|
0
|
1 <?php
|
|
|
2
|
|
|
3 // {{{ license
|
|
|
4
|
|
|
5 /* vim: set expandtab tabstop=4 shiftwidth=4 softtabstop=4 foldmethod=marker: */
|
|
|
6 //
|
|
|
7 // +----------------------------------------------------------------------+
|
|
|
8 // | This library is free software; you can redistribute it and/or modify |
|
|
|
9 // | it under the terms of the GNU Lesser General Public License as |
|
|
|
10 // | published by the Free Software Foundation; either version 2.1 of the |
|
|
|
11 // | License, or (at your option) any later version. |
|
|
|
12 // | |
|
|
|
13 // | This library is distributed in the hope that it will be useful, but |
|
|
|
14 // | WITHOUT ANY WARRANTY; without even the implied warranty of |
|
|
|
15 // | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
|
|
|
16 // | Lesser General Public License for more details. |
|
|
|
17 // | |
|
|
|
18 // | You should have received a copy of the GNU Lesser General Public |
|
|
|
19 // | License along with this library; if not, write to the Free Software |
|
|
|
20 // | Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 |
|
|
|
21 // | USA. |
|
|
|
22 // +----------------------------------------------------------------------+
|
|
|
23 //
|
|
|
24
|
|
|
25 // }}}
|
|
|
26 require_once 'Net/IDNA2/Exception.php';
|
|
|
27 require_once 'Net/IDNA2/Exception/Nameprep.php';
|
|
|
28
|
|
|
29 /**
|
|
|
30 * Encode/decode Internationalized Domain Names.
|
|
|
31 *
|
|
|
32 * The class allows one to convert internationalized domain names
|
|
|
33 * (see RFC 3490 for details) as they can be used with various registries worldwide
|
|
|
34 * to be translated between their original (localized) form and their encoded form
|
|
|
35 * as it will be used in the DNS (Domain Name System).
|
|
|
36 *
|
|
|
37 * The class provides two public methods, encode() and decode(), which do exactly
|
|
|
38 * what you would expect them to do. You are allowed to use complete domain names,
|
|
|
39 * simple strings and complete email addresses as well. That means, that you might
|
|
|
40 * use any of the following notations:
|
|
|
41 *
|
|
|
42 * - www.n�rgler.com
|
|
|
43 * - xn--nrgler-wxa
|
|
|
44 * - xn--brse-5qa.xn--knrz-1ra.info
|
|
|
45 *
|
|
|
46 * Unicode input might be given as either UTF-8 string, UCS-4 string or UCS-4
|
|
|
47 * array. Unicode output is available in the same formats.
|
|
|
48 * You can select your preferred format via {@link set_paramter()}.
|
|
|
49 *
|
|
|
50 * ACE input and output is always expected to be ASCII.
|
|
|
51 *
|
|
|
52 * @package Net
|
|
|
53 * @author Markus Nix <mnix@docuverse.de>
|
|
|
54 * @author Matthias Sommerfeld <mso@phlylabs.de>
|
|
|
55 * @author Stefan Neufeind <pear.neufeind@speedpartner.de>
|
|
|
56 * @version $Id$
|
|
|
57 */
|
|
|
58 class Net_IDNA2
|
|
|
59 {
|
|
|
60 // {{{ npdata
|
|
|
61 /**
|
|
|
62 * These Unicode codepoints are
|
|
|
63 * mapped to nothing, See RFC3454 for details
|
|
|
64 *
|
|
|
65 * @static
|
|
|
66 * @var array
|
|
|
67 * @access private
|
|
|
68 */
|
|
|
69 private static $_np_map_nothing = array(
|
|
|
70 0xAD,
|
|
|
71 0x34F,
|
|
|
72 0x1806,
|
|
|
73 0x180B,
|
|
|
74 0x180C,
|
|
|
75 0x180D,
|
|
|
76 0x200B,
|
|
|
77 0x200C,
|
|
|
78 0x200D,
|
|
|
79 0x2060,
|
|
|
80 0xFE00,
|
|
|
81 0xFE01,
|
|
|
82 0xFE02,
|
|
|
83 0xFE03,
|
|
|
84 0xFE04,
|
|
|
85 0xFE05,
|
|
|
86 0xFE06,
|
|
|
87 0xFE07,
|
|
|
88 0xFE08,
|
|
|
89 0xFE09,
|
|
|
90 0xFE0A,
|
|
|
91 0xFE0B,
|
|
|
92 0xFE0C,
|
|
|
93 0xFE0D,
|
|
|
94 0xFE0E,
|
|
|
95 0xFE0F,
|
|
|
96 0xFEFF
|
|
|
97 );
|
|
|
98
|
|
|
99 /**
|
|
|
100 * Prohibited codepints
|
|
|
101 *
|
|
|
102 * @static
|
|
|
103 * @var array
|
|
|
104 * @access private
|
|
|
105 */
|
|
|
106 private static $_general_prohibited = array(
|
|
|
107 0,
|
|
|
108 1,
|
|
|
109 2,
|
|
|
110 3,
|
|
|
111 4,
|
|
|
112 5,
|
|
|
113 6,
|
|
|
114 7,
|
|
|
115 8,
|
|
|
116 9,
|
|
|
117 0xA,
|
|
|
118 0xB,
|
|
|
119 0xC,
|
|
|
120 0xD,
|
|
|
121 0xE,
|
|
|
122 0xF,
|
|
|
123 0x10,
|
|
|
124 0x11,
|
|
|
125 0x12,
|
|
|
126 0x13,
|
|
|
127 0x14,
|
|
|
128 0x15,
|
|
|
129 0x16,
|
|
|
130 0x17,
|
|
|
131 0x18,
|
|
|
132 0x19,
|
|
|
133 0x1A,
|
|
|
134 0x1B,
|
|
|
135 0x1C,
|
|
|
136 0x1D,
|
|
|
137 0x1E,
|
|
|
138 0x1F,
|
|
|
139 0x20,
|
|
|
140 0x21,
|
|
|
141 0x22,
|
|
|
142 0x23,
|
|
|
143 0x24,
|
|
|
144 0x25,
|
|
|
145 0x26,
|
|
|
146 0x27,
|
|
|
147 0x28,
|
|
|
148 0x29,
|
|
|
149 0x2A,
|
|
|
150 0x2B,
|
|
|
151 0x2C,
|
|
|
152 0x2F,
|
|
|
153 0x3B,
|
|
|
154 0x3C,
|
|
|
155 0x3D,
|
|
|
156 0x3E,
|
|
|
157 0x3F,
|
|
|
158 0x40,
|
|
|
159 0x5B,
|
|
|
160 0x5C,
|
|
|
161 0x5D,
|
|
|
162 0x5E,
|
|
|
163 0x5F,
|
|
|
164 0x60,
|
|
|
165 0x7B,
|
|
|
166 0x7C,
|
|
|
167 0x7D,
|
|
|
168 0x7E,
|
|
|
169 0x7F,
|
|
|
170 0x3002
|
|
|
171 );
|
|
|
172
|
|
|
173 /**
|
|
|
174 * Codepints prohibited by Nameprep
|
|
|
175 * @static
|
|
|
176 * @var array
|
|
|
177 * @access private
|
|
|
178 */
|
|
|
179 private static $_np_prohibit = array(
|
|
|
180 0xA0,
|
|
|
181 0x1680,
|
|
|
182 0x2000,
|
|
|
183 0x2001,
|
|
|
184 0x2002,
|
|
|
185 0x2003,
|
|
|
186 0x2004,
|
|
|
187 0x2005,
|
|
|
188 0x2006,
|
|
|
189 0x2007,
|
|
|
190 0x2008,
|
|
|
191 0x2009,
|
|
|
192 0x200A,
|
|
|
193 0x200B,
|
|
|
194 0x202F,
|
|
|
195 0x205F,
|
|
|
196 0x3000,
|
|
|
197 0x6DD,
|
|
|
198 0x70F,
|
|
|
199 0x180E,
|
|
|
200 0x200C,
|
|
|
201 0x200D,
|
|
|
202 0x2028,
|
|
|
203 0x2029,
|
|
|
204 0xFEFF,
|
|
|
205 0xFFF9,
|
|
|
206 0xFFFA,
|
|
|
207 0xFFFB,
|
|
|
208 0xFFFC,
|
|
|
209 0xFFFE,
|
|
|
210 0xFFFF,
|
|
|
211 0x1FFFE,
|
|
|
212 0x1FFFF,
|
|
|
213 0x2FFFE,
|
|
|
214 0x2FFFF,
|
|
|
215 0x3FFFE,
|
|
|
216 0x3FFFF,
|
|
|
217 0x4FFFE,
|
|
|
218 0x4FFFF,
|
|
|
219 0x5FFFE,
|
|
|
220 0x5FFFF,
|
|
|
221 0x6FFFE,
|
|
|
222 0x6FFFF,
|
|
|
223 0x7FFFE,
|
|
|
224 0x7FFFF,
|
|
|
225 0x8FFFE,
|
|
|
226 0x8FFFF,
|
|
|
227 0x9FFFE,
|
|
|
228 0x9FFFF,
|
|
|
229 0xAFFFE,
|
|
|
230 0xAFFFF,
|
|
|
231 0xBFFFE,
|
|
|
232 0xBFFFF,
|
|
|
233 0xCFFFE,
|
|
|
234 0xCFFFF,
|
|
|
235 0xDFFFE,
|
|
|
236 0xDFFFF,
|
|
|
237 0xEFFFE,
|
|
|
238 0xEFFFF,
|
|
|
239 0xFFFFE,
|
|
|
240 0xFFFFF,
|
|
|
241 0x10FFFE,
|
|
|
242 0x10FFFF,
|
|
|
243 0xFFF9,
|
|
|
244 0xFFFA,
|
|
|
245 0xFFFB,
|
|
|
246 0xFFFC,
|
|
|
247 0xFFFD,
|
|
|
248 0x340,
|
|
|
249 0x341,
|
|
|
250 0x200E,
|
|
|
251 0x200F,
|
|
|
252 0x202A,
|
|
|
253 0x202B,
|
|
|
254 0x202C,
|
|
|
255 0x202D,
|
|
|
256 0x202E,
|
|
|
257 0x206A,
|
|
|
258 0x206B,
|
|
|
259 0x206C,
|
|
|
260 0x206D,
|
|
|
261 0x206E,
|
|
|
262 0x206F,
|
|
|
263 0xE0001
|
|
|
264 );
|
|
|
265
|
|
|
266 /**
|
|
|
267 * Codepoint ranges prohibited by nameprep
|
|
|
268 *
|
|
|
269 * @static
|
|
|
270 * @var array
|
|
|
271 * @access private
|
|
|
272 */
|
|
|
273 private static $_np_prohibit_ranges = array(
|
|
|
274 array(0x80, 0x9F ),
|
|
|
275 array(0x2060, 0x206F ),
|
|
|
276 array(0x1D173, 0x1D17A ),
|
|
|
277 array(0xE000, 0xF8FF ),
|
|
|
278 array(0xF0000, 0xFFFFD ),
|
|
|
279 array(0x100000, 0x10FFFD),
|
|
|
280 array(0xFDD0, 0xFDEF ),
|
|
|
281 array(0xD800, 0xDFFF ),
|
|
|
282 array(0x2FF0, 0x2FFB ),
|
|
|
283 array(0xE0020, 0xE007F )
|
|
|
284 );
|
|
|
285
|
|
|
286 /**
|
|
|
287 * Replacement mappings (casemapping, replacement sequences, ...)
|
|
|
288 *
|
|
|
289 * @static
|
|
|
290 * @var array
|
|
|
291 * @access private
|
|
|
292 */
|
|
|
293 private static $_np_replacemaps = array(
|
|
|
294 0x41 => array(0x61),
|
|
|
295 0x42 => array(0x62),
|
|
|
296 0x43 => array(0x63),
|
|
|
297 0x44 => array(0x64),
|
|
|
298 0x45 => array(0x65),
|
|
|
299 0x46 => array(0x66),
|
|
|
300 0x47 => array(0x67),
|
|
|
301 0x48 => array(0x68),
|
|
|
302 0x49 => array(0x69),
|
|
|
303 0x4A => array(0x6A),
|
|
|
304 0x4B => array(0x6B),
|
|
|
305 0x4C => array(0x6C),
|
|
|
306 0x4D => array(0x6D),
|
|
|
307 0x4E => array(0x6E),
|
|
|
308 0x4F => array(0x6F),
|
|
|
309 0x50 => array(0x70),
|
|
|
310 0x51 => array(0x71),
|
|
|
311 0x52 => array(0x72),
|
|
|
312 0x53 => array(0x73),
|
|
|
313 0x54 => array(0x74),
|
|
|
314 0x55 => array(0x75),
|
|
|
315 0x56 => array(0x76),
|
|
|
316 0x57 => array(0x77),
|
|
|
317 0x58 => array(0x78),
|
|
|
318 0x59 => array(0x79),
|
|
|
319 0x5A => array(0x7A),
|
|
|
320 0xB5 => array(0x3BC),
|
|
|
321 0xC0 => array(0xE0),
|
|
|
322 0xC1 => array(0xE1),
|
|
|
323 0xC2 => array(0xE2),
|
|
|
324 0xC3 => array(0xE3),
|
|
|
325 0xC4 => array(0xE4),
|
|
|
326 0xC5 => array(0xE5),
|
|
|
327 0xC6 => array(0xE6),
|
|
|
328 0xC7 => array(0xE7),
|
|
|
329 0xC8 => array(0xE8),
|
|
|
330 0xC9 => array(0xE9),
|
|
|
331 0xCA => array(0xEA),
|
|
|
332 0xCB => array(0xEB),
|
|
|
333 0xCC => array(0xEC),
|
|
|
334 0xCD => array(0xED),
|
|
|
335 0xCE => array(0xEE),
|
|
|
336 0xCF => array(0xEF),
|
|
|
337 0xD0 => array(0xF0),
|
|
|
338 0xD1 => array(0xF1),
|
|
|
339 0xD2 => array(0xF2),
|
|
|
340 0xD3 => array(0xF3),
|
|
|
341 0xD4 => array(0xF4),
|
|
|
342 0xD5 => array(0xF5),
|
|
|
343 0xD6 => array(0xF6),
|
|
|
344 0xD8 => array(0xF8),
|
|
|
345 0xD9 => array(0xF9),
|
|
|
346 0xDA => array(0xFA),
|
|
|
347 0xDB => array(0xFB),
|
|
|
348 0xDC => array(0xFC),
|
|
|
349 0xDD => array(0xFD),
|
|
|
350 0xDE => array(0xFE),
|
|
|
351 0xDF => array(0x73, 0x73),
|
|
|
352 0x100 => array(0x101),
|
|
|
353 0x102 => array(0x103),
|
|
|
354 0x104 => array(0x105),
|
|
|
355 0x106 => array(0x107),
|
|
|
356 0x108 => array(0x109),
|
|
|
357 0x10A => array(0x10B),
|
|
|
358 0x10C => array(0x10D),
|
|
|
359 0x10E => array(0x10F),
|
|
|
360 0x110 => array(0x111),
|
|
|
361 0x112 => array(0x113),
|
|
|
362 0x114 => array(0x115),
|
|
|
363 0x116 => array(0x117),
|
|
|
364 0x118 => array(0x119),
|
|
|
365 0x11A => array(0x11B),
|
|
|
366 0x11C => array(0x11D),
|
|
|
367 0x11E => array(0x11F),
|
|
|
368 0x120 => array(0x121),
|
|
|
369 0x122 => array(0x123),
|
|
|
370 0x124 => array(0x125),
|
|
|
371 0x126 => array(0x127),
|
|
|
372 0x128 => array(0x129),
|
|
|
373 0x12A => array(0x12B),
|
|
|
374 0x12C => array(0x12D),
|
|
|
375 0x12E => array(0x12F),
|
|
|
376 0x130 => array(0x69, 0x307),
|
|
|
377 0x132 => array(0x133),
|
|
|
378 0x134 => array(0x135),
|
|
|
379 0x136 => array(0x137),
|
|
|
380 0x139 => array(0x13A),
|
|
|
381 0x13B => array(0x13C),
|
|
|
382 0x13D => array(0x13E),
|
|
|
383 0x13F => array(0x140),
|
|
|
384 0x141 => array(0x142),
|
|
|
385 0x143 => array(0x144),
|
|
|
386 0x145 => array(0x146),
|
|
|
387 0x147 => array(0x148),
|
|
|
388 0x149 => array(0x2BC, 0x6E),
|
|
|
389 0x14A => array(0x14B),
|
|
|
390 0x14C => array(0x14D),
|
|
|
391 0x14E => array(0x14F),
|
|
|
392 0x150 => array(0x151),
|
|
|
393 0x152 => array(0x153),
|
|
|
394 0x154 => array(0x155),
|
|
|
395 0x156 => array(0x157),
|
|
|
396 0x158 => array(0x159),
|
|
|
397 0x15A => array(0x15B),
|
|
|
398 0x15C => array(0x15D),
|
|
|
399 0x15E => array(0x15F),
|
|
|
400 0x160 => array(0x161),
|
|
|
401 0x162 => array(0x163),
|
|
|
402 0x164 => array(0x165),
|
|
|
403 0x166 => array(0x167),
|
|
|
404 0x168 => array(0x169),
|
|
|
405 0x16A => array(0x16B),
|
|
|
406 0x16C => array(0x16D),
|
|
|
407 0x16E => array(0x16F),
|
|
|
408 0x170 => array(0x171),
|
|
|
409 0x172 => array(0x173),
|
|
|
410 0x174 => array(0x175),
|
|
|
411 0x176 => array(0x177),
|
|
|
412 0x178 => array(0xFF),
|
|
|
413 0x179 => array(0x17A),
|
|
|
414 0x17B => array(0x17C),
|
|
|
415 0x17D => array(0x17E),
|
|
|
416 0x17F => array(0x73),
|
|
|
417 0x181 => array(0x253),
|
|
|
418 0x182 => array(0x183),
|
|
|
419 0x184 => array(0x185),
|
|
|
420 0x186 => array(0x254),
