Subversion-Projekte lars-tiefland.laravel_shop

Revision

Details | Letzte Änderung | Log anzeigen | RSS feed

Revision Autor Zeilennr. Zeile
148 lars 1
<?php
2
 
3
// https://en.wikipedia.org/wiki/List_of_ISO_639-1_codes
4
 
5
return [
6
    // Dutch (Flemish)
7
    'nl' => [
8
        'Á' => 'A',
9
        'á' => 'a',
10
        'Ä' => 'A',
11
        'ä' => 'a',
12
        'À' => 'A',
13
        'à' => 'a',
14
        'Â' => 'A',
15
        'â' => 'a',
16
        'É' => 'E',
17
        'é' => 'e',
18
        'Ë' => 'E',
19
        'ë' => 'e',
20
        'È' => 'E',
21
        'è' => 'e',
22
        'Ê' => 'E',
23
        'ê' => 'e',
24
        'Í' => 'I',
25
        'í' => 'i',
26
        'Ï' => 'I',
27
        'ï' => 'i',
28
        'Ì' => 'I',
29
        'ì' => 'i',
30
        'Î' => 'I',
31
        'î' => 'i',
32
        'Ó' => 'O',
33
        'ó' => 'o',
34
        'Ö' => 'O',
35
        'ö' => 'o',
36
        'Ò' => 'O',
37
        'ò' => 'o',
38
        'Ô' => 'O',
39
        'ô' => 'o',
40
        'Ú' => 'U',
41
        'ú' => 'u',
42
        'Ü' => 'U',
43
        'ü' => 'u',
44
        'Ù' => 'U',
45
        'ù' => 'u',
46
        'Û' => 'U',
47
        'û' => 'u',
48
        'Ý' => 'Y',
49
        'ý' => 'y',
50
        'Ÿ' => 'Y',
51
    ],
52
    // Italian
53
    'it' => [
54
        'à' => 'a',
55
        'À' => 'A',
56
        'é' => 'e',
57
        'É' => 'E',
58
        'è' => 'e',
59
        'È' => 'E',
60
        'ì' => 'i',
61
        'Ì' => 'I',
62
        'Ò' => 'O',
63
        'ò' => 'o',
64
        'ù' => 'u',
65
        'Ù' => 'U',
66
    ],
67
    // Macedonian
68
    'mk' => [
69
        'А' => 'A',
70
        'Б' => 'B',
71
        'В' => 'V',
72
        'Г' => 'G',
73
        'Д' => 'D',
74
        'Ѓ' => 'Gj',
75
        'Е' => 'E',
76
        'Ж' => 'Zh',
77
        'З' => 'Z',
78
        'Ѕ' => 'Dz',
79
        'И' => 'I',
80
        'Ј' => 'J',
81
        'К' => 'K',
82
        'Л' => 'L',
83
        'Љ' => 'Lj',
84
        'М' => 'M',
85
        'Н' => 'N',
86
        'Њ' => 'Nj',
87
        'О' => 'O',
88
        'П' => 'P',
89
        'Р' => 'R',
90
        'С' => 'S',
91
        'Т' => 'T',
92
        'Ќ' => 'Kj',
93
        'У' => 'U',
94
        'Ф' => 'F',
95
        'Х' => 'H',
96
        'Ц' => 'C',
97
        'Ч' => 'Ch',
98
        'Џ' => 'Dj',
99
        'Ш' => 'Sh',
100
        'а' => 'a',
101
        'б' => 'b',
102
        'в' => 'v',
103
        'г' => 'g',
104
        'д' => 'd',
105
        'ѓ' => 'gj',
106
        'е' => 'e',
107
        'ж' => 'zh',
108
        'з' => 'z',
109
        'ѕ' => 'dz',
110
        'и' => 'i',
111
        'ј' => 'j',
112
        'к' => 'k',
113
        'л' => 'l',
114
        'љ' => 'lj',
115
        'м' => 'm',
116
        'н' => 'n',
117
        'њ' => 'nj',
118
        'о' => 'o',
119
        'п' => 'p',
120
        'р' => 'r',
121
        'с' => 's',
122
        'т' => 't',
123
        'ќ' => 'kj',
124
        'у' => 'u',
125
        'ф' => 'f',
126
        'х' => 'h',
127
        'ц' => 'c',
128
        'ч' => 'ch',
129
        'џ' => 'dj',
130
        'ш' => 'sh',
131
    ],
132
    // Portuguese (Brazil)
133
    'pt' => [
134
        'æ' => 'ae',
135
        'ǽ' => 'ae',
136
        'À' => 'A',
137
        'Á' => 'A',
138
        'Â' => 'A',
139
        'Ã' => 'A',
140
        'Å' => 'AA',
141
        'Ǻ' => 'A',
142
        'Ä‚' => 'A',
143
        'Ǎ' => 'A',
144
        'Æ' => 'AE',
145
        'Ǽ' => 'AE',
146
        'à' => 'a',
147
        'á' => 'a',
148
        'â' => 'a',
149
        'ã' => 'a',
150
        'å' => 'aa',
151
        'Ç»' => 'a',
152
        'ă' => 'a',
153
        'ÇŽ' => 'a',
154
        'ª' => 'a',
155
        'Ĉ' => 'C',
156
        'ÄŠ' => 'C',
157
        'Ç' => 'C',
158
        'ç' => 'c',
159
        'ĉ' => 'c',
160
        'ċ' => 'c',
161
        'Ð' => 'Dj',
162
        'Đ' => 'D',
163
        'ð' => 'dj',
164
        'Ä‘' => 'd',
165
        'È' => 'E',
166
        'É' => 'E',
167
        'Ê' => 'E',
168
        'Ë' => 'E',
169
        'Ä”' => 'E',
170
        'Ä–' => 'E',
171
        'è' => 'e',
172
        'é' => 'e',
173
        'ê' => 'e',
174
        'ë' => 'e',
175
        'Ä•' => 'e',
176
        'Ä—' => 'e',
177
        'Æ’' => 'f',
178
        'Ĝ' => 'G',
179
        'Ġ' => 'G',
180
        'ĝ' => 'g',
181
        'Ä¡' => 'g',
182
        'Ĥ' => 'H',
183
        'Ħ' => 'H',
184
        'Ä¥' => 'h',
185
        'ħ' => 'h',
186
        'Ì' => 'I',
187
        'Í' => 'I',
188
        'Î' => 'I',
189
        'Ï' => 'I',
190
        'Ĩ' => 'I',
191
        'Ĭ' => 'I',
192
        'Ǐ' => 'I',
193
        'Ä®' => 'I',
194
        'IJ' => 'IJ',
195
        'ì' => 'i',
196
        'í' => 'i',
197
        'î' => 'i',
198
        'ï' => 'i',
199
        'Ä©' => 'i',
200
        'ĭ' => 'i',
201
        'ǐ' => 'i',
202
        'į' => 'i',
203
        'ij' => 'ij',
204
        'Ä´' => 'J',
205
        'ĵ' => 'j',
206
        'Ĺ' => 'L',
207
        'Ľ' => 'L',
208
        'Ä¿' => 'L',
209
        'ĺ' => 'l',
210
        'ľ' => 'l',
211
        'Å€' => 'l',
212
        'Ñ' => 'N',
213
        'ñ' => 'n',
214
        'ʼn' => 'n',
215
        'Ò' => 'O',
216
        'Ó' => 'O',
217
        'Ô' => 'O',
218
        'Õ' => 'O',
219
        'Ō' => 'O',
220
        'ÅŽ' => 'O',
221
        'Ç‘' => 'O',
222
        'Ő' => 'O',
223
        'Ơ' => 'O',
224
        'Ø' => 'OE',
225
        'Ǿ' => 'O',
226
        'Å’' => 'OE',
227
        'ò' => 'o',
228
        'ó' => 'o',
229
        'ô' => 'o',
230
        'õ' => 'o',
231
        'ō' => 'o',
232
        'ŏ' => 'o',
233
        'Ç’' => 'o',
234
        'ő' => 'o',
235
        'Æ¡' => 'o',
236
        'ø' => 'oe',
237
        'Ç¿' => 'o',
238
        'º' => 'o',
239
        'Å“' => 'oe',
240
        'Å”' => 'R',
241
        'Å–' => 'R',
242
        'Å•' => 'r',
243
        'Å—' => 'r',
244
        'Ŝ' => 'S',
245
        'Ș' => 'S',
246
        'ŝ' => 's',
247
        'ș' => 's',
248
        'Å¿' => 's',
249
        'Å¢' => 'T',
250
        'Èš' => 'T',
251
        'Ŧ' => 'T',
252
        'Þ' => 'TH',
253
        'Å£' => 't',
254
        'ț' => 't',
255
        'ŧ' => 't',
256
        'þ' => 'th',
257
        'Ù' => 'U',
258
        'Ú' => 'U',
259
        'Û' => 'U',
260
        'Ü' => 'U',
261
        'Ũ' => 'U',
262
        'Ŭ' => 'U',
263
        'Ű' => 'U',
264
        'Ų' => 'U',
265
        'Ư' => 'U',
266
        'Ç“' => 'U',
267
        'Ç•' => 'U',
268
        'Ç—' => 'U',
269
        'Ǚ' => 'U',
270
        'Ǜ' => 'U',
271
        'ù' => 'u',
272
        'ú' => 'u',
273
        'û' => 'u',
274
        'ü' => 'u',
275
        'Å©' => 'u',
276
        'ŭ' => 'u',
277
        'ű' => 'u',
278
        'ų' => 'u',
279
        'ư' => 'u',
280
        'Ç”' => 'u',
281
        'Ç–' => 'u',
282
        'ǘ' => 'u',
283
        'Çš' => 'u',
284
        'ǜ' => 'u',
285
        'Å´' => 'W',
286
        'ŵ' => 'w',
287
        'Ý' => 'Y',
288
        'Ÿ' => 'Y',
289
        'Ŷ' => 'Y',
290
        'ý' => 'y',
291
        'ÿ' => 'y',
292
        'Å·' => 'y',
293
    ],
294
    // Greek(lish) (Elláda)
295
    'el__greeklish' => [
296
        'ΑΥ' => 'AU',
297
        'ΑΎ' => 'AU',
298
        'Αυ' => 'Au',
299
        'Αύ' => 'Au',
300
        'ΕΊ' => 'EI',
301
        'ΕΙ' => 'EI',
302
        'Ει' => 'EI',
303
        'ΕΥ' => 'EU',
304
        'ΕΎ' => 'EU',
305
        'Εί' => 'Ei',
306
        'Ευ' => 'Eu',
307
        'Εύ' => 'Eu',
308
        'ΟΙ' => 'OI',
309
        'ΟΊ' => 'OI',
310
        'ΟΥ' => 'OU',
311
        'ΟΎ' => 'OU',
312
        'Οι' => 'Oi',
313
        'Οί' => 'Oi',
314
        'Ου' => 'Ou',
315
        'Ού' => 'Ou',
316
        'ΥΙ' => 'YI',
317
        'ΎΙ' => 'YI',
318
        'Υι' => 'Yi',
319
        'Ύι' => 'Yi',
320
        'ΥΊ' => 'Yi',
321
        'Υί' => 'Yi',
322
        'αυ' => 'au',
323
        'αύ' => 'au',
324
        'εί' => 'ei',
325
        'ει' => 'ei',
326
        'ευ' => 'eu',
327
        'εύ' => 'eu',
328
        'οι' => 'oi',
329
        'οί' => 'oi',
330
        'ου' => 'ou',
331
        'ού' => 'ou',
332
        'υι' => 'yi',
333
        'ύι' => 'yi',
334
        'υί' => 'yi',
335
        'Α'  => 'A',
336
        'Ά'  => 'A',
337
        'Β'  => 'B',
338
        'Δ'  => 'D',
339
        'Ε'  => 'E',
340
        'Έ'  => 'E',
341
        'Φ'  => 'F',
342
        'Γ'  => 'G',
343
        'Η'  => 'H',
344
        'Ή'  => 'H',
345
        'Ι'  => 'I',
346
        'Ί'  => 'I',
347
        'Ϊ'  => 'I',
348
        'Κ'  => 'K',
349
        'Ξ'  => 'Ks',
350
        'Λ'  => 'L',
351
        'Μ'  => 'M',
352
        'Ν'  => 'N',
353
        'Π'  => 'N',
354
        'Ο'  => 'O',
355
        'Ό'  => 'O',
356
        'Ψ'  => 'Ps',
357
        'Ρ'  => 'R',
358
        'Σ'  => 'S',
359
        'Τ'  => 'T',
360
        'Θ'  => 'Th',
361
        'Ω'  => 'W',
362
        'Ώ'  => 'W',
363
        'Χ'  => 'X',
364
        'ϒ'  => 'Y',
365
        'Υ'  => 'Y',
366
        'Ύ'  => 'Y',
367
        'Ϋ'  => 'Y',
368
        'Ζ'  => 'Z',
369
        'α'  => 'a',
370
        'ά'  => 'a',
371
        'β'  => 'b',
372
        'δ'  => 'd',
373
        'ε'  => 'e',
374
        'έ'  => 'e',
375
        'φ'  => 'f',
376
        'γ'  => 'g',
377
        'η'  => 'h',
378
        'ή'  => 'h',
379
        'ι'  => 'i',
380
        'ί'  => 'i',
381
        'ϊ'  => 'i',
382
        'ΐ'  => 'i',
383
        'κ'  => 'k',
384
        'ξ'  => 'ks',
385
        'λ'  => 'l',
386
        'μ'  => 'm',
387
        'ν'  => 'n',
388
        'ο'  => 'o',
389
        'ό'  => 'o',
390
        'π'  => 'p',
391
        'ψ'  => 'ps',
392
        'ρ'  => 'r',
393
        'σ'  => 's',
394
        'ς'  => 's',
395
        'τ'  => 't',
396
        'ϑ'  => 'th',
397
        'θ'  => 'th',
398
        'ϐ'  => 'v',
399
        'ω'  => 'w',
400
        'ώ'  => 'w',
