1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* Created by gerk on 30.11.17 05:56 |
4
|
|
|
*/ |
5
|
|
|
|
6
|
|
|
namespace PeekAndPoke\Component\Psi\Psi\Str; |
7
|
|
|
|
8
|
|
|
use PeekAndPoke\Component\Psi\UnaryFunction; |
9
|
|
|
|
10
|
|
|
/** |
11
|
|
|
* Replaces all special characters with their "normal" form |
12
|
|
|
* |
13
|
|
|
* Special thanks to WORDPRESS and their authors for this... things were taken from there. |
14
|
|
|
* |
15
|
|
|
* @author Karsten J. Gerber <[email protected]> |
16
|
|
|
*/ |
17
|
|
|
class WithoutAccents implements UnaryFunction |
18
|
|
|
{ |
19
|
|
|
/** @var string[]|null */ |
20
|
|
|
private static $utf8Map; |
21
|
|
|
/** @var string[]|null */ |
22
|
|
|
private static $isoIn; |
23
|
|
|
/** @var string[]|null */ |
24
|
|
|
private static $isoOut; |
25
|
|
|
|
26
|
631 |
|
public function __invoke($input) |
27
|
|
|
{ |
28
|
631 |
|
if ($input === null || ! is_scalar($input)) { |
29
|
1 |
|
return null; |
30
|
|
|
} |
31
|
|
|
|
32
|
|
|
// nothing to replace? |
33
|
630 |
|
if (! preg_match('/[\x80-\xff]/', $input)) { |
34
|
6 |
|
return $input; |
35
|
|
|
} |
36
|
|
|
|
37
|
624 |
|
if ($this->seemsUtf8($input)) { |
38
|
623 |
|
return $this->removeAccentsUtf8($input); |
39
|
|
|
} |
40
|
|
|
|
41
|
|
|
// Assume ISO-8859-1 if not UTF-8 |
42
|
1 |
|
return $this->removeAccentsIso($input); |
43
|
|
|
} |
44
|
|
|
|
45
|
|
|
/** |
46
|
|
|
* @return string[] |
47
|
|
|
*/ |
48
|
1 |
|
public static function getUtf8Map() |
49
|
|
|
{ |
50
|
1 |
|
self::setupUtf8Map(); |
51
|
|
|
|
52
|
1 |
|
return self::$utf8Map; |
53
|
|
|
} |
54
|
|
|
|
55
|
624 |
|
private function seemsUtf8($string) |
56
|
|
|
{ |
57
|
624 |
|
return mb_detect_encoding($string, 'UTF-8', true) === 'UTF-8'; |
58
|
|
|
} |
59
|
|
|
|
60
|
623 |
|
private function removeAccentsUtf8($input) |
61
|
|
|
{ |
62
|
623 |
|
self::setupUtf8Map(); |
63
|
|
|
|
64
|
623 |
|
return strtr($input, self::$utf8Map); |
|
|
|
|
65
|
|
|
} |
66
|
|
|
|
67
|
1 |
|
private function removeAccentsIso($string) |
68
|
|
|
{ |
69
|
1 |
|
self::setupIsoMap(); |
70
|
|
|
|
71
|
1 |
|
$string = strtr($string, self::$isoIn, self::$isoOut); |
|
|
|
|
72
|
1 |
|
$doubleIn = [chr(140), chr(156), chr(198), chr(208), chr(222), chr(223), chr(230), chr(240), chr(254)]; |
73
|
1 |
|
$doubleOut = ['OE', 'oe', 'AE', 'DH', 'TH', 'ss', 'ae', 'dh', 'th']; |
74
|
|
|
|
75
|
1 |
|
return str_replace($doubleIn, $doubleOut, $string); |
76
|
|
|
} |
77
|
|
|
|
78
|
|
|
/** |
79
|
|
|
* @codeCoverageIgnore |
80
|
|
|
*/ |
81
|
|
|
public static function setupIsoMap() |
82
|
|
|
{ |
83
