    '\u028d': "w",
557
    '\u028e': "y",
558
    '\u028f': "Y",
559
    '\u0290': "z",
560
    '\u0291': "z",
561
    '\u0297': "C",
562
    '\u0299': "B",
563
    '\u029a': "e",
564
    '\u029b': "G",
565
    '\u029c': "H",
566
    '\u029d': "j",
567
    '\u029e': "k",
568
    '\u029f': "L",
569
    '\u02a0': "q",
570
    '\u02a3': "DZ",
571
    '\u02a5': "dz",
572
    '\u02a6': "ts",
573
    '\u02a8': "tc",
574
    '\u02aa': "ls",
575
    '\u02ab': "lz",
576
    '\u02ae': "h",
577
    '\u02af': "h",
578
    '\u02c6': "^",
579
    '\u02dc': "~",
580
    '\u037e': ";",
581
    '\u0386': "Alpha",
582
    '\u0388': "Epsilon",
583
    '\u0389': "Eta",
584
    '\u038a': "Iota",
585
    '\u038c': "Omicron",
586
    '\u038e': "Upsilon",
587
    '\u038f': "Omega",
588
    '\u0390': "iota",
589
    '\u0391': "Alpha",
590
    '\u0392': "Beta",
591
    '\u0393': "Gamma",
592
    '\u0394': "Delta",
593
    '\u0395': "Epsilon",
594
    '\u0396': "Zeta",
595
    '\u0397': "Eta",
596
    '\u0398': "Theta",
597
    '\u0399': "Iota",
598
    '\u039a': "Kappa",
599
    '\u039b': "Lambda",
600
    '\u039c': "Mu",
601
    '\u039d': "Nu",
602
    '\u039e': "Xi",
603
    '\u039f': "Omicron",
604
    '\u03a0': "Pi",
605
    '\u03a1': "Rho",
606
    '\u03a3': "Sigma",
607
    '\u03a4': "Tau",
608
    '\u03a5': "Upsilon",
609
    '\u03a6': "Phi",
610
    '\u03a7': "Chi",
611
    '\u03a8': "Psi",
612
    '\u03a9': "Omega",
613
    '\u03aa': "Iota",
614
    '\u03ab': "Upsilon",
615
    '\u03ac': "alpha",
616
    '\u03ad': "epsilon",
617
    '\u03ae': "eta",
618
    '\u03af': "iota",
619
    '\u03b0': "upsilon",
620
    '\u03b1': "alpha",
621
    '\u03b2': "beta",
622
    '\u03b3': "gamma",
623
    '\u03b4': "delta",
624
    '\u03b5': "epsilon",
625
    '\u03b6': "zeta",
626
    '\u03b7': "eta",
627
    '\u03b8': "theta",
628
    '\u03b9': "iota",
629
    '\u03ba': "kappa",
630
    '\u03bb': "lambda",
631
    '\u03bc': "mu",
632
    '\u03bd': "nu",
633
    '\u03be': "xi",
634
    '\u03bf': "omicron",
635
    '\u03c0': "pi",
636
    '\u03c1': "rho",
637
    '\u03c2': "sigma",
638
    '\u03c3': "sigma",
639
    '\u03c4': "tau",
640
    '\u03c5': "upsilon",
641
    '\u03c6': "phi",
642
    '\u03c7': "chi",
643
    '\u03c8': "psi",
644
    '\u03c9': "omega",
645
    '\u03ca': "iota",
646
    '\u03cb': "upsilon",
647
    '\u03cc': "omicron",
648
    '\u03cd': "upsilon",
649
    '\u03ce': "omega",
650
    '\u03d0': "beta",
651
    '\u03d1': "theta",
652
    '\u03d2': "Upsilon",
653
    '\u03d3': "Upsilon",
654
    '\u03d4': "Upsilon",
655
    '\u03d5': "phi",
656
    '\u03d6': "pi",
657
    '\u03f0': "kappa",
658
    '\u03f1': "rho",
659
    '\u03f2': "sigma",
660
    '\u03f4': "Theta",
661
    '\u03f5': "epsilon",
662
    '\u03f9': "Sigma",
663
    '\u03fc': "Rho",
664
    '\u1e00': "A",
665
    '\u1e01': "a",
666
    '\u1e02': "B",
667
    '\u1e03': "b",
668
    '\u1e04': "B",
669
    '\u1e05': "b",
670
    '\u1e06': "B",
671
    '\u1e07': "b",
672
    '\u1e08': "C",
673
    '\u1e09': "c",
674
    '\u1e0a': "D",
675
    '\u1e0b': "d",
676
    '\u1e0c': "D",
677
    '\u1e0d': "d",
678
    '\u1e0e': "D",
679
    '\u1e0f': "d",