401
        'χ'  => 'x',
402
        'υ'  => 'y',
403
        'ύ'  => 'y',
404
        'ΰ'  => 'y',
405
        'ϋ'  => 'y',
406
        'ζ'  => 'z',
407
    ],
408
    // Greek (Elláda)
409
    'el' => [
410
        'ΑΥ' => 'AU',
411
        'Αυ' => 'Au',
412
        'ΟΥ' => 'U',
413
        'Ου' => 'u',
414
        'ΕΥ' => 'EF',
415
        'Ευ' => 'Ef',
416
        'ΕΙ' => 'I',
417
        'Ει' => 'I',
418
        'ΟΙ' => 'I',
419
        'Οι' => 'I',
420
        'ΥΙ' => 'I',
421
        'Υι' => 'I',
422
        'ΑΎ' => 'AU',
423
        'Αύ' => 'Au',
424
        'ΟΎ' => 'OU',
425
        'Ού' => 'Ou',
426
        'ΕΎ' => 'EU',
427
        'Εύ' => 'Eu',
428
        'ΕΊ' => 'I',
429
        'Εί' => 'I',
430
        'ΟΊ' => 'I',
431
        'Οί' => 'I',
432
        'ΎΙ' => 'I',
433
        'Ύι' => 'I',
434
        'ΥΊ' => 'I',
435
        'Υί' => 'I',
436
        'αυ' => 'au',
437
        'ου' => 'u',
438
        'ευ' => 'ef',
439
        'ει' => 'i',
440
        'οι' => 'i',
441
        'υι' => 'i',
442
        'αύ' => 'au',
443
        'ού' => 'ou',
444
        'εύ' => 'eu',
445
        'εί' => 'i',
446
        'οί' => 'i',
447
        'ύι' => 'i',
448
        'υί' => 'i',
449
        'α'  => 'a',
450
        'β'  => 'v',
451
        'γ'  => 'gh',
452
        'δ'  => 'd',
453
        'ε'  => 'e',
454
        'ζ'  => 'z',
455
        'η'  => 'i',
456
        'θ'  => 'th',
457
        'ι'  => 'i',
458
        'κ'  => 'k',
459
        'λ'  => 'l',
460
        'μ'  => 'm',
461
        'ν'  => 'n',
462
        'ξ'  => 'ks',
463
        'ο'  => 'o',
464
        'π'  => 'p',
465
        'ρ'  => 'r',
466
        'σ'  => 's',
467
        'τ'  => 't',
468
        'υ'  => 'i',
469
        'φ'  => 'f',
470
        'χ'  => 'kh',
471
        'ψ'  => 'ps',
472
        'ω'  => 'o',
473
        'ά'  => 'a',
474
        'έ'  => 'e',
475
        'ί'  => 'i',
476
        'ό'  => 'o',
477
        'ϒ'  => 'Y',
478
        'ύ'  => 'y',
479
        'ή'  => 'i',
480
        'ώ'  => 'w',
481
        'ς'  => 's',
482
        'ϊ'  => 'i',
483
        'ΰ'  => 'y',
484
        'ϋ'  => 'y',
485
        'ΐ'  => 'i',
486
        'Α'  => 'A',
487
        'Β'  => 'B',
488
        'Γ'  => 'G',
489
        'Δ'  => 'D',
490
        'Ε'  => 'E',
491
        'Ζ'  => 'Z',
492
        'Η'  => 'H',
493
        'Θ'  => 'Th',
494
        'Ι'  => 'I',
495
        'Κ'  => 'K',
496
        'Λ'  => 'L',
497
        'Μ'  => 'M',
498
        'Ν'  => 'N',
499
        'Ξ'  => 'Ks',
500
        'Ο'  => 'O',
501
        'Π'  => 'P',
502
        'Ρ'  => 'R',
503
        'Σ'  => 'S',
504
        'Τ'  => 'T',
505
        'Υ'  => 'Y',
506
        'Φ'  => 'F',
507
        'Χ'  => 'X',
508
        'Ψ'  => 'Ps',
509
        'Ω'  => 'O',
510
        'Ά'  => 'A',
511
        'Έ'  => 'E',
512
        'Ί'  => 'I',
513
        'Ό'  => 'O',
514
        'Ύ'  => 'Y',
515
        'Ή'  => 'I',
516
        'Ώ'  => 'W',
517
        'Ϊ'  => 'I',
518
        'Ϋ'  => 'Y',
519
        'ϐ'  => 'v',
520
        'ϑ'  => 'th',
521
    ],
522
    // Hindi
523
    'hi' => [
524
        'अ'  => 'a',
525
        'आ'  => 'aa',
526
        'ए'  => 'e',
527
        'ई'  => 'ii',
528
        'ऍ'  => 'ei',
529
        'ऎ'  => 'ae',
530
        'ऐ'  => 'ai',
531
        'इ'  => 'i',
532
        'ओ'  => 'o',
533
        'ऑ'  => 'oi',
534
        'ऒ'  => 'oii',
535
        'ऊ'  => 'uu',
536
        'औ'  => 'ou',
537
        'उ'  => 'u',
538
        'ब'  => 'B',
539
        'भ'  => 'Bha',
540
        'च'  => 'Ca',
541
        'छ'  => 'Chha',
542
        'ड'  => 'Da',
543
        'ढ'  => 'Dha',
544
        'फ'  => 'Fa',
545
        'फ़' => 'Fi',
546
        'ग'  => 'Ga',
547
        'घ'  => 'Gha',
548
        'ग़' => 'Ghi',
549
        'ह'  => 'Ha',
550
        'ज'  => 'Ja',
551
        'झ'  => 'Jha',
552
        'क'  => 'Ka',
553
        'ख'  => 'Kha',
554
        'ख़' => 'Khi',
555
        'ल'  => 'L',
556
        'ळ'  => 'Li',
557
        'ऌ'  => 'Li',
558
        'ऴ'  => 'Lii',
559
        'ॡ'  => 'Lii',
560
        'म'  => 'Ma',
561
        'न'  => 'Na',
562
        'ङ'  => 'Na',
563
        'ञ'  => 'Nia',
564
        'ण'  => 'Nae',
565
        'ऩ'  => 'Ni',
566
        'ॐ'  => 'oms',
567
        'प'  => 'Pa',
568
        'क़' => 'Qi',
569
        'र'  => 'Ra',
570
        'ऋ'  => 'Ri',
571
        'ॠ'  => 'Ri',
572
        'ऱ'  => 'Ri',
573
        'स'  => 'Sa',
574
        'श'  => 'Sha',
575
        'ष'  => 'Shha',
576
        'ट'  => 'Ta',
577
        'त'  => 'Ta',
578
        'ठ'  => 'Tha',
579
        'द'  => 'Tha',
580
        'थ'  => 'Tha',
581
        'ध'  => 'Thha',
582
        'ड़' => 'ugDha',
583
        'ढ़' => 'ugDhha',
584
        'व'  => 'Va',
585
        'य'  => 'Ya',
586
        'य़' => 'Yi',
587
        'ज़' => 'Za',
588
    ],
589
    // Armenian
590
    'hy' => [
591
        'Ô±' => 'A',
592
        'Ô²' => 'B',
593
        'Ô³' => 'G',
594
        'Ô´' => 'D',
595
        'Ôµ' => 'E',
596
        'Ô¶' => 'Z',
597
        'Ô·' => 'E',
598
        'Ô¸' => 'Y',
599
        'Ô¹' => 'Th',
600
        'Ôº' => 'Zh',
601
        'Ô»' => 'I',
602
        'Ô¼' => 'L',
603
        'Ô½' => 'Kh',
604
        'Ô¾' => 'Ts',
605
        'Ô¿' => 'K',
606
        'Õ€' => 'H',
607
        'Ձ' => 'Dz',
608
        'Õ‚' => 'Gh',
609
        'Ճ' => 'Tch',
610
        'Õ„' => 'M',
611
        'Õ…' => 'Y',
612
        'Õ†' => 'N',
613
        'Õ‡' => 'Sh',
614
        'Ո' => 'Vo',
615
        'Չ' => 'Ch',
616
        'ÕŠ' => 'P',
617
        'Ջ' => 'J',
618
        'Ռ' => 'R',
619
        'Ս' => 'S',
620
        'ÕŽ' => 'V',
621
        'Տ' => 'T',
622
        'Ր' => 'R',
623
        'Õ‘' => 'C',
624
        'Õ’' => 'u',
625
        'Õ“' => 'Ph',
626
        'Õ”' => 'Q',
627
        'Ö‡' => 'ev',
628
        'Õ•' => 'O',
629
        'Õ–' => 'F',
630
        'Õ¡' => 'a',
631
        'Õ¢' => 'b',
632
        'Õ£' => 'g',
633
        'Õ¤' => 'd',
634
        'Õ¥' => 'e',
635
        'Õ¦' => 'z',
636
        'Õ§' => 'e',
637
        'Õ¨' => 'y',
638
        'Õ©' => 'th',
639
        'Õª' => 'zh',
640
        'Õ«' => 'i',
641
        'Õ¬' => 'l',
642
        'խ' => 'kh',
643
        'Õ®' => 'ts',
644
        'Õ¯' => 'k',
645
        'Õ°' => 'h',
646
        'Õ±' => 'dz',
647
        'Õ²' => 'gh',
648
        'Õ³' => 'tch',
649
        'Õ´' => 'm',
650
        'Õµ' => 'y',
651
        'Õ¶' => 'n',
652
        'Õ·' => 'sh',
653
        'Õ¸' => 'vo',
654
        'Õ¹' => 'ch',
655
        'Õº' => 'p',
656
        'Õ»' => 'j',
657
        'Õ¼' => 'r',
658
        'Õ½' => 's',
659
        'Õ¾' => 'v',
660
        'Õ¿' => 't',
661
        'Ö€' => 'r',
662
        'ց' => 'c',
663
        'Ö‚' => 'u',
664
        'փ' => 'ph',
665
        'Ö„' => 'q',
666
        'Ö…' => 'o',
667
        'Ö†' => 'f',
668
    ],
669
    // Swedish
670
    'sv' => [
671
        'Ä' => 'A',
672
        'ä' => 'a',
673
        'Å' => 'A',
674
        'å' => 'a',
675
        'Ö' => 'O',
676
        'ö' => 'o',
677
    ],
678
    // Turkmen
679
    'tk' => [
680
        'Ç' => 'C',
681
        'Ä' => 'A',
682
        'Ž' => 'Z',
683
        'Ň' => 'N',
684
        'Ö' => 'O',
685
        'Åž' => 'S',
686
        'Ü' => 'U',
687
        'Ý' => 'Y',
688
        'ç' => 'c',
689
        'ä' => 'a',
690
        'ž' => 'z',
691
        'ň' => 'n',
692
        'ö' => 'o',
693
        'ş' => 's',
694
        'ü' => 'u',
695
        'ý' => 'y',
696
    ],
697
    // Turkish
698
    'tr' => [
699
        'ň' => 'n',
700
        'Ň' => 'N',
701
        'ş' => 's',
702
        'Åž' => 'S',
703
        'ı' => 'i',
704
        'İ' => 'I',
705
        'ç' => 'c',
706
        'Ç' => 'C',
707
        'ä' => 'a',
708
        'Ä' => 'A',
709
        'ü' => 'u',
710
        'Ü' => 'U',
711
        'ö' => 'o',
712
        'Ö' => 'O',
713
        'ğ' => 'g',
714
        'Äž' => 'G',
715
        'ý' => 'y',
716
        'Ý' => 'Y',
717
        'ž' => 'z',
718
        'Ž' => 'Z',
719
    ],
720
    // Bulgarian
721
    'bg' => [
722
        'ьо' => 'yo',
723
        'А'  => 'A',
724
        'Б'  => 'B',
725
        'В'  => 'V',
726
        'Г'  => 'G',
727
        'Д'  => 'D',
728
        'Е'  => 'E',
729
        'Ж'  => 'Zh',
730
        'З'  => 'Z',
731
        'И'  => 'I',
732
        'Й'  => 'Y',
733
        'К'  => 'K',
734
        'Л'  => 'L',
735
        'М'  => 'M',
736
        'Н'  => 'N',
737
        'О'  => 'O',
738
        'П'  => 'P',
739
        'Р'  => 'R',
740
        'С'  => 'S',
741
        'Т'  => 'T',
742
        'У'  => 'U',
743
        'Ф'  => 'F',
744
        'Х'  => 'H',
745
        'Ц'  => 'C',
746
        'Ч'  => 'Ch',
747
        'Ш'  => 'Sh',
748
        'Щ'  => 'Sht',
749
        'Ъ'  => 'A',
750
        'Ь'  => '',
751
        'Ю'  => 'Yu',
752
        'Я'  => 'Ya',
753
        'а'  => 'a',
754
        'б'  => 'b',
755
        'в'  => 'v',
756
        'г'  => 'g',
757
        'д'  => 'd',
758
        'е'  => 'e',
759
        'ж'  => 'zh',
760
        'з'  => 'z',
761
        'и'  => 'i',
762
        'й'  => 'y',
763
        'к'  => 'k',
764
        'л'  => 'l',
765
        'м'  => 'm',
766
        'н'  => 'n',
767
        'о'  => 'o',
768
        'п'  => 'p',
769
        'р'  => 'r',
770
        'с'  => 's',
771
        'т'  => 't',
772
        'у'  => 'u',
773
        'ф'  => 'f',
774
        'х'  => 'h',
775
        'ц'  => 'c',
776
        'ч'  => 'ch',
777
        'ш'  => 'sh',
778
        'щ'  => 'sht',
779
        'ъ'  => 'a',
780
        'ь'  => '',
781
        'ю'  => 'yu',
782
        'я'  => 'ya',
783
    ],
784
    // Hungarian
785
    'hu' => [
786
        'Á' => 'A',
787
        'Ä’' => 'E',
788
        'É' => 'E',
789
        'Í' => 'I',
790
        'Ó' => 'O',
791
        'Ö' => 'O',
792
        'Ő' => 'O',
793
        'Ú' => 'U',
794
        'Ü' => 'U',
795
        'Ű' => 'U',
796
        'á' => 'a',