|
|
|
421 0x187 => array(0x188),
|
|
|
422 0x189 => array(0x256),
|
|
|
423 0x18A => array(0x257),
|
|
|
424 0x18B => array(0x18C),
|
|
|
425 0x18E => array(0x1DD),
|
|
|
426 0x18F => array(0x259),
|
|
|
427 0x190 => array(0x25B),
|
|
|
428 0x191 => array(0x192),
|
|
|
429 0x193 => array(0x260),
|
|
|
430 0x194 => array(0x263),
|
|
|
431 0x196 => array(0x269),
|
|
|
432 0x197 => array(0x268),
|
|
|
433 0x198 => array(0x199),
|
|
|
434 0x19C => array(0x26F),
|
|
|
435 0x19D => array(0x272),
|
|
|
436 0x19F => array(0x275),
|
|
|
437 0x1A0 => array(0x1A1),
|
|
|
438 0x1A2 => array(0x1A3),
|
|
|
439 0x1A4 => array(0x1A5),
|
|
|
440 0x1A6 => array(0x280),
|
|
|
441 0x1A7 => array(0x1A8),
|
|
|
442 0x1A9 => array(0x283),
|
|
|
443 0x1AC => array(0x1AD),
|
|
|
444 0x1AE => array(0x288),
|
|
|
445 0x1AF => array(0x1B0),
|
|
|
446 0x1B1 => array(0x28A),
|
|
|
447 0x1B2 => array(0x28B),
|
|
|
448 0x1B3 => array(0x1B4),
|
|
|
449 0x1B5 => array(0x1B6),
|
|
|
450 0x1B7 => array(0x292),
|
|
|
451 0x1B8 => array(0x1B9),
|
|
|
452 0x1BC => array(0x1BD),
|
|
|
453 0x1C4 => array(0x1C6),
|
|
|
454 0x1C5 => array(0x1C6),
|
|
|
455 0x1C7 => array(0x1C9),
|
|
|
456 0x1C8 => array(0x1C9),
|
|
|
457 0x1CA => array(0x1CC),
|
|
|
458 0x1CB => array(0x1CC),
|
|
|
459 0x1CD => array(0x1CE),
|
|
|
460 0x1CF => array(0x1D0),
|
|
|
461 0x1D1 => array(0x1D2),
|
|
|
462 0x1D3 => array(0x1D4),
|
|
|
463 0x1D5 => array(0x1D6),
|
|
|
464 0x1D7 => array(0x1D8),
|
|
|
465 0x1D9 => array(0x1DA),
|
|
|
466 0x1DB => array(0x1DC),
|
|
|
467 0x1DE => array(0x1DF),
|
|
|
468 0x1E0 => array(0x1E1),
|
|
|
469 0x1E2 => array(0x1E3),
|
|
|
470 0x1E4 => array(0x1E5),
|
|
|
471 0x1E6 => array(0x1E7),
|
|
|
472 0x1E8 => array(0x1E9),
|
|
|
473 0x1EA => array(0x1EB),
|
|
|
474 0x1EC => array(0x1ED),
|
|
|
475 0x1EE => array(0x1EF),
|
|
|
476 0x1F0 => array(0x6A, 0x30C),
|
|
|
477 0x1F1 => array(0x1F3),
|
|
|
478 0x1F2 => array(0x1F3),
|
|
|
479 0x1F4 => array(0x1F5),
|
|
|
480 0x1F6 => array(0x195),
|
|
|
481 0x1F7 => array(0x1BF),
|
|
|
482 0x1F8 => array(0x1F9),
|
|
|
483 0x1FA => array(0x1FB),
|
|
|
484 0x1FC => array(0x1FD),
|
|
|
485 0x1FE => array(0x1FF),
|
|
|
486 0x200 => array(0x201),
|
|
|
487 0x202 => array(0x203),
|
|
|
488 0x204 => array(0x205),
|
|
|
489 0x206 => array(0x207),
|
|
|
490 0x208 => array(0x209),
|
|
|
491 0x20A => array(0x20B),
|
|
|
492 0x20C => array(0x20D),
|
|
|
493 0x20E => array(0x20F),
|
|
|
494 0x210 => array(0x211),
|
|
|
495 0x212 => array(0x213),
|
|
|
496 0x214 => array(0x215),
|
|
|
497 0x216 => array(0x217),
|
|
|
498 0x218 => array(0x219),
|
|
|
499 0x21A => array(0x21B),
|
|
|
500 0x21C => array(0x21D),
|
|
|
501 0x21E => array(0x21F),
|
|
|
502 0x220 => array(0x19E),
|
|
|
503 0x222 => array(0x223),
|
|
|
504 0x224 => array(0x225),
|
|
|
505 0x226 => array(0x227),
|
|
|
506 0x228 => array(0x229),
|
|
|
507 0x22A => array(0x22B),
|
|
|
508 0x22C => array(0x22D),
|
|
|
509 0x22E => array(0x22F),
|
|
|
510 0x230 => array(0x231),
|
|
|
511 0x232 => array(0x233),
|
|
|
512 0x345 => array(0x3B9),
|
|
|
513 0x37A => array(0x20, 0x3B9),
|
|
|
514 0x386 => array(0x3AC),
|
|
|
515 0x388 => array(0x3AD),
|
|
|
516 0x389 => array(0x3AE),
|
|
|
517 0x38A => array(0x3AF),
|
|
|
518 0x38C => array(0x3CC),
|
|
|
519 0x38E => array(0x3CD),
|
|
|
520 0x38F => array(0x3CE),
|
|
|
521 0x390 => array(0x3B9, 0x308, 0x301),
|
|
|
522 0x391 => array(0x3B1),
|
|
|
523 0x392 => array(0x3B2),
|
|
|
524 0x393 => array(0x3B3),
|
|
|
525 0x394 => array(0x3B4),
|
|
|
526 0x395 => array(0x3B5),
|
|
|
527 0x396 => array(0x3B6),
|
|
|
528 0x397 => array(0x3B7),
|
|
|
529 0x398 => array(0x3B8),
|
|
|
530 0x399 => array(0x3B9),
|
|
|
531 0x39A => array(0x3BA),
|
|
|
532 0x39B => array(0x3BB),
|
|
|
533 0x39C => array(0x3BC),
|
|
|
534 0x39D => array(0x3BD),
|
|
|
535 0x39E => array(0x3BE),
|
|
|
536 0x39F => array(0x3BF),
|
|
|
537 0x3A0 => array(0x3C0),
|
|
|
538 0x3A1 => array(0x3C1),
|
|
|
539 0x3A3 => array(0x3C3),
|
|
|
540 0x3A4 => array(0x3C4),
|
|
|
541 0x3A5 => array(0x3C5),
|
|
|
542 0x3A6 => array(0x3C6),
|
|
|
543 0x3A7 => array(0x3C7),
|
|
|
544 0x3A8 => array(0x3C8),
|
|
|
545 0x3A9 => array(0x3C9),
|
|
|
546 0x3AA => array(0x3CA),
|
|
|
547 0x3AB => array(0x3CB),
|
|
|
548 0x3B0 => array(0x3C5, 0x308, 0x301),
|
|
|
549 0x3C2 => array(0x3C3),
|
|
|
550 0x3D0 => array(0x3B2),
|
|
|
551 0x3D1 => array(0x3B8),
|
|
|
552 0x3D2 => array(0x3C5),
|
|
|
553 0x3D3 => array(0x3CD),
|
|
|
554 0x3D4 => array(0x3CB),
|
|
|
555 0x3D5 => array(0x3C6),
|
|
|
556 0x3D6 => array(0x3C0),
|
|
|
557 0x3D8 => array(0x3D9),
|
|
|
558 0x3DA => array(0x3DB),
|
|
|
559 0x3DC => array(0x3DD),
|
|
|
560 0x3DE => array(0x3DF),
|
|
|
561 0x3E0 => array(0x3E1),
|
|
|
562 0x3E2 => array(0x3E3),
|
|
|
563 0x3E4 => array(0x3E5),
|
|
|
564 0x3E6 => array(0x3E7),
|
|
|
565 0x3E8 => array(0x3E9),
|
|
|
566 0x3EA => array(0x3EB),
|
|
|
567 0x3EC => array(0x3ED),
|
|
|
568 0x3EE => array(0x3EF),
|
|
|
569 0x3F0 => array(0x3BA),
|
|
|
570 0x3F1 => array(0x3C1),
|
|
|
571 0x3F2 => array(0x3C3),
|
|
|
572 0x3F4 => array(0x3B8),
|
|
|
573 0x3F5 => array(0x3B5),
|
|
|
574 0x400 => array(0x450),
|
|
|
575 0x401 => array(0x451),
|
|
|
576 0x402 => array(0x452),
|
|
|
577 0x403 => array(0x453),
|
|
|
578 0x404 => array(0x454),
|
|
|
579 0x405 => array(0x455),
|
|
|
580 0x406 => array(0x456),
|
|
|
581 0x407 => array(0x457),
|
|
|
582 0x408 => array(0x458),
|
|
|
583 0x409 => array(0x459),
|
|
|
584 0x40A => array(0x45A),
|
|
|
585 0x40B => array(0x45B),
|
|
|
586 0x40C => array(0x45C),
|
|
|
587 0x40D => array(0x45D),
|
|
|
588 0x40E => array(0x45E),
|
|
|
589 0x40F => array(0x45F),
|
|
|
590 0x410 => array(0x430),
|
|
|
591 0x411 => array(0x431),
|
|
|
592 0x412 => array(0x432),
|
|
|
593 0x413 => array(0x433),
|
|
|
594 0x414 => array(0x434),
|
|
|
595 0x415 => array(0x435),
|
|
|
596 0x416 => array(0x436),
|
|
|
597 0x417 => array(0x437),
|
|
|
598 0x418 => array(0x438),
|
|
|
599 0x419 => array(0x439),
|
|
|
600 0x41A => array(0x43A),
|
|
|
601 0x41B => array(0x43B),
|
|
|
602 0x41C => array(0x43C),
|
|
|
603 0x41D => array(0x43D),
|
|
|
604 0x41E => array(0x43E),
|
|
|
605 0x41F => array(0x43F),
|
|
|
606 0x420 => array(0x440),
|
|
|
607 0x421 => array(0x441),
|
|
|
608 0x422 => array(0x442),
|
|
|
609 0x423 => array(0x443),
|
|
|
610 0x424 => array(0x444),
|
|
|
611 0x425 => array(0x445),
|
|
|
612 0x426 => array(0x446),
|
|
|
613 0x427 => array(0x447),
|
|
|
614 0x428 => array(0x448),
|
|
|
615 0x429 => array(0x449),
|
|
|
616 0x42A => array(0x44A),
|
|
|
617 0x42B => array(0x44B),
|
|
|
618 0x42C => array(0x44C),
|
|
|
619 0x42D => array(0x44D),
|
|
|
620 0x42E => array(0x44E),
|
|
|
621 0x42F => array(0x44F),
|
|
|
622 0x460 => array(0x461),
|
|
|
623 0x462 => array(0x463),
|
|
|
624 0x464 => array(0x465),
|
|
|
625 0x466 => array(0x467),
|
|
|
626 0x468 => array(0x469),
|
|
|
627 0x46A => array(0x46B),
|
|
|
628 0x46C => array(0x46D),
|
|
|
629 0x46E => array(0x46F),
|
|
|
630 0x470 => array(0x471),
|
|
|
631 0x472 => array(0x473),
|
|
|
632 0x474 => array(0x475),
|
|
|
633 0x476 => array(0x477),
|
|
|
634 0x478 => array(0x479),
|
|
|
635 0x47A => array(0x47B),
|
|
|
636 0x47C => array(0x47D),
|
|
|
637 0x47E => array(0x47F),
|
|
|
638 0x480 => array(0x481),
|
|
|
639 0x48A => array(0x48B),
|
|
|
640 0x48C => array(0x48D),
|
|
|
641 0x48E => array(0x48F),
|
|
|
642 0x490 => array(0x491),
|
|
|
643 0x492 => array(0x493),
|
|
|
644 0x494 => array(0x495),
|
|
|
645 0x496 => array(0x497),
|
|
|
646 0x498 => array(0x499),
|
|
|
647 0x49A => array(0x49B),
|
|
|
648 0x49C => array(0x49D),
|
|
|
649 0x49E => array(0x49F),
|
|
|
650 0x4A0 => array(0x4A1),
|
|
|
651 0x4A2 => array(0x4A3),
|
|
|
652 0x4A4 => array(0x4A5),
|
|
|
653 0x4A6 => array(0x4A7),
|
|
|
654 0x4A8 => array(0x4A9),
|
|
|
655 0x4AA => array(0x4AB),
|
|
|
656 0x4AC => array(0x4AD),
|
|
|
657 0x4AE => array(0x4AF),
|
|
|
658 0x4B0 => array(0x4B1),
|
|
|
659 0x4B2 => array(0x4B3),
|
|
|
660 0x4B4 => array(0x4B5),
|
|
|
661 0x4B6 => array(0x4B7),
|
|
|
662 0x4B8 => array(0x4B9),
|
|
|
663 0x4BA => array(0x4BB),
|
|
|
664 0x4BC => array(0x4BD),
|
|
|
665 0x4BE => array(0x4BF),
|
|
|
666 0x4C1 => array(0x4C2),
|
|
|
667 0x4C3 => array(0x4C4),
|
|
|
668 0x4C5 => array(0x4C6),
|
|
|
669 0x4C7 => array(0x4C8),
|
|
|
670 0x4C9 => array(0x4CA),
|
|
|
671 0x4CB => array(0x4CC),
|
|
|
672 0x4CD => array(0x4CE),
|
|
|
673 0x4D0 => array(0x4D1),
|
|
|
674 0x4D2 => array(0x4D3),
|
|
|
675 0x4D4 => array(0x4D5),
|
|
|
676 0x4D6 => array(0x4D7),
|
|
|
677 0x4D8 => array(0x4D9),
|
|
|
678 0x4DA => array(0x4DB),
|
|
|
679 0x4DC => array(0x4DD),
|
|
|
680 0x4DE => array(0x4DF),
|
|
|
681 0x4E0 => array(0x4E1),
|
|
|
682 0x4E2 => array(0x4E3),
|
|
|
683 0x4E4 => array(0x4E5),
|
|
|
684 0x4E6 => array(0x4E7),
|
|
|
685 0x4E8 => array(0x4E9),
|
|
|
686 0x4EA => array(0x4EB),
|
|
|
687 0x4EC => array(0x4ED),
|
|
|
688 0x4EE => array(0x4EF),
|
|
|
689 0x4F0 => array(0x4F1),
|
|
|
690 0x4F2 => array(0x4F3),
|
|
|
691 0x4F4 => array(0x4F5),
|
|
|
692 0x4F8 => array(0x4F9),
|
|
|
693 0x500 => array(0x501),
|
|
|
694 0x502 => array(0x503),
|
|
|
695 0x504 => array(0x505),
|
|
|
696 0x506 => array(0x507),
|
|
|
697 0x508 => array(0x509),
|
|
|
698 0x50A => array(0x50B),
|
|
|
699 0x50C => array(0x50D),
|
|
|
700 0x50E => array(0x50F),
|
|
|
701 0x531 => array(0x561),
|
|
|
702 0x532 => array(0x562),
|
|
|
703 0x533 => array(0x563),
|
|
|
704 0x534 => array(0x564),
|
|
|
705 0x535 => array(0x565),
|
|
|
706 0x536 => array(0x566),
|
|
|
707 0x537 => array(0x567),
|
|
|
708 0x538 => array(0x568),
|
|
|
709 0x539 => array(0x569),
|
|
|
710 0x53A => array(0x56A),
|
|
|
711 0x53B => array(0x56B),
|
|
|
712 0x53C => array(0x56C),
|
|
|
713 0x53D => array(0x56D),
|
|
|
714 0x53E => array(0x56E),
|
|
|
715 0x53F => array(0x56F),
|
|
|
716 0x540 => array(0x570),
|
|
|
717 0x541 => array(0x571),
|
|
|
718 0x542 => array(0x572),
|
|
|
719 0x543 => array(0x573),
|
|
|
720 0x544 => array(0x574),
|
|
|
721 0x545 => array(0x575),
|
|
|
722 0x546 => array(0x576),
|
|
|
723 0x547 => array(0x577),
|
|
|
724 0x548 => array(0x578),
|
|
|
725 0x549 => array(0x579),
|
|
|
726 0x54A => array(0x57A),
|
|
|
727 0x54B => array(0x57B),
|
|
|
728 0x54C => array(0x57C),
|
|
|
729 0x54D => array(0x57D),
|
|
|
730 0x54E => array(0x57E),
|
|
|
731 0x54F => array(0x57F),
|
|
|
732 0x550 => array(0x580),
|
|
|
733 0x551 => array(0x581),
|
|
|
734 0x552 => array(0x582),
|
|
|
735 0x553 => array(0x583),
|
|
|
736 0x554 => array(0x584),
|
|
|
737 0x555 => array(0x585),
|
|
|
738 0x556 => array(0x586),
|
|
|
739 0x587 => array(0x565, 0x582),
|
|
|
740 0x1E00 => array(0x1E01),
|
|
|
741 0x1E02 => array(0x1E03),
|
|
|
742 0x1E04 => array(0x1E05),
|
|
|
743 0x1E06 => array(0x1E07),
|
|
|
744 0x1E08 => array(0x1E09),
|
|
|
745 0x1E0A => array(0x1E0B),
|
|
|
746 0x1E0C => array(0x1E0D),
|
|
|
747 0x1E0E => array(0x1E0F),
|
|
|
748 0x1E10 => array(0x1E11),
|
|
|
749 0x1E12 => array(0x1E13),
|
|
|
750 0x1E14 => array(0x1E15),
|
|
|
751 0x1E16 => array(0x1E17),
|
|
|
752 0x1E18 => array(0x1E19),
|
|
|
753 0x1E1A => array(0x1E1B),
|
|
|
754 0x1E1C => array(0x1E1D),
|
|
|
755 0x1E1E => array(0x1E1F),
|
|
|
756 0x1E20 => array(0x1E21),
|
|
|
757 0x1E22 => array(0x1E23),
|
|
|
758 0x1E24 => array(0x1E25),
|
|
|
759 0x1E26 => array(0x1E27),
|
|
|
760 0x1E28 => array(0x1E29),
|
|
|
761 0x1E2A => array(0x1E2B),
|
|
|
762 0x1E2C => array(0x1E2D),
|
|
|
763 0x1E2E => array(0x1E2F),