680
    '\u1e10': "D",
681
    '\u1e11': "d",
682
    '\u1e12': "D",
683
    '\u1e13': "d",
684
    '\u1e14': "E",
685
    '\u1e15': "e",
686
    '\u1e16': "E",
687
    '\u1e17': "e",
688
    '\u1e18': "E",
689
    '\u1e19': "e",
690
    '\u1e1a': "E",
691
    '\u1e1b': "e",
692
    '\u1e1c': "E",
693
    '\u1e1d': "e",
694
    '\u1e1e': "F",
695
    '\u1e1f': "f",
696
    '\u1e20': "G",
697
    '\u1e21': "g",
698
    '\u1e22': "H",
699
    '\u1e23': "h",
700
    '\u1e24': "H",
701
    '\u1e25': "h",
702
    '\u1e26': "H",
703
    '\u1e27': "h",
704
    '\u1e28': "H",
705
    '\u1e29': "h",
706
    '\u1e2a': "H",
707
    '\u1e2b': "h",
708
    '\u1e2c': "I",
709
    '\u1e2d': "i",
710
    '\u1e2e': "I",
711
    '\u1e2f': "i",
712
    '\u1e30': "K",
713
    '\u1e31': "k",
714
    '\u1e32': "K",
715
    '\u1e33': "k",
716
    '\u1e34': "K",
717
    '\u1e35': "k",
718
    '\u1e36': "L",
719
    '\u1e37': "l",
720
    '\u1e38': "L",
721
    '\u1e39': "l",
722
    '\u1e3a': "L",
723
    '\u1e3b': "l",
724
    '\u1e3c': "L",
725
    '\u1e3d': "l",
726
    '\u1e3e': "M",
727
    '\u1e3f': "m",
728
    '\u1e40': "M",
729
    '\u1e41': "m",
730
    '\u1e42': "M",
731
    '\u1e43': "m",
732
    '\u1e44': "N",
733
    '\u1e45': "n",
734
    '\u1e46': "N",
735
    '\u1e47': "n",
736
    '\u1e48': "N",
737
    '\u1e49': "n",
738
    '\u1e4a': "N",
739
    '\u1e4b': "n",
740
    '\u1e4c': "O",
741
    '\u1e4d': "o",
742
    '\u1e4e': "O",
743
    '\u1e4f': "o",
744
    '\u1e50': "O",
745
    '\u1e51': "o",
746
    '\u1e52': "O",
747
    '\u1e53': "o",
748
    '\u1e54': "P",
749
    '\u1e55': "p",
750
    '\u1e56': "P",
751
    '\u1e57': "p",
752
    '\u1e58': "R",
753
    '\u1e59': "r",
754
    '\u1e5a': "R",
755
    '\u1e5b': "r",
756
    '\u1e5c': "R",
757
    '\u1e5d': "r",
758
    '\u1e5e': "R",
759
    '\u1e5f': "r",
760
    '\u1e60': "S",
761
    '\u1e61': "s",
762
    '\u1e62': "S",
763
    '\u1e63': "s",
764
    '\u1e64': "S",
765
    '\u1e65': "s",
766
    '\u1e66': "S",
767
    '\u1e67': "s",
768
    '\u1e68': "S",
769
    '\u1e69': "s",
770
    '\u1e6a': "T",
771
    '\u1e6b': "t",
772
    '\u1e6c': "T",
773
    '\u1e6d': "t",
774
    '\u1e6e': "T",
775
    '\u1e6f': "t",
776
    '\u1e70': "T",
777
    '\u1e71': "t",
778
    '\u1e72': "U",
779
    '\u1e73': "u",
780
    '\u1e74': "U",
781
    '\u1e75': "u",
782
    '\u1e76': "U",
783
    '\u1e77': "u",
784
    '\u1e78': "U",
785
    '\u1e79': "u",
786
    '\u1e7a': "U",
787
    '\u1e7b': "u",
788
    '\u1e7c': "V",
789
    '\u1e7d': "v",
790
    '\u1e7e': "V",
791
    '\u1e7f': "v",
792
    '\u1e80': "W",
793
    '\u1e81': "w",
794
    '\u1e82': "W",
795
    '\u1e83': "w",
796
    '\u1e84': "W",
797
    '\u1e85': "w",
798
    '\u1e86': "W",
799
    '\u1e87': "w",
800
    '\u1e88': "W",
801
    '\u1e89': "w",
802
    '\u1e8a': "X",
803
    '\u1e8b': "x",
804
    '\u1e8c': "X",
805
    '\u1e8d': "x",
806
    '\u1e8e': "Y",
807
    '\u1e8f': "y",
808
    '\u1e90': "Z",
809
    '\u1e91': "z",
810
    '\u1e92': "Z",
811
    '\u1e93': "z",
812
    '\u1e94': "Z",
813
    '\u1e95': "z",
814
    '\u1e96': "h",
815
    '\u1e97': "t",
816