797
        'Ä“' => 'e',
798
        'é' => 'e',
799
        'í' => 'i',
800
        'ó' => 'o',
801
        'ö' => 'o',
802
        'ő' => 'o',
803
        'ú' => 'u',
804
        'ü' => 'u',
805
        'ű' => 'u',
806
    ],
807
    // Myanmar (Burmese)
808
    'my' => [
809
        'န်ုပ်' => 'nub',
810
        'ောင်'  => 'aung',
811
        'ိုက်'  => 'aik',
812
        'ိုဒ်'  => 'ok',
813
        'ိုင်'  => 'aing',
814
        'ိုလ်'  => 'ol',
815
        'ေါင်'  => 'aung',
816
        'သြော'  => 'aw',
817
        'ောက်'  => 'auk',
818
        'ိတ်'   => 'eik',
819
        'ုတ်'   => 'ok',
820
        'ုန်'   => 'on',
821
        'ေတ်'   => 'it',
822
        'ုဒ်'   => 'ait',
823
        'ာန်'   => 'an',
824
        'ိန်'   => 'ein',
825
        'ွတ်'   => 'ut',
826
        'ေါ်'   => 'aw',
827
        'ွန်'   => 'un',
828
        'ိပ်'   => 'eik',
829
        'ုပ်'   => 'ok',
830
        'ွပ်'   => 'ut',
831
        'ိမ်'   => 'ein',
832
        'ုမ်'   => 'on',
833
        'ော်'   => 'aw',
834
        'ွမ်'   => 'un',
835
        'က်'    => 'et',
836
        'ေါ'    => 'aw',
837
        'ော'    => 'aw',
838
        'ျွ'    => 'ywa',
839
        'ြွ'    => 'yw',
840
        'ို'    => 'o',
841
        'ုံ'    => 'on',
842
        'တ်'    => 'at',
843
        'င်'    => 'in',
844
        'ည်'    => 'i',
845
        'ဒ်'    => 'd',
846
        'န်'    => 'an',
847
        'ပ်'    => 'at',
848
        'မ်'    => 'an',
849
        'စျ'    => 'za',
850
        'ယ်'    => 'e',
851
        'ဉ်'    => 'in',
852
        'စ်'    => 'it',
853
        'ိံ'    => 'ein',
854
        'ဲ'     => 'e',
855
        'း'     => '',
856
        'ာ'     => 'a',
857
        'ါ'     => 'a',
858
        'ေ'     => 'e',
859
        'ံ'     => 'an',
860
        'ိ'     => 'i',
861
        'ီ'     => 'i',
862
        'ု'     => 'u',
863
        'ူ'     => 'u',
864
        '်'     => 'at',
865
        '္'     => '',
866
        '့'     => '',
867
        'က'     => 'k',
868
        '၉'     => '9',
869
        'တ'     => 't',
870
        'ရ'     => 'ya',
871
        'ယ'     => 'y',
872
        'မ'     => 'm',
873
        'ဘ'     => 'ba',
874
        'ဗ'     => 'b',
875
        'ဖ'     => 'pa',
876
        'ပ'     => 'p',
877
        'န'     => 'n',
878
        'ဓ'     => 'da',
879
        'ဒ'     => 'd',
880
        'ထ'     => 'ta',
881
        'ဏ'     => 'na',
882
        'ဝ'     => 'w',
883
        'ဎ'     => 'da',
884
        'ဍ'     => 'd',
885
        'ဌ'     => 'ta',
886
        'ဋ'     => 't',
887
        'ည'     => 'ny',
888
        'ဇ'     => 'z',
889
        'ဆ'     => 'sa',
890
        'စ'     => 's',
891
        'င'     => 'ng',
892
        'ဃ'     => 'ga',
893
        'ဂ'     => 'g',
894
        'လ'     => 'l',
895
        'သ'     => 'th',
896
        '၈'     => '8',
897
        'ဩ'     => 'aw',
898
        'ခ'     => 'kh',
899
        '၆'     => '6',
900
        '၅'     => '5',
901
        '၄'     => '4',
902
        '၃'     => '3',
903
        '၂'     => '2',
904
        '၁'     => '1',
905
        '၀'     => '0',
906
        '၌'     => 'hnaik',
907
        '၍'     => 'ywae',
908
        'ဪ'     => 'aw',
909
        'ဦ'     => '-u',
910
        'ဟ'     => 'h',
911
        'ဉ'     => 'u',
912
        'ဤ'     => '-i',
913
        'ဣ'     => 'i',
914
        '၏'     => '-e',
915
        'ဧ'     => 'e',
916
        'ှ'     => 'h',
917
        'ွ'     => 'w',
918
        'ျ'     => 'ya',
919
        'ြ'     => 'y',
920
        'အ'     => 'a',
921
        'ဠ'     => 'la',
922
        '၇'     => '7',
923
    ],
924
    // Croatian (Hrvatska)
925
    'hr' => [
926
        'DŽ' => 'DZ',
927
        'Dž' => 'Dz',
928
        'dž' => 'dz',
929
        'DZ'  => 'DZ',
930
        'Dz'  => 'Dz',
931
        'dz'  => 'dz',
932
        'IJ'  => 'IJ',
933
        'ij'  => 'ij',
934
        'LJ'  => 'LJ',
935
        'Lj'  => 'Lj',
936
        'lj'  => 'lj',
937
        'ÇŠ'  => 'NJ',
938
        'Nj'  => 'Nj',
939
        'nj'  => 'nj',
940
        'ž'  => 'z',
941
        'Ž'  => 'Z',
942
        'Ä‘'  => 'dj',
943
        'Đ'  => 'Dj',
944
        'č'  => 'c',
945
        'Č'  => 'C',
946
        'ć'  => 'c',
947
        'Ć'  => 'C',
948
        'š'  => 's',
949
        'Š'  => 'S',
950
    ],
951
    // Finnish
952
    'fi' => [
953
        'Ä' => 'A',
954
        'Ö' => 'O',
955
        'ä' => 'a',
956
        'ö' => 'o',
957
    ],
958
    // Georgian (Kartvelian)
959
    'ka' => [
960
        'ა' => 'a',
961
        'ბ' => 'b',
962
        'გ' => 'g',
963
        'დ' => 'd',
964
        'ე' => 'e',
965
        'ვ' => 'v',
966
        'ზ' => 'z',
967
        'თ' => 't',
968
        'ი' => 'i',
969
        'კ' => 'k',
970
        'ლ' => 'l',
971
        'მ' => 'm',
972
        'ნ' => 'n',
973
        'ო' => 'o',
974
        'პ' => 'p',
975
        'ჟ' => 'zh',
976
        'რ' => 'r',
977
        'ს' => 's',
978
        'ტ' => 't',
979
        'უ' => 'u',
980
        'ფ' => 'f',
981
        'ქ' => 'q',
982
        'ღ' => 'gh',
983
        'ყ' => 'y',
984
        'შ' => 'sh',
985
        'ჩ' => 'ch',
986
        'ც' => 'ts',
987
        'ძ' => 'dz',
988
        'წ' => 'ts',
989
        'ჭ' => 'ch',
990
        'ხ' => 'kh',
991
        'ჯ' => 'j',
992
        'ჰ' => 'h',
993
    ],
994
    // Russian
995
    'ru' => [
996
        'А' => 'A',
997
        'а' => 'a',
998
        'Б' => 'B',
999
        'б' => 'b',
1000
        'В' => 'V',
1001
        'в' => 'v',
1002
        'Г' => 'G',
1003
        'г' => 'g',
1004
        'Д' => 'D',
1005
        'д' => 'd',
1006
        'Е' => 'E',
1007
        'е' => 'e',
1008
        'Ё' => 'Yo',
1009
        'ё' => 'yo',
1010
        'Ж' => 'Zh',
1011
        'ж' => 'zh',
1012
        'З' => 'Z',
1013
        'з' => 'z',
1014
        'И' => 'I',
1015
        'и' => 'i',
1016
        'Й' => 'Y',
1017
        'й' => 'y',
1018
        'К' => 'K',
1019
        'к' => 'k',
1020
        'Л' => 'L',
1021
        'л' => 'l',
1022
        'М' => 'M',
1023
        'м' => 'm',
1024
        'Н' => 'N',
1025
        'н' => 'n',
1026
        'О' => 'O',
1027
        'о' => 'o',
1028
        'П' => 'P',
1029
        'п' => 'p',
1030
        'Р' => 'R',
1031
        'р' => 'r',
1032
        'С' => 'S',
1033
        'с' => 's',
1034
        'Т' => 'T',
1035
        'т' => 't',
1036
        'У' => 'U',
1037
        'у' => 'u',
1038
        'Ф' => 'F',
1039
        'ф' => 'f',
1040
        'Х' => 'H',
1041
        'х' => 'h',
1042
        'Ц' => 'Ts',
1043
        'ц' => 'ts',
1044
        'Ч' => 'Ch',
1045
        'ч' => 'ch',
1046
        'ш' => 'sh',
1047
        'Ш' => 'Sh',
1048
        'Щ' => 'Sch',
1049
        'щ' => 'sch',
1050
        'Ъ' => '',
1051
        'ъ' => '',
1052
        'Ы' => 'Y',
1053
        'ы' => 'y',
1054
        'Ь' => '',
1055
        'ь' => '',
1056
        'Э' => 'E',
1057
        'э' => 'e',
1058
        'Ю' => 'Yu',
1059
        'ю' => 'yu',
1060
        'Я' => 'Ya',
1061
        'я' => 'ya',
1062
    ],
1063
    // Russian - GOST 7.79-2000(B)
1064
    // -> https://en.m.wikipedia.org/wiki/Romanization_of_Russian#content-collapsible-block-1
1065
    'ru__gost_2000_b' => [
1066
        'А' => 'A',
1067
        'а' => 'a',
1068
        'Б' => 'B',
1069
        'б' => 'b',
1070
        'В' => 'V',
1071
        'в' => 'v',
1072
        'Г' => 'G',
1073
        'г' => 'g',
1074
        'Д' => 'D',
1075
        'д' => 'd',
1076
        'Е' => 'E',
1077
        'е' => 'e',
1078
        'Ё' => 'Yo',
1079
        'ё' => 'yo',
1080
        'Ж' => 'Zh',
1081
        'ж' => 'zh',
1082
        'З' => 'Z',
1083
        'з' => 'z',
1084
        'И' => 'i',
1085
        'и' => 'i',
1086
        'Й' => 'i',
1087
        'й' => 'i',
1088
        'К' => 'K',
1089
        'к' => 'k',
1090
        'Л' => 'L',
1091
        'л' => 'l',
1092
        'М' => 'M',
1093
        'м' => 'm',
1094
        'Н' => 'N',
1095
        'н' => 'n',
1096
        'О' => 'O',
1097
        'о' => 'o',
1098
        'П' => 'P',
1099
        'п' => 'p',
1100
        'Р' => 'R',
1101
        'р' => 'r',
1102
        'С' => 'S',
1103
        'с' => 's',
1104
        'Т' => 'T',
1105
        'т' => 't',
1106
        'У' => 'U',
1107
        'у' => 'u',
1108
        'Ф' => 'F',
1109
        'ф' => 'f',
1110
        'Х' => 'X',
1111
        'х' => 'x',
1112
        'Ц' => 'Cz',
1113
        'ц' => 'cz',
1114
        'Ч' => 'Ch',
1115
        'ч' => 'ch',
1116
        'ш' => 'sh',
1117
        'Ш' => 'Sh',
1118
        'Щ' => 'Shh',
1119
        'щ' => 'shh',
1120
        'Ъ' => '',
1121
        'ъ' => '',
1122
        'Ы' => 'Y\'',
1123
        'ы' => 'y\'',
1124
        'Ь' => '',
1125
        'ь' => '',
1126
        'Э' => 'E\'',
1127
        'э' => 'e\'',
1128
        'Ю' => 'Yu',
1129
        'ю' => 'yu',
1130
        'Я' => 'Ya',
1131
        'я' => 'ya',
1132
        'І' => 'I',
1133
        'і' => 'i',
1134
        'Ѳ' => 'Fh',
1135
        'ѳ' => 'fh',
1136
        'Ѣ' => 'Ye',
1137
        'ѣ' => 'ye',
1138
        'Ѵ' => 'Yh',
1139
        'ѵ' => 'yh',
1140
        'Є' => '',
1141
        'є' => '',
1142
        'Ѥ' => '',
1143
        'ѥ' => '',
1144
        'Ѕ' => 'Js',
1145
        'ѕ' => 'js',
1146
        'Ꙋ' => '',
1147
        'ꙋ' => '',
1148
        'Ѡ' => '',
1149
        'ѡ' => '',
1150
        'Ѿ' => '',
1151
        'ѿ' => '',
1152
        'Ѫ' => '',
1153
        'ѫ' => '',
1154
        'Ѧ' => '',
1155
        'ѧ' => '',
1156
        'Ѭ' => '',
1157
        'ѭ' => '',
1158
        'Ѩ' => '',
1159
        'ѩ' => '',
1160
        'Ѯ' => '',
1161
        'ѯ' => '',
1162
        'Ѱ' => '',
1163
        'ѱ' => '',
1164
    ],
1165
    // Russian - Passport (2013), ICAO
1166
    // -> https://en.m.wikipedia.org/wiki/Romanization_of_Russian#content-collapsible-block-1
1167
    'ru__passport_2013' => [
1168