|
|
|
if (self::$isoIn !== null) { |
84
|
|
|
return; |
85
|
|
|
} |
86
|
|
|
|
87
|
|
|
self::$isoIn = |
88
|
|
|
chr(128) . chr(131) . chr(138) . chr(142) . chr(154) . chr(158) |
|
|
|
|
89
|
|
|
. chr(159) . chr(162) . chr(165) . chr(181) . chr(192) . chr(193) . chr(194) |
90
|
|
|
. chr(195) . chr(196) . chr(197) . chr(199) . chr(200) . chr(201) . chr(202) |
91
|
|
|
. chr(203) . chr(204) . chr(205) . chr(206) . chr(207) . chr(209) . chr(210) |
92
|
|
|
. chr(211) . chr(212) . chr(213) . chr(214) . chr(216) . chr(217) . chr(218) |
93
|
|
|
. chr(219) . chr(220) . chr(221) . chr(224) . chr(225) . chr(226) . chr(227) |
94
|
|
|
. chr(228) . chr(229) . chr(231) . chr(232) . chr(233) . chr(234) . chr(235) |
95
|
|
|
. chr(236) . chr(237) . chr(238) . chr(239) . chr(241) . chr(242) . chr(243) |
96
|
|
|
. chr(244) . chr(245) . chr(246) . chr(248) . chr(249) . chr(250) . chr(251) |
97
|
|
|
. chr(252) . chr(253) . chr(255); |
98
|
|
|
|
99
|
|
|
self::$isoOut = 'EfSZszYcYuAAAAAACEEEEIIIINOOOOOOUUUUYaaaaaaceeeeiiiinoooooouuuuyy'; |
|
|
|
|
100
|
|
|
} |
101
|
|
|
|
102
|
|
|
/** |
103
|
|
|
* @codeCoverageIgnore |
104
|
|
|
*/ |
105
|
|
|
public static function setupUtf8Map() |
106
|
|
|
{ |
107
|
|
|
if (self::$utf8Map !== null) { |
108
|
|
|
return; |
109
|
|
|
} |
110
|
|
|
|
111
|
|
|
self::$utf8Map = [ |
112
|
|
|
// Decompositions for Latin-1 Supplement |
113
|
|
|
chr(194) . chr(170) => 'a', |
114
|
|
|
chr(194) . chr(186) => 'o', |
115
|
|
|
chr(195) . chr(128) => 'A', |
116
|
|
|
chr(195) . chr(129) => 'A', |
117
|
|
|
chr(195) . chr(130) => 'A', |
118
|
|
|
chr(195) . chr(131) => 'A', |
119
|
|
|
chr(195) . chr(132) => 'A', |
120
|
|
|
chr(195) . chr(133) => 'A', |
121
|
|
|
chr(195) . chr(134) => 'AE', |
122
|
|
|
chr(195) . chr(135) => 'C', |
123
|
|
|
chr(195) . chr(136) => 'E', |
124
|
|
|
chr(195) . chr(137) => 'E', |
125
|
|
|
chr(195) . chr(138) => 'E', |
126
|
|
|
chr(195) . chr(139) => 'E', |
127
|
|
|
chr(195) . chr(140) => 'I', |
128
|
|
|
chr(195) . chr(141) => 'I', |
129
|
|
|
chr(195) . chr(142) => 'I', |
130
|
|
|
chr(195) . chr(143) => 'I', |
131
|
|
|
chr(195) . chr(144) => 'D', |
132
|
|
|
chr(195) . chr(145) => 'N', |
133
|
|
|
chr(195) . chr(146) => 'O', |
134
|
|
|
chr(195) . chr(147) => 'O', |
135
|
|
|
chr(195) . chr(148) => 'O', |
136
|
|
|
chr(195) . chr(149) => 'O', |
137
|
|
|
chr(195) . chr(150) => 'O', |
138
|
|
|
chr(195) . chr(153) => 'U', |
139
|
|
|
chr(195) . chr(154) => 'U', |
140
|
|
|
chr(195) . chr(155) => 'U', |
141
|
|
|
chr(195) . chr(156) => 'U', |
142
|
|
|
chr(195) . chr(157) => 'Y', |
143
|
|
|