|
|
|
764 0x1E30 => array(0x1E31),
|
|
|
765 0x1E32 => array(0x1E33),
|
|
|
766 0x1E34 => array(0x1E35),
|
|
|
767 0x1E36 => array(0x1E37),
|
|
|
768 0x1E38 => array(0x1E39),
|
|
|
769 0x1E3A => array(0x1E3B),
|
|
|
770 0x1E3C => array(0x1E3D),
|
|
|
771 0x1E3E => array(0x1E3F),
|
|
|
772 0x1E40 => array(0x1E41),
|
|
|
773 0x1E42 => array(0x1E43),
|
|
|
774 0x1E44 => array(0x1E45),
|
|
|
775 0x1E46 => array(0x1E47),
|
|
|
776 0x1E48 => array(0x1E49),
|
|
|
777 0x1E4A => array(0x1E4B),
|
|
|
778 0x1E4C => array(0x1E4D),
|
|
|
779 0x1E4E => array(0x1E4F),
|
|
|
780 0x1E50 => array(0x1E51),
|
|
|
781 0x1E52 => array(0x1E53),
|
|
|
782 0x1E54 => array(0x1E55),
|
|
|
783 0x1E56 => array(0x1E57),
|
|
|
784 0x1E58 => array(0x1E59),
|
|
|
785 0x1E5A => array(0x1E5B),
|
|
|
786 0x1E5C => array(0x1E5D),
|
|
|
787 0x1E5E => array(0x1E5F),
|
|
|
788 0x1E60 => array(0x1E61),
|
|
|
789 0x1E62 => array(0x1E63),
|
|
|
790 0x1E64 => array(0x1E65),
|
|
|
791 0x1E66 => array(0x1E67),
|
|
|
792 0x1E68 => array(0x1E69),
|
|
|
793 0x1E6A => array(0x1E6B),
|
|
|
794 0x1E6C => array(0x1E6D),
|
|
|
795 0x1E6E => array(0x1E6F),
|
|
|
796 0x1E70 => array(0x1E71),
|
|
|
797 0x1E72 => array(0x1E73),
|
|
|
798 0x1E74 => array(0x1E75),
|
|
|
799 0x1E76 => array(0x1E77),
|
|
|
800 0x1E78 => array(0x1E79),
|
|
|
801 0x1E7A => array(0x1E7B),
|
|
|
802 0x1E7C => array(0x1E7D),
|
|
|
803 0x1E7E => array(0x1E7F),
|
|
|
804 0x1E80 => array(0x1E81),
|
|
|
805 0x1E82 => array(0x1E83),
|
|
|
806 0x1E84 => array(0x1E85),
|
|
|
807 0x1E86 => array(0x1E87),
|
|
|
808 0x1E88 => array(0x1E89),
|
|
|
809 0x1E8A => array(0x1E8B),
|
|
|
810 0x1E8C => array(0x1E8D),
|
|
|
811 0x1E8E => array(0x1E8F),
|
|
|
812 0x1E90 => array(0x1E91),
|
|
|
813 0x1E92 => array(0x1E93),
|
|
|
814 0x1E94 => array(0x1E95),
|
|
|
815 0x1E96 => array(0x68, 0x331),
|
|
|
816 0x1E97 => array(0x74, 0x308),
|
|
|
817 0x1E98 => array(0x77, 0x30A),
|
|
|
818 0x1E99 => array(0x79, 0x30A),
|
|
|
819 0x1E9A => array(0x61, 0x2BE),
|
|
|
820 0x1E9B => array(0x1E61),
|
|
|
821 0x1EA0 => array(0x1EA1),
|
|
|
822 0x1EA2 => array(0x1EA3),
|
|
|
823 0x1EA4 => array(0x1EA5),
|
|
|
824 0x1EA6 => array(0x1EA7),
|
|
|
825 0x1EA8 => array(0x1EA9),
|
|
|
826 0x1EAA => array(0x1EAB),
|
|
|
827 0x1EAC => array(0x1EAD),
|
|
|
828 0x1EAE => array(0x1EAF),
|
|
|
829 0x1EB0 => array(0x1EB1),
|
|
|
830 0x1EB2 => array(0x1EB3),
|
|
|
831 0x1EB4 => array(0x1EB5),
|
|
|
832 0x1EB6 => array(0x1EB7),
|
|
|
833 0x1EB8 => array(0x1EB9),
|
|
|
834 0x1EBA => array(0x1EBB),
|
|
|
835 0x1EBC => array(0x1EBD),
|
|
|
836 0x1EBE => array(0x1EBF),
|
|
|
837 0x1EC0 => array(0x1EC1),
|
|
|
838 0x1EC2 => array(0x1EC3),
|
|
|
839 0x1EC4 => array(0x1EC5),
|
|
|
840 0x1EC6 => array(0x1EC7),
|
|
|
841 0x1EC8 => array(0x1EC9),
|
|
|
842 0x1ECA => array(0x1ECB),
|
|
|
843 0x1ECC => array(0x1ECD),
|
|
|
844 0x1ECE => array(0x1ECF),
|
|
|
845 0x1ED0 => array(0x1ED1),
|
|
|
846 0x1ED2 => array(0x1ED3),
|
|
|
847 0x1ED4 => array(0x1ED5),
|
|
|
848 0x1ED6 => array(0x1ED7),
|
|
|
849 0x1ED8 => array(0x1ED9),
|
|
|
850 0x1EDA => array(0x1EDB),
|
|
|
851 0x1EDC => array(0x1EDD),
|
|
|
852 0x1EDE => array(0x1EDF),
|
|
|
853 0x1EE0 => array(0x1EE1),
|
|
|
854 0x1EE2 => array(0x1EE3),
|
|
|
855 0x1EE4 => array(0x1EE5),
|
|
|
856 0x1EE6 => array(0x1EE7),
|
|
|
857 0x1EE8 => array(0x1EE9),
|
|
|
858 0x1EEA => array(0x1EEB),
|
|
|
859 0x1EEC => array(0x1EED),
|
|
|
860 0x1EEE => array(0x1EEF),
|
|
|
861 0x1EF0 => array(0x1EF1),
|
|
|
862 0x1EF2 => array(0x1EF3),
|
|
|
863 0x1EF4 => array(0x1EF5),
|
|
|
864 0x1EF6 => array(0x1EF7),
|
|
|
865 0x1EF8 => array(0x1EF9),
|
|
|
866 0x1F08 => array(0x1F00),
|
|
|
867 0x1F09 => array(0x1F01),
|
|
|
868 0x1F0A => array(0x1F02),
|
|
|
869 0x1F0B => array(0x1F03),
|
|
|
870 0x1F0C => array(0x1F04),
|
|
|
871 0x1F0D => array(0x1F05),
|
|
|
872 0x1F0E => array(0x1F06),
|
|
|
873 0x1F0F => array(0x1F07),
|
|
|
874 0x1F18 => array(0x1F10),
|
|
|
875 0x1F19 => array(0x1F11),
|
|
|
876 0x1F1A => array(0x1F12),
|
|
|
877 0x1F1B => array(0x1F13),
|
|
|
878 0x1F1C => array(0x1F14),
|
|
|
879 0x1F1D => array(0x1F15),
|
|
|
880 0x1F28 => array(0x1F20),
|
|
|
881 0x1F29 => array(0x1F21),
|
|
|
882 0x1F2A => array(0x1F22),
|
|
|
883 0x1F2B => array(0x1F23),
|
|
|
884 0x1F2C => array(0x1F24),
|
|
|
885 0x1F2D => array(0x1F25),
|
|
|
886 0x1F2E => array(0x1F26),
|
|
|
887 0x1F2F => array(0x1F27),
|
|
|
888 0x1F38 => array(0x1F30),
|
|
|
889 0x1F39 => array(0x1F31),
|
|
|
890 0x1F3A => array(0x1F32),
|
|
|
891 0x1F3B => array(0x1F33),
|
|
|
892 0x1F3C => array(0x1F34),
|
|
|
893 0x1F3D => array(0x1F35),
|
|
|
894 0x1F3E => array(0x1F36),
|
|
|
895 0x1F3F => array(0x1F37),
|
|
|
896 0x1F48 => array(0x1F40),
|
|
|
897 0x1F49 => array(0x1F41),
|
|
|
898 0x1F4A => array(0x1F42),
|
|
|
899 0x1F4B => array(0x1F43),
|
|
|
900 0x1F4C => array(0x1F44),
|
|
|
901 0x1F4D => array(0x1F45),
|
|
|
902 0x1F50 => array(0x3C5, 0x313),
|
|
|
903 0x1F52 => array(0x3C5, 0x313, 0x300),
|
|
|
904 0x1F54 => array(0x3C5, 0x313, 0x301),
|
|
|
905 0x1F56 => array(0x3C5, 0x313, 0x342),
|
|
|
906 0x1F59 => array(0x1F51),
|
|
|
907 0x1F5B => array(0x1F53),
|
|
|
908 0x1F5D => array(0x1F55),
|
|
|
909 0x1F5F => array(0x1F57),
|
|
|
910 0x1F68 => array(0x1F60),
|
|
|
911 0x1F69 => array(0x1F61),
|
|
|
912 0x1F6A => array(0x1F62),
|
|
|
913 0x1F6B => array(0x1F63),
|
|
|
914 0x1F6C => array(0x1F64),
|
|
|
915 0x1F6D => array(0x1F65),
|
|
|
916 0x1F6E => array(0x1F66),
|
|
|
917 0x1F6F => array(0x1F67),
|
|
|
918 0x1F80 => array(0x1F00, 0x3B9),
|
|
|
919 0x1F81 => array(0x1F01, 0x3B9),
|
|
|
920 0x1F82 => array(0x1F02, 0x3B9),
|
|
|
921 0x1F83 => array(0x1F03, 0x3B9),
|
|
|
922 0x1F84 => array(0x1F04, 0x3B9),
|
|
|
923 0x1F85 => array(0x1F05, 0x3B9),
|
|
|
924 0x1F86 => array(0x1F06, 0x3B9),
|
|
|
925 0x1F87 => array(0x1F07, 0x3B9),
|
|
|
926 0x1F88 => array(0x1F00, 0x3B9),
|
|
|
927 0x1F89 => array(0x1F01, 0x3B9),
|
|
|
928 0x1F8A => array(0x1F02, 0x3B9),
|
|
|
929 0x1F8B => array(0x1F03, 0x3B9),
|
|
|
930 0x1F8C => array(0x1F04, 0x3B9),
|
|
|
931 0x1F8D => array(0x1F05, 0x3B9),
|
|
|
932 0x1F8E => array(0x1F06, 0x3B9),
|
|
|
933 0x1F8F => array(0x1F07, 0x3B9),
|
|
|
934 0x1F90 => array(0x1F20, 0x3B9),
|
|
|
935 0x1F91 => array(0x1F21, 0x3B9),
|
|
|
936 0x1F92 => array(0x1F22, 0x3B9),
|
|
|
937 0x1F93 => array(0x1F23, 0x3B9),
|
|
|
938 0x1F94 => array(0x1F24, 0x3B9),
|
|
|
939 0x1F95 => array(0x1F25, 0x3B9),
|
|
|
940 0x1F96 => array(0x1F26, 0x3B9),
|
|
|
941 0x1F97 => array(0x1F27, 0x3B9),
|
|
|
942 0x1F98 => array(0x1F20, 0x3B9),
|
|
|
943 0x1F99 => array(0x1F21, 0x3B9),
|
|
|
944 0x1F9A => array(0x1F22, 0x3B9),
|
|
|
945 0x1F9B => array(0x1F23, 0x3B9),
|
|
|
946 0x1F9C => array(0x1F24, 0x3B9),
|
|
|
947 0x1F9D => array(0x1F25, 0x3B9),
|
|
|
948 0x1F9E => array(0x1F26, 0x3B9),
|
|
|
949 0x1F9F => array(0x1F27, 0x3B9),
|
|
|
950 0x1FA0 => array(0x1F60, 0x3B9),
|
|
|
951 0x1FA1 => array(0x1F61, 0x3B9),
|
|
|
952 0x1FA2 => array(0x1F62, 0x3B9),
|
|
|
953 0x1FA3 => array(0x1F63, 0x3B9),
|
|
|
954 0x1FA4 => array(0x1F64, 0x3B9),
|
|
|
955 0x1FA5 => array(0x1F65, 0x3B9),
|
|
|
956 0x1FA6 => array(0x1F66, 0x3B9),
|
|
|
957 0x1FA7 => array(0x1F67, 0x3B9),
|
|
|
958 0x1FA8 => array(0x1F60, 0x3B9),
|
|
|
959 0x1FA9 => array(0x1F61, 0x3B9),
|
|
|
960 0x1FAA => array(0x1F62, 0x3B9),
|
|
|
961 0x1FAB => array(0x1F63, 0x3B9),
|
|
|
962 0x1FAC => array(0x1F64, 0x3B9),
|
|
|
963 0x1FAD => array(0x1F65, 0x3B9),
|
|
|
964 0x1FAE => array(0x1F66, 0x3B9),
|
|
|
965 0x1FAF => array(0x1F67, 0x3B9),
|
|
|
966 0x1FB2 => array(0x1F70, 0x3B9),
|
|
|
967 0x1FB3 => array(0x3B1, 0x3B9),
|
|
|
968 0x1FB4 => array(0x3AC, 0x3B9),
|
|
|
969 0x1FB6 => array(0x3B1, 0x342),
|
|
|
970 0x1FB7 => array(0x3B1, 0x342, 0x3B9),
|
|
|
971 0x1FB8 => array(0x1FB0),
|
|
|
972 0x1FB9 => array(0x1FB1),
|
|
|
973 0x1FBA => array(0x1F70),
|
|
|
974 0x1FBB => array(0x1F71),
|
|
|
975 0x1FBC => array(0x3B1, 0x3B9),
|
|
|
976 0x1FBE => array(0x3B9),
|
|
|
977 0x1FC2 => array(0x1F74, 0x3B9),
|
|
|
978 0x1FC3 => array(0x3B7, 0x3B9),
|
|
|
979 0x1FC4 => array(0x3AE, 0x3B9),
|
|
|
980 0x1FC6 => array(0x3B7, 0x342),
|
|
|
981 0x1FC7 => array(0x3B7, 0x342, 0x3B9),
|
|
|
982 0x1FC8 => array(0x1F72),
|
|
|
983 0x1FC9 => array(0x1F73),
|
|
|
984 0x1FCA => array(0x1F74),
|
|
|
985 0x1FCB => array(0x1F75),
|
|
|
986 0x1FCC => array(0x3B7, 0x3B9),
|
|
|
987 0x1FD2 => array(0x3B9, 0x308, 0x300),
|
|
|
988 0x1FD3 => array(0x3B9, 0x308, 0x301),
|
|
|
989 0x1FD6 => array(0x3B9, 0x342),
|
|
|
990 0x1FD7 => array(0x3B9, 0x308, 0x342),
|
|
|
991 0x1FD8 => array(0x1FD0),
|
|
|
992 0x1FD9 => array(0x1FD1),
|
|
|
993 0x1FDA => array(0x1F76),
|
|
|
994 0x1FDB => array(0x1F77),
|
|
|
995 0x1FE2 => array(0x3C5, 0x308, 0x300),
|
|
|
996 0x1FE3 => array(0x3C5, 0x308, 0x301),
|
|
|
997 0x1FE4 => array(0x3C1, 0x313),
|
|
|
998 0x1FE6 => array(0x3C5, 0x342),
|
|
|
999 0x1FE7 => array(0x3C5, 0x308, 0x342),
|
|
|
1000 0x1FE8 => array(0x1FE0),
|
|
|
1001 0x1FE9 => array(0x1FE1),
|
|
|
1002 0x1FEA => array(0x1F7A),
|
|
|
1003 0x1FEB => array(0x1F7B),
|
|
|
1004 0x1FEC => array(0x1FE5),
|
|
|
1005 0x1FF2 => array(0x1F7C, 0x3B9),
|
|
|
1006 0x1FF3 => array(0x3C9, 0x3B9),
|
|
|
1007 0x1FF4 => array(0x3CE, 0x3B9),
|
|
|
1008 0x1FF6 => array(0x3C9, 0x342),
|
|
|
1009 0x1FF7 => array(0x3C9, 0x342, 0x3B9),
|
|
|
1010 0x1FF8 => array(0x1F78),
|
|
|
1011 0x1FF9 => array(0x1F79),
|
|
|
1012 0x1FFA => array(0x1F7C),
|
|
|
1013 0x1FFB => array(0x1F7D),
|
|
|
1014 0x1FFC => array(0x3C9, 0x3B9),
|
|
|
1015 0x20A8 => array(0x72, 0x73),
|
|
|
1016 0x2102 => array(0x63),
|
|
|
1017 0x2103 => array(0xB0, 0x63),
|
|
|
1018 0x2107 => array(0x25B),
|
|
|
1019 0x2109 => array(0xB0, 0x66),
|
|
|
1020 0x210B => array(0x68),
|
|
|
1021 0x210C => array(0x68),
|
|
|
1022 0x210D => array(0x68),
|
|
|
1023 0x2110 => array(0x69),
|
|
|
1024 0x2111 => array(0x69),
|
|
|
1025 0x2112 => array(0x6C),
|
|
|
1026 0x2115 => array(0x6E),
|
|
|
1027 0x2116 => array(0x6E, 0x6F),
|
|
|
1028 0x2119 => array(0x70),
|
|
|
1029 0x211A => array(0x71),
|
|
|
1030 0x211B => array(0x72),
|
|
|
1031 0x211C => array(0x72),
|
|
|
1032 0x211D => array(0x72),
|
|
|
1033 0x2120 => array(0x73, 0x6D),
|
|
|
1034 0x2121 => array(0x74, 0x65, 0x6C),
|
|
|
1035 0x2122 => array(0x74, 0x6D),
|
|
|
1036 0x2124 => array(0x7A),
|
|
|
1037 0x2126 => array(0x3C9),
|
|
|
1038 0x2128 => array(0x7A),
|
|
|
1039 0x212A => array(0x6B),
|
|
|
1040 0x212B => array(0xE5),
|
|
|
1041 0x212C => array(0x62),
|
|
|
1042 0x212D => array(0x63),
|
|
|
1043 0x2130 => array(0x65),
|
|
|
1044 0x2131 => array(0x66),
|
|
|
1045 0x2133 => array(0x6D),
|
|
|
1046 0x213E => array(0x3B3),
|
|
|
1047 0x213F => array(0x3C0),
|
|
|
1048 0x2145 => array(0x64),
|
|
|
1049 0x2160 => array(0x2170),
|
|
|
1050 0x2161 => array(0x2171),
|
|
|
1051 0x2162 => array(0x2172),
|
|
|
1052 0x2163 => array(0x2173),
|
|
|
1053 0x2164 => array(0x2174),
|
|
|
1054 0x2165 => array(0x2175),
|
|
|
1055 0x2166 => array(0x2176),
|
|
|
1056 0x2167 => array(0x2177),
|
|
|
1057 0x2168 => array(0x2178),
|
|
|
1058 0x2169 => array(0x2179),
|
|
|
1059 0x216A => array(0x217A),
|
|
|
1060 0x216B => array(0x217B),
|
|
|
1061 0x216C => array(0x217C),
|
|
|
1062 0x216D => array(0x217D),
|
|
|
1063 0x216E => array(0x217E),
|
|
|
1064 0x216F => array(0x217F),
|
|
|
1065 0x24B6 => array(0x24D0),
|
|
|
1066 0x24B7 => array(0x24D1),
|
|
|
1067 0x24B8 => array(0x24D2),
|
|
|
1068 0x24B9 => array(0x24D3),
|
|
|
1069 0x24BA => array(0x24D4),
|
|
|