        'А' => 'A',
1169
        'а' => 'a',
1170
        'Б' => 'B',
1171
        'б' => 'b',
1172
        'В' => 'V',
1173
        'в' => 'v',
1174
        'Г' => 'G',
1175
        'г' => 'g',
1176
        'Д' => 'D',
1177
        'д' => 'd',
1178
        'Е' => 'E',
1179
        'е' => 'e',
1180
        'Ё' => 'E',
1181
        'ё' => 'e',
1182
        'Ж' => 'Zh',
1183
        'ж' => 'zh',
1184
        'З' => 'Z',
1185
        'з' => 'z',
1186
        'И' => 'i',
1187
        'и' => 'i',
1188
        'Й' => 'i',
1189
        'й' => 'i',
1190
        'К' => 'K',
1191
        'к' => 'k',
1192
        'Л' => 'L',
1193
        'л' => 'l',
1194
        'М' => 'M',
1195
        'м' => 'm',
1196
        'Н' => 'N',
1197
        'н' => 'n',
1198
        'О' => 'O',
1199
        'о' => 'o',
1200
        'П' => 'P',
1201
        'п' => 'p',
1202
        'Р' => 'R',
1203
        'р' => 'r',
1204
        'С' => 'S',
1205
        'с' => 's',
1206
        'Т' => 'T',
1207
        'т' => 't',
1208
        'У' => 'U',
1209
        'у' => 'u',
1210
        'Ф' => 'F',
1211
        'ф' => 'f',
1212
        'Х' => 'Kh',
1213
        'х' => 'kh',
1214
        'Ц' => 'Ts',
1215
        'ц' => 'ts',
1216
        'Ч' => 'Ch',
1217
        'ч' => 'ch',
1218
        'ш' => 'sh',
1219
        'Ш' => 'Sh',
1220
        'Щ' => 'Shch',
1221
        'щ' => 'shch',
1222
        'Ъ' => 'Ie',
1223
        'ъ' => 'ie',
1224
        'Ы' => 'Y',
1225
        'ы' => 'y',
1226
        'Ь' => '',
1227
        'ь' => '',
1228
        'Э' => 'E',
1229
        'э' => 'e',
1230
        'Ю' => 'Iu',
1231
        'ю' => 'iu',
1232
        'Я' => 'Ia',
1233
        'я' => 'ia',
1234
        'І' => '',
1235
        'і' => '',
1236
        'Ѳ' => '',
1237
        'ѳ' => '',
1238
        'Ѣ' => '',
1239
        'ѣ' => '',
1240
        'Ѵ' => '',
1241
        'ѵ' => '',
1242
        'Є' => '',
1243
        'є' => '',
1244
        'Ѥ' => '',
1245
        'ѥ' => '',
1246
        'Ѕ' => '',
1247
        'ѕ' => '',
1248
        'Ꙋ' => '',
1249
        'ꙋ' => '',
1250
        'Ѡ' => '',
1251
        'ѡ' => '',
1252
        'Ѿ' => '',
1253
        'ѿ' => '',
1254
        'Ѫ' => '',
1255
        'ѫ' => '',
1256
        'Ѧ' => '',
1257
        'ѧ' => '',
1258
        'Ѭ' => '',
1259
        'ѭ' => '',
1260
        'Ѩ' => '',
1261
        'ѩ' => '',
1262
        'Ѯ' => '',
1263
        'ѯ' => '',
1264
        'Ѱ' => '',
1265
        'ѱ' => '',
1266
    ],
1267
    // Ukrainian
1268
    // -> https://zakon.rada.gov.ua/laws/show/55-2010-%D0%BF?lang=en
1269
    'uk' => [
1270
        'Г' => 'H',
1271
        'г' => 'h',
1272
        'Ґ' => 'G',
1273
        'ґ' => 'g',
1274
        'Є' => 'Ye',
1275
        'є' => 'ye',
1276
        'И' => 'Y',
1277
        'и' => 'y',
1278
        'І' => 'I',
1279
        'і' => 'i',
1280
        'Ї' => 'Yi',
1281
        'ї' => 'yi',
1282
        'Й' => 'Y',
1283
        'й' => 'y',
1284
        'Х' => 'Kh',
1285
        'х' => 'kh',
1286
        'Ц' => 'Ts',
1287
        'ц' => 'ts',
1288
        'Ч' => 'Ch',
1289
        'ч' => 'ch',
1290
        'Ш' => 'Sh',
1291
        'ш' => 'sh',
1292
        'Щ' => 'Shch',
1293
        'щ' => 'shch',
1294
    ],
1295
    // Kazakh
1296
    'kk' => [
1297
        'Ә' => 'A',
1298
        'Ғ' => 'G',
1299
        'Қ' => 'Q',
1300
        'Ң' => 'N',
1301
        'Ө' => 'O',
1302
        'Ұ' => 'U',
1303
        'Ү' => 'U',
1304
        'Һ' => 'H',
1305
        'ә' => 'a',
1306
        'ғ' => 'g',
1307
        'қ' => 'q',
1308
        'ң' => 'n',
1309
        'ө' => 'o',
1310
        'ұ' => 'u',
1311
        'ү' => 'u',
1312
        'һ' => 'h',
1313
    ],
1314
    // Czech
1315
    'cs' => [
1316
        'á' => 'a',
1317
        'Á' => 'A',
1318
        'č' => 'c',
1319
        'Č' => 'C',
1320
        'ď' => 'd',
1321
        'Ď' => 'D',
1322
        'é' => 'e',
1323
        'É' => 'E',
1324
        'ě' => 'e',
1325
        'Ě' => 'E',
1326
        'í' => 'i',
1327
        'Í' => 'I',
1328
        'ň' => 'n',
1329
        'Ň' => 'N',
1330
        'ó' => 'o',
1331
        'Ó' => 'O',
1332
        'ř' => 'r',
1333
        'Ř' => 'R',
1334
        'š' => 's',
1335
        'Š' => 'S',
1336
        'ť' => 't',
1337
        'Ť' => 'T',
1338
        'ú' => 'u',
1339
        'Ú' => 'U',
1340
        'ů' => 'u',
1341
        'Ů' => 'U',
1342
        'ý' => 'y',
1343
        'Ý' => 'Y',
1344
        'ž' => 'z',
1345
        'Ž' => 'Z',
1346
    ],
1347
    // Danish
1348
    'da' => [
1349
        'Æ' => 'Ae',
1350
        'æ' => 'ae',
1351
        'Ø' => 'Oe',
1352
        'ø' => 'oe',
1353
        'Å' => 'Aa',
1354
        'å' => 'aa',
1355
        'É' => 'E',
1356
        'é' => 'e',
1357
    ],
1358
    // Polish
1359
    'pl' => [
1360
        'ą' => 'a',
1361
        'ć' => 'c',
1362
        'ę' => 'e',
1363
        'ł' => 'l',
1364
        'ń' => 'n',
1365
        'ó' => 'o',
1366
        'ś' => 's',
1367
        'ź' => 'z',
1368
        'ż' => 'z',
1369
        'Ą' => 'A',
1370
        'Ć' => 'C',
1371
        'Ę' => 'E',
1372
        'Ł' => 'L',
1373
        'Ń' => 'N',
1374
        'Ó' => 'O',
1375
        'Ś' => 'S',
1376
        'Ź' => 'Z',
1377
        'Ż' => 'Z',
1378
    ],
1379
    // Romanian
1380
    'ro' => [
1381
        'ă' => 'a',
1382
        'â' => 'a',
1383
        'Ä‚' => 'A',
1384
        'Â' => 'A',
1385
        'î' => 'i',
1386
        'Î' => 'I',
1387
        'ș' => 's',
1388
        'ş' => 's',
1389
        'Åž' => 'S',
1390
        'Ș' => 'S',
1391
        'ț' => 't',
1392
        'Å£' => 't',
1393
        'Å¢' => 'T',
1394
        'Èš' => 'T',
1395
    ],
1396
    // Esperanto
1397
    'eo' => [
1398
        'ĉ' => 'cx',
1399
        'ĝ' => 'gx',
1400
        'Ä¥' => 'hx',
1401
        'ĵ' => 'jx',
1402
        'ŝ' => 'sx',
1403
        'ŭ' => 'ux',
1404
        'Ĉ' => 'CX',
1405
        'Ĝ' => 'GX',
1406
        'Ĥ' => 'HX',
1407
        'Ä´' => 'JX',
1408
        'Ŝ' => 'SX',
1409
        'Ŭ' => 'UX',
1410
    ],
1411
    // Estonian
1412
    'et' => [
1413
        'Š' => 'S',
1414
        'Ž' => 'Z',
1415
        'Õ' => 'O',
1416
        'Ä' => 'A',
1417
        'Ö' => 'O',
1418
        'Ü' => 'U',
1419
        'š' => 's',
1420
        'ž' => 'z',
1421
        'õ' => 'o',
1422
        'ä' => 'a',
1423
        'ö' => 'o',
1424
        'ü' => 'u',
1425
    ],
1426
    // Latvian
1427
    'lv' => [
1428
        'ā' => 'a',
1429
        'č' => 'c',
1430
        'Ä“' => 'e',
1431
        'Ä£' => 'g',
1432
        'Ä«' => 'i',
1433
        'Ä·' => 'k',
1434
        'ļ' => 'l',
1435
        'ņ' => 'n',
1436
        'š' => 's',
1437
        'Å«' => 'u',
1438
        'ž' => 'z',
1439
        'Ä€' => 'A',
1440
        'Č' => 'C',
1441
        'Ä’' => 'E',
1442
        'Ä¢' => 'G',
1443
        'Ī' => 'i',
1444
        'Ķ' => 'k',
1445
        'Ä»' => 'L',
1446
        'Å…' => 'N',
1447
        'Š' => 'S',
1448
        'Ū' => 'u',
1449
        'Ž' => 'Z',
1450
    ],
1451
    // Lithuanian
1452
    'lt' => [
1453
        'ą' => 'a',
1454
        'č' => 'c',
1455
        'ę' => 'e',
1456
        'Ä—' => 'e',
1457
        'į' => 'i',
1458
        'š' => 's',
1459
        'ų' => 'u',
1460
        'Å«' => 'u',
1461
        'ž' => 'z',
1462
        'Ą' => 'A',
1463
        'Č' => 'C',
1464
        'Ę' => 'E',
1465
        'Ä–' => 'E',
1466
        'Ä®' => 'I',
1467
        'Š' => 'S',
1468
        'Ų' => 'U',
1469
        'Ū' => 'U',
1470
        'Ž' => 'Z',
1471
    ],
1472
    // Norwegian
1473
    'no' => [
1474
        'Æ' => 'AE',
1475
        'æ' => 'ae',
1476
        'Ø' => 'OE',
1477
        'ø' => 'oe',
1478
        'Å' => 'AA',
1479
        'å' => 'aa',
1480
    ],
1481
    // Vietnamese
1482
    'vi' => [
1483
        'Á' => 'A',
1484
        'À' => 'A',
1485
        'Ả' => 'A',
1486
        'Ã' => 'A',
1487
        'Ạ' => 'A',
1488
        'Ä‚' => 'A',
1489
        'Ắ' => 'A',
1490
        'Ằ' => 'A',
1491
        'Ẳ' => 'A',
1492
        'Ẵ' => 'A',
1493
        'Ặ' => 'A',
1494
        'Â' => 'A',
1495
        'Ấ' => 'A',
1496
        'Ầ' => 'A',
1497
        'Ẩ' => 'A',
1498
        'Ẫ' => 'A',
1499
        'Ậ' => 'A',
1500
        'á' => 'a',
1501
        'à' => 'a',
1502
        'ả' => 'a',
1503
        'ã' => 'a',
1504
        'ạ' => 'a',
1505
        'ă' => 'a',
1506
        'ắ' => 'a',
1507
        'ằ' => 'a',
1508
        'ẳ' => 'a',
1509
        'ẵ' => 'a',
1510
        'ặ' => 'a',
1511
        'â' => 'a',
1512
        'ấ' => 'a',
1513
        'ầ' => 'a',
1514
        'ẩ' => 'a',
1515
        'ẫ' => 'a',
1516
        'ậ' => 'a',
1517
        'É' => 'E',
1518
        'È' => 'E',
1519
        'Ẻ' => 'E',
1520
        'Ẽ' => 'E',
1521
        'Ẹ' => 'E',
1522
        'Ê' => 'E',
1523
        'Ế' => 'E',
1524
        'Ề' => 'E',
1525
        'Ể' => 'E',
1526
        'Ễ' => 'E',
1527
        'Ệ' => 'E',
1528
        'é' => 'e',
1529
        'è' => 'e',
1530
        'ẻ' => 'e',
1531
        'ẽ' => 'e',
1532
        'ẹ' => 'e',
1533
        'ê' => 'e',
1534
        'ế' => 'e',
1535
        'ề' => 'e',
1536
        'ể' => 'e',
1537
        'á»…' => 'e',
1538
        'ệ' => 'e',
1539
        'Í' => 'I',
1540
        'Ì' => 'I',
1541
        'Ỉ' => 'I',
1542
        'Ĩ' => 'I',
1543
        'Ị' => 'I',
1544
        'í' => 'i',
1545
        'ì' => 'i',
1546
        'ỉ' => 'i',
1547
        'Ä©' => 'i',
1548
        'ị' => 'i',
1549
        'Ó' => 'O',
1550
        'Ò' => 'O',
1551
        'Ỏ' => 'O',
1552
        'Õ' => 'O',
1553
        'Ọ' => 'O',
1554
        'Ô' => 'O',
1555
        'Ố' => 'O',
1556
        'á»’' => 'O',
1557
        'á»”' => 'O',
1558
        'á»–' => 'O',
1559
        'Ộ' => 'O',
1560
        'Ơ' => 'O',
1561
        'Ớ' => 'O',
1562
        'Ờ' => 'O',
1563
        'Ở' => 'O',
1564
        'Ỡ' => 'O',
1565
        'Ợ' => 'O',
1566