chr(195) . chr(158) => 'TH', |
144
|
|
|
chr(195) . chr(159) => 's', |
145
|
|
|
chr(195) . chr(160) => 'a', |
146
|
|
|
chr(195) . chr(161) => 'a', |
147
|
|
|
chr(195) . chr(162) => 'a', |
148
|
|
|
chr(195) . chr(163) => 'a', |
149
|
|
|
chr(195) . chr(164) => 'a', |
150
|
|
|
chr(195) . chr(165) => 'a', |
151
|
|
|
chr(195) . chr(166) => 'ae', |
152
|
|
|
chr(195) . chr(167) => 'c', |
153
|
|
|
chr(195) . chr(168) => 'e', |
154
|
|
|
chr(195) . chr(169) => 'e', |
155
|
|
|
chr(195) . chr(170) => 'e', |
156
|
|
|
chr(195) . chr(171) => 'e', |
157
|
|
|
chr(195) . chr(172) => 'i', |
158
|
|
|
chr(195) . chr(173) => 'i', |
159
|
|
|
chr(195) . chr(174) => 'i', |
160
|
|
|
chr(195) . chr(175) => 'i', |
161
|
|
|
chr(195) . chr(176) => 'd', |
162
|
|
|
chr(195) . chr(177) => 'n', |
163
|
|
|
chr(195) . chr(178) => 'o', |
164
|
|
|
chr(195) . chr(179) => 'o', |
165
|
|
|
chr(195) . chr(180) => 'o', |
166
|
|
|
chr(195) . chr(181) => 'o', |
167
|
|
|
chr(195) . chr(182) => 'o', |
168
|
|
|
chr(195) . chr(184) => 'o', |
169
|
|
|
chr(195) . chr(185) => 'u', |
170
|
|
|
chr(195) . chr(186) => 'u', |
171
|
|
|
chr(195) . chr(187) => 'u', |
172
|
|
|
chr(195) . chr(188) => 'u', |
173
|
|
|
chr(195) . chr(189) => 'y', |
174
|
|
|
chr(195) . chr(190) => 'th', |
175
|
|
|
chr(195) . chr(191) => 'y', |
176
|
|
|
chr(195) . chr(152) => 'O', |
177
|
|
|
// Decompositions for Latin Extended-A |
178
|
|
|
chr(196) . chr(128) => 'A', |
179
|
|
|
chr(196) . chr(129) => 'a', |
180
|
|
|
chr(196) . chr(130) => 'A', |
181
|
|
|
chr(196) . chr(131) => 'a', |
182
|
|
|
chr(196) . chr(132) => 'A', |
183
|
|
|
chr(196) . chr(133) => 'a', |
184
|
|
|
chr(196) . chr(134) => 'C', |
185
|
|
|
chr(196) . chr(135) => 'c', |
186
|
|
|
chr(196) . chr(136) => 'C', |
187
|
|
|
chr(196) . chr(137) => 'c', |
188
|
|
|
chr(196) . chr(138) => 'C', |
189
|
|
|
chr(196) . chr(139) => 'c', |
190
|
|
|
chr(196) . chr(140) => 'C', |
191
|
|
|
chr(196) . chr(141) => 'c', |
192
|
|
|
chr(196) . chr(142) => 'D', |
193
|
|
|
chr(196) . chr(143) => 'd', |
194
|
|
|
chr(196) . chr(144) => 'D', |
195
|
|
|
chr(196) . chr(145) => 'd', |
196
|
|
|
chr(196) . chr(146) => 'E', |
197
|
|
|
chr(196) . chr(147) => 'e', |
198
|
|
|
chr(196) . chr(148) => 'E', |
199
|
|
|
chr(196) . chr(149) => 'e', |
200
|
|
|
chr(196) . chr(150) => 'E', |
201
|
|
|
chr(196) . chr(151) => 'e', |
202
|
|
|
chr(196) . chr(152) => 'E', |
203
|
|
|
chr(196) . chr(153) => 'e', |
204
|
|
|
chr(196) . chr(154) => 'E', |
205
|
|
|
chr(196) . chr(155) => 'e', |
206
|
|
|
chr(196) . chr(156) => 'G', |
207
|
|
|
chr(196) . chr(157) => 'g', |
208
|
|
|