1070 0x24BB => array(0x24D5),
|
|
|
1071 0x24BC => array(0x24D6),
|
|
|
1072 0x24BD => array(0x24D7),
|
|
|
1073 0x24BE => array(0x24D8),
|
|
|
1074 0x24BF => array(0x24D9),
|
|
|
1075 0x24C0 => array(0x24DA),
|
|
|
1076 0x24C1 => array(0x24DB),
|
|
|
1077 0x24C2 => array(0x24DC),
|
|
|
1078 0x24C3 => array(0x24DD),
|
|
|
1079 0x24C4 => array(0x24DE),
|
|
|
1080 0x24C5 => array(0x24DF),
|
|
|
1081 0x24C6 => array(0x24E0),
|
|
|
1082 0x24C7 => array(0x24E1),
|
|
|
1083 0x24C8 => array(0x24E2),
|
|
|
1084 0x24C9 => array(0x24E3),
|
|
|
1085 0x24CA => array(0x24E4),
|
|
|
1086 0x24CB => array(0x24E5),
|
|
|
1087 0x24CC => array(0x24E6),
|
|
|
1088 0x24CD => array(0x24E7),
|
|
|
1089 0x24CE => array(0x24E8),
|
|
|
1090 0x24CF => array(0x24E9),
|
|
|
1091 0x3371 => array(0x68, 0x70, 0x61),
|
|
|
1092 0x3373 => array(0x61, 0x75),
|
|
|
1093 0x3375 => array(0x6F, 0x76),
|
|
|
1094 0x3380 => array(0x70, 0x61),
|
|
|
1095 0x3381 => array(0x6E, 0x61),
|
|
|
1096 0x3382 => array(0x3BC, 0x61),
|
|
|
1097 0x3383 => array(0x6D, 0x61),
|
|
|
1098 0x3384 => array(0x6B, 0x61),
|
|
|
1099 0x3385 => array(0x6B, 0x62),
|
|
|
1100 0x3386 => array(0x6D, 0x62),
|
|
|
1101 0x3387 => array(0x67, 0x62),
|
|
|
1102 0x338A => array(0x70, 0x66),
|
|
|
1103 0x338B => array(0x6E, 0x66),
|
|
|
1104 0x338C => array(0x3BC, 0x66),
|
|
|
1105 0x3390 => array(0x68, 0x7A),
|
|
|
1106 0x3391 => array(0x6B, 0x68, 0x7A),
|
|
|
1107 0x3392 => array(0x6D, 0x68, 0x7A),
|
|
|
1108 0x3393 => array(0x67, 0x68, 0x7A),
|
|
|
1109 0x3394 => array(0x74, 0x68, 0x7A),
|
|
|
1110 0x33A9 => array(0x70, 0x61),
|
|
|
1111 0x33AA => array(0x6B, 0x70, 0x61),
|
|
|
1112 0x33AB => array(0x6D, 0x70, 0x61),
|
|
|
1113 0x33AC => array(0x67, 0x70, 0x61),
|
|
|
1114 0x33B4 => array(0x70, 0x76),
|
|
|
1115 0x33B5 => array(0x6E, 0x76),
|
|
|
1116 0x33B6 => array(0x3BC, 0x76),
|
|
|
1117 0x33B7 => array(0x6D, 0x76),
|
|
|
1118 0x33B8 => array(0x6B, 0x76),
|
|
|
1119 0x33B9 => array(0x6D, 0x76),
|
|
|
1120 0x33BA => array(0x70, 0x77),
|
|
|
1121 0x33BB => array(0x6E, 0x77),
|
|
|
1122 0x33BC => array(0x3BC, 0x77),
|
|
|
1123 0x33BD => array(0x6D, 0x77),
|
|
|
1124 0x33BE => array(0x6B, 0x77),
|
|
|
1125 0x33BF => array(0x6D, 0x77),
|
|
|
1126 0x33C0 => array(0x6B, 0x3C9),
|
|
|
1127 0x33C1 => array(0x6D, 0x3C9),
|
|
|
1128 /* 0x33C2 => array(0x61, 0x2E, 0x6D, 0x2E), */
|
|
|
1129 0x33C3 => array(0x62, 0x71),
|
|
|
1130 0x33C6 => array(0x63, 0x2215, 0x6B, 0x67),
|
|
|
1131 0x33C7 => array(0x63, 0x6F, 0x2E),
|
|
|
1132 0x33C8 => array(0x64, 0x62),
|
|
|
1133 0x33C9 => array(0x67, 0x79),
|
|
|
1134 0x33CB => array(0x68, 0x70),
|
|
|
1135 0x33CD => array(0x6B, 0x6B),
|
|
|
1136 0x33CE => array(0x6B, 0x6D),
|
|
|
1137 0x33D7 => array(0x70, 0x68),
|
|
|
1138 0x33D9 => array(0x70, 0x70, 0x6D),
|
|
|
1139 0x33DA => array(0x70, 0x72),
|
|
|
1140 0x33DC => array(0x73, 0x76),
|
|
|
1141 0x33DD => array(0x77, 0x62),
|
|
|
1142 0xFB00 => array(0x66, 0x66),
|
|
|
1143 0xFB01 => array(0x66, 0x69),
|
|
|
1144 0xFB02 => array(0x66, 0x6C),
|
|
|
1145 0xFB03 => array(0x66, 0x66, 0x69),
|
|
|
1146 0xFB04 => array(0x66, 0x66, 0x6C),
|
|
|
1147 0xFB05 => array(0x73, 0x74),
|
|
|
1148 0xFB06 => array(0x73, 0x74),
|
|
|
1149 0xFB13 => array(0x574, 0x576),
|
|
|
1150 0xFB14 => array(0x574, 0x565),
|
|
|
1151 0xFB15 => array(0x574, 0x56B),
|
|
|
1152 0xFB16 => array(0x57E, 0x576),
|
|
|
1153 0xFB17 => array(0x574, 0x56D),
|
|
|
1154 0xFF21 => array(0xFF41),
|
|
|
1155 0xFF22 => array(0xFF42),
|
|
|
1156 0xFF23 => array(0xFF43),
|
|
|
1157 0xFF24 => array(0xFF44),
|
|
|
1158 0xFF25 => array(0xFF45),
|
|
|
1159 0xFF26 => array(0xFF46),
|
|
|
1160 0xFF27 => array(0xFF47),
|
|
|
1161 0xFF28 => array(0xFF48),
|
|
|
1162 0xFF29 => array(0xFF49),
|
|
|
1163 0xFF2A => array(0xFF4A),
|
|
|
1164 0xFF2B => array(0xFF4B),
|
|
|
1165 0xFF2C => array(0xFF4C),
|
|
|
1166 0xFF2D => array(0xFF4D),
|
|
|
1167 0xFF2E => array(0xFF4E),
|
|
|
1168 0xFF2F => array(0xFF4F),
|
|
|
1169 0xFF30 => array(0xFF50),
|
|
|
1170 0xFF31 => array(0xFF51),
|
|
|
1171 0xFF32 => array(0xFF52),
|
|
|
1172 0xFF33 => array(0xFF53),
|
|
|
1173 0xFF34 => array(0xFF54),
|
|
|
1174 0xFF35 => array(0xFF55),
|
|
|
1175 0xFF36 => array(0xFF56),
|
|
|
1176 0xFF37 => array(0xFF57),
|
|
|
1177 0xFF38 => array(0xFF58),
|
|
|
1178 0xFF39 => array(0xFF59),
|
|
|
1179 0xFF3A => array(0xFF5A),
|
|
|
1180 0x10400 => array(0x10428),
|
|
|
1181 0x10401 => array(0x10429),
|
|
|
1182 0x10402 => array(0x1042A),
|
|
|
1183 0x10403 => array(0x1042B),
|
|
|
1184 0x10404 => array(0x1042C),
|
|
|
1185 0x10405 => array(0x1042D),
|
|
|
1186 0x10406 => array(0x1042E),
|
|
|
1187 0x10407 => array(0x1042F),
|
|
|
1188 0x10408 => array(0x10430),
|
|
|
1189 0x10409 => array(0x10431),
|
|
|
1190 0x1040A => array(0x10432),
|
|
|
1191 0x1040B => array(0x10433),
|
|
|
1192 0x1040C => array(0x10434),
|
|
|
1193 0x1040D => array(0x10435),
|
|
|
1194 0x1040E => array(0x10436),
|
|
|
1195 0x1040F => array(0x10437),
|
|
|
1196 0x10410 => array(0x10438),
|
|
|
1197 0x10411 => array(0x10439),
|
|
|
1198 0x10412 => array(0x1043A),
|
|
|
1199 0x10413 => array(0x1043B),
|
|
|
1200 0x10414 => array(0x1043C),
|
|
|
1201 0x10415 => array(0x1043D),
|
|
|
1202 0x10416 => array(0x1043E),
|
|
|
1203 0x10417 => array(0x1043F),
|
|
|
1204 0x10418 => array(0x10440),
|
|
|
1205 0x10419 => array(0x10441),
|
|
|
1206 0x1041A => array(0x10442),
|
|
|
1207 0x1041B => array(0x10443),
|
|
|
1208 0x1041C => array(0x10444),
|
|
|
1209 0x1041D => array(0x10445),
|
|
|
1210 0x1041E => array(0x10446),
|
|
|
1211 0x1041F => array(0x10447),
|
|
|
1212 0x10420 => array(0x10448),
|
|
|
1213 0x10421 => array(0x10449),
|
|
|
1214 0x10422 => array(0x1044A),
|
|
|
1215 0x10423 => array(0x1044B),
|
|
|
1216 0x10424 => array(0x1044C),
|
|
|
1217 0x10425 => array(0x1044D),
|
|
|
1218 0x1D400 => array(0x61),
|
|
|
1219 0x1D401 => array(0x62),
|
|
|
1220 0x1D402 => array(0x63),
|
|
|
1221 0x1D403 => array(0x64),
|
|
|
1222 0x1D404 => array(0x65),
|
|
|
1223 0x1D405 => array(0x66),
|
|
|
1224 0x1D406 => array(0x67),
|
|
|
1225 0x1D407 => array(0x68),
|
|
|
1226 0x1D408 => array(0x69),
|
|
|
1227 0x1D409 => array(0x6A),
|
|
|
1228 0x1D40A => array(0x6B),
|
|
|
1229 0x1D40B => array(0x6C),
|
|
|
1230 0x1D40C => array(0x6D),
|
|
|
1231 0x1D40D => array(0x6E),
|
|
|
1232 0x1D40E => array(0x6F),
|
|
|
1233 0x1D40F => array(0x70),
|
|
|
1234 0x1D410 => array(0x71),
|
|
|
1235 0x1D411 => array(0x72),
|
|
|
1236 0x1D412 => array(0x73),
|
|
|
1237 0x1D413 => array(0x74),
|
|
|
1238 0x1D414 => array(0x75),
|
|
|
1239 0x1D415 => array(0x76),
|
|
|
1240 0x1D416 => array(0x77),
|
|
|
1241 0x1D417 => array(0x78),
|
|
|
1242 0x1D418 => array(0x79),
|
|
|
1243 0x1D419 => array(0x7A),
|
|
|
1244 0x1D434 => array(0x61),
|
|
|
1245 0x1D435 => array(0x62),
|
|
|
1246 0x1D436 => array(0x63),
|
|
|
1247 0x1D437 => array(0x64),
|
|
|
1248 0x1D438 => array(0x65),
|
|
|
1249 0x1D439 => array(0x66),
|
|
|
1250 0x1D43A => array(0x67),
|
|
|
1251 0x1D43B => array(0x68),
|
|
|
1252 0x1D43C => array(0x69),
|
|
|
1253 0x1D43D => array(0x6A),
|
|
|
1254 0x1D43E => array(0x6B),
|
|
|
1255 0x1D43F => array(0x6C),
|
|
|
1256 0x1D440 => array(0x6D),
|
|
|
1257 0x1D441 => array(0x6E),
|
|
|
1258 0x1D442 => array(0x6F),
|
|
|
1259 0x1D443 => array(0x70),
|
|
|
1260 0x1D444 => array(0x71),
|
|
|
1261 0x1D445 => array(0x72),
|
|
|
1262 0x1D446 => array(0x73),
|
|
|
1263 0x1D447 => array(0x74),
|
|
|
1264 0x1D448 => array(0x75),
|
|
|
1265 0x1D449 => array(0x76),
|
|
|
1266 0x1D44A => array(0x77),
|
|
|
1267 0x1D44B => array(0x78),
|
|
|
1268 0x1D44C => array(0x79),
|
|
|
1269 0x1D44D => array(0x7A),
|
|
|
1270 0x1D468 => array(0x61),
|
|
|
1271 0x1D469 => array(0x62),
|
|
|
1272 0x1D46A => array(0x63),
|
|
|
1273 0x1D46B => array(0x64),
|
|
|
1274 0x1D46C => array(0x65),
|
|
|
1275 0x1D46D => array(0x66),
|
|
|
1276 0x1D46E => array(0x67),
|
|
|
1277 0x1D46F => array(0x68),
|
|
|
1278 0x1D470 => array(0x69),
|
|
|
1279 0x1D471 => array(0x6A),
|
|
|
1280 0x1D472 => array(0x6B),
|
|
|
1281 0x1D473 => array(0x6C),
|
|
|
1282 0x1D474 => array(0x6D),
|
|
|
1283 0x1D475 => array(0x6E),
|
|
|
1284 0x1D476 => array(0x6F),
|
|
|
1285 0x1D477 => array(0x70),
|
|
|
1286 0x1D478 => array(0x71),
|
|
|
1287 0x1D479 => array(0x72),
|
|
|
1288 0x1D47A => array(0x73),
|
|
|
1289 0x1D47B => array(0x74),
|
|
|
1290 0x1D47C => array(0x75),
|
|
|
1291 0x1D47D => array(0x76),
|
|
|
1292 0x1D47E => array(0x77),
|
|
|
1293 0x1D47F => array(0x78),
|
|
|
1294 0x1D480 => array(0x79),
|
|
|
1295 0x1D481 => array(0x7A),
|
|
|
1296 0x1D49C => array(0x61),
|
|
|
1297 0x1D49E => array(0x63),
|
|
|
1298 0x1D49F => array(0x64),
|
|
|
1299 0x1D4A2 => array(0x67),
|
|
|
1300 0x1D4A5 => array(0x6A),
|
|
|
1301 0x1D4A6 => array(0x6B),
|
|
|
1302 0x1D4A9 => array(0x6E),
|
|
|
1303 0x1D4AA => array(0x6F),
|
|
|
1304 0x1D4AB => array(0x70),
|
|
|
1305 0x1D4AC => array(0x71),
|
|
|
1306 0x1D4AE => array(0x73),
|
|
|
1307 0x1D4AF => array(0x74),
|
|
|
1308 0x1D4B0 => array(0x75),
|
|
|
1309 0x1D4B1 => array(0x76),
|
|
|
1310 0x1D4B2 => array(0x77),
|
|
|
1311 0x1D4B3 => array(0x78),
|
|
|
1312 0x1D4B4 => array(0x79),
|
|
|
1313 0x1D4B5 => array(0x7A),
|
|
|
1314 0x1D4D0 => array(0x61),
|
|
|
1315 0x1D4D1 => array(0x62),
|
|
|
1316 0x1D4D2 => array(0x63),
|
|
|
1317 0x1D4D3 => array(0x64),
|
|
|
1318 0x1D4D4 => array(0x65),
|
|
|
1319 0x1D4D5 => array(0x66),
|
|
|
1320 0x1D4D6 => array(0x67),
|
|
|
1321 0x1D4D7 => array(0x68),
|
|
|
1322 0x1D4D8 => array(0x69),
|
|
|
1323 0x1D4D9 => array(0x6A),
|
|
|
1324 0x1D4DA => array(0x6B),
|
|
|
1325 0x1D4DB => array(0x6C),
|
|
|
1326 0x1D4DC => array(0x6D),
|
|
|
1327 0x1D4DD => array(0x6E),
|
|
|
1328 0x1D4DE => array(0x6F),
|
|
|
1329 0x1D4DF => array(0x70),
|
|
|
1330 0x1D4E0 => array(0x71),
|
|
|
1331 0x1D4E1 => array(0x72),
|
|
|
1332 0x1D4E2 => array(0x73),
|
|
|
1333 0x1D4E3 => array(0x74),
|
|
|
1334 0x1D4E4 => array(0x75),
|
|
|
1335 0x1D4E5 => array(0x76),
|
|
|
1336 0x1D4E6 => array(0x77),
|
|
|
1337 0x1D4E7 => array(0x78),
|
|
|
1338 0x1D4E8 => array(0x79),
|
|
|
1339 0x1D4E9 => array(0x7A),
|
|
|
1340 0x1D504 => array(0x61),
|
|
|
1341 0x1D505 => array(0x62),
|
|
|
1342 0x1D507 => array(0x64),
|
|
|
1343 0x1D508 => array(0x65),
|
|
|
1344 0x1D509 => array(0x66),
|
|
|
1345 0x1D50A => array(0x67),
|
|
|
1346 0x1D50D => array(0x6A),
|
|
|
1347 0x1D50E => array(0x6B),
|
|
|
1348 0x1D50F => array(0x6C),
|
|
|
1349 0x1D510 => array(0x6D),
|
|
|
1350 0x1D511 => array(0x6E),
|
|
|
1351 0x1D512 => array(0x6F),
|
|
|
1352 0x1D513 => array(0x70),
|
|
|
1353 0x1D514 => array(0x71),
|
|
|
1354 0x1D516 => array(0x73),
|
|
|
1355 0x1D517 => array(0x74),
|
|
|
1356 0x1D518 => array(0x75),
|
|
|
1357 0x1D519 => array(0x76),
|
|
|
1358 0x1D51A => array(0x77),
|
|
|
1359 0x1D51B => array(0x78),
|
|
|
1360 0x1D51C => array(0x79),
|
|
|
1361 0x1D538 => array(0x61),
|
|
|
1362 0x1D539 => array(0x62),
|
|
|
1363 0x1D53B => array(0x64),
|
|
|
1364 0x1D53C => array(0x65),
|
|
|
1365 0x1D53D => array(0x66),
|
|
|
1366 0x1D53E => array(0x67),
|
|
|
1367 0x1D540 => array(0x69),
|
|
|
1368 0x1D541 => array(0x6A),
|
|
|
1369 0x1D542 => array(0x6B),
|
|
|
1370 0x1D543 => array(0x6C),
|
|
|
1371 0x1D544 => array(0x6D),
|
|
|
1372 0x1D546 => array(0x6F),
|
|
|
1373 0x1D54A => array(0x73),
|
|
|
1374 0x1D54B => array(0x74),
|
|
|
1375 0x1D54C => array(0x75),
|
|
|
1376 0x1D54D => array(0x76),
|
|
|
1377 0x1D54E => array(0x77),
|
|
|
1378 0x1D54F => array(0x78),
|
|
|