        'ó' => 'o',
1567
        'ò' => 'o',
1568
        'ỏ' => 'o',
1569
        'õ' => 'o',
1570
        'ọ' => 'o',
1571
        'ô' => 'o',
1572
        'ố' => 'o',
1573
        'ồ' => 'o',
1574
        'ổ' => 'o',
1575
        'á»—' => 'o',
1576
        'ộ' => 'o',
1577
        'Æ¡' => 'o',
1578
        'ớ' => 'o',
1579
        'ờ' => 'o',
1580
        'ở' => 'o',
1581
        'ỡ' => 'o',
1582
        'ợ' => 'o',
1583
        'Ú' => 'U',
1584
        'Ù' => 'U',
1585
        'Ủ' => 'U',
1586
        'Ũ' => 'U',
1587
        'Ụ' => 'U',
1588
        'Ư' => 'U',
1589
        'Ứ' => 'U',
1590
        'Ừ' => 'U',
1591
        'Ử' => 'U',
1592
        'á»®' => 'U',
1593
        'á»°' => 'U',
1594
        'ú' => 'u',
1595
        'ù' => 'u',
1596
        'á»§' => 'u',
1597
        'Å©' => 'u',
1598
        'ụ' => 'u',
1599
        'ư' => 'u',
1600
        'ứ' => 'u',
1601
        'ừ' => 'u',
1602
        'ử' => 'u',
1603
        'ữ' => 'u',
1604
        'á»±' => 'u',
1605
        'Ý' => 'Y',
1606
        'Ỳ' => 'Y',
1607
        'á»¶' => 'Y',
1608
        'Ỹ' => 'Y',
1609
        'á»´' => 'Y',
1610
        'ý' => 'y',
1611
        'ỳ' => 'y',
1612
        'á»·' => 'y',
1613
        'ỹ' => 'y',
1614
        'ỵ' => 'y',
1615
        'Đ' => 'D',
1616
        'Ä‘' => 'd',
1617
    ],
1618
    // Persian (Farsi)
1619
    'fa' => [
1620
        'ا' => 'a',
1621
        'ب' => 'b',
1622
        'پ' => 'p',
1623
        'ت' => 't',
1624
        'ث' => 's',
1625
        'ج' => 'j',
1626
        'چ' => 'ch',
1627
        'ح' => 'h',
1628
        'خ' => 'kh',
1629
        'د' => 'd',
1630
        'ذ' => 'z',
1631
        'ر' => 'r',
1632
        'ز' => 'z',
1633
        'س' => 's',
1634
        'ش' => 'sh',
1635
        'ص' => 's',
1636
        'ض' => 'z',
1637
        'ط' => 't',
1638
        'ظ' => 'z',
1639
        'ع' => 'a',
1640
        'غ' => 'gh',
1641
        'ف' => 'f',
1642
        'ق' => 'gh',
1643
        'ک' => 'k',
1644
        'گ' => 'g',
1645
        'ل' => 'l',
1646
        'ژ' => 'zh',
1647
        'ك' => 'k',
1648
        'م' => 'm',
1649
        'ن' => 'n',
1650
        'ه' => 'h',
1651
        'و' => 'o',
1652
        'ی' => 'y',
1653
        'آ' => 'a',
1654
        '٠' => '0',
1655
        '١' => '1',
1656
        '٢' => '2',
1657
        '٣' => '3',
1658
        '٤' => '4',
1659
        '٥' => '5',
1660
        '٦' => '6',
1661
        '٧' => '7',
1662
        '٨' => '8',
1663
        '٩' => '9',
1664
    ],
1665
    // Arabic
1666
    'ar' => [
1667
        'أ' => 'a',
1668
        'ب' => 'b',
1669
        'ت' => 't',
1670
        'ث' => 'th',
1671
        'ج' => 'g',
1672
        'ح' => 'h',
1673
        'خ' => 'kh',
1674
        'د' => 'd',
1675
        'ذ' => 'th',
1676
        'ر' => 'r',
1677
        'ز' => 'z',
1678
        'س' => 's',
1679
        'ش' => 'sh',
1680
        'ص' => 's',
1681
        'ض' => 'd',
1682
        'ط' => 't',
1683
        'ظ' => 'th',
1684
        'ع' => 'aa',
1685
        'غ' => 'gh',
1686
        'ف' => 'f',
1687
        'ق' => 'k',
1688
        'ك' => 'k',
1689
        'ل' => 'l',
1690
        'م' => 'm',
1691
        'ن' => 'n',
1692
        'ه' => 'h',
1693
        'و' => 'o',
1694
        'ي' => 'y',
1695
        'ا' => 'a',
1696
        'إ' => 'a',
1697
        'آ' => 'a',
1698
        'ؤ' => 'o',
1699
        'ئ' => 'y',
1700
        'ء' => 'aa',
1701
        '٠' => '0',
1702
        '١' => '1',
1703
        '٢' => '2',
1704
        '٣' => '3',
1705
        '٤' => '4',
1706
        '٥' => '5',
1707
        '٦' => '6',
1708
        '٧' => '7',
1709
        '٨' => '8',
1710
        '٩' => '9',
1711
    ],
1712
    // Serbian
1713
    'sr' => [
1714
        'Ä‘' => 'dj',
1715
        'ž' => 'z',
1716
        'ć' => 'c',
1717
        'č' => 'c',
1718
        'š' => 's',
1719
        'Đ' => 'Dj',
1720
        'Ž' => 'Z',
1721
        'Ć' => 'C',
1722
        'Č' => 'C',
1723
        'Š' => 'S',
1724
        'а' => 'a',
1725
        'б' => 'b',
1726
        'в' => 'v',
1727
        'г' => 'g',
1728
        'д' => 'd',
1729
        'ђ' => 'dj',
1730
        'е' => 'e',
1731
        'ж' => 'z',
1732
        'з' => 'z',
1733
        'и' => 'i',
1734
        'ј' => 'j',
1735
        'к' => 'k',
1736
        'л' => 'l',
1737
        'љ' => 'lj',
1738
        'м' => 'm',
1739
        'н' => 'n',
1740
        'њ' => 'nj',
1741
        'о' => 'o',
1742
        'п' => 'p',
1743
        'р' => 'r',
1744
        'с' => 's',
1745
        'т' => 't',
1746
        'ћ' => 'c',
1747
        'у' => 'u',
1748
        'ф' => 'f',
1749
        'х' => 'h',
1750
        'ц' => 'c',
1751
        'ч' => 'c',
1752
        'џ' => 'dz',
1753
        'ш' => 's',
1754
        'А' => 'A',
1755
        'Б' => 'B',
1756
        'В' => 'V',
1757
        'Г' => 'G',
1758
        'Д' => 'D',
1759
        'Ђ' => 'Dj',
1760
        'Е' => 'E',
1761
        'Ж' => 'Z',
1762
        'З' => 'Z',
1763
        'И' => 'I',
1764
        'Ј' => 'j',
1765
        'К' => 'K',
1766
        'Л' => 'L',
1767
        'Љ' => 'Lj',
1768
        'М' => 'M',
1769
        'Н' => 'N',
1770
        'Њ' => 'Nj',
1771
        'О' => 'O',
1772
        'П' => 'P',
1773
        'Р' => 'R',
1774
        'С' => 'S',
1775
        'Т' => 'T',
1776
        'Ћ' => 'C',
1777
        'У' => 'U',
1778
        'Ф' => 'F',
1779
        'Х' => 'H',
1780
        'Ц' => 'C',
1781
        'Ч' => 'C',
1782
        'Џ' => 'Dz',
1783
        'Ш' => 'S',
1784
    ],
1785
    // Serbian - Cyrillic
1786
    'sr__cyr' => [
1787
        'а' => 'a',
1788
        'б' => 'b',
1789
        'в' => 'v',
1790
        'г' => 'g',
1791
        'д' => 'd',
1792
        'ђ' => 'dj',
1793
        'е' => 'e',
1794
        'ж' => 'z',
1795
        'з' => 'z',
1796
        'и' => 'i',
1797
        'ј' => 'j',
1798
        'к' => 'k',
1799
        'л' => 'l',
1800
        'љ' => 'lj',
1801
        'м' => 'm',
1802
        'н' => 'n',
1803
        'њ' => 'nj',
1804
        'о' => 'o',
1805
        'п' => 'p',
1806
        'р' => 'r',
1807
        'с' => 's',
1808
        'т' => 't',
1809
        'ћ' => 'c',
1810
        'у' => 'u',
1811
        'ф' => 'f',
1812
        'х' => 'h',
1813
        'ц' => 'c',
1814
        'ч' => 'c',
1815
        'џ' => 'dz',
1816
        'ш' => 's',
1817
        'А' => 'A',
1818
        'Б' => 'B',
1819
        'В' => 'V',
1820
        'Г' => 'G',
1821
        'Д' => 'D',
1822
        'Ђ' => 'Dj',
1823
        'Е' => 'E',
1824
        'Ж' => 'Z',
1825
        'З' => 'Z',
1826
        'И' => 'I',
1827
        'Ј' => 'j',
1828
        'К' => 'K',
1829
        'Л' => 'L',
1830
        'Љ' => 'Lj',
1831
        'М' => 'M',
1832
        'Н' => 'N',
1833
        'Њ' => 'Nj',
1834
        'О' => 'O',
1835
        'П' => 'P',
1836
        'Р' => 'R',
1837
        'С' => 'S',
1838
        'Т' => 'T',
1839
        'Ћ' => 'C',
1840
        'У' => 'U',
1841
        'Ф' => 'F',
1842
        'Х' => 'H',
1843
        'Ц' => 'C',
1844
        'Ч' => 'C',
1845
        'Џ' => 'Dz',
1846
        'Ш' => 'S',
1847
    ],
1848
    // Serbian - Latin
1849
    'sr__lat' => [
1850
        'Ä‘' => 'dj',
1851
        'ž' => 'z',
1852
        'ć' => 'c',
1853
        'č' => 'c',
1854
        'š' => 's',
1855
        'Đ' => 'Dj',
1856
        'Ž' => 'Z',
1857
        'Ć' => 'C',
1858
        'Č' => 'C',
1859
        'Š' => 'S',
1860
    ],
1861
    // Azerbaijani
1862
    'az' => [
1863
        'ç' => 'c',
1864
        'ə' => 'e',
1865
        'ğ' => 'g',
1866
        'ı' => 'i',
1867
        'ö' => 'o',
1868
        'ş' => 's',
1869
        'ü' => 'u',
1870
        'Ç' => 'C',
1871
        'Ə' => 'E',
1872
        'Äž' => 'G',
1873
        'İ' => 'I',
1874
        'Ö' => 'O',
1875
        'Åž' => 'S',
1876
        'Ü' => 'U',
1877
    ],
1878
    // Slovak
1879
    'sk' => [
1880
        'á' => 'a',
1881
        'ä' => 'a',
1882
        'č' => 'c',
1883
        'ď' => 'd',
1884
        'é' => 'e',
1885
        'í' => 'i',
1886
        'ľ' => 'l',
1887
        'ĺ' => 'l',
1888
        'ň' => 'n',
1889
        'ó' => 'o',
1890
        'ô' => 'o',
1891
        'Å•' => 'r',
1892
        'š' => 's',
1893
        'ť' => 't',
1894
        'ú' => 'u',
1895
        'ý' => 'y',
1896
        'ž' => 'z',
1897
        'Á' => 'A',
1898
        'Ä' => 'A',
1899
        'Č' => 'C',
1900
        'Ď' => 'D',
1901
        'É' => 'E',
1902
        'Í' => 'I',
1903
        'Ľ' => 'L',
1904
        'Ĺ' => 'L',
1905
        'Ň' => 'N',
1906
        'Ó' => 'O',
1907
        'Ô' => 'O',
1908
        'Å”' => 'R',
1909
        'Š' => 'S',
1910
        'Ť' => 'T',
1911
        'Ú' => 'U',
1912
        'Ý' => 'Y',
1913
        'Ž' => 'Z',
1914
    ],
1915
    // French
1916
    'fr' => [
1917
        'Æ' => 'AE',
1918
        'æ' => 'ae',
1919
        'Å’' => 'OE',
1920
        'Å“' => 'oe',
1921
        'â' => 'a',
1922
        'Â' => 'A',
1923
        'à' => 'a',
1924
        'À' => 'A',
1925
        'ä' => 'a',
1926
        'Ä' => 'A',
1927
        'ç' => 'c',
1928
        'Ç' => 'C',
1929
        'é' => 'e',
1930
        'É' => 'E',
1931
        'ê' => 'e',
1932
        'Ê' => 'E',
1933
        'ë' => 'e',
1934
        'Ë' => 'E',
1935
        'è' => 'e',
1936
        'È' => 'E',
1937
        'ï' => 'i',
1938
        'î' => 'i',
1939
        'Ï' => 'I',
1940
        'Î' => 'I',
1941
        'ÿ' => 'y',
1942
        'Ÿ' => 'Y',
1943
        'ô' => 'o',
1944
        'Ô' => 'O',
1945
        'ö' => 'o',
1946
        'Ö' => 'O',
1947
        'û' => 'u',
1948
        'Û' => 'U',
1949
        'ù' => 'u',
1950
        'Ù' => 'U',
1951
        'ü' => 'u',
1952
        'Ü' => 'U',
1953
    ],
1954
    // Austrian (French)
1955
    'fr_at' => [
1956
        'ß' => 'sz',
1957
        'ẞ' => 'SZ',
1958
        'Æ' => 'AE',
1959
        'æ' => 'ae',
1960
        'Å’' => 'OE',
1961