chr(196) . chr(158) => 'G', |
209
|
|
|
chr(196) . chr(159) => 'g', |
210
|
|
|
chr(196) . chr(160) => 'G', |
211
|
|
|
chr(196) . chr(161) => 'g', |
212
|
|
|
chr(196) . chr(162) => 'G', |
213
|
|
|
chr(196) . chr(163) => 'g', |
214
|
|
|
chr(196) . chr(164) => 'H', |
215
|
|
|
chr(196) . chr(165) => 'h', |
216
|
|
|
chr(196) . chr(166) => 'H', |
217
|
|
|
chr(196) . chr(167) => 'h', |
218
|
|
|
chr(196) . chr(168) => 'I', |
219
|
|
|
chr(196) . chr(169) => 'i', |
220
|
|
|
chr(196) . chr(170) => 'I', |
221
|
|
|
chr(196) . chr(171) => 'i', |
222
|
|
|
chr(196) . chr(172) => 'I', |
223
|
|
|
chr(196) . chr(173) => 'i', |
224
|
|
|
chr(196) . chr(174) => 'I', |
225
|
|
|
chr(196) . chr(175) => 'i', |
226
|
|
|
chr(196) . chr(176) => 'I', |
227
|
|
|
chr(196) . chr(177) => 'i', |
228
|
|
|
chr(196) . chr(178) => 'IJ', |
229
|
|
|
chr(196) . chr(179) => 'ij', |
230
|
|
|
chr(196) . chr(180) => 'J', |
231
|
|
|
chr(196) . chr(181) => 'j', |
232
|
|
|
chr(196) . chr(182) => 'K', |
233
|
|
|
chr(196) . chr(183) => 'k', |
234
|
|
|
chr(196) . chr(184) => 'k', |
235
|
|
|
chr(196) . chr(185) => 'L', |
236
|
|
|
chr(196) . chr(186) => 'l', |
237
|
|
|
chr(196) . chr(187) => 'L', |
238
|
|
|
chr(196) . chr(188) => 'l', |
239
|
|
|
chr(196) . chr(189) => 'L', |
240
|
|
|
chr(196) . chr(190) => 'l', |
241
|
|
|
chr(196) . chr(191) => 'L', |
242
|
|
|
chr(197) . chr(128) => 'l', |
243
|
|
|
chr(197) . chr(129) => 'L', |
244
|
|
|
chr(197) . chr(130) => 'l', |
245
|
|
|
chr(197) . chr(131) => 'N', |
246
|
|
|
chr(197) . chr(132) => 'n', |
247
|
|
|
chr(197) . chr(133) => 'N', |
248
|
|
|
chr(197) . chr(134) => 'n', |
249
|
|
|
chr(197) . chr(135) => 'N', |
250
|
|
|
chr(197) . chr(136) => 'n', |
251
|
|
|
chr(197) . chr(137) => 'N', |
252
|
|
|
chr(197) . chr(138) => 'n', |
253
|
|
|
chr(197) . chr(139) => 'N', |
254
|
|
|
chr(197) . chr(140) => 'O', |
255
|
|
|
chr(197) . chr(141) => 'o', |
256
|
|
|
chr(197) . chr(142) => 'O', |
257
|
|
|
chr(197) . chr(143) => 'o', |
258
|
|
|
chr(197) . chr(144) => 'O', |
259
|
|
|
chr(197) . chr(145) => 'o', |
260
|
|
|
chr(197) . chr(146) => 'OE', |
261
|
|
|
chr(197) . chr(147) => 'oe', |
262
|
|
|
chr(197) . chr(148) => 'R', |
263
|
|
|
chr(197) . chr(149) => 'r', |
264
|
|
|
chr(197) . chr(150) => 'R', |
265
|
|
|
chr(197) . chr(151) => 'r', |
266
|
|
|
chr(197) . chr(152) => 'R', |
267
|
|
|
chr(197) . chr(153) => 'r', |
268
|
|
|
chr(197) . chr(154) => 'S', |
269
|
|
|
chr(197) . chr(155) => 's', |
270
|
|
|
chr(197) . chr(156) => 'S', |
271
|
|
|
chr(197) . chr(157) => 's', |
272
|
|
|
chr(197) . chr(158) => 'S', |
273
|
|
|