1379 0x1D550 => array(0x79),
|
|
|
1380 0x1D56C => array(0x61),
|
|
|
1381 0x1D56D => array(0x62),
|
|
|
1382 0x1D56E => array(0x63),
|
|
|
1383 0x1D56F => array(0x64),
|
|
|
1384 0x1D570 => array(0x65),
|
|
|
1385 0x1D571 => array(0x66),
|
|
|
1386 0x1D572 => array(0x67),
|
|
|
1387 0x1D573 => array(0x68),
|
|
|
1388 0x1D574 => array(0x69),
|
|
|
1389 0x1D575 => array(0x6A),
|
|
|
1390 0x1D576 => array(0x6B),
|
|
|
1391 0x1D577 => array(0x6C),
|
|
|
1392 0x1D578 => array(0x6D),
|
|
|
1393 0x1D579 => array(0x6E),
|
|
|
1394 0x1D57A => array(0x6F),
|
|
|
1395 0x1D57B => array(0x70),
|
|
|
1396 0x1D57C => array(0x71),
|
|
|
1397 0x1D57D => array(0x72),
|
|
|
1398 0x1D57E => array(0x73),
|
|
|
1399 0x1D57F => array(0x74),
|
|
|
1400 0x1D580 => array(0x75),
|
|
|
1401 0x1D581 => array(0x76),
|
|
|
1402 0x1D582 => array(0x77),
|
|
|
1403 0x1D583 => array(0x78),
|
|
|
1404 0x1D584 => array(0x79),
|
|
|
1405 0x1D585 => array(0x7A),
|
|
|
1406 0x1D5A0 => array(0x61),
|
|
|
1407 0x1D5A1 => array(0x62),
|
|
|
1408 0x1D5A2 => array(0x63),
|
|
|
1409 0x1D5A3 => array(0x64),
|
|
|
1410 0x1D5A4 => array(0x65),
|
|
|
1411 0x1D5A5 => array(0x66),
|
|
|
1412 0x1D5A6 => array(0x67),
|
|
|
1413 0x1D5A7 => array(0x68),
|
|
|
1414 0x1D5A8 => array(0x69),
|
|
|
1415 0x1D5A9 => array(0x6A),
|
|
|
1416 0x1D5AA => array(0x6B),
|
|
|
1417 0x1D5AB => array(0x6C),
|
|
|
1418 0x1D5AC => array(0x6D),
|
|
|
1419 0x1D5AD => array(0x6E),
|
|
|
1420 0x1D5AE => array(0x6F),
|
|
|
1421 0x1D5AF => array(0x70),
|
|
|
1422 0x1D5B0 => array(0x71),
|
|
|
1423 0x1D5B1 => array(0x72),
|
|
|
1424 0x1D5B2 => array(0x73),
|
|
|
1425 0x1D5B3 => array(0x74),
|
|
|
1426 0x1D5B4 => array(0x75),
|
|
|
1427 0x1D5B5 => array(0x76),
|
|
|
1428 0x1D5B6 => array(0x77),
|
|
|
1429 0x1D5B7 => array(0x78),
|
|
|
1430 0x1D5B8 => array(0x79),
|
|
|
1431 0x1D5B9 => array(0x7A),
|
|
|
1432 0x1D5D4 => array(0x61),
|
|
|
1433 0x1D5D5 => array(0x62),
|
|
|
1434 0x1D5D6 => array(0x63),
|
|
|
1435 0x1D5D7 => array(0x64),
|
|
|
1436 0x1D5D8 => array(0x65),
|
|
|
1437 0x1D5D9 => array(0x66),
|
|
|
1438 0x1D5DA => array(0x67),
|
|
|
1439 0x1D5DB => array(0x68),
|
|
|
1440 0x1D5DC => array(0x69),
|
|
|
1441 0x1D5DD => array(0x6A),
|
|
|
1442 0x1D5DE => array(0x6B),
|
|
|
1443 0x1D5DF => array(0x6C),
|
|
|
1444 0x1D5E0 => array(0x6D),
|
|
|
1445 0x1D5E1 => array(0x6E),
|
|
|
1446 0x1D5E2 => array(0x6F),
|
|
|
1447 0x1D5E3 => array(0x70),
|
|
|
1448 0x1D5E4 => array(0x71),
|
|
|
1449 0x1D5E5 => array(0x72),
|
|
|
1450 0x1D5E6 => array(0x73),
|
|
|
1451 0x1D5E7 => array(0x74),
|
|
|
1452 0x1D5E8 => array(0x75),
|
|
|
1453 0x1D5E9 => array(0x76),
|
|
|
1454 0x1D5EA => array(0x77),
|
|
|
1455 0x1D5EB => array(0x78),
|
|
|
1456 0x1D5EC => array(0x79),
|
|
|
1457 0x1D5ED => array(0x7A),
|
|
|
1458 0x1D608 => array(0x61),
|
|
|
1459 0x1D609 => array(0x62),
|
|
|
1460 0x1D60A => array(0x63),
|
|
|
1461 0x1D60B => array(0x64),
|
|
|
1462 0x1D60C => array(0x65),
|
|
|
1463 0x1D60D => array(0x66),
|
|
|
1464 0x1D60E => array(0x67),
|
|
|
1465 0x1D60F => array(0x68),
|
|
|
1466 0x1D610 => array(0x69),
|
|
|
1467 0x1D611 => array(0x6A),
|
|
|
1468 0x1D612 => array(0x6B),
|
|
|
1469 0x1D613 => array(0x6C),
|
|
|
1470 0x1D614 => array(0x6D),
|
|
|
1471 0x1D615 => array(0x6E),
|
|
|
1472 0x1D616 => array(0x6F),
|
|
|
1473 0x1D617 => array(0x70),
|
|
|
1474 0x1D618 => array(0x71),
|
|
|
1475 0x1D619 => array(0x72),
|
|
|
1476 0x1D61A => array(0x73),
|
|
|
1477 0x1D61B => array(0x74),
|
|
|
1478 0x1D61C => array(0x75),
|
|
|
1479 0x1D61D => array(0x76),
|
|
|
1480 0x1D61E => array(0x77),
|
|
|
1481 0x1D61F => array(0x78),
|
|
|
1482 0x1D620 => array(0x79),
|
|
|
1483 0x1D621 => array(0x7A),
|
|
|
1484 0x1D63C => array(0x61),
|
|
|
1485 0x1D63D => array(0x62),
|
|
|
1486 0x1D63E => array(0x63),
|
|
|
1487 0x1D63F => array(0x64),
|
|
|
1488 0x1D640 => array(0x65),
|
|
|
1489 0x1D641 => array(0x66),
|
|
|
1490 0x1D642 => array(0x67),
|
|
|
1491 0x1D643 => array(0x68),
|
|
|
1492 0x1D644 => array(0x69),
|
|
|
1493 0x1D645 => array(0x6A),
|
|
|
1494 0x1D646 => array(0x6B),
|
|
|
1495 0x1D647 => array(0x6C),
|
|
|
1496 0x1D648 => array(0x6D),
|
|
|
1497 0x1D649 => array(0x6E),
|
|
|
1498 0x1D64A => array(0x6F),
|
|
|
1499 0x1D64B => array(0x70),
|
|
|
1500 0x1D64C => array(0x71),
|
|
|
1501 0x1D64D => array(0x72),
|
|
|
1502 0x1D64E => array(0x73),
|
|
|
1503 0x1D64F => array(0x74),
|
|
|
1504 0x1D650 => array(0x75),
|
|
|
1505 0x1D651 => array(0x76),
|
|
|
1506 0x1D652 => array(0x77),
|
|
|
1507 0x1D653 => array(0x78),
|
|
|
1508 0x1D654 => array(0x79),
|
|
|
1509 0x1D655 => array(0x7A),
|
|
|
1510 0x1D670 => array(0x61),
|
|
|
1511 0x1D671 => array(0x62),
|
|
|
1512 0x1D672 => array(0x63),
|
|
|
1513 0x1D673 => array(0x64),
|
|
|
1514 0x1D674 => array(0x65),
|
|
|
1515 0x1D675 => array(0x66),
|
|
|
1516 0x1D676 => array(0x67),
|
|
|
1517 0x1D677 => array(0x68),
|
|
|
1518 0x1D678 => array(0x69),
|
|
|
1519 0x1D679 => array(0x6A),
|
|
|
1520 0x1D67A => array(0x6B),
|
|
|
1521 0x1D67B => array(0x6C),
|
|
|
1522 0x1D67C => array(0x6D),
|
|
|
1523 0x1D67D => array(0x6E),
|
|
|
1524 0x1D67E => array(0x6F),
|
|
|
1525 0x1D67F => array(0x70),
|
|
|
1526 0x1D680 => array(0x71),
|
|
|
1527 0x1D681 => array(0x72),
|
|
|
1528 0x1D682 => array(0x73),
|
|
|
1529 0x1D683 => array(0x74),
|
|
|
1530 0x1D684 => array(0x75),
|
|
|
1531 0x1D685 => array(0x76),
|
|
|
1532 0x1D686 => array(0x77),
|
|
|
1533 0x1D687 => array(0x78),
|
|
|
1534 0x1D688 => array(0x79),
|
|
|
1535 0x1D689 => array(0x7A),
|
|
|
1536 0x1D6A8 => array(0x3B1),
|
|
|
1537 0x1D6A9 => array(0x3B2),
|
|
|
1538 0x1D6AA => array(0x3B3),
|
|
|
1539 0x1D6AB => array(0x3B4),
|
|
|
1540 0x1D6AC => array(0x3B5),
|
|
|
1541 0x1D6AD => array(0x3B6),
|
|
|
1542 0x1D6AE => array(0x3B7),
|
|
|
1543 0x1D6AF => array(0x3B8),
|
|
|
1544 0x1D6B0 => array(0x3B9),
|
|
|
1545 0x1D6B1 => array(0x3BA),
|
|
|
1546 0x1D6B2 => array(0x3BB),
|
|
|
1547 0x1D6B3 => array(0x3BC),
|
|
|
1548 0x1D6B4 => array(0x3BD),
|
|
|
1549 0x1D6B5 => array(0x3BE),
|
|
|
1550 0x1D6B6 => array(0x3BF),
|
|
|
1551 0x1D6B7 => array(0x3C0),
|
|
|
1552 0x1D6B8 => array(0x3C1),
|
|
|
1553 0x1D6B9 => array(0x3B8),
|
|
|
1554 0x1D6BA => array(0x3C3),
|
|
|
1555 0x1D6BB => array(0x3C4),
|
|
|
1556 0x1D6BC => array(0x3C5),
|
|
|
1557 0x1D6BD => array(0x3C6),
|
|
|
1558 0x1D6BE => array(0x3C7),
|
|
|
1559 0x1D6BF => array(0x3C8),
|
|
|
1560 0x1D6C0 => array(0x3C9),
|
|
|
1561 0x1D6D3 => array(0x3C3),
|
|
|
1562 0x1D6E2 => array(0x3B1),
|
|
|
1563 0x1D6E3 => array(0x3B2),
|
|
|
1564 0x1D6E4 => array(0x3B3),
|
|
|
1565 0x1D6E5 => array(0x3B4),
|
|
|
1566 0x1D6E6 => array(0x3B5),
|
|
|
1567 0x1D6E7 => array(0x3B6),
|
|
|
1568 0x1D6E8 => array(0x3B7),
|
|
|
1569 0x1D6E9 => array(0x3B8),
|
|
|
1570 0x1D6EA => array(0x3B9),
|
|
|
1571 0x1D6EB => array(0x3BA),
|
|
|
1572 0x1D6EC => array(0x3BB),
|
|
|
1573 0x1D6ED => array(0x3BC),
|
|
|
1574 0x1D6EE => array(0x3BD),
|
|
|
1575 0x1D6EF => array(0x3BE),
|
|
|
1576 0x1D6F0 => array(0x3BF),
|
|
|
1577 0x1D6F1 => array(0x3C0),
|
|
|
1578 0x1D6F2 => array(0x3C1),
|
|
|
1579 0x1D6F3 => array(0x3B8),
|
|
|
1580 0x1D6F4 => array(0x3C3),
|
|
|
1581 0x1D6F5 => array(0x3C4),
|
|
|
1582 0x1D6F6 => array(0x3C5),
|
|
|
1583 0x1D6F7 => array(0x3C6),
|
|
|
1584 0x1D6F8 => array(0x3C7),
|
|
|
1585 0x1D6F9 => array(0x3C8),
|
|
|
1586 0x1D6FA => array(0x3C9),
|
|
|
1587 0x1D70D => array(0x3C3),
|
|
|
1588 0x1D71C => array(0x3B1),
|
|
|
1589 0x1D71D => array(0x3B2),
|
|
|
1590 0x1D71E => array(0x3B3),
|
|
|
1591 0x1D71F => array(0x3B4),
|
|
|
1592 0x1D720 => array(0x3B5),
|
|
|
1593 0x1D721 => array(0x3B6),
|
|
|
1594 0x1D722 => array(0x3B7),
|
|
|
1595 0x1D723 => array(0x3B8),
|
|
|
1596 0x1D724 => array(0x3B9),
|
|
|
1597 0x1D725 => array(0x3BA),
|
|
|
1598 0x1D726 => array(0x3BB),
|
|
|
1599 0x1D727 => array(0x3BC),
|
|
|
1600 0x1D728 => array(0x3BD),
|
|
|
1601 0x1D729 => array(0x3BE),
|
|
|
1602 0x1D72A => array(0x3BF),
|
|
|
1603 0x1D72B => array(0x3C0),
|
|
|
1604 0x1D72C => array(0x3C1),
|
|
|
1605 0x1D72D => array(0x3B8),
|
|
|
1606 0x1D72E => array(0x3C3),
|
|
|
1607 0x1D72F => array(0x3C4),
|
|
|
1608 0x1D730 => array(0x3C5),
|
|
|
1609 0x1D731 => array(0x3C6),
|
|
|
1610 0x1D732 => array(0x3C7),
|
|
|
1611 0x1D733 => array(0x3C8),
|
|
|
1612 0x1D734 => array(0x3C9),
|
|
|
1613 0x1D747 => array(0x3C3),
|
|
|
1614 0x1D756 => array(0x3B1),
|
|
|
1615 0x1D757 => array(0x3B2),
|
|
|
1616 0x1D758 => array(0x3B3),
|
|
|
1617 0x1D759 => array(0x3B4),
|
|
|
1618 0x1D75A => array(0x3B5),
|
|
|
1619 0x1D75B => array(0x3B6),
|
|
|
1620 0x1D75C => array(0x3B7),
|
|
|
1621 0x1D75D => array(0x3B8),
|
|
|
1622 0x1D75E => array(0x3B9),
|
|
|
1623 0x1D75F => array(0x3BA),
|
|
|
1624 0x1D760 => array(0x3BB),
|
|
|
1625 0x1D761 => array(0x3BC),
|
|
|
1626 0x1D762 => array(0x3BD),
|
|
|
1627 0x1D763 => array(0x3BE),
|
|
|
1628 0x1D764 => array(0x3BF),
|
|
|
1629 0x1D765 => array(0x3C0),
|
|
|
1630 0x1D766 => array(0x3C1),
|
|
|
1631 0x1D767 => array(0x3B8),
|
|
|
1632 0x1D768 => array(0x3C3),
|
|
|
1633 0x1D769 => array(0x3C4),
|
|
|
1634 0x1D76A => array(0x3C5),
|
|
|
1635 0x1D76B => array(0x3C6),
|
|
|
1636 0x1D76C => array(0x3C7),
|
|
|
1637 0x1D76D => array(0x3C8),
|
|
|
1638 0x1D76E => array(0x3C9),
|
|
|
1639 0x1D781 => array(0x3C3),
|
|
|
1640 0x1D790 => array(0x3B1),
|
|
|
1641 0x1D791 => array(0x3B2),
|
|
|
1642 0x1D792 => array(0x3B3),
|
|
|
1643 0x1D793 => array(0x3B4),
|
|
|
1644 0x1D794 => array(0x3B5),
|
|
|
1645 0x1D795 => array(0x3B6),
|
|
|
1646 0x1D796 => array(0x3B7),
|
|
|
1647 0x1D797 => array(0x3B8),
|
|
|
1648 0x1D798 => array(0x3B9),
|
|
|
1649 0x1D799 => array(0x3BA),
|
|
|
1650 0x1D79A => array(0x3BB),
|
|
|
1651 0x1D79B => array(0x3BC),
|
|
|
1652 0x1D79C => array(0x3BD),
|
|
|
1653 0x1D79D => array(0x3BE),
|
|
|
1654 0x1D79E => array(0x3BF),
|
|
|
1655 0x1D79F => array(0x3C0),
|
|
|
1656 0x1D7A0 => array(0x3C1),
|
|
|
1657 0x1D7A1 => array(0x3B8),
|
|
|
1658 0x1D7A2 => array(0x3C3),
|
|
|
1659 0x1D7A3 => array(0x3C4),
|
|
|
1660 0x1D7A4 => array(0x3C5),
|
|
|
1661 0x1D7A5 => array(0x3C6),
|
|
|
1662 0x1D7A6 => array(0x3C7),
|
|
|
1663 0x1D7A7 => array(0x3C8),
|
|
|
1664 0x1D7A8 => array(0x3C9),
|
|
|
1665 0x1D7BB => array(0x3C3),
|
|
|
1666 0x3F9 => array(0x3C3),
|
|
|
1667 0x1D2C => array(0x61),
|
|
|
1668 0x1D2D => array(0xE6),
|
|
|
1669 0x1D2E => array(0x62),
|
|
|
1670 0x1D30 => array(0x64),
|
|
|
1671 0x1D31 => array(0x65),
|
|
|
1672 0x1D32 => array(0x1DD),
|
|
|
1673 0x1D33 => array(0x67),
|
|
|
1674 0x1D34 => array(0x68),
|
|
|
1675 0x1D35 => array(0x69),
|
|
|
1676 0x1D36 => array(0x6A),
|
|
|
1677 0x1D37 => array(0x6B),
|
|
|
1678 0x1D38 => array(0x6C),
|
|
|
1679 0x1D39 => array(0x6D),
|
|
|
1680 0x1D3A => array(0x6E),
|
|
|
1681 0x1D3C => array(0x6F),
|
|
|
1682 0x1D3D => array(0x223),
|
|
|
1683 0x1D3E => array(0x70),
|
|
|
1684 0x1D3F => array(0x72),
|
|
|
1685 0x1D40 => array(0x74),
|
|
|
1686 0x1D41 => array(0x75),
|
|
|
1687 0x1D42 => array(0x77),
|
|
|
1688 0x213B => array(0x66, 0x61, 0x78),
|
|
|
1689 0x3250 => array(0x70, 0x74, 0x65),
|
|
|
1690 0x32CC => array(0x68, 0x67),
|
|
|
1691 0x32CE => array(0x65, 0x76),
|
|
|
1692 0x32CF => array(0x6C, 0x74, 0x64),
|
|
|
1693 0x337A => array(0x69, 0x75),
|
|
|
1694 0x33DE => array(0x76, 0x2215, 0x6D),
|
|
|
1695 0x33DF => array(0x61, 0x2215, 0x6D)
|
|
|
1696 );
|
|
|
1697
|
|
|
1698 /**
|
|
|
1699 * Normalization Combining Classes; Code Points not listed
|
|
|
1700 * got Combining Class 0.