        'Å“' => 'oe',
1962
        'â' => 'a',
1963
        'Â' => 'A',
1964
        'à' => 'a',
1965
        'À' => 'A',
1966
        'ä' => 'a',
1967
        'Ä' => 'A',
1968
        'ç' => 'c',
1969
        'Ç' => 'C',
1970
        'é' => 'e',
1971
        'É' => 'E',
1972
        'ê' => 'e',
1973
        'Ê' => 'E',
1974
        'ë' => 'e',
1975
        'Ë' => 'E',
1976
        'è' => 'e',
1977
        'È' => 'E',
1978
        'ï' => 'i',
1979
        'î' => 'i',
1980
        'Ï' => 'I',
1981
        'Î' => 'I',
1982
        'ÿ' => 'y',
1983
        'Ÿ' => 'Y',
1984
        'ô' => 'o',
1985
        'Ô' => 'O',
1986
        'ö' => 'o',
1987
        'Ö' => 'O',
1988
        'û' => 'u',
1989
        'Û' => 'U',
1990
        'ù' => 'u',
1991
        'Ù' => 'U',
1992
        'ü' => 'u',
1993
        'Ü' => 'U',
1994
    ],
1995
    // Switzerland (French)
1996
    'fr_ch' => [
1997
        'ß' => 'ss',
1998
        'ẞ' => 'SS',
1999
        'Æ' => 'AE',
2000
        'æ' => 'ae',
2001
        'Å’' => 'OE',
2002
        'Å“' => 'oe',
2003
        'â' => 'a',
2004
        'Â' => 'A',
2005
        'à' => 'a',
2006
        'À' => 'A',
2007
        'ä' => 'a',
2008
        'Ä' => 'A',
2009
        'ç' => 'c',
2010
        'Ç' => 'C',
2011
        'é' => 'e',
2012
        'É' => 'E',
2013
        'ê' => 'e',
2014
        'Ê' => 'E',
2015
        'ë' => 'e',
2016
        'Ë' => 'E',
2017
        'è' => 'e',
2018
        'È' => 'E',
2019
        'ï' => 'i',
2020
        'î' => 'i',
2021
        'Ï' => 'I',
2022
        'Î' => 'I',
2023
        'ÿ' => 'y',
2024
        'Ÿ' => 'Y',
2025
        'ô' => 'o',
2026
        'Ô' => 'O',
2027
        'ö' => 'o',
2028
        'Ö' => 'O',
2029
        'û' => 'u',
2030
        'Û' => 'U',
2031
        'ù' => 'u',
2032
        'Ù' => 'U',
2033
        'ü' => 'u',
2034
        'Ü' => 'U',
2035
    ],
2036
    // German
2037
    'de' => [
2038
        'Ä' => 'Ae',
2039
        'Ö' => 'Oe',
2040
        'Ü' => 'Ue',
2041
        'ä' => 'ae',
2042
        'ö' => 'oe',
2043
        'ü' => 'ue',
2044
        'ß' => 'ss',
2045
        'ẞ' => 'SS',
2046
    ],
2047
    // Austrian (German)
2048
    'de_at' => [
2049
        'Ä' => 'Ae',
2050
        'Ö' => 'Oe',
2051
        'Ü' => 'Ue',
2052
        'ä' => 'ae',
2053
        'ö' => 'oe',
2054
        'ü' => 'ue',
2055
        'ß' => 'sz',
2056
        'ẞ' => 'SZ',
2057
    ],
2058
    // Switzerland (German)
2059
    'de_ch' => [
2060
        'Ä' => 'Ae',
2061
        'Ö' => 'Oe',
2062
        'Ü' => 'Ue',
2063
        'ä' => 'ae',
2064
        'ö' => 'oe',
2065
        'ü' => 'ue',
2066
        'ß' => 'ss',
2067
        'ẞ' => 'SS',
2068
    ],
2069
    // Bengali (Bangla)
2070
    'bn' => [
2071
        'ভ্ল'   => 'vl',
2072
        'পশ'    => 'psh',
2073
        'ব্ধ'   => 'bdh',
2074
        'ব্জ'   => 'bj',
2075
        'ব্দ'   => 'bd',
2076
        'ব্ব'   => 'bb',
2077
        'ব্ল'   => 'bl',
2078
        'ভ'     => 'v',
2079
        'ব'     => 'b',
2080
        'চ্ঞ'   => 'cNG',
2081
        'চ্ছ'   => 'cch',
2082
        'চ্চ'   => 'cc',
2083
        'ছ'     => 'ch',
2084
        'চ'     => 'c',
2085
        'ধ্ন'   => 'dhn',
2086
        'ধ্ম'   => 'dhm',
2087
        'দ্ঘ'   => 'dgh',
2088
        'দ্ধ'   => 'ddh',
2089
        'দ্ভ'   => 'dv',
2090
        'দ্ম'   => 'dm',
2091
        'ড্ড'   => 'DD',
2092
        'ঢ'     => 'Dh',
2093
        'ধ'     => 'dh',
2094
        'দ্গ'   => 'dg',
2095
        'দ্দ'   => 'dd',
2096
        'ড'     => 'D',
2097
        'দ'     => 'd',
2098
        '।'     => '.',
2099
        'ঘ্ন'   => 'Ghn',
2100
        'গ্ধ'   => 'Gdh',
2101
        'গ্ণ'   => 'GN',
2102
        'গ্ন'   => 'Gn',
2103
        'গ্ম'   => 'Gm',
2104
        'গ্ল'   => 'Gl',
2105
        'জ্ঞ'   => 'jNG',
2106
        'ঘ'     => 'Gh',
2107
        'গ'     => 'g',
2108
        'হ্ণ'   => 'hN',
2109
        'হ্ন'   => 'hn',
2110
        'হ্ম'   => 'hm',
2111
        'হ্ল'   => 'hl',
2112
        'হ'     => 'h',
2113
        'জ্ঝ'   => 'jjh',
2114
        'ঝ'     => 'jh',
2115
        'জ্জ'   => 'jj',
2116
        'জ'     => 'j',
2117
        'ক্ষ্ণ' => 'kxN',
2118
        'ক্ষ্ম' => 'kxm',
2119
        'ক্ষ'   => 'ksh',
2120
        'কশ'    => 'ksh',
2121
        'ক্ক'   => 'kk',
2122
        'ক্ট'   => 'kT',
2123
        'ক্ত'   => 'kt',
2124
        'ক্ল'   => 'kl',
2125
        'ক্স'   => 'ks',
2126
        'খ'     => 'kh',
2127
        'ক'     => 'k',
2128
        'ল্ভ'   => 'lv',
2129
        'ল্ধ'   => 'ldh',
2130
        'লখ'    => 'lkh',
2131
        'লঘ'    => 'lgh',
2132
        'লফ'    => 'lph',
2133
        'ল্ক'   => 'lk',
2134
        'ল্গ'   => 'lg',
2135
        'ল্ট'   => 'lT',
2136
        'ল্ড'   => 'lD',
2137
        'ল্প'   => 'lp',
2138
        'ল্ম'   => 'lm',
2139
        'ল্ল'   => 'll',
2140
        'ল্ব'   => 'lb',
2141
        'ল'     => 'l',
2142
        'ম্থ'   => 'mth',
2143
        'ম্ফ'   => 'mf',
2144
        'ম্ভ'   => 'mv',
2145
        'মপ্ল'  => 'mpl',
2146
        'ম্ন'   => 'mn',
2147
        'ম্প'   => 'mp',
2148
        'ম্ম'   => 'mm',
2149
        'ম্ল'   => 'ml',
2150
        'ম্ব'   => 'mb',
2151
        'ম'     => 'm',
2152
        '০'     => '0',
2153
        'à§§'     => '1',
2154
        '২'     => '2',
2155
        'à§©'     => '3',
2156
        '৪'     => '4',
2157
        'à§«'     => '5',
2158
        '৬'     => '6',
2159
        '৭'     => '7',
2160
        'à§®'     => '8',
2161
        '৯'     => '9',
2162
        'ঙ্ক্ষ' => 'Ngkx',
2163
        'ঞ্ছ'   => 'nch',
2164
        'ঙ্ঘ'   => 'ngh',
2165
        'ঙ্খ'   => 'nkh',
2166
        'ঞ্ঝ'   => 'njh',
2167
        'ঙ্গৌ'  => 'ngOU',
2168
        'ঙ্গৈ'  => 'ngOI',
2169
        'ঞ্চ'   => 'nc',
2170
        'ঙ্ক'   => 'nk',
2171
        'ঙ্ষ'   => 'Ngx',
2172
        'ঙ্গ'   => 'ngo',
2173
        'ঙ্ম'   => 'Ngm',
2174
        'ঞ্জ'   => 'nj',
2175
        'ন্ধ'   => 'ndh',
2176
        'ন্ঠ'   => 'nTh',
2177
        'ণ্ঠ'   => 'NTh',
2178
        'ন্থ'   => 'nth',
2179
        'ঙ্গা'  => 'nga',
2180
        'ঙ্গি'  => 'ngi',
2181
        'ঙ্গী'  => 'ngI',
2182
        'ঙ্গু'  => 'ngu',
2183
        'ঙ্গূ'  => 'ngU',
2184
        'ঙ্গে'  => 'nge',
2185
        'ঙ্গো'  => 'ngO',
2186
        'ণ্ঢ'   => 'NDh',
2187
        'নশ'    => 'nsh',
2188
        'ঙর'    => 'Ngr',
2189
        'ঞর'    => 'NGr',
2190
        'ংর'    => 'ngr',
2191
        'ঙ'     => 'Ng',
2192
        'ঞ'     => 'NG',
2193
        'ং'     => 'ng',
2194
        'ন্ন'   => 'nn',
2195
        'ণ্ণ'   => 'NN',
2196
        'ণ্ন'   => 'Nn',
2197
        'ন্ম'   => 'nm',
2198
        'ণ্ম'   => 'Nm',
2199
        'ন্দ'   => 'nd',
2200
        'ন্ট'   => 'nT',
2201
        'ণ্ট'   => 'NT',
2202
        'ন্ড'   => 'nD',
2203
        'ণ্ড'   => 'ND',
2204
        'ন্ত'   => 'nt',
2205
        'ন্স'   => 'ns',
2206
        'ন'     => 'n',
2207
        'ণ'     => 'N',
2208
        'ৈ'     => 'OI',
2209
        'ৌ'     => 'OU',
2210
        'ো'     => 'O',
2211
        'ঐ'     => 'OI',
2212
        'ঔ'     => 'OU',
2213
        'অ'     => 'o',
2214
        'ও'     => 'oo',
2215
        'ফ্ল'   => 'fl',
2216
        'প্ট'   => 'pT',
2217
        'প্ত'   => 'pt',
2218
        'প্ন'   => 'pn',
2219
        'প্প'   => 'pp',
2220
        'প্ল'   => 'pl',
2221
        'প্স'   => 'ps',
2222
        'ফ'     => 'f',
2223
        'প'     => 'p',
2224
        'ৃ'     => 'rri',
2225
        'ঋ'     => 'rri',
2226
        'রর‍্য' => 'rry',
2227
        '্র্য'  => 'ry',
2228
        '্রর'   => 'rr',
2229
        'ড়্গ'   => 'Rg',
2230
        'ঢ়'     => 'Rh',
2231
        'ড়'     => 'R',
2232
        'র'     => 'r',
2233
        '্র'    => 'r',
2234
        'শ্ছ'   => 'Sch',
2235
        'ষ্ঠ'   => 'ShTh',
2236
        'ষ্ফ'   => 'Shf',
2237
        'স্ক্ল' => 'skl',
2238
        'স্খ'   => 'skh',
2239
        'স্থ'   => 'sth',
2240
        'স্ফ'   => 'sf',
2241
        'শ্চ'   => 'Sc',
2242
        'শ্ত'   => 'St',
2243
        'শ্ন'   => 'Sn',
2244
        'শ্ম'   => 'Sm',
2245
        'শ্ল'   => 'Sl',
2246
        'ষ্ক'   => 'Shk',
2247
        'ষ্ট'   => 'ShT',
2248
        'ষ্ণ'   => 'ShN',
2249
        'ষ্প'   => 'Shp',
2250
        'ষ্ম'   => 'Shm',
2251
        'স্প্ল' => 'spl',
2252
        'স্ক'   => 'sk',
2253
        'স্ট'   => 'sT',
2254
        'স্ত'   => 'st',
2255
        'স্ন'   => 'sn',
2256
        'স্প'   => 'sp',
2257
        'স্ম'   => 'sm',
2258
        'স্ল'   => 'sl',
2259
        'শ'     => 'S',
2260
        'ষ'     => 'Sh',
2261
        'স'     => 's',
2262
        'ু'     => 'u',
2263
        'উ'     => 'u',
2264
        'অ্য'   => 'oZ',
2265
        'ত্থ'   => 'tth',
2266
        'à§Ž'     => 'tt',
2267
        'ট্ট'   => 'TT',
2268
        'ট্ম'   => 'Tm',
2269
        'ঠ'     => 'Th',
2270
        'ত্ন'   => 'tn',
2271
        'ত্ম'   => 'tm',
2272
        'থ'     => 'th',
2273
        'ত্ত'   => 'tt',
2274
        'ট'     => 'T',
2275
        'ত'     => 't',
2276
        'অ্যা'  => 'AZ',
2277
        'া'     => 'a',
2278
        'আ'     => 'a',
2279
        'য়া'    => 'ya',
2280
        'য়'     => 'y',
2281
        'ি'     => 'i',
2282
        'ই'     => 'i',
2283
        'à§€'     => 'ee',
2284
        'ঈ'     => 'ee',
2285
        'à§‚'     => 'uu',
2286
        'ঊ'     => 'uu',
2287
        'ে'     => 'e',
2288
        'এ'     => 'e',
2289
        'য'     => 'z',
2290
        '্য'    => 'Z',
2291
        'ইয়'    => 'y',
2292
        'ওয়'    => 'w',
2293
        '্ব'    => 'w',
2294
        'এক্স'  => 'x',
2295
        'ঃ'     => ':',
2296
        'ঁ'     => 'nn',
2297
        '্‌'    => '',
2298
    ],
2299
    // English
2300
    'en' => [
2301
    ],
2302
    // Latin (+ Cyrillic ?) chars
2303
    //
2304
    // -> Mix of languages, but we need to keep this here, so that different languages can handle there own behavior.