    '\u1e98': "w",
817
    '\u1e99': "y",
818
    '\u1e9a': "a",
819
    '\u1e9b': "s",
820
    '\u1ea0': "A",
821
    '\u1ea1': "a",
822
    '\u1ea2': "A",
823
    '\u1ea3': "a",
824
    '\u1ea4': "A",
825
    '\u1ea5': "a",
826
    '\u1ea6': "A",
827
    '\u1ea7': "a",
828
    '\u1ea8': "A",
829
    '\u1ea9': "a",
830
    '\u1eaa': "A",
831
    '\u1eab': "a",
832
    '\u1eac': "A",
833
    '\u1ead': "a",
834
    '\u1eae': "A",
835
    '\u1eaf': "a",
836
    '\u1eb0': "A",
837
    '\u1eb1': "a",
838
    '\u1eb2': "A",
839
    '\u1eb3': "a",
840
    '\u1eb4': "A",
841
    '\u1eb5': "a",
842
    '\u1eb6': "A",
843
    '\u1eb7': "a",
844
    '\u1eb8': "E",
845
    '\u1eb9': "e",
846
    '\u1eba': "E",
847
    '\u1ebb': "e",
848
    '\u1ebc': "E",
849
    '\u1ebd': "e",
850
    '\u1ebe': "E",
851
    '\u1ebf': "e",
852
    '\u1ec0': "E",
853
    '\u1ec1': "e",
854
    '\u1ec2': "E",
855
    '\u1ec3': "e",
856
    '\u1ec4': "E",
857
    '\u1ec5': "e",
858
    '\u1ec6': "E",
859
    '\u1ec7': "e",
860
    '\u1ec8': "I",
861
    '\u1ec9': "i",
862
    '\u1eca': "I",
863
    '\u1ecb': "i",
864
    '\u1ecc': "O",
865
    '\u1ecd': "o",
866
    '\u1ece': "O",
867
    '\u1ecf': "o",
868
    '\u1ed0': "O",
869
    '\u1ed1': "o",
870
    '\u1ed2': "O",
871
    '\u1ed3': "o",
872
    '\u1ed4': "O",
873
    '\u1ed5': "o",
874
    '\u1ed6': "O",
875
    '\u1ed7': "o",
876
    '\u1ed8': "O",
877
    '\u1ed9': "o",
878
    '\u1eda': "O",
879
    '\u1edb': "o",
880
    '\u1edc': "O",
881
    '\u1edd': "o",
882
    '\u1ede': "O",
883
    '\u1edf': "o",
884
    '\u1ee0': "O",
885
    '\u1ee1': "o",
886
    '\u1ee2': "O",
887
    '\u1ee3': "o",
888
    '\u1ee4': "U",
889
    '\u1ee5': "u",
890
    '\u1ee6': "U",
891
    '\u1ee7': "u",
892
    '\u1ee8': "U",
893
    '\u1ee9': "u",
894
    '\u1eea': "U",
895
    '\u1eeb': "u",
896
    '\u1eec': "U",
897
    '\u1eed': "u",
898
    '\u1eee': "U",
899
    '\u1eef': "u",
900
    '\u1ef0': "U",
901
    '\u1ef1': "u",
902
    '\u1ef2': "Y",
903
    '\u1ef3': "y",
904
    '\u1ef4': "Y",
905
    '\u1ef5': "y",
906
    '\u1ef6': "Y",
907
    '\u1ef7': "y",
908
    '\u1ef8': "Y",
909
    '\u1ef9': "y",
910
    '\u2002': " ",
911
    '\u2003': " ",
912
    '\u2009': " ",
913
    '\u2010': "-",
914
    '\u2011': "-",
915
    '\u2012': "-",
916
    '\u2013': "-",
917
    '\u2014': "-",
918
    '\u2015': "--",
919
    '\u2016': "||",
920
    '\u2017': "_",
921
    '\u2018': "'",
922
    '\u2019': "'",
923
    '\u201a': ",",
924
    '\u201b': "'",
925
    '\u2024': ".",
926
    '\u2025': "..",
927
    '\u2026': "...",
928
    '\u202f': " ",
929
    '\u2030': "0/00",
930
    '\u2031': "0/000",
931
    '\u2032': "'",
932
    '\u2034': "'''",
933
    '\u2035': "`",
934
    '\u2037': "'''",
935
    '\u2038': "^",
936
    '\u2039': "<",
937
    '\u203a': ">",
938
    '\u203c': "!!",
939
    '\u203d': "?",
940
    '\u2044': "/",
941
    '\u2045': "[",
942
    '\u2046': "]",
943
    '\u2047': "??",
944
    '\u2048': "?!",
945
    '\u2049': "!?",
946
    '\u204e': "*",
947
    '\u2052': "%",
948
    '\u2053': "~",
949
    '\u2056': "...",
950
    '\u2057': "''''",
951
    '\u2058': "....",