|
|
|
1701 *
|
|
|
1702 * @static
|
|
|
1703 * @var array
|
|
|
1704 * @access private
|
|
|
1705 */
|
|
|
1706 private static $_np_norm_combcls = array(
|
|
|
1707 0x334 => 1,
|
|
|
1708 0x335 => 1,
|
|
|
1709 0x336 => 1,
|
|
|
1710 0x337 => 1,
|
|
|
1711 0x338 => 1,
|
|
|
1712 0x93C => 7,
|
|
|
1713 0x9BC => 7,
|
|
|
1714 0xA3C => 7,
|
|
|
1715 0xABC => 7,
|
|
|
1716 0xB3C => 7,
|
|
|
1717 0xCBC => 7,
|
|
|
1718 0x1037 => 7,
|
|
|
1719 0x3099 => 8,
|
|
|
1720 0x309A => 8,
|
|
|
1721 0x94D => 9,
|
|
|
1722 0x9CD => 9,
|
|
|
1723 0xA4D => 9,
|
|
|
1724 0xACD => 9,
|
|
|
1725 0xB4D => 9,
|
|
|
1726 0xBCD => 9,
|
|
|
1727 0xC4D => 9,
|
|
|
1728 0xCCD => 9,
|
|
|
1729 0xD4D => 9,
|
|
|
1730 0xDCA => 9,
|
|
|
1731 0xE3A => 9,
|
|
|
1732 0xF84 => 9,
|
|
|
1733 0x1039 => 9,
|
|
|
1734 0x1714 => 9,
|
|
|
1735 0x1734 => 9,
|
|
|
1736 0x17D2 => 9,
|
|
|
1737 0x5B0 => 10,
|
|
|
1738 0x5B1 => 11,
|
|
|
1739 0x5B2 => 12,
|
|
|
1740 0x5B3 => 13,
|
|
|
1741 0x5B4 => 14,
|
|
|
1742 0x5B5 => 15,
|
|
|
1743 0x5B6 => 16,
|
|
|
1744 0x5B7 => 17,
|
|
|
1745 0x5B8 => 18,
|
|
|
1746 0x5B9 => 19,
|
|
|
1747 0x5BB => 20,
|
|
|
1748 0x5Bc => 21,
|
|
|
1749 0x5BD => 22,
|
|
|
1750 0x5BF => 23,
|
|
|
1751 0x5C1 => 24,
|
|
|
1752 0x5C2 => 25,
|
|
|
1753 0xFB1E => 26,
|
|
|
1754 0x64B => 27,
|
|
|
1755 0x64C => 28,
|
|
|
1756 0x64D => 29,
|
|
|
1757 0x64E => 30,
|
|
|
1758 0x64F => 31,
|
|
|
1759 0x650 => 32,
|
|
|
1760 0x651 => 33,
|
|
|
1761 0x652 => 34,
|
|
|
1762 0x670 => 35,
|
|
|
1763 0x711 => 36,
|
|
|
1764 0xC55 => 84,
|
|
|
1765 0xC56 => 91,
|
|
|
1766 0xE38 => 103,
|
|
|
1767 0xE39 => 103,
|
|
|
1768 0xE48 => 107,
|
|
|
1769 0xE49 => 107,
|
|
|
1770 0xE4A => 107,
|
|
|
1771 0xE4B => 107,
|
|
|
1772 0xEB8 => 118,
|
|
|
1773 0xEB9 => 118,
|
|
|
1774 0xEC8 => 122,
|
|
|
1775 0xEC9 => 122,
|
|
|
1776 0xECA => 122,
|
|
|
1777 0xECB => 122,
|
|
|
1778 0xF71 => 129,
|
|
|
1779 0xF72 => 130,
|
|
|
1780 0xF7A => 130,
|
|
|
1781 0xF7B => 130,
|
|
|
1782 0xF7C => 130,
|
|
|
1783 0xF7D => 130,
|
|
|
1784 0xF80 => 130,
|
|
|
1785 0xF74 => 132,
|
|
|
1786 0x321 => 202,
|
|
|
1787 0x322 => 202,
|
|
|
1788 0x327 => 202,
|
|
|
1789 0x328 => 202,
|
|
|
1790 0x31B => 216,
|
|
|
1791 0xF39 => 216,
|
|
|
1792 0x1D165 => 216,
|
|
|
1793 0x1D166 => 216,
|
|
|
1794 0x1D16E => 216,
|
|
|
1795 0x1D16F => 216,
|
|
|
1796 0x1D170 => 216,
|
|
|
1797 0x1D171 => 216,
|
|
|
1798 0x1D172 => 216,
|
|
|
1799 0x302A => 218,
|
|
|
1800 0x316 => 220,
|
|
|
1801 0x317 => 220,
|
|
|
1802 0x318 => 220,
|
|
|
1803 0x319 => 220,
|
|
|
1804 0x31C => 220,
|
|
|
1805 0x31D => 220,
|
|
|
1806 0x31E => 220,
|
|
|
1807 0x31F => 220,
|
|
|
1808 0x320 => 220,
|
|
|
1809 0x323 => 220,
|
|
|
1810 0x324 => 220,
|
|
|
1811 0x325 => 220,
|
|
|
1812 0x326 => 220,
|
|
|
1813 0x329 => 220,
|
|
|
1814 0x32A => 220,
|
|
|
1815 0x32B => 220,
|
|
|
1816 0x32C => 220,
|
|
|
1817 0x32D => 220,
|
|
|
1818 0x32E => 220,
|
|
|
1819 0x32F => 220,
|
|
|
1820 0x330 => 220,
|
|
|
1821 0x331 => 220,
|
|
|
1822 0x332 => 220,
|
|
|
1823 0x333 => 220,
|
|
|
1824 0x339 => 220,
|
|
|
1825 0x33A => 220,
|
|
|
1826 0x33B => 220,
|
|
|
1827 0x33C => 220,
|
|
|
1828 0x347 => 220,
|
|
|
1829 0x348 => 220,
|
|
|
1830 0x349 => 220,
|
|
|
1831 0x34D => 220,
|
|
|
1832 0x34E => 220,
|
|
|
1833 0x353 => 220,
|
|
|
1834 0x354 => 220,
|
|
|
1835 0x355 => 220,
|
|
|
1836 0x356 => 220,
|
|
|
1837 0x591 => 220,
|
|
|
1838 0x596 => 220,
|
|
|
1839 0x59B => 220,
|
|
|
1840 0x5A3 => 220,
|
|
|
1841 0x5A4 => 220,
|
|
|
1842 0x5A5 => 220,
|
|
|
1843 0x5A6 => 220,
|
|
|
1844 0x5A7 => 220,
|
|
|
1845 0x5AA => 220,
|
|
|
1846 0x655 => 220,
|
|
|
1847 0x656 => 220,
|
|
|
1848 0x6E3 => 220,
|
|
|
1849 0x6EA => 220,
|
|
|
1850 0x6ED => 220,
|
|
|
1851 0x731 => 220,
|
|
|
1852 0x734 => 220,
|
|
|
1853 0x737 => 220,
|
|
|
1854 0x738 => 220,
|
|
|
1855 0x739 => 220,
|
|
|
1856 0x73B => 220,
|
|
|
1857 0x73C => 220,
|
|
|
1858 0x73E => 220,
|
|
|
1859 0x742 => 220,
|
|
|
1860 0x744 => 220,
|
|
|
1861 0x746 => 220,
|
|
|
1862 0x748 => 220,
|
|
|
1863 0x952 => 220,
|
|
|
1864 0xF18 => 220,
|
|
|
1865 0xF19 => 220,
|
|
|
1866 0xF35 => 220,
|
|
|
1867 0xF37 => 220,
|
|
|
1868 0xFC6 => 220,
|
|
|
1869 0x193B => 220,
|
|
|
1870 0x20E8 => 220,
|
|
|
1871 0x1D17B => 220,
|
|
|
1872 0x1D17C => 220,
|
|
|
1873 0x1D17D => 220,
|
|
|
1874 0x1D17E => 220,
|
|
|
1875 0x1D17F => 220,
|
|
|
1876 0x1D180 => 220,
|
|
|
1877 0x1D181 => 220,
|
|
|
1878 0x1D182 => 220,
|
|
|
1879 0x1D18A => 220,
|
|
|
1880 0x1D18B => 220,
|
|
|
1881 0x59A => 222,
|
|
|
1882 0x5AD => 222,
|
|
|
1883 0x1929 => 222,
|
|
|
1884 0x302D => 222,
|
|
|
1885 0x302E => 224,
|
|
|
1886 0x302F => 224,
|
|
|
1887 0x1D16D => 226,
|
|
|
1888 0x5AE => 228,
|
|
|
1889 0x18A9 => 228,
|
|
|
1890 0x302B => 228,
|
|
|
1891 0x300 => 230,
|
|
|
1892 0x301 => 230,
|
|
|
1893 0x302 => 230,
|
|
|
1894 0x303 => 230,
|
|
|
1895 0x304 => 230,
|
|
|
1896 0x305 => 230,
|
|
|
1897 0x306 => 230,
|
|
|
1898 0x307 => 230,
|
|
|
1899 0x308 => 230,
|
|
|
1900 0x309 => 230,
|
|
|
1901 0x30A => 230,
|
|
|
1902 0x30B => 230,
|
|
|
1903 0x30C => 230,
|
|
|
1904 0x30D => 230,
|
|
|
1905 0x30E => 230,
|
|
|
1906 0x30F => 230,
|
|
|
1907 0x310 => 230,
|
|
|
1908 0x311 => 230,
|
|
|
1909 0x312 => 230,
|
|
|
1910 0x313 => 230,
|
|
|
1911 0x314 => 230,
|
|
|
1912 0x33D => 230,
|
|
|
1913 0x33E => 230,
|
|
|
1914 0x33F => 230,
|
|
|
1915 0x340 => 230,
|
|
|
1916 0x341 => 230,
|
|
|
1917 0x342 => 230,
|
|
|
1918 0x343 => 230,
|
|
|
1919 0x344 => 230,
|
|
|
1920 0x346 => 230,
|
|
|
1921 0x34A => 230,
|
|
|
1922 0x34B => 230,
|
|
|
1923 0x34C => 230,
|
|
|
1924 0x350 => 230,
|
|
|
1925 0x351 => 230,
|
|
|
1926 0x352 => 230,
|
|
|
1927 0x357 => 230,
|
|
|
1928 0x363 => 230,
|
|
|
1929 0x364 => 230,
|
|
|
1930 0x365 => 230,
|
|
|
1931 0x366 => 230,
|
|
|
1932 0x367 => 230,
|
|
|
1933 0x368 => 230,
|
|
|
1934 0x369 => 230,
|
|
|
1935 0x36A => 230,
|
|
|
1936 0x36B => 230,
|
|
|
1937 0x36C => 230,
|
|
|
1938 0x36D => 230,
|
|
|
1939 0x36E => 230,
|
|
|
1940 0x36F => 230,
|
|
|
1941 0x483 => 230,
|
|
|
1942 0x484 => 230,
|
|
|
1943 0x485 => 230,
|
|
|
1944 0x486 => 230,
|
|
|
1945 0x592 => 230,
|
|
|
1946 0x593 => 230,
|
|
|
1947 0x594 => 230,
|
|
|
1948 0x595 => 230,
|
|
|
1949 0x597 => 230,
|
|
|
1950 0x598 => 230,
|
|
|
1951 0x599 => 230,
|
|
|
1952 0x59C => 230,
|
|
|
1953 0x59D => 230,
|
|
|
1954 0x59E => 230,
|
|
|
1955 0x59F => 230,
|
|
|
1956 0x5A0 => 230,
|
|
|
1957 0x5A1 => 230,
|
|
|
1958 0x5A8 => 230,
|
|
|
1959 0x5A9 => 230,
|
|
|
1960 0x5AB => 230,
|
|
|
1961 0x5AC => 230,
|
|
|
1962 0x5AF => 230,
|
|
|
1963 0x5C4 => 230,
|
|
|
1964 0x610 => 230,
|
|
|
1965 0x611 => 230,
|
|
|
1966 0x612 => 230,
|
|
|
1967 0x613 => 230,
|
|
|
1968 0x614 => 230,
|
|
|
1969 0x615 => 230,
|
|
|
1970 0x653 => 230,
|
|
|
1971 0x654 => 230,
|
|
|
1972 0x657 => 230,
|
|
|
1973 0x658 => 230,
|
|
|
1974 0x6D6 => 230,
|
|
|
1975 0x6D7 => 230,
|
|
|
1976 0x6D8 => 230,
|
|
|
1977 0x6D9 => 230,
|
|
|
1978 0x6DA => 230,
|
|
|
1979 0x6DB => 230,
|
|
|
1980 0x6DC => 230,
|
|
|
1981 0x6DF => 230,
|
|
|
1982 0x6E0 => 230,
|
|
|
1983 0x6E1 => 230,
|
|
|
1984 0x6E2 => 230,
|
|
|
1985 0x6E4 => 230,
|
|
|
1986 0x6E7 => 230,
|
|
|
1987 0x6E8 => 230,
|
|
|
1988 0x6EB => 230,
|
|
|
1989 0x6EC => 230,
|
|
|
1990 0x730 => 230,
|
|
|
1991 0x732 => 230,
|
|
|
1992 0x733 => 230,
|
|
|
1993 0x735 => 230,
|
|
|
1994 0x736 => 230,
|
|
|
1995 0x73A => 230,
|
|
|
1996 0x73D => 230,
|
|
|
1997 0x73F => 230,
|
|
|
1998 0x740 => 230,
|
|
|
1999 0x741 => 230,
|
|
|
2000 0x743 => 230,
|
|
|
2001 0x745 => 230,
|
|
|
2002 0x747 => 230,
|
|
|
2003 0x749 => 230,
|
|
|
2004 0x74A => 230,
|
|
|
2005 0x951 => 230,
|
|
|
2006 0x953 => 230,
|
|
|
2007 0x954 => 230,
|
|
|
2008 0xF82 => 230,
|
|
|
2009 0xF83 => 230,
|
|
|
2010 0xF86 => 230,
|
|
|
2011 0xF87 => 230,
|
|
|
2012 0x170D => 230,
|
|
|
2013 0x193A => 230,
|
|
|
2014 0x20D0 => 230,
|
|
|
2015 0x20D1 => 230,
|
|
|
2016 0x20D4 => 230,
|
|
|
2017 0x20D5 => 230,
|
|
|
2018 0x20D6 => 230,
|
|
|
2019 0x20D7 => 230,
|
|
|
2020 0x20DB => 230,
|
|
|
2021 0x20DC => 230,
|
|
|
2022 0x20E1 => 230,
|
|
|
2023 0x20E7 => 230,
|
|
|
2024 0x20E9 => 230,
|
|
|
2025 0xFE20 => 230,
|
|
|
2026 0xFE21 => 230,
|
|
|
2027 0xFE22 => 230,
|
|
|
2028 0xFE23 => 230,
|
|
|
2029 0x1D185 => 230,
|
|
|
2030 0x1D186 => 230,
|
|
|
2031 0x1D187 => 230,
|
|
|
2032 0x1D189 => 230,
|
|
|
2033 0x1D188 => 230,
|
|
|
2034 0x1D1AA => 230,
|
|
|
2035 0x1D1AB => 230,
|
|
|
2036 0x1D1AC => 230,
|
|
|
2037 0x1D1AD => 230,
|
|
|
2038 0x315 => 232,
|
|
|
2039 0x31A => 232,
|
|
|
2040 0x302C => 232,
|
|
|
2041 0x35F => 233,
|
|
|
2042 0x362 => 233,
|
|
|
2043 0x35D => 234,
|
|
|
2044 0x35E => 234,
|
|
|
2045 0x360 => 234,
|
|
|
2046 0x361 => 234,
|
|
|
2047 0x345 => 240
|
|
|
2048 );
|
|
|
2049 // }}}
|
|
|
2050
|
|
|
2051 // {{{ properties
|
|
|
2052 /**
|
|
|
2053 * @var string
|
|
|
2054 * @access private
|
|
|
2055 */
|
|
|
2056 private $_punycode_prefix = 'xn--';
|
|
|
2057
|
|
|
2058 /**
|
|
|
2059 * @access private
|
|
|
2060 */
|
|
|
2061 private $_invalid_ucs = 0x80000000;
|
|
|
2062
|
|
|
2063 /**
|
|
|
2064 * @access private
|
|
|
2065 */
|
|
|
2066 private $_max_ucs = 0x10FFFF;
|
|
|
2067
|
|
|
2068 /**
|
|
|
2069 * @var int
|
|
|
2070 * @access private
|
|
|
2071 */
|
|
|
2072 private $_base = 36;
|
|
|
2073
|
|
|
2074 /**
|
|
|
2075 * @var int
|
|
|
2076 * @access private
|
|
|
2077 */
|
|
|
2078 private $_tmin = 1;
|
|
|
2079
|
|
|
2080 /**
|
|
|
2081 * @var int
|
|
|
2082 * @access private
|
|
|
2083 */
|
|
|
2084 private $_tmax = 26;
|
|
|
2085
|
|
|
2086 /**
|
|
|
2087 * @var int
|
|
|
2088 * @access private
|
|
|
2089 */
|
|
|
2090 private $_skew = 38;
|
|
|
2091
|
|
|
2092 /**
|
|
|
2093 * @var int
|
|
|
2094 * @access private
|
|
|
2095 */
|