2305
    'latin' => [
2306
        'Ëš'   => '0',
2307
        '¹'   => '1',
2308
        '²'   => '2',
2309
        '³'   => '3',
2310
        '⁴'   => '4',
2311
        '⁵'   => '5',
2312
        '⁶'   => '6',
2313
        '⁷'   => '7',
2314
        '⁸'   => '8',
2315
        '⁹'   => '9',
2316
        '₀'   => '0',
2317
        '₁'   => '1',
2318
        '₂'   => '2',
2319
        '₃'   => '3',
2320
        '₄'   => '4',
2321
        '₅'   => '5',
2322
        '₆'   => '6',
2323
        '₇'   => '7',
2324
        '₈'   => '8',
2325
        '₉'   => '9',
2326
        '௦'   => '0',
2327
        '௧'   => '1',
2328
        '௨'   => '2',
2329
        '௩'   => '3',
2330
        '௪'   => '4',
2331
        '௫'   => '5',
2332
        '௬'   => '6',
2333
        '௭'   => '7',
2334
        '௮'   => '8',
2335
        '௯'   => '9',
2336
        '௰'   => '10',
2337
        '௱'   => '100',
2338
        '௲'   => '1000',
2339
        'Ꜳ'   => 'AA',
2340
        'ꜳ'   => 'aa',
2341
        'Æ'   => 'AE',
2342
        'æ'   => 'ae',
2343
        'Ǽ'   => 'AE',
2344
        'ǽ'   => 'ae',
2345
        'Ꜵ'   => 'AO',
2346
        'ꜵ'   => 'ao',
2347
        'Ꜷ'   => 'AU',
2348
        'ꜷ'   => 'au',
2349
        'Ꜹ'   => 'AV',
2350
        'ꜹ'   => 'av',
2351
        'Ꜻ'   => 'av',
2352
        'ꜻ'   => 'av',
2353
        'Ꜽ'   => 'AY',
2354
        'ꜽ'   => 'ay',
2355
        'ȸ'   => 'db',
2356
        'Ê£'   => 'dz',
2357
        'Ê¥'   => 'dz',
2358
        'ʤ'   => 'dezh',
2359
        '🙰'   => 'et',
2360
        'ff'   => 'ff',
2361
        'ffi'   => 'ffi',
2362
        'ffl'   => 'ffl',
2363
        'fi'   => 'fi',
2364
        'fl'   => 'fl',
2365
        'Ê©'   => 'feng',
2366
        'IJ'   => 'IJ',
2367
        'ij'   => 'ij',
2368
        'ʪ'   => 'ls',
2369
        'Ê«'   => 'lz',
2370
        'É®'   => 'lezh',
2371
        'ȹ'   => 'qp',
2372
        'ʨ'   => 'tc',
2373
        'ʦ'   => 'ts',
2374
        'ʧ'   => 'tesh',
2375
        'Å’'   => 'OE',
2376
        'Å“'   => 'oe',
2377
        'Ꝏ'   => 'OO',
2378
        'ꝏ'   => 'oo',
2379
        'ẞ'   => 'SS',
2380
        'ß'   => 'ss',
2381
        'st'   => 'st',
2382
        'ſt'   => 'st',
2383
        'Ꜩ'   => 'TZ',
2384
        'ꜩ'   => 'tz',
2385
        'ᵫ'   => 'ue',
2386
        'Aι'  => 'Ai',
2387
        'αι'  => 'ai',
2388
        'Ει'  => 'Ei',
2389
        'ει'  => 'ei',
2390
        'Οι'  => 'Oi',
2391
        'οι'  => 'oi',
2392
        'Ου'  => 'Oy',
2393
        'ου'  => 'oy',
2394
        'Υι'  => 'Yi',
2395
        'υι'  => 'yi',
2396
        'á¼€'   => 'a',
2397
        'ἁ'   => 'a',
2398
        'ἂ'   => 'a',
2399
        'ἃ'   => 'a',
2400
        'ἄ'   => 'a',
2401
        'á¼…'   => 'a',
2402
        'ἆ'   => 'a',
2403
        'ἇ'   => 'a',
2404
        'Ἀ'   => 'A',
2405
        'Ἁ'   => 'A',
2406
        'Ἂ'   => 'A',
2407
        'Ἃ'   => 'A',
2408
        'Ἄ'   => 'A',
2409
        'Ἅ'   => 'A',
2410
        'Ἆ'   => 'A',
2411
        'Ἇ'   => 'A',
2412
        'á¾°'   => 'a',
2413
        'á¾±'   => 'a',
2414
        'á¾²'   => 'a',
2415
        'á¾³'   => 'a',
2416
        'á¾´'   => 'a',
2417
        'á¾¶'   => 'a',
2418
        'á¾·'   => 'a',
2419
        'Ᾰ'   => 'A',
2420
        'á¾¹'   => 'A',
2421
        'Ὰ'   => 'A',
2422
        'á¾»'   => 'A',
2423
        'á¾¼'   => 'A',
2424
        'Ä'   => 'A',
2425
        'ä'   => 'a',
2426
        'À'   => 'A',
2427
        'à'   => 'a',
2428
        'Á'   => 'A',
2429
        'á'   => 'a',
2430
        'Â'   => 'A',
2431
        'â'   => 'a',
2432
        'Ã'   => 'A',
2433
        'ã'   => 'a',
2434
        'A̧'  => 'A',
2435
        'a̧'  => 'a',
2436
        'Ą'   => 'A',
2437
        'ą'   => 'a',
2438
        'Ⱥ'   => 'A',
2439
        'â±¥'   => 'a',
2440
        'Å'   => 'A',
2441
        'å'   => 'a',
2442
        'Ǻ'   => 'A',
2443
        'Ç»'   => 'a',
2444
        'Ä‚'   => 'A',
2445
        'ă'   => 'a',
2446
        'Ǎ'   => 'A',
2447
        'ÇŽ'   => 'a',
2448
        'Ȧ'   => 'A',
2449
        'ȧ'   => 'a',
2450
        'Ạ'   => 'A',
2451
        'ạ'   => 'a',
2452
        'Ä€'   => 'A',
2453
        'ā'   => 'a',
2454
        'ª'   => 'a',
2455
        'Ɓ'   => 'B',
2456
        'Ѣ'   => 'E',
2457
        'ѣ'   => 'e',
2458
        'Ç'   => 'C',
2459
        'ç'   => 'c',
2460
        'Ĉ'   => 'C',
2461
        'ĉ'   => 'c',
2462
        'C̈'  => 'C',
2463
        'c̈'  => 'c',
2464
        'C̨'  => 'C',
2465
        'c̨'  => 'c',
2466
        'È»'   => 'C',
2467
        'ȼ'   => 'c',
2468
        'Č'   => 'C',
2469
        'č'   => 'c',
2470
        'Ć'   => 'C',
2471
        'ć'   => 'c',
2472
        'C̀'  => 'C',
2473
        'c̀'  => 'c',
2474
        'ÄŠ'   => 'C',
2475
        'ċ'   => 'c',
2476
        'CÌ£'  => 'C',
2477
        'c̣'  => 'c',
2478
        'C̄'  => 'C',
2479
        'c̄'  => 'c',
2480
        'C̃'  => 'C',
2481
        'c̃'  => 'c',
2482
        'Ð'   => 'D',
2483
        'Đ'   => 'D',
2484
        'ð'   => 'd',
2485
        'Ä‘'   => 'd',
2486
        'È'   => 'E',
2487
        'É'   => 'E',
2488
        'Ê'   => 'E',
2489
        'Ë'   => 'E',
2490
        'Ä”'   => 'E',
2491
        'Ä–'   => 'E',
2492
        'Ȩ'   => 'E',
2493
        'È©'   => 'e',
2494
        'Ę'   => 'E',
2495
        'ę'   => 'e',
2496
        'Ɇ'   => 'E',
2497
        'ɇ'   => 'e',
2498
        'Ě'   => 'E',
2499
        'ě'   => 'e',
2500
        'Ẹ'   => 'E',
2501
        'ẹ'   => 'e',
2502
        'Ä’'   => 'E',
2503
        'Ä“'   => 'e',
2504
        'Ẽ'   => 'E',
2505
        'ẽ'   => 'e',
2506
        'è'   => 'e',
2507
        'é'   => 'e',
2508
        'ê'   => 'e',
2509
        'ë'   => 'e',
2510
        'Ä•'   => 'e',
2511
        'Ä—'   => 'e',
2512
        'Æ’'   => 'f',
2513
        'Ѳ'   => 'F',
2514
        'ѳ'   => 'f',
2515
        'Ĝ'   => 'G',
2516
        'Ġ'   => 'G',
2517
        'ĝ'   => 'g',
2518
        'Ä¡'   => 'g',
2519
        'Ĥ'   => 'H',
2520
        'Ħ'   => 'H',
2521
        'Ä¥'   => 'h',
2522
        'ħ'   => 'h',
2523
        'Ì'   => 'I',
2524
        'Í'   => 'I',
2525
        'Î'   => 'I',
2526
        'Ï'   => 'I',
2527
        'Ĩ'   => 'I',
2528
        'Ĭ'   => 'I',
2529
        'Ǐ'   => 'I',
2530
        'Ä®'   => 'I',
2531
        'ì'   => 'i',
2532
        'í'   => 'i',
2533
        'î'   => 'i',
2534
        'ï'   => 'i',
2535
        'Ä©'   => 'i',
2536
        'ĭ'   => 'i',
2537
        'ǐ'   => 'i',
2538
        'į'   => 'i',
2539
        'І'   => 'I',
2540
        'і'   => 'i',
2541
        'I̧'  => 'I',
2542
        'i̧'  => 'i',
2543
        'Æ—'   => 'I',
2544
        'ɨ'   => 'i',
2545
        'İ'   => 'I',
2546
        'i'   => 'i',
2547
        'Ị'   => 'I',
2548
        'ị'   => 'i',
2549
        'Ī'   => 'I',
2550
        'Ä«'   => 'i',
2551
        'Ä´'   => 'J',
2552
        'ĵ'   => 'j',
2553
        'J́́' => 'J',
2554
        'j́'  => 'j',
2555
        'J̀̀' => 'J',
2556
        'j̀'  => 'j',
2557
        'J̈'  => 'J',
2558
        'j̈'  => 'j',
2559
        'J̧'  => 'J',
2560
        'j̧'  => 'j',
2561
        'J̨'  => 'J',
2562
        'j̨'  => 'j',
2563
        'Ɉ'   => 'J',
2564
        'ɉ'   => 'j',
2565
        'J̌'  => 'J',
2566
        'ǰ'   => 'j',
2567
        'J̇'  => 'J',
2568
        'j'   => 'j',
2569
        'JÌ£'  => 'J',
2570
        'j̣'  => 'j',
2571
        'J̄'  => 'J',
2572
        'j̄'  => 'j',
2573
        'J̃'  => 'J',
2574
        'j̃'  => 'j',
2575
        'Й' => 'i',
2576
        'й' => 'i',
2577
        'ĸ'   => 'k',
2578
        'Ĺ'   => 'L',
2579
        'Ľ'   => 'L',
2580
        'Ä¿'   => 'L',
2581
        'ĺ'   => 'l',
2582
        'ľ'   => 'l',
2583
        'Å€'   => 'l',
2584
        'L̀'  => 'L',
2585
        'l̀'  => 'l',
2586
        'L̂'  => 'L',
2587
        'l̂'  => 'l',
2588
        'L̈'  => 'L',
2589
        'l̈'  => 'l',
2590
        'Ä»'   => 'L',
2591
        'ļ'   => 'l',
2592
        'L̨'  => 'L',
2593
        'l̨'  => 'l',
2594
        'Ł'   => 'L',
2595
        'ł'   => 'l',
2596
        'Ƚ'   => 'L',
2597
        'Æš'   => 'l',
2598
        'L̇'  => 'L',
2599
        'l̇'  => 'l',
2600
        'Ḷ'   => 'L',
2601
        'ḷ'   => 'l',
2602
        'L̄'  => 'L',
2603
        'l̄'  => 'l',
2604
        'L̃'  => 'L',
2605
        'l̃'  => 'l',
2606
        'Ñ'   => 'N',
2607
        'ñ'   => 'n',
2608
        'ÅŠ'   => 'N',
2609
        'ŋ'   => 'n',
2610
        'ʼn'   => 'n',
2611
        'Ń'   => 'N',
2612
        'ń'   => 'n',
2613
        'Ǹ'   => 'N',
2614
        'ǹ'   => 'n',
2615
        'N̂'  => 'N',
2616
        'n̂'  => 'n',
2617
        'N̈'  => 'N',
2618
        'n̈'  => 'n',
2619
        'Å…'   => 'N',
2620
        'ņ'   => 'n',
2621
        'N̨'  => 'N',
2622
        'n̨'  => 'n',
2623
        'Ꞥ'   => 'N',
2624
        'ꞥ'   => 'n',
2625
        'Ň'   => 'N',
2626
        'ň'   => 'n',
2627
        'Ṅ'   => 'N',
2628
        'á¹…'   => 'n',
2629
        'Ṇ'   => 'N',
2630
        'ṇ'   => 'n',
2631
        'N̄'  => 'N',
2632
        'n̄'  => 'n',
2633
        'Ö'   => 'O',
2634
        'Ò'   => 'O',
2635
        'Ó'   => 'O',
2636
        'Ô'   => 'O',
2637
        'Õ'   => 'O',
2638
        'Ō'   => 'O',
2639
        'ÅŽ'   => 'O',
2640
        'Ç‘'   => 'O',
2641
        'Ő'   => 'O',
2642
        'Ơ'   => 'O',
2643
        'Ø'   => 'O',
2644
        'Ǿ'   => 'O',
2645
        'ö'   => 'o',