952
    '\u2059': ".....",
953
    '\u205a': "..",
954
    '\u205f': " ",
955
    '\u2070': "0", # superscript 0
956
    '\u2071': "i",
957
    '\u2072': "2",
958
    '\u2073': "3",
959
    '\u2074': "4",
960
    '\u2075': "5",
961
    '\u2076': "6",
962
    '\u2077': "7",
963
    '\u2078': "8",
964
    '\u2079': "9",
965
    '\u207A': "+",
966
    '\u207B': "-",
967
    '\u207C': "=",
968
    '\u207D': "(",
969
    '\u207E': ")",
970
    '\u207F': "n",
971
    '\u2080': "0", # subscript 0
972
    '\u2081': "1", # subscript 1
973
    '\u2082': "2", # subscript 2
974
    '\u2083': "3", # subscript 3
975
    '\u2084': "4", # subscript 4
976
    '\u2085': "5", # subscript 5
977
    '\u2086': "6", # subscript 6
978
    '\u2087': "7", # subscript 7
979
    '\u2088': "8", # subscript 8
980
    '\u2089': "9", # subscript 9
981
    '\u208A': "+",
982
    '\u208B': "-",
983
    '\u208C': "=",
984
    '\u208D': "(",
985
    '\u208E': ")",
986
    '\u2090': "a",
987
    '\u2091': "e",
988
    '\u2092': "o",
989
    '\u2093': "x",
990
    '\u2094': "e", # subscript schwa
991
    '\u2095': "h",
992
    '\u2096': "k",
993
    '\u2097': "l",
994
    '\u2098': "m",
995
    '\u2099': "n",
996
    '\u209A': "p",
997
    '\u209B': "s",
998
    '\u209C': "t",
999
    '\u2100': "a/c",
1000
    '\u2101': "a/s",
1001
    '\u2102': "C",
1002
    '\u2103': "C",
1003
    '\u2105': "c/o",
1004
    '\u2106': "c/u",
1005
    '\u2107': "E",
1006
    '\u2109': "F",
1007
    '\u210a': "g",
1008
    '\u210b': "H",
1009
    '\u210c': "H",
1010
    '\u210d': "H",
1011
    '\u210e': "h",
1012
    '\u210f': "h",
1013
    '\u2110': "I",
1014
    '\u2111': "I",
1015
    '\u2112': "L",
1016
    '\u2113': "l",
1017
    '\u2115': "N",
1018
    '\u2116': "No",
1019
    '\u2117': "(p)",
1020
    '\u2118': "P",
1021
    '\u2119': "P",
1022
    '\u211a': "Q",
1023
    '\u211b': "R",
1024
    '\u211c': "R",
1025
    '\u211d': "R",
1026
    '\u211e': "Px",
1027
    '\u2120': "SM",
1028
    '\u2121': "TEL",
1029
    '\u2122': "TM",
1030
    '\u2124': "Z",
1031
    '\u2126': "Omega",
1032
    '\u2128': "Z",
1033
    '\u2129': "iota",
1034
    '\u212a': "K",
1035
    '\u212b': "A",
1036
    '\u212c': "B",
1037
    '\u212d': "C",
1038
    '\u212e': "e",
1039
    '\u212f': "e",
1040
    '\u2130': "E",
1041
    '\u2131': "F",
1042
    '\u2132': "F",
1043
    '\u2133': "M",
1044
    '\u2134': "o",
1045
    '\u2139': "i",
1046
    '\u213a': "Q",
1047
    '\u213b': "FAX",
1048
    '\u213c': "pi",
1049
    '\u213d': "gamma",
1050
    '\u213e': "Gamma",
1051
    '\u213f': "PI",
1052
    '\u2141': "G",
1053
    '\u2142': "L",
1054
    '\u2144': "Y",
1055
    '\u2145': "D",
1056
    '\u2146': "d",
1057
    '\u2147': "e",
1058
    '\u2148': "i",
1059
    '\u2149': "j",
1060
    '\u214b': "&",
1061
    '\u2153': "1/3",
1062
    '\u2154': "2/3",
1063
    '\u2155': "1/5",
1064
    '\u2156': "2/5",
1065
    '\u2157': "3/5",
1066
    '\u2158': "4/5",
1067
    '\u2159': "1/6",
1068
    '\u215a': "5/6",
1069
    '\u215b': "1/8",
1070
    '\u215c': "3/8",
1071
    '\u215d': "5/8",
1072
    '\u215e': "7/8",
1073
    '\u215f': "1/",
1074
    '\u2160': "I",