|
|
2096 private $_damp = 700;
|
|
|
2097
|
|
|
2098 /**
|
|
|
2099 * @var int
|
|
|
2100 * @access private
|
|
|
2101 */
|
|
|
2102 private $_initial_bias = 72;
|
|
|
2103
|
|
|
2104 /**
|
|
|
2105 * @var int
|
|
|
2106 * @access private
|
|
|
2107 */
|
|
|
2108 private $_initial_n = 0x80;
|
|
|
2109
|
|
|
2110 /**
|
|
|
2111 * @var int
|
|
|
2112 * @access private
|
|
|
2113 */
|
|
|
2114 private $_slast;
|
|
|
2115
|
|
|
2116 /**
|
|
|
2117 * @access private
|
|
|
2118 */
|
|
|
2119 private $_sbase = 0xAC00;
|
|
|
2120
|
|
|
2121 /**
|
|
|
2122 * @access private
|
|
|
2123 */
|
|
|
2124 private $_lbase = 0x1100;
|
|
|
2125
|
|
|
2126 /**
|
|
|
2127 * @access private
|
|
|
2128 */
|
|
|
2129 private $_vbase = 0x1161;
|
|
|
2130
|
|
|
2131 /**
|
|
|
2132 * @access private
|
|
|
2133 */
|
|
|
2134 private $_tbase = 0x11a7;
|
|
|
2135
|
|
|
2136 /**
|
|
|
2137 * @var int
|
|
|
2138 * @access private
|
|
|
2139 */
|
|
|
2140 private $_lcount = 19;
|
|
|
2141
|
|
|
2142 /**
|
|
|
2143 * @var int
|
|
|
2144 * @access private
|
|
|
2145 */
|
|
|
2146 private $_vcount = 21;
|
|
|
2147
|
|
|
2148 /**
|
|
|
2149 * @var int
|
|
|
2150 * @access private
|
|
|
2151 */
|
|
|
2152 private $_tcount = 28;
|
|
|
2153
|
|
|
2154 /**
|
|
|
2155 * vcount * tcount
|
|
|
2156 *
|
|
|
2157 * @var int
|
|
|
2158 * @access private
|
|
|
2159 */
|
|
|
2160 private $_ncount = 588;
|
|
|
2161
|
|
|
2162 /**
|
|
|
2163 * lcount * tcount * vcount
|
|
|
2164 *
|
|
|
2165 * @var int
|
|
|
2166 * @access private
|
|
|
2167 */
|
|
|
2168 private $_scount = 11172;
|
|
|
2169
|
|
|
2170 /**
|
|
|
2171 * Default encoding for encode()'s input and decode()'s output is UTF-8;
|
|
|
2172 * Other possible encodings are ucs4_string and ucs4_array
|
|
|
2173 * See {@link setParams()} for how to select these
|
|
|
2174 *
|
|
|
2175 * @var bool
|
|
|
2176 * @access private
|
|
|
2177 */
|
|
|
2178 private $_api_encoding = 'utf8';
|
|
|
2179
|
|
|
2180 /**
|
|
|
2181 * Overlong UTF-8 encodings are forbidden
|
|
|
2182 *
|
|
|
2183 * @var bool
|
|
|
2184 * @access private
|
|
|
2185 */
|
|
|
2186 private $_allow_overlong = false;
|
|
|
2187
|
|
|
2188 /**
|
|
|
2189 * Behave strict or not
|
|
|
2190 *
|
|
|
2191 * @var bool
|
|
|
2192 * @access private
|
|
|
2193 */
|
|
|
2194 private $_strict_mode = false;
|
|
|
2195
|
|
|
2196 /**
|
|
|
2197 * IDNA-version to use
|
|
|
2198 *
|
|
|
2199 * Values are "2003" and "2008".
|
|
|
2200 * Defaults to "2003", since that was the original version and for
|
|
|
2201 * compatibility with previous versions of this library.
|
|
|
2202 * If you need to encode "new" characters like the German "Eszett",
|
|
|
2203 * please switch to 2008 first before encoding.
|
|
|
2204 *
|
|
|
2205 * @var bool
|
|
|
2206 * @access private
|
|
|
2207 */
|
|
|
2208 private $_version = '2003';
|
|
|
2209
|
|
|
2210 /**
|
|
|
2211 * Cached value indicating whether or not mbstring function overloading is
|
|
|
2212 * on for strlen
|
|
|
2213 *
|
|
|
2214 * This is cached for optimal performance.
|
|
|
2215 *
|
|
|
2216 * @var boolean
|
|
|
2217 * @see Net_IDNA2::_byteLength()
|
|
|
2218 */
|
|
|
2219 private static $_mb_string_overload = null;
|
|
|
2220 // }}}
|
|
|
2221
|
|
|
2222
|
|
|
2223 // {{{ constructor
|
|
|
2224 /**
|
|
|
2225 * Constructor
|
|
|
2226 *
|
|
|
2227 * @param array $options Options to initialise the object with
|
|
|
2228 *
|
|
|
2229 * @access public
|
|
|
2230 * @see setParams()
|
|
|
2231 */
|
|
|
2232 public function __construct($options = null)
|
|
|
2233 {
|
|
|
2234 $this->_slast = $this->_sbase + $this->_lcount * $this->_vcount * $this->_tcount;
|
|
|
2235
|
|
|
2236 if (is_array($options)) {
|
|
|
2237 $this->setParams($options);
|
|
|
2238 }
|
|
|
2239
|
|
|
2240 // populate mbstring overloading cache if not set
|
|
|
2241 if (self::$_mb_string_overload === null) {
|
|
|
2242 self::$_mb_string_overload = (extension_loaded('mbstring')
|
|
|
2243 && (ini_get('mbstring.func_overload') & 0x02) === 0x02);
|
|
|
2244 }
|
|
|
2245 }
|
|
|
2246 // }}}
|
|
|
2247
|
|
|
2248
|
|
|
2249 /**
|
|
|
2250 * Sets a new option value. Available options and values:
|
|
|
2251 *
|
|
|
2252 * [utf8 - Use either UTF-8 or ISO-8859-1 as input (true for UTF-8, false
|
|
|
2253 * otherwise); The output is always UTF-8]
|
|
|
2254 * [overlong - Unicode does not allow unnecessarily long encodings of chars,
|
|
|
2255 * to allow this, set this parameter to true, else to false;
|
|
|
2256 * default is false.]
|
|
|
2257 * [strict - true: strict mode, good for registration purposes - Causes errors
|
|
|
2258 * on failures; false: loose mode, ideal for "wildlife" applications
|
|
|
2259 * by silently ignoring errors and returning the original input instead]
|
|
|
2260 *
|
|
|
2261 * @param mixed $option Parameter to set (string: single parameter; array of Parameter => Value pairs)
|
|
|
2262 * @param string $value Value to use (if parameter 1 is a string)
|
|
|
2263 *
|
|
|
2264 * @return boolean true on success, false otherwise
|
|
|
2265 * @access public
|
|
|
2266 */
|
|
|
2267 public function setParams($option, $value = false)
|
|
|
2268 {
|
|
|
2269 if (!is_array($option)) {
|
|
|
2270 $option = array($option => $value);
|
|
|
2271 }
|
|
|
2272
|
|
|
2273 foreach ($option as $k => $v) {
|
|
|
2274 switch ($k) {
|
|
|
2275 case 'encoding':
|
|
|
2276 switch ($v) {
|
|
|
2277 case 'utf8':
|
|
|
2278 case 'ucs4_string':
|
|
|
2279 case 'ucs4_array':
|
|
|
2280 $this->_api_encoding = $v;
|
|
|
2281 break;
|
|
|
2282
|
|
|
2283 default:
|
|
|
2284 throw new InvalidArgumentException('Set Parameter: Unknown parameter '.$v.' for option '.$k);
|
|
|
2285 }
|
|
|
2286
|
|
|
2287 break;
|
|
|
2288
|
|
|
2289 case 'overlong':
|
|
|
2290 $this->_allow_overlong = ($v) ? true : false;
|
|
|
2291 break;
|
|
|
2292
|
|
|
2293 case 'strict':
|
|
|
2294 $this->_strict_mode = ($v) ? true : false;
|
|
|
2295 break;
|
|
|
2296
|
|
|
2297 case 'version':
|
|
|
2298 if (in_array($v, array('2003', '2008'))) {
|
|
|
2299 $this->_version = $v;
|
|
|
2300 } else {
|
|
|
2301 throw new InvalidArgumentException('Set Parameter: Invalid parameter '.$v.' for option '.$k);
|
|
|
2302 }
|
|
|
2303 break;
|
|
|
2304
|
|
|
2305 default:
|
|
|
2306 return false;
|
|
|
2307 }
|
|
|
2308 }
|
|
|
2309
|
|
|
2310 return true;
|
|
|
2311 }
|
|
|
2312
|
|
|
2313 /**
|
|
|
2314 * Encode a given UTF-8 domain name.
|
|
|
2315 *
|
|
|
2316 * @param string $decoded Domain name (UTF-8 or UCS-4)
|
|
|
2317 * @param string $one_time_encoding Desired input encoding, see {@link set_parameter}
|
|
|
2318 * If not given will use default-encoding
|
|
|
2319 *
|
|
|
2320 * @return string Encoded Domain name (ACE string)
|
|
|
2321 * @return mixed processed string
|
|
|
2322 * @throws Exception
|
|
|
2323 * @access public
|
|
|
2324 */
|
|
|
2325 public function encode($decoded, $one_time_encoding = false)
|
|
|
2326 {
|
|
|
2327 // Forcing conversion of input to UCS4 array
|
|
|
2328 // If one time encoding is given, use this, else the objects property
|
|
|
2329 switch (($one_time_encoding) ? $one_time_encoding : $this->_api_encoding) {
|
|
|
2330 case 'utf8':
|
|
|
2331 $decoded = $this->_utf8_to_ucs4($decoded);
|
|
|
2332 break;
|
|
|
2333 case 'ucs4_string':
|
|
|
2334 $decoded = $this->_ucs4_string_to_ucs4($decoded);
|
|
|
2335 case 'ucs4_array': // No break; before this line. Catch case, but do nothing
|
|
|
2336 break;
|
|
|
2337 default:
|
|
|
2338 throw new InvalidArgumentException('Unsupported input format');
|
|
|
2339 }
|
|
|
2340
|
|
|
2341 // No input, no output, what else did you expect?
|
|
|
2342 if (empty($decoded)) return '';
|
|
|
2343
|
|
|
2344 // Anchors for iteration
|
|
|
2345 $last_begin = 0;
|
|
|
2346 // Output string
|
|
|
2347 $output = '';
|
|
|
2348
|
|
|
2349 foreach ($decoded as $k => $v) {
|
|
|
2350 // Make sure to use just the plain dot
|
|
|
2351 switch($v) {
|
|
|
2352 case 0x3002:
|
|
|
2353 case 0xFF0E:
|
|
|
2354 case 0xFF61:
|
|
|
2355 $decoded[$k] = 0x2E;
|
|
|
2356 // It's right, no break here
|
|
|
2357 // The codepoints above have to be converted to dots anyway
|
|
|
2358
|
|
|
2359 // Stumbling across an anchoring character
|
|
|
2360 case 0x2E:
|
|
|
2361 case 0x2F:
|
|
|
2362 case 0x3A:
|
|
|
2363 case 0x3F:
|
|
|
2364 case 0x40:
|
|
|
2365 // Neither email addresses nor URLs allowed in strict mode
|
|
|
2366 if ($this->_strict_mode) {
|
|
|
2367 throw new InvalidArgumentException('Neither email addresses nor URLs are allowed in strict mode.');
|
|
|
2368 }
|
|
|
2369 // Skip first char
|
|
|
2370 if ($k) {
|
|
|
2371 $encoded = '';
|
|
|
2372 $encoded = $this->_encode(array_slice($decoded, $last_begin, (($k)-$last_begin)));
|
|
|
2373 if ($encoded) {
|
|
|
2374 $output .= $encoded;
|
|
|
2375 } else {
|
|
|
2376 $output .= $this->_ucs4_to_utf8(array_slice($decoded, $last_begin, (($k)-$last_begin)));
|
|
|
2377 }
|
|
|
2378 $output .= chr($decoded[$k]);
|
|
|
2379 }
|
|
|
2380 $last_begin = $k + 1;
|
|
|
2381 }
|
|
|
2382 }
|
|
|
2383 // Catch the rest of the string
|
|
|
2384 if ($last_begin) {
|
|
|
2385 $inp_len = sizeof($decoded);
|
|
|
2386 $encoded = '';
|
|
|
2387 $encoded = $this->_encode(array_slice($decoded, $last_begin, (($inp_len)-$last_begin)));
|
|
|
2388 if ($encoded) {
|
|
|
2389 $output .= $encoded;
|
|
|
2390 } else {
|
|
|
2391 $output .= $this->_ucs4_to_utf8(array_slice($decoded, $last_begin, (($inp_len)-$last_begin)));
|
|
|
2392 }
|
|
|
2393 return $output;
|
|
|
2394 }
|
|
|
2395
|
|
|
2396 if ($output = $this->_encode($decoded)) {
|
|
|
2397 return $output;
|
|
|
2398 }
|
|
|
2399
|
|
|
2400 return $this->_ucs4_to_utf8($decoded);
|
|
|
2401 }
|
|
|
2402
|
|
|
2403 /**
|
|
|
2404 * Decode a given ACE domain name.