2646
        'ò'   => 'o',
2647
        'ó'   => 'o',
2648
        'ô'   => 'o',
2649
        'õ'   => 'o',
2650
        'ō'   => 'o',
2651
        'ŏ'   => 'o',
2652
        'Ç’'   => 'o',
2653
        'ő'   => 'o',
2654
        'Æ¡'   => 'o',
2655
        'ø'   => 'o',
2656
        'Ç¿'   => 'o',
2657
        'º'   => 'o',
2658
        'O̧'  => 'O',
2659
        'o̧'  => 'o',
2660
        'Ǫ'   => 'O',
2661
        'Ç«'   => 'o',
2662
        'Ɵ'   => 'O',
2663
        'ɵ'   => 'o',
2664
        'È®'   => 'O',
2665
        'ȯ'   => 'o',
2666
        'Ọ'   => 'O',
2667
        'ọ'   => 'o',
2668
        'Å”'   => 'R',
2669
        'Å–'   => 'R',
2670
        'Å•'   => 'r',
2671
        'Å—'   => 'r',
2672
        'Ŝ'   => 'S',
2673
        'Ș'   => 'S',
2674
        'ș'   => 's',
2675
        'Ś'   => 'S',
2676
        'ś'   => 's',
2677
        'S̀'  => 'S',
2678
        's̀'  => 's',
2679
        'Ŝ̀'  => 'S',
2680
        'ŝ'   => 's',
2681
        'S̈'  => 'S',
2682
        's̈'  => 's',
2683
        'Åž'   => 'S',
2684
        'ş'   => 's',
2685
        'S̨'  => 'S',
2686
        's̨'  => 's',
2687
        'Ꞩ'   => 'S',
2688
        'êž©'   => 's',
2689
        'Š'   => 'S',
2690
        'š'   => 's',
2691
        'Ṡ'   => 'S',
2692
        'ṡ'   => 's',
2693
        'á¹¢'   => 'S',
2694
        'á¹£'   => 's',
2695
        'S̄'  => 'S',
2696
        's̄'  => 's',
2697
        'S̃'  => 'S',
2698
        's̃'  => 's',
2699
        'Å¿'   => 's',
2700
        'Å¢'   => 'T',
2701
        'Èš'   => 'T',
2702
        'Ŧ'   => 'T',
2703
        'Þ'   => 'TH',
2704
        'Å£'   => 't',
2705
        'ț'   => 't',
2706
        'ŧ'   => 't',
2707
        'þ'   => 'th',
2708
        'T́'  => 'T',
2709
        't́'  => 't',
2710
        'T̀'  => 'T',
2711
        't̀'  => 't',
2712
        'T̂'  => 'T',
2713
        't̂'  => 't',
2714
        'T̈'  => 'T',
2715
        'ẗ'   => 't',
2716
        'T̨'  => 'T',
2717
        't̨'  => 't',
2718
        'Ⱦ'   => 'T',
2719
        'ⱦ'   => 't',
2720
        'Ť'   => 'T',
2721
        'ť'   => 't',
2722
        'Ṫ'   => 'T',
2723
        'ṫ'   => 't',
2724
        'Ṭ'   => 'T',
2725
        'ṭ'   => 't',
2726
        'T̄'  => 'T',
2727
        't̄'  => 't',
2728
        'T̃'  => 'T',
2729
        't̃'  => 't',
2730
        'Ü'   => 'U',
2731
        'Ù'   => 'U',
2732
        'Ú'   => 'U',
2733
        'Û'   => 'U',
2734
        'Ũ'   => 'U',
2735
        'Ŭ'   => 'U',
2736
        'Ű'   => 'U',
2737
        'Ų'   => 'U',
2738
        'Ư'   => 'U',
2739
        'Ç“'   => 'U',
2740
        'Ç•'   => 'U',
2741
        'Ç—'   => 'U',
2742
        'Ǚ'   => 'U',
2743
        'Ǜ'   => 'U',
2744
        'ü'   => 'u',
2745
        'ù'   => 'u',
2746
        'ú'   => 'u',
2747
        'û'   => 'u',
2748
        'Å©'   => 'u',
2749
        'ŭ'   => 'u',
2750
        'ű'   => 'u',
2751
        'ų'   => 'u',
2752
        'ư'   => 'u',
2753
        'Ç”'   => 'u',
2754
        'Ç–'   => 'u',
2755
        'ǘ'   => 'u',
2756
        'Çš'   => 'u',
2757
        'ǜ'   => 'u',
2758
        'U̧'  => 'U',
2759
        'u̧'  => 'u',
2760
        'É„'   => 'U',
2761
        'ʉ'   => 'u',
2762
        'U̇'  => 'U',
2763
        'u̇'  => 'u',
2764
        'Ụ'   => 'U',
2765
        'ụ'   => 'u',
2766
        'Ū'   => 'U',
2767
        'Å«'   => 'u',
2768
        'Ʊ'   => 'U',
2769
        'ÊŠ'   => 'u',
2770
        'Å´'   => 'W',
2771
        'ŵ'   => 'w',
2772
        'Ẁ'   => 'W',
2773
        'ẁ'   => 'w',
2774
        'Ẃ'   => 'W',
2775
        'ẃ'   => 'w',
2776
        'Ẅ'   => 'W',
2777
        'ẅ'   => 'w',
2778
        'Ѵ'   => 'I',
2779
        'ѵ'   => 'i',
2780
        'Ꙗ'   => 'Ja',
2781
        'ꙗ'   => 'ja',
2782
        'Є'   => 'Je',
2783
        'є'   => 'je',
2784
        'Ѥ'   => 'Je',
2785
        'ѥ'   => 'je',
2786
        'Ѕ'   => 'Dz',
2787
        'ѕ'   => 'dz',
2788
        'Ꙋ'   => 'U',
2789
        'ꙋ'   => 'u',
2790
        'Ѡ'   => 'O',
2791
        'ѡ'   => 'o',
2792
        'Ѿ'   => 'Ot',
2793
        'ѿ'   => 'ot',
2794
        'Ѫ'   => 'U',
2795
        'ѫ'   => 'u',
2796
        'Ѧ'   => 'Ja',
2797
        'ѧ'   => 'ja',
2798
        'Ѭ'   => 'Ju',
2799
        'ѭ'   => 'ju',
2800
        'Ѩ'   => 'Ja',
2801
        'ѩ'   => 'Ja',
2802
        'Ѯ'   => 'Ks',
2803
        'ѯ'   => 'ks',
2804
        'Ѱ'   => 'Ps',
2805
        'ѱ'   => 'ps',
2806
        'Х'   => 'X',
2807
        'х'   => 'x',
2808
        'Ý'   => 'Y',
2809
        'Ÿ'   => 'Y',
2810
        'Ŷ'   => 'Y',
2811
        'ý'   => 'y',
2812
        'ÿ'   => 'y',
2813
        'Å·'   => 'y',
2814
        'Ỳ'   => 'Y',
2815
        'ỳ'   => 'y',
2816
        'Y̧'  => 'Y',
2817
        'y̧'  => 'y',
2818
        'Y̨'  => 'Y',
2819
        'y̨'  => 'y',
2820
        'ÉŽ'   => 'Y',
2821
        'ɏ'   => 'y',
2822
        'Y̌'  => 'Y',
2823
        'y̌'  => 'y',
2824
        'Ẏ'   => 'Y',
2825
        'ẏ'   => 'y',
2826
        'á»´'   => 'Y',
2827
        'ỵ'   => 'y',
2828
        'Ȳ'   => 'Y',
2829
        'ȳ'   => 'y',
2830
        'Ỹ'   => 'Y',
2831
        'ỹ'   => 'y',
2832
        'Щ' => 'Shh',
2833
        'щ' => 'shh',
2834
        'Ź'   => 'Z',
2835
        'ź'   => 'z',
2836
        'Z̀'  => 'Z',
2837
        'z̀'  => 'z',
2838
        'Ẑ'   => 'Z',
2839
        'ẑ'   => 'z',
2840
        'Z̈'  => 'Z',
2841
        'z̈'  => 'z',
2842
        'Z̧'  => 'Z',
2843
        'z̧'  => 'z',
2844
        'Z̨'  => 'Z',
2845
        'z̨'  => 'z',
2846
        'Ƶ'   => 'Z',
2847
        'ƶ'   => 'z',
2848
        'Ž'   => 'Z',
2849
        'ž'   => 'z',
2850
        'Ż'   => 'Z',
2851
        'ż'   => 'z',
2852
        'Ẓ'   => 'Z',
2853
        'ẓ'   => 'z',
2854
        'Z̄'  => 'Z',
2855
        'z̄'  => 'z',
2856
        'Z̃'  => 'Z',
2857
        'z̃'  => 'z',
2858
    ],
2859
    // whitespace chars
2860
    ' ' => [
2861
        "\xc2\xa0"     => ' ', // 'NO-BREAK SPACE'
2862
        "\xe1\x9a\x80" => ' ', // 'OGHAM SPACE MARK'
2863
        "\xe2\x80\x80" => ' ', // 'EN QUAD'
2864
        "\xe2\x80\x81" => ' ', // 'EM QUAD'
2865
        "\xe2\x80\x82" => ' ', // 'EN SPACE'
2866
        "\xe2\x80\x83" => ' ', // 'EM SPACE'
2867
        "\xe2\x80\x84" => ' ', // 'THREE-PER-EM SPACE'
2868
        "\xe2\x80\x85" => ' ', // 'FOUR-PER-EM SPACE'
2869
        "\xe2\x80\x86" => ' ', // 'SIX-PER-EM SPACE'
2870
        "\xe2\x80\x87" => ' ', // 'FIGURE SPACE'
2871
        "\xe2\x80\x88" => ' ', // 'PUNCTUATION SPACE'
2872
        "\xe2\x80\x89" => ' ', // 'THIN SPACE'
2873
        "\xe2\x80\x8a" => ' ', // 'HAIR SPACE'
2874
        "\xe2\x80\xa8" => ' ', // 'LINE SEPARATOR'
2875
        "\xe2\x80\xa9" => ' ', // 'PARAGRAPH SEPARATOR'
2876
        "\xe2\x80\x8b" => ' ', // 'ZERO WIDTH SPACE'
2877
        "\xe2\x80\xaf" => ' ', // 'NARROW NO-BREAK SPACE'
2878
        "\xe2\x81\x9f" => ' ', // 'MEDIUM MATHEMATICAL SPACE'
2879
        "\xe3\x80\x80" => ' ', // 'IDEOGRAPHIC SPACE'
2880
        "\xef\xbe\xa0" => ' ', // 'HALFWIDTH HANGUL FILLER'
2881
    ],
2882
    // commonly used in Word documents
2883
    'msword' => [
2884
        "\xc2\xab"     => '<<', // « (U+00AB) in UTF-8
2885
        "\xc2\xbb"     => '>>', // » (U+00BB) in UTF-8
2886
        "\xe2\x80\x98" => "'", // ‘ (U+2018) in UTF-8
2887
        "\xe2\x80\x99" => "'", // ’ (U+2019) in UTF-8
2888
        "\xe2\x80\x9a" => "'", // ‚ (U+201A) in UTF-8
2889
        "\xe2\x80\x9b" => "'", // ‛ (U+201B) in UTF-8
2890
        "\xe2\x80\x9c" => '"', // “ (U+201C) in UTF-8
2891
        "\xe2\x80\x9d" => '"', // ” (U+201D) in UTF-8
2892
        "\xe2\x80\x9e" => '"', // „ (U+201E) in UTF-8
2893
        "\xe2\x80\x9f" => '"', // ‟ (U+201F) in UTF-8
2894
        "\xe2\x80\xb9" => "'", // ‹ (U+2039) in UTF-8
2895
        "\xe2\x80\xba" => "'", // › (U+203A) in UTF-8
2896
        "\xe2\x80\x93" => '-', // – (U+2013) in UTF-8
2897
        "\xe2\x80\x94" => '-', // — (U+2014) in UTF-8
2898
        "\xe2\x80\xa6" => '...', // … (U+2026) in UTF-8
2899
    ],
2900
    // Currency
2901
    //
2902
    // url => https://en.wikipedia.org/wiki/Currency_symbol
2903
    'currency_short' => [
2904
        '€'  => 'EUR',
2905
        '$'  => '$',
2906
        'â‚¢'  => 'Cr',
2907
        'â‚£'  => 'Fr.',
2908
        '£'  => 'PS',
2909
        '₤'  => 'L.',
2910
        'ℳ'  => 'M',
2911
        'â‚¥'  => 'mil',
2912
        '₦'  => 'N',
2913
        'â‚§'  => 'Pts',
2914
        '₨'  => 'Rs',
2915
        'රු' => 'LKR',
2916
        'ரூ' => 'LKR',
2917
        '௹'  => 'Rs',
2918
        'रू' => 'NPR',
2919
        '₹'  => 'Rs',
2920
        '૱'  => 'Rs',
2921
        'â‚©'  => 'W',
2922
        '₪'  => 'NS',
2923
        '₸'  => 'KZT',
2924
        'â‚«'  => 'D',
2925
        '֏'  => 'AMD',
2926
        '₭'  => 'K',
2927
        '₺'  => 'TL',
2928
        '₼'  => 'AZN',
2929
        'â‚®'  => 'T',
2930
        '₯'  => 'Dr',
2931
        '₲'  => 'PYG',
2932
        '₾'  => 'GEL',
2933
        '₳'  => 'ARA',
2934
        'â‚´'  => 'UAH',
2935
        '₽'  => 'RUB',
2936
        '₵'  => 'GHS',
2937
        'â‚¡'  => 'CL',
2938
        '¢'  => 'c',
2939
        '¥'  => 'YEN',
2940
        '円'  => 'JPY',
2941
        'à§³'  => 'BDT',
2942
        '元'  => 'CNY',
2943
        'ï·¼'  => 'SAR',
2944
        '៛'  => 'KR',
2945
        '₠'  => 'ECU',
2946
        '¤'  => '$?',
2947
        '฿'  => 'THB',
2948
        '؋'  => 'AFN',
2949
    ],
2950
];