chr(197) . chr(159) => 's', |
274
|
|
|
chr(197) . chr(160) => 'S', |
275
|
|
|
chr(197) . chr(161) => 's', |
276
|
|
|
chr(197) . chr(162) => 'T', |
277
|
|
|
chr(197) . chr(163) => 't', |
278
|
|
|
chr(197) . chr(164) => 'T', |
279
|
|
|
chr(197) . chr(165) => 't', |
280
|
|
|
chr(197) . chr(166) => 'T', |
281
|
|
|
chr(197) . chr(167) => 't', |
282
|
|
|
chr(197) . chr(168) => 'U', |
283
|
|
|
chr(197) . chr(169) => 'u', |
284
|
|
|
chr(197) . chr(170) => 'U', |
285
|
|
|
chr(197) . chr(171) => 'u', |
286
|
|
|
chr(197) . chr(172) => 'U', |
287
|
|
|
chr(197) . chr(173) => 'u', |
288
|
|
|
chr(197) . chr(174) => 'U', |
289
|
|
|
chr(197) . chr(175) => 'u', |
290
|
|
|
chr(197) . chr(176) => 'U', |
291
|
|
|
chr(197) . chr(177) => 'u', |
292
|
|
|
chr(197) . chr(178) => 'U', |
293
|
|
|
chr(197) . chr(179) => 'u', |
294
|
|
|
chr(197) . chr(180) => 'W', |
295
|
|
|
chr(197) . chr(181) => 'w', |
296
|
|
|
chr(197) . chr(182) => 'Y', |
297
|
|
|
chr(197) . chr(183) => 'y', |
298
|
|
|
chr(197) . chr(184) => 'Y', |
299
|
|
|
chr(197) . chr(185) => 'Z', |
300
|
|
|
chr(197) . chr(186) => 'z', |
301
|
|
|
chr(197) . chr(187) => 'Z', |
302
|
|
|
chr(197) . chr(188) => 'z', |
303
|
|
|
chr(197) . chr(189) => 'Z', |
304
|
|
|
chr(197) . chr(190) => 'z', |
305
|
|
|
chr(197) . chr(191) => 's', |
306
|
|
|
// Decompositions for Latin Extended-B |
307
|
|
|
chr(200) . chr(152) => 'S', |
308
|
|
|
chr(200) . chr(153) => 's', |
309
|
|
|
chr(200) . chr(154) => 'T', |
310
|
|
|
chr(200) . chr(155) => 't', |
311
|
|
|
// Euro Sign |
312
|
|
|
chr(226) . chr(130) . chr(172) => 'E', |
313
|
|
|
// GBP (Pound) Sign |
314
|
|
|
chr(194) . chr(163) => '', |
315
|
|
|
// Vowels with diacritic (Vietnamese) |
316
|
|
|
// unmarked |
317
|
|
|
chr(198) . chr(160) => 'O', |
318
|
|
|
chr(198) . chr(161) => 'o', |
319
|
|
|
chr(198) . chr(175) => 'U', |
320
|
|
|
chr(198) . chr(176) => 'u', |
321
|
|
|
// grave accent |
322
|
|
|
chr(225) . chr(186) . chr(166) => 'A', |
323
|
|
|
chr(225) . chr(186) . chr(167) => 'a', |
324
|
|
|
chr(225) . chr(186) . chr(176) => 'A', |
325
|
|
|
chr(225) . chr(186) . chr(177) => 'a', |
326
|
|
|
chr(225) . chr(187) . chr(128) => 'E', |
327
|
|
|
chr(225) . chr(187) . chr(129) => 'e', |
328
|
|
|
chr(225) . chr(187) . chr(146) => 'O', |
329
|
|
|
chr(225) . chr(187) . chr(147) => 'o', |
330
|
|
|
chr(225) . chr(187) . chr(156) => 'O', |
331
|
|
|
chr(225) . chr(187) . chr(157) => 'o', |
332
|
|
|
chr(225) . chr(187) . chr(170) => 'U', |
333
|
|
|
chr(225) . chr(187) . chr(171) => 'u', |
334
|
|
|
chr(225) . chr(187) . chr(178) => 'Y', |
335
|
|
|