1075
    '\u2161': "II",
1076
    '\u2162': "III",
1077
    '\u2163': "IV",
1078
    '\u2164': "V",
1079
    '\u2165': "VI",
1080
    '\u2166': "VII",
1081
    '\u2167': "VIII",
1082
    '\u2168': "IX",
1083
    '\u2169': "X",
1084
    '\u216a': "XI",
1085
    '\u216b': "XII",
1086
    '\u216c': "L",
1087
    '\u216d': "C",
1088
    '\u216e': "D",
1089
    '\u216f': "M",
1090
    '\u2170': "i",
1091
    '\u2171': "ii",
1092
    '\u2172': "iii",
1093
    '\u2173': "iv",
1094
    '\u2174': "v",
1095
    '\u2175': "vi",
1096
    '\u2176': "vii",
1097
    '\u2177': "viii",
1098
    '\u2178': "ix",
1099
    '\u2179': "x",
1100
    '\u217a': "xi",
1101
    '\u217b': "xii",
1102
    '\u217c': "l",
1103
    '\u217d': "c",
1104
    '\u217e': "d",
1105
    '\u217f': "m",
1106
    '\u2180': "CD",
1107
    '\u2190': "<-",
1108
    '\u2192': "->",
1109
    '\u2194': "<->",
1110
    '\u219a': "<-",
1111
    '\u219b': "->",
1112
    '\u21ae': "<->",
1113
    '\u21cd': "<=",
1114
    '\u21ce': "<=>",
1115
    '\u21cf': "=>",
1116
    '\u21d0': "<=",
1117
    '\u21d2': "=>",
1118
    '\u21d4': "<=>",
1119
    '\u2303': "^",
1120
    '\u2329': "<",
1121
    '\u232a': ">",
1122
    '\u239b': "(",
1123
    '\u239c': "(",
1124
    '\u239d': "(",
1125
    '\u239e': ")",
1126
    '\u239f': ")",
1127
    '\u23a0': ")",
1128
    '\u23a1': "[",
1129
    '\u23a2': "[",
1130
    '\u23a3': "[",
1131
    '\u23a4': "]",
1132
    '\u23a5': "]",
1133
    '\u23a6': "]",
1134
    '\u23a7': "{",
1135
    '\u23a8': "{",
1136
    '\u23a9': "{",
1137
    '\u23ab': "}",
1138
    '\u23ac': "}",
1139
    '\u23ad': "}",
1140
    '\u2460': "1",
1141
    '\u2461': "2",
1142
    '\u2462': "3",
1143
    '\u2463': "4",
1144
    '\u2464': "5",
1145
    '\u2465': "6",
1146
    '\u2466': "7",
1147
    '\u2467': "8",
1148
    '\u2468': "9",
1149
    '\u2469': "10",
1150
    '\u246a': "11",
1151
    '\u246b': "12",
1152
    '\u246c': "13",
1153
    '\u246d': "14",
1154
    '\u246e': "15",
1155
    '\u246f': "16",
1156
    '\u2470': "17",
1157
    '\u2471': "18",
1158
    '\u2472': "19",
1159
    '\u2473': "20",
1160
    '\u2474': "(1)",
1161
    '\u2475': "(2)",
1162
    '\u2476': "(3)",
1163
    '\u2477': "(4)",
1164
    '\u2478': "(5)",
1165
    '\u2479': "(6)",
1166
    '\u247a': "(7)",
1167
    '\u247b': "(8)",
1168
    '\u247c': "(9)",
1169
    '\u247d': "(10)",
1170
    '\u247e': "(11)",
1171
    '\u247f': "(12)",
1172
    '\u2480': "(13)",
1173
    '\u2481': "(14)",
1174
    '\u2482': "(15)",
1175
    '\u2483': "(16)",
1176
    '\u2484': "(17)",
1177
    '\u2485': "(18)",
1178
    '\u2486': "(19)",
1179
    '\u2487': "(20)",
1180
    '\u2488': "1.",
1181
    '\u2489': "2.",
1182
    '\u248a': "3.",
1183
    '\u248b': "4.",
1184
    '\u248c': "5.",
1185
    '\u248d': "6.",
1186
    '\u248e': "7.",
1187
    '\u248f': "8.",
1188
    '\u2490': "9.",
1189
    '\u2491': "10.",
1190
    '\u2492': "11.",
1191
    '\u2493': "12.",
1192
    '\u2494': "13.",
1193
    '\u2495': "14.",
1194
    '\u2496': "15.",
1195
    '\u2497': "16.",
1196
    '\u2498': "17.",
1197
    '\u2499': "18.",
1198
    '\u249a': "19.",
1199
    '\u249b': "20.",
1200