|
|
|
2405 *
|
|
|
2406 * @param string $input Domain name (ACE string)
|
|
|
2407 * @param string $one_time_encoding Desired output encoding, see {@link set_parameter}
|
|
|
2408 *
|
|
|
2409 * @return string Decoded Domain name (UTF-8 or UCS-4)
|
|
|
2410 * @throws Exception
|
|
|
2411 * @access public
|
|
|
2412 */
|
|
|
2413 public function decode($input, $one_time_encoding = false)
|
|
|
2414 {
|
|
|
2415 // Optionally set
|
|
|
2416 if ($one_time_encoding) {
|
|
|
2417 switch ($one_time_encoding) {
|
|
|
2418 case 'utf8':
|
|
|
2419 case 'ucs4_string':
|
|
|
2420 case 'ucs4_array':
|
|
|
2421 break;
|
|
|
2422 default:
|
|
|
2423 throw new InvalidArgumentException('Unknown encoding '.$one_time_encoding);
|
|
|
2424 }
|
|
|
2425 }
|
|
|
2426 // Make sure to drop any newline characters around
|
|
|
2427 $input = trim($input);
|
|
|
2428
|
|
|
2429 // Negotiate input and try to determine, whether it is a plain string,
|
|
|
2430 // an email address or something like a complete URL
|
|
|
2431 if (strpos($input, '@')) { // Maybe it is an email address
|
|
|
2432 // No no in strict mode
|
|
|
2433 if ($this->_strict_mode) {
|
|
|
2434 throw new InvalidArgumentException('Only simple domain name parts can be handled in strict mode');
|
|
|
2435 }
|
|
|
2436 list($email_pref, $input) = explode('@', $input, 2);
|
|
|
2437 $arr = explode('.', $input);
|
|
|
2438 foreach ($arr as $k => $v) {
|
|
|
2439 $conv = $this->_decode($v);
|
|
|
2440 if ($conv) $arr[$k] = $conv;
|
|
|
2441 }
|
|
|
2442 $return = $email_pref . '@' . join('.', $arr);
|
|
|
2443 } elseif (preg_match('![:\./]!', $input)) { // Or a complete domain name (with or without paths / parameters)
|
|
|
2444 // No no in strict mode
|
|
|
2445 if ($this->_strict_mode) {
|
|
|
2446 throw new InvalidArgumentException('Only simple domain name parts can be handled in strict mode');
|
|
|
2447 }
|
|
|
2448
|
|
|
2449 $parsed = parse_url($input);
|
|
|
2450 if (isset($parsed['host'])) {
|
|
|
2451 $arr = explode('.', $parsed['host']);
|
|
|
2452 foreach ($arr as $k => $v) {
|
|
|
2453 $conv = $this->_decode($v);
|
|
|
2454 if ($conv) $arr[$k] = $conv;
|
|
|
2455 }
|
|
|
2456 $parsed['host'] = join('.', $arr);
|
|
|
2457 if (isset($parsed['scheme'])) {
|
|
|
2458 $parsed['scheme'] .= (strtolower($parsed['scheme']) == 'mailto') ? ':' : '://';
|
|
|
2459 }
|
|
|
2460 $return = $this->_unparse_url($parsed);
|
|
|
2461 } else { // parse_url seems to have failed, try without it
|
|
|
2462 $arr = explode('.', $input);
|
|
|
2463 foreach ($arr as $k => $v) {
|
|
|
2464 $conv = $this->_decode($v);
|
|
|
2465 if ($conv) $arr[$k] = $conv;
|
|
|
2466 }
|
|
|
2467 $return = join('.', $arr);
|
|
|
2468 }
|
|
|
2469 } else { // Otherwise we consider it being a pure domain name string
|
|
|
2470 $return = $this->_decode($input);
|
|
|
2471 }
|
|
|
2472 // The output is UTF-8 by default, other output formats need conversion here
|
|
|
2473 // If one time encoding is given, use this, else the objects property
|
|
|
2474 switch (($one_time_encoding) ? $one_time_encoding : $this->_api_encoding) {
|
|
|
2475 case 'utf8':
|
|
|
2476 return $return;
|
|
|
2477 break;
|
|
|
2478 case 'ucs4_string':
|
|
|
2479 return $this->_ucs4_to_ucs4_string($this->_utf8_to_ucs4($return));
|
|
|
2480 break;
|
|
|
2481 case 'ucs4_array':
|
|
|
2482 return $this->_utf8_to_ucs4($return);
|
|
|
2483 break;
|
|
|
2484 default:
|
|
|
2485 throw new InvalidArgumentException('Unsupported output format');
|
|
|
2486 }
|
|
|
2487 }
|
|
|
2488
|
|
|
2489
|
|
|
2490 // {{{ private
|
|
|
2491 /**
|
|
|
2492 * Opposite function to parse_url()
|
|
|
2493 *
|
|
|
2494 * Inspired by code from comments of php.net-documentation for parse_url()
|
|
|
2495 *
|
|
|
2496 * @param array $parts_arr parts (strings) as returned by parse_url()
|
|
|
2497 *
|
|
|
2498 * @return string
|
|
|
2499 * @access private
|
|
|
2500 */
|
|
|
2501 private function _unparse_url($parts_arr)
|
|
|
2502 {
|
|
|
2503 if (!empty($parts_arr['scheme'])) {
|
|
|
2504 $ret_url = $parts_arr['scheme'];
|
|
|
2505 }
|
|
|
2506 if (!empty($parts_arr['user'])) {
|
|
|
2507 $ret_url .= $parts_arr['user'];
|
|
|
2508 if (!empty($parts_arr['pass'])) {
|
|
|
2509 $ret_url .= ':' . $parts_arr['pass'];
|
|
|
2510 }
|
|
|
2511 $ret_url .= '@';
|
|
|
2512 }
|
|
|
2513 $ret_url .= $parts_arr['host'];
|
|
|
2514 if (!empty($parts_arr['port'])) {
|
|
|
2515 $ret_url .= ':' . $parts_arr['port'];
|
|
|
2516 }
|
|
|
2517 $ret_url .= $parts_arr['path'];
|
|
|
2518 if (!empty($parts_arr['query'])) {
|
|
|
2519 $ret_url .= '?' . $parts_arr['query'];
|
|
|
2520 }
|
|
|
2521 if (!empty($parts_arr['fragment'])) {
|
|
|
2522 $ret_url .= '#' . $parts_arr['fragment'];
|
|
|
2523 }
|
|
|
2524 return $ret_url;
|
|
|
2525 }
|
|
|
2526
|
|
|
2527 /**
|
|
|
2528 * The actual encoding algorithm.
|
|
|
2529 *
|
|
|
2530 * @param string $decoded Decoded string which should be encoded
|
|
|
2531 *
|
|
|
2532 * @return string Encoded string
|
|
|
2533 * @throws Exception
|
|
|
2534 * @access private
|
|
|
2535 */
|
|
|
2536 private function _encode($decoded)
|
|
|
2537 {
|
|
|
2538 // We cannot encode a domain name containing the Punycode prefix
|
|
|
2539 $extract = self::_byteLength($this->_punycode_prefix);
|
|
|
2540 $check_pref = $this->_utf8_to_ucs4($this->_punycode_prefix);
|
|
|
2541 $check_deco = array_slice($decoded, 0, $extract);
|
|
|
2542
|
|
|
2543 if ($check_pref == $check_deco) {
|
|
|
2544 throw new InvalidArgumentException('This is already a punycode string');
|
|
|
2545 }
|
|
|
2546
|
|
|
2547 // We will not try to encode strings consisting of basic code points only
|
|
|
2548 $encodable = false;
|
|
|
2549 foreach ($decoded as $k => $v) {
|
|
|
2550 if ($v > 0x7a) {
|
|
|
2551 $encodable = true;
|
|
|
2552 break;
|
|
|
2553 }
|
|
|
2554 }
|
|
|
2555 if (!$encodable) {
|
|
|
2556 if ($this->_strict_mode) {
|
|
|
2557 throw new InvalidArgumentException('The given string does not contain encodable chars');
|
|
|
2558 }
|
|
|
2559
|
|
|
2560 return false;
|
|
|
2561 }
|
|
|
2562
|
|
|
2563 // Do NAMEPREP
|
|
|
2564 $decoded = $this->_nameprep($decoded);
|
|
|
2565
|
|
|
2566 $deco_len = count($decoded);
|
|
|
2567
|
|
|
2568 // Empty array
|
|
|
2569 if (!$deco_len) {
|
|
|
2570 return false;
|
|
|
2571 }
|
|
|
2572
|
|
|
2573 // How many chars have been consumed
|
|
|
2574 $codecount = 0;
|
|
|
2575
|
|
|
2576 // Start with the prefix; copy it to output
|
|
|
2577 $encoded = $this->_punycode_prefix;
|
|
|
2578
|
|
|
2579 $encoded = '';
|
|
|
2580 // Copy all basic code points to output
|
|
|
2581 for ($i = 0; $i < $deco_len; ++$i) {
|
|
|
2582 $test = $decoded[$i];
|
|
|
2583 // Will match [0-9a-zA-Z-]
|
|
|
2584 if ((0x2F < $test && $test < 0x40)
|
|
|
2585 || (0x40 < $test && $test < 0x5B)
|
|
|
2586 || (0x60 < $test && $test <= 0x7B)
|
|
|
2587 || (0x2D == $test)
|
|
|
2588 ) {
|
|
|
2589 $encoded .= chr($decoded[$i]);
|
|
|
2590 $codecount++;
|
|
|
2591 }
|
|
|
2592 }
|
|
|
2593
|
|
|
2594 // All codepoints were basic ones
|
|
|
2595 if ($codecount == $deco_len) {
|
|
|
2596 return $encoded;
|
|
|
2597 }
|
|
|
2598
|
|
|
2599 // Start with the prefix; copy it to output
|
|
|
2600 $encoded = $this->_punycode_prefix . $encoded;
|
|
|
2601
|
|
|
2602 // If we have basic code points in output, add an hyphen to the end
|
|
|
2603 if ($codecount) {
|
|
|
2604 $encoded .= '-';
|
|
|
2605 }
|
|
|
2606
|
|
|
2607 // Now find and encode all non-basic code points
|
|
|
2608 $is_first = true;
|
|
|
2609 $cur_code = $this->_initial_n;
|
|
|
2610 $bias = $this->_initial_bias;
|
|
|
2611 $delta = 0;
|
|
|
2612
|
|
|
2613 while ($codecount < $deco_len) {
|
|
|
2614 // Find the smallest code point >= the current code point and
|
|
|
2615 // remember the last ouccrence of it in the input
|
|
|
2616 for ($i = 0, $next_code = $this->_max_ucs; $i < $deco_len; $i++) {
|
|
|
2617 if ($decoded[$i] >= $cur_code && $decoded[$i] <= $next_code) {
|
|
|
2618 $next_code = $decoded[$i];
|
|
|
2619 }
|
|
|
2620 }
|
|
|
2621
|
|
|
2622 $delta += ($next_code - $cur_code) * ($codecount + 1);
|
|
|
2623 $cur_code = $next_code;
|
|
|
2624
|
|
|
2625 // Scan input again and encode all characters whose code point is $cur_code
|
|
|
2626 for ($i = 0; $i < $deco_len; $i++) {
|
|
|
2627 if ($decoded[$i] < $cur_code) {
|
|
|
2628 $delta++;
|
|
|
2629 } else if ($decoded[$i] == $cur_code) {
|
|
|
2630 for ($q = $delta, $k = $this->_base; 1; $k += $this->_base) {
|
|
|
2631 $t = ($k <= $bias)?
|
|
|
2632 $this->_tmin :
|
|
|
2633 (($k >= $bias + $this->_tmax)? $this->_tmax : $k - $bias);
|
|
|
2634
|
|
|
2635 if ($q < $t) {
|
|
|
2636 break;
|
|
|
2637 }
|
|
|
2638
|
|
|
2639 $encoded .= $this->_encodeDigit(ceil($t + (($q - $t) % ($this->_base - $t))));
|
|
|
2640 $q = ($q - $t) / ($this->_base - $t);
|
|
|
2641 }
|
|
|
2642
|
|
|
2643 $encoded .= $this->_encodeDigit($q);
|
|
|
2644 $bias = $this->_adapt($delta, $codecount + 1, $is_first);
|
|
|
2645 $codecount++;
|
|
|
2646 $delta = 0;
|
|
|
2647 $is_first = false;
|
|
|
2648 }
|
|
|
2649 }
|
|
|
2650
|
|
|
2651 $delta++;
|
|
|
2652 $cur_code++;
|
|
|
2653 }
|
|
|
2654
|
|
|
2655 return $encoded;
|
|
|
2656 }
|
|
|
2657
|
|
|
2658 /**
|
|
|
2659 * The actual decoding algorithm.
|
|
|
2660 *
|
|
|
2661 * @param string $encoded Encoded string which should be decoded
|
|
|
2662 *
|
|
|
2663 * @return string Decoded string
|
|
|
2664 * @throws Exception
|
|
|
2665 * @access private
|
|
|
2666 */
|
|
|
2667 private function _decode($encoded)
|
|
|
2668 {
|
|
|
2669 // We do need to find the Punycode prefix
|
|
|
2670 if (!preg_match('!^' . preg_quote($this->_punycode_prefix, '!') . '!', $encoded)) {
|
|
|
2671 return false;
|
|
|
2672 }
|
|
|
2673
|
|
|
2674 $encode_test = preg_replace('!^' . preg_quote($this->_punycode_prefix, '!') . '!', '', $encoded);
|
|
|
2675
|
|
|
2676 // If nothing left after removing the prefix, it is hopeless
|
|
|
2677 if (!$encode_test) {
|
|
|
2678 return false;
|
|
|
2679 }
|
|
|
2680
|
|
|
2681 // Find last occurrence of the delimiter
|
|
|
2682 $delim_pos = strrpos($encoded, '-');
|
|
|
2683
|
|
|
2684 if ($delim_pos > self::_byteLength($this->_punycode_prefix)) {
|
|
|
2685 for ($k = self::_byteLength($this->_punycode_prefix); $k < $delim_pos; ++$k) {
|
|
|
2686 $decoded[] = ord($encoded{$k});
|
|
|
2687 }
|
|
|
2688 } else {
|
|
|
2689 $decoded = array();
|
|
|
2690 }
|
|
|
2691
|
|
|
2692 $deco_len = count($decoded);
|
|
|
2693 $enco_len = self::_byteLength($encoded);
|
|
|
2694
|
|
|
2695 // Wandering through the strings; init
|
|
|
2696 $is_first = true;
|
|
|
2697 $bias = $this->_initial_bias;
|
|
|
2698 $idx = 0;
|
|
|
2699 $char = $this->_initial_n;
|
|
|
2700
|
|
|
2701 for ($enco_idx = ($delim_pos)? ($delim_pos + 1) : 0; $enco_idx < $enco_len; ++$deco_len) {
|
|
|
2702 for ($old_idx = $idx, $w = 1, $k = $this->_base; 1 ; $k += $this->_base) {
|
|
|
2703 $digit = $this->_decodeDigit($encoded{$enco_idx++});
|
|
|
2704 $idx += $digit * $w;
|
|
|
2705
|
|
|
2706 $t = ($k <= $bias) ?
|
|
|
2707 $this->_tmin :
|
|
|
2708 (($k >= $bias + $this->_tmax)? $this->_tmax : ($k - $bias));
|
|
|
2709
|
|
|
2710 if ($digit < $t) {
|
|
|
2711 break;
|
|
|
2712 }
|
|
|
2713
|
|
|
2714 $w = (int)($w * ($this->_base - $t));
|
|
|
2715 }
|
|
|
2716
|
|
|
2717 $bias = $this->_adapt($idx - $old_idx, $deco_len + 1, $is_first);
|
|
|
2718 $is_first = false;
|
|
|
2719 $char += (int) ($idx / ($deco_len + 1));
|
|
|
2720 $idx %= ($deco_len + 1);
|
|
|
2721
|
|
|
2722 if ($deco_len > 0) {
|
|
|
2723 // Make room for the decoded char
|
|
|
2724 for ($i = $deco_len; $i > $idx; $i--) {
|
|
|
2725 $decoded[$i] = $decoded[($i - 1)];
|
|
|
2726 }
|
|
|
2727 }
|
|
|
2728
|
|
|
2729 $decoded[$idx++] = $char;
|
|
|
2730 }
|
|
|
2731
|
|
|
2732 return $this->_ucs4_to_utf8($decoded);
|
|
|
2733 }
|
|
|
2734
|
|
|
2735 /**
|
|
|
2736 * Adapt the bias according to the current code point and position.
|
|
|
2737 *
|
|
|
2738 * @param int $delta ...
|
|
|
2739 * @param int $npoints ...
|
|
|
2740 * @param boolean $is_first ...
|
|
|
2741 *
|
|
|
2742 * @return int
|
|
|
2743 * @access private
|
|
|
2744 */
|
|
|
2745 private function _adapt($delta, $npoints, $is_first)
|
|
|
2746 {
|
|
|
2747 $delta = (int) ($is_first ? ($delta / $this->_damp) : ($delta / 2));
|
|
|
2748 $delta += (int) ($delta / $npoints);
|
|
|
2749
|
|
|
2750 for ($k = 0; $delta > (($this->_base - $this->_tmin) * $this->_tmax) / 2; $k += $this->_base) {
|
|
|
2751 $delta = (int) ($delta / ($this->_base - $this->_tmin));
|
|
|
2752 }
|
|
|
2753
|
|
|
2754 return (int) ($k + ($this->_base - $this->_tmin + 1) * $delta / ($delta + $this->_skew));
|
|
|
2755 }
|
|
|
2756
|
|
|
2757 /**
|
|
|
2758 * Encoding a certain digit.
|
|
|
2759 *
|
|
|
2760 * @param int $d One digit to encode
|
|
|
2761 *
|
|
|
2762 * @return char Encoded digit
|
|
|
2763 * @access private
|
|
|
2764 */
|
|
|
2765 private function _encodeDigit($d)
|
|
|
2766 {
|
|
|
2767 return chr($d + 22 + 75 * ($d < 26));
|
|
|
2768 }
|
|
|
2769
|
|
|
2770 /**
|
|
|
2771 * Decode a certain digit.
|
|
|
2772 *
|
|
|
2773 * @param char $cp One digit (character) to decode
|
|
|
2774 *
|
|
|
2775 * @return int Decoded digit
|
|
|
2776 * @access private
|
|
|
2777 */
|
|
|
2778 private function _decodeDigit($cp)
|
|
|
2779 {
|
|
|
2780 $cp = ord($cp);
|
|
|
2781 return ($cp - 48 < 10)? $cp - 22 : (($cp - 65 < 26)? $cp - 65 : (($cp - 97 < 26)? $cp - 97 : $this->_base));
|
|
|
2782 }
|
|
|
2783
|
|
|
2784 /**
|
|
|
2785 * Do Nameprep according to