chr(225) . chr(187) . chr(179) => 'y', |
336
|
|
|
// hook |
337
|
|
|
chr(225) . chr(186) . chr(162) => 'A', |
338
|
|
|
chr(225) . chr(186) . chr(163) => 'a', |
339
|
|
|
chr(225) . chr(186) . chr(168) => 'A', |
340
|
|
|
chr(225) . chr(186) . chr(169) => 'a', |
341
|
|
|
chr(225) . chr(186) . chr(178) => 'A', |
342
|
|
|
chr(225) . chr(186) . chr(179) => 'a', |
343
|
|
|
chr(225) . chr(186) . chr(186) => 'E', |
344
|
|
|
chr(225) . chr(186) . chr(187) => 'e', |
345
|
|
|
chr(225) . chr(187) . chr(130) => 'E', |
346
|
|
|
chr(225) . chr(187) . chr(131) => 'e', |
347
|
|
|
chr(225) . chr(187) . chr(136) => 'I', |
348
|
|
|
chr(225) . chr(187) . chr(137) => 'i', |
349
|
|
|
chr(225) . chr(187) . chr(142) => 'O', |
350
|
|
|
chr(225) . chr(187) . chr(143) => 'o', |
351
|
|
|
chr(225) . chr(187) . chr(148) => 'O', |
352
|
|
|
chr(225) . chr(187) . chr(149) => 'o', |
353
|
|
|
chr(225) . chr(187) . chr(158) => 'O', |
354
|
|
|
chr(225) . chr(187) . chr(159) => 'o', |
355
|
|
|
chr(225) . chr(187) . chr(166) => 'U', |
356
|
|
|
chr(225) . chr(187) . chr(167) => 'u', |
357
|
|
|
chr(225) . chr(187) . chr(172) => 'U', |
358
|
|
|
chr(225) . chr(187) . chr(173) => 'u', |
359
|
|
|
chr(225) . chr(187) . chr(182) => 'Y', |
360
|
|
|
chr(225) . chr(187) . chr(183) => 'y', |
361
|
|
|
// tilde |
362
|
|
|
chr(225) . chr(186) . chr(170) => 'A', |
363
|
|
|
chr(225) . chr(186) . chr(171) => 'a', |
364
|
|
|
chr(225) . chr(186) . chr(180) => 'A', |
365
|
|
|
chr(225) . chr(186) . chr(181) => 'a', |
366
|
|
|
chr(225) . chr(186) . chr(188) => 'E', |
367
|
|
|
chr(225) . chr(186) . chr(189) => 'e', |
368
|
|
|
chr(225) . chr(187) . chr(132) => 'E', |
369
|
|
|
chr(225) . chr(187) . chr(133) => 'e', |
370
|
|
|
chr(225) . chr(187) . chr(150) => 'O', |
371
|
|
|
chr(225) . chr(187) . chr(151) => 'o', |
372
|
|
|
chr(225) . chr(187) . chr(160) => 'O', |
373
|
|
|
chr(225) . chr(187) . chr(161) => 'o', |
374
|
|
|
chr(225) . chr(187) . chr(174) => 'U', |
375
|
|
|
chr(225) . chr(187) . chr(175) => 'u', |
376
|
|
|
chr(225) . chr(187) . chr(184) => 'Y', |
377
|
|
|
chr(225) . chr(187) . chr(185) => 'y', |
378
|
|
|
// acute accent |
379
|
|
|
chr(225) . chr(186) . chr(164) => 'A', |
380
|
|
|
chr(225) . chr(186) . chr(165) => 'a', |
381
|
|
|
chr(225) . chr(186) . chr(174) => 'A', |
382
|
|
|
chr(225) . chr(186) . chr(175) => 'a', |
383
|
|
|
chr(225) . chr(186) . chr(190) => 'E', |
384
|
|
|
chr(225) . chr(186) . chr(191) => 'e', |
385
|
|
|
chr(225) . chr(187) . chr(144) => 'O', |
386
|
|
|
chr(225) . chr(187) . chr(145) => 'o', |
387
|
|
|
chr(225) . chr(187) . chr(154) => 'O', |
388
|
|