    '\u249c': "(a)",
1201
    '\u249d': "(b)",
1202
    '\u249e': "(c)",
1203
    '\u249f': "(d)",
1204
    '\u24a0': "(e)",
1205
    '\u24a1': "(f)",
1206
    '\u24a2': "(g)",
1207
    '\u24a3': "(h)",
1208
    '\u24a4': "(i)",
1209
    '\u24a5': "(j)",
1210
    '\u24a6': "(k)",
1211
    '\u24a7': "(l)",
1212
    '\u24a8': "(m)",
1213
    '\u24a9': "(n)",
1214
    '\u24aa': "(o)",
1215
    '\u24ab': "(p)",
1216
    '\u24ac': "(q)",
1217
    '\u24ad': "(r)",
1218
    '\u24ae': "(s)",
1219
    '\u24af': "(t)",
1220
    '\u24b0': "(u)",
1221
    '\u24b1': "(v)",
1222
    '\u24b2': "(w)",
1223
    '\u24b3': "(x)",
1224
    '\u24b4': "(y)",
1225
    '\u24b5': "(z)",
1226
    '\u24b6': "A",
1227
    '\u24b7': "B",
1228
    '\u24b8': "C",
1229
    '\u24b9': "D",
1230
    '\u24ba': "E",
1231
    '\u24bb': "F",
1232
    '\u24bc': "G",
1233
    '\u24bd': "H",
1234
    '\u24be': "I",
1235
    '\u24bf': "J",
1236
    '\u24c0': "K",
1237
    '\u24c1': "L",
1238
    '\u24c2': "M",
1239
    '\u24c3': "N",
1240
    '\u24c4': "O",
1241
    '\u24c5': "P",
1242
    '\u24c6': "Q",
1243
    '\u24c7': "R",
1244
    '\u24c8': "S",
1245
    '\u24c9': "T",
1246
    '\u24ca': "U",
1247
    '\u24cb': "V",
1248
    '\u24cc': "W",
1249
    '\u24cd': "X",
1250
    '\u24ce': "Y",
1251
    '\u24cf': "Z",
1252
    '\u24d0': "a",
1253
    '\u24d1': "b",
1254
    '\u24d2': "c",
1255
    '\u24d3': "d",
1256
    '\u24d4': "e",
1257
    '\u24d5': "f",
1258
    '\u24d6': "g",
1259
    '\u24d7': "h",
1260
    '\u24d8': "i",
1261
    '\u24d9': "j",
1262
    '\u24da': "k",
1263
    '\u24db': "l",
1264
    '\u24dc': "m",
1265
    '\u24dd': "n",
1266
    '\u24de': "o",
1267
    '\u24df': "p",
1268
    '\u24e0': "q",
1269
    '\u24e1': "r",
1270
    '\u24e2': "s",
1271
    '\u24e3': "t",
1272
    '\u24e4': "u",
1273
    '\u24e5': "v",
1274
    '\u24e6': "w",
1275
    '\u24e7': "x",
1276
    '\u24e8': "y",
1277
    '\u24e9': "z",
1278
    '\u24ea': "0",
1279
    '\u24eb': "(11)",
1280
    '\u24ec': "(12)",
1281
    '\u24ed': "(13)",
1282
    '\u24ee': "(14)",
1283
    '\u24ef': "(15)",
1284
    '\u24f0': "(16)",
1285
    '\u24f1': "(17)",
1286
    '\u24f2': "(18)",
1287
    '\u24f3': "(19)",
1288
    '\u24f4': "(20)",
1289
    '\u24f5': "(1)",
1290
    '\u24f6': "(2)",
1291
    '\u24f7': "(3)",
1292
    '\u24f8': "(4)",
1293
    '\u24f9': "(5)",
1294
    '\u24fa': "(6)",
1295
    '\u24fb': "(7)",
1296
    '\u24fc': "(8)",
1297
    '\u24fd': "(9)",
1298
    '\u24fe': "(10)",
1299
    '\u24ff': "(0)",
1300
    '\u3000': " ",
1301
    '\u3001': ",",
1302
    '\u3007': "0",
1303
    '\u3008': "<",
1304
    '\u3009': ">",
1305
    '\u300a': "<<",
1306
    '\u300b': ">>",
1307
    '\u301b': "]",
1308
    '\u301c': "~",
1309
    '\u3021': "1",
1310
    '\u3022': "2",
1311
    '\u3023': "3",
1312
    '\u3024': "4",
1313
    '\u3025': "5",
1314
    '\u3026': "6",
1315
    '\u3027': "7",
1316
    '\u3028': "8",
1317
    '\u3029': "9",
1318
    '\ufb00': "ff",
1319
    '\ufb01': "fi",
1320
    '\ufb02': "fl",
1321
    '\ufb03': "ffi",
1322
    '\ufb04': "ffl",
1323
    '\ufb05': "st",
1324
    '\ufb06': "st",
1325
    '\ufb29': "+",
1326