|
chr(225) . chr(187) . chr(155) => 'o', |
389
|
|
|
chr(225) . chr(187) . chr(168) => 'U', |
390
|
|
|
chr(225) . chr(187) . chr(169) => 'u', |
391
|
|
|
// dot below |
392
|
|
|
chr(225) . chr(186) . chr(160) => 'A', |
393
|
|
|
chr(225) . chr(186) . chr(161) => 'a', |
394
|
|
|
chr(225) . chr(186) . chr(172) => 'A', |
395
|
|
|
chr(225) . chr(186) . chr(173) => 'a', |
396
|
|
|
chr(225) . chr(186) . chr(182) => 'A', |
397
|
|
|
chr(225) . chr(186) . chr(183) => 'a', |
398
|
|
|
chr(225) . chr(186) . chr(184) => 'E', |
399
|
|
|
chr(225) . chr(186) . chr(185) => 'e', |
400
|
|
|
chr(225) . chr(187) . chr(134) => 'E', |
401
|
|
|
chr(225) . chr(187) . chr(135) => 'e', |
402
|
|
|
chr(225) . chr(187) . chr(138) => 'I', |
403
|
|
|
chr(225) . chr(187) . chr(139) => 'i', |
404
|
|
|
chr(225) . chr(187) . chr(140) => 'O', |
405
|
|
|
chr(225) . chr(187) . chr(141) => 'o', |
406
|
|
|
chr(225) . chr(187) . chr(152) => 'O', |
407
|
|
|
chr(225) . chr(187) . chr(153) => 'o', |
408
|
|
|
chr(225) . chr(187) . chr(162) => 'O', |
409
|
|
|
chr(225) . chr(187) . chr(163) => 'o', |
410
|
|
|
chr(225) . chr(187) . chr(164) => 'U', |
411
|
|
|
chr(225) . chr(187) . chr(165) => 'u', |
412
|
|
|
chr(225) . chr(187) . chr(176) => 'U', |
413
|
|
|
chr(225) . chr(187) . chr(177) => 'u', |
414
|
|
|
chr(225) . chr(187) . chr(180) => 'Y', |
415
|
|
|
chr(225) . chr(187) . chr(181) => 'y', |
416
|
|
|
// Vowels with diacritic (Chinese, Hanyu Pinyin) |
417
|
|
|
chr(201) . chr(145) => 'a', |
418
|
|
|
// macron |
419
|
|
|
chr(199) . chr(149) => 'U', |
420
|
|
|
chr(199) . chr(150) => 'u', |
421
|
|
|
// acute accent |
422
|
|
|
chr(199) . chr(151) => 'U', |
423
|
|
|
chr(199) . chr(152) => 'u', |
424
|
|
|
// caron |
425
|
|
|
chr(199) . chr(141) => 'A', |
426
|
|
|
chr(199) . chr(142) => 'a', |
427
|
|
|
chr(199) . chr(143) => 'I', |
428
|
|
|
chr(199) . chr(144) => 'i', |
429
|
|
|
chr(199) . chr(145) => 'O', |
430
|
|
|
chr(199) . chr(146) => 'o', |
431
|
|
|
chr(199) . chr(147) => 'U', |
432
|
|
|
chr(199) . chr(148) => 'u', |
433
|
|
|
chr(199) . chr(153) => 'U', |
434
|
|
|
chr(199) . chr(154) => 'u', |
435
|
|
|
// grave accent |
436
|
|
|
chr(199) . chr(155) => 'U', |
437
|
|
|
chr(199) . chr(156) => 'u', |
438
|
|
|
// german umlauts |
439
|
|
|
chr(195) . chr(132) => 'Ae', |
440
|
|
|
chr(195) . chr(164) => 'ae', |
441
|
|
|
chr(195) . chr(150) => 'Oe', |
442
|
|
|
chr(195) . chr(182) => 'oe', |
443
|
|
|
chr(195) . chr(156) => 'Ue', |
444
|
|
|
chr(195) . chr(188) => 'ue', |
445
|
|
|
chr(195) . chr(159) => 'ss', |
446
|
|
|
]; |
447
|
|
|
} |
448
|
|
|
} |
449
|
|
|
|