    '\ufe50': ",",
1327
    '\ufe51': ",",
1328
    '\ufe52': ".",
1329
    '\ufe54': ";",
1330
    '\ufe55': ":",
1331
    '\ufe56': "?",
1332
    '\ufe57': "!",
1333
    '\ufe58': "-",
1334
    '\ufe59': "(",
1335
    '\ufe5a': ")",
1336
    '\ufe5b': "{",
1337
    '\ufe5c': "}",
1338
    '\ufe5d': "(",
1339
    '\ufe5e': ")",
1340
    '\ufe5f': "#",
1341
    '\ufe60': "&",
1342
    '\ufe61': "*",
1343
    '\ufe62': "+",
1344
    '\ufe63': "-",
1345
    '\ufe64': "<",
1346
    '\ufe65': ">",
1347
    '\ufe66': "=",
1348
    '\ufe69': "$",
1349
    '\ufe6a': "%",
1350
    '\ufe6b': "@",
1351
    '\uff01': "!",
1352
    '\uff03': "#",
1353
    '\uff04': "$",
1354
    '\uff05': "%",
1355
    '\uff06': "&",
1356
    '\uff07': "'",
1357
    '\uff08': "(",
1358
    '\uff09': ")",
1359
    '\uff0a': "*",
1360
    '\uff0b': "+",
1361
    '\uff0c': ",",
1362
    '\uff0d': "-",
1363
    '\uff0e': ".",
1364
    '\uff0f': "/",
1365
    '\uff10': "0",
1366
    '\uff11': "1",
1367
    '\uff12': "2",
1368
    '\uff13': "3",
1369
    '\uff14': "4",
1370
    '\uff15': "5",
1371
    '\uff16': "6",
1372
    '\uff17': "7",
1373
    '\uff18': "8",
1374
    '\uff19': "9",
1375
    '\uff1a': ":",
1376
    '\uff1b': ";",
1377
    '\uff1c': "<",
1378
    '\uff1d': "=",
1379
    '\uff1e': ">",
1380
    '\uff1f': "?",
1381
    '\uff20': "@",
1382
    '\uff21': "A",
1383
    '\uff22': "B",
1384
    '\uff23': "C",
1385
    '\uff24': "D",
1386
    '\uff25': "E",
1387
    '\uff26': "F",
1388
    '\uff27': "G",
1389
    '\uff28': "H",
1390
    '\uff29': "I",
1391
    '\uff2a': "J",
1392
    '\uff2b': "K",
1393
    '\uff2c': "L",
1394
    '\uff2d': "M",
1395
    '\uff2e': "N",
1396
    '\uff2f': "O",
1397
    '\uff30': "P",
1398
    '\uff31': "Q",
1399
    '\uff32': "R",
1400
    '\uff33': "S",
1401
    '\uff34': "T",
1402
    '\uff35': "U",
1403
    '\uff36': "V",
1404
    '\uff37': "W",
1405
    '\uff38': "X",
1406
    '\uff39': "Y",
1407
    '\uff3a': "Z",
1408
    '\uff3b': "[",
1409
    '\uff3d': "]",
1410
    '\uff3e': "^",
1411
    '\uff3f': "_",
1412
    '\uff40': "`",
1413
    '\uff41': "a",
1414
    '\uff42': "b",
1415
    '\uff43': "c",
1416
    '\uff44': "d",
1417
    '\uff45': "e",
1418
    '\uff46': "f",
1419
    '\uff47': "g",
1420
    '\uff48': "h",
1421
    '\uff49': "i",
1422
    '\uff4a': "j",
1423
    '\uff4b': "k",
1424
    '\uff4c': "l",
1425
    '\uff4d': "m",
1426
    '\uff4e': "n",
1427
    '\uff4f': "o",
1428
    '\uff50': "p",
1429
    '\uff51': "q",
1430
    '\uff52': "r",
1431
    '\uff53': "s",
1432
    '\uff54': "t",
1433
    '\uff55': "u",
1434
    '\uff56': "v",
1435
    '\uff57': "w",
1436
    '\uff58': "x",
1437
    '\uff59': "y",
1438
    '\uff5a': "z",
1439
    '\uff5b': "{",
1440
    '\uff5c': "|",
1441
    '\uff5d': "}",
1442
    '\uff5e': "~",
1443
    '\uff5f': "(",
1444
    '\uff60': ")",
1445
    '\uff64': ",",
1446
    '\uffe9': "<-",
1447
    '\uffeb': "->",
1448
    '\u2000': " ",
1449
    '\u2001': " ",
1450
    '\u2004': " ",
1451
    '\u2005': " ",
1452
    '\u2006': " ",
1453
    '\u2007': " ",
1454
    '\u2008': " ",
1455
    '\u200A': " ",
1456
    '\u200B': " ",
1457
    '\u2215': "/",
1458
    '\u207b': "-",
1459
    '\u208b': "-",
1460
    '\u2212': "-",
1461
    '\u2264': "<=",
1462
    '\u2265': ">=",
1463
    '\uc2a0': " ",
1464
    '\ufeff': " ",
1465
    '\u200E': "",
1466
    '\u2020': "",
1467
    '\u2217': "*",
1468
    '\u0092': "",
1469
    '\u00f0': "d", # Decimal=240 (Latin Small Letter Eth)
1470
    '\u0138': "K", # Decimal=312 (Latin Small Letter KRA)
1471
    '\u0150': "O", # Decimal=330 (Latin Capital Letter O With Double Acute)
1472
    '\u018d': "d", # Decimal=397 (Latin Small Letter Turned Delta)
1473
    '\u018f': "e", # Decimal=399 (Latin Capital Letter Schwa)
1474
    '\u0189': "D", # Decimal=425 (Latin Capital Letter African D)
1475
    '\u0190': "E", # Decimal=426 (Latin Capital Letter Open E)
1476
    '\u01b7': "3", # Decimal=439 (Latin Capital Letter Ezh)
1477
    '\u01b8': "E", # Decimal=440 (Latin Capital Letter Ezh Reversed)
1478
    '\u01b9': "e", # Decimal=441 (Latin Small Letter Ezh Reversed)
1479
    '\u01ba': "e", # Decimal=442 (Latin Small Letter Ezh With Tail)
1480
    '\u01be': "t", # Decimal=446 (Latin Letter Inverted Glottal Stop With Stroke)
1481
    '\u01bf': "p", # Decimal=447 (Latin Letter Wynn)
1482
    '\u01c1': "||", # Decimal=449 (Latin Letter Lateral Click)
1483
    '\u01ee': "3", # Decimal=494 (Latin Capital Letter Ezh With Caron)
1484
    '\u01ef': "3", # Decimal=495 (Latin Small Letter Ezh With Caron)
1485
    '\u0323': "",
1486
    '\u0410': "",
1487
    '\u0411': "",
1488
    '\u0412': "",
1489
    '\u0413': "",
1490
    '\u0414': "",
1491
    '\u0415': "",
1492
    '\u0416': "",
1493
    '\u0417': "",
1494
    '\u0418': "",
1495
    '\u0419': "",
1496
    '\u041a': "",
1497
    '\u041b': "",
1498
    '\u041c': "",
1499
    '\u041d': "",
1500
    '\u041e': "",
1501
    '\u041f': "",
1502
    '\u0420': "",
1503
    '\u0421': "",
1504
    '\u0422': "",
1505
    '\u0423': "",
1506
    '\u0424': "",
1507
    '\u0425': "",
1508
    '\u0426': "",
1509
    '\u0427': "",
1510
    '\u0428': "",
1511
    '\u0429': "",
1512
    '\u042a': "",
1513
    '\u042b': "",
1514
    '\u042c': "",
1515
    '\u042d': "",
1516
    '\u042e': "",
1517
    '\u042f': "",
1518
    '\u0430': "",
1519
    '\u0431': "",
1520
    '\u0432': "",
1521
    '\u0433': "",
1522
    '\u0434': "",
1523
    '\u0435': "",
1524
    '\u0436': "",
1525
    '\u0437': "",
1526
    '\u0438': "",
1527
    '\u0439': "",
1528
    '\u043a': "",
1529
    '\u043b': "",
1530
    '\u043c': "",
1531
    '\u043d': "",
1532
    '\u043e': "",
1533
    '\u043f': "",
1534
    '\u0440': "",
1535
    '\u0441': "",
1536
    '\u0442': "",
1537
    '\u0443': "",
1538
    '\u0444': "",
1539
    '\u0445': "",
1540
    '\u0446': "",
1541
    '\u0447': "",
1542
    '\u0448': "",
1543
    '\u0449': "",
1544
    '\u044a': "",
1545
    '\u044b': "",
1546
    '\u044c': "",
1547
    '\u044d': "",
1548
    '\u044e': "",
1549
    '\u044f': "",
1550
    '\u0451': "",
1551
    '\u200b': "",
1552
    '\u200e': "",
1553
    '\u2022': "",
1554
    '\u2261': "",
1555
    '\u8166': "",
1556
    '\u982d': "",
1557
    '\ub1cc': "",
1558
    '\ub450': "",
1559
    '\uf030': "",
1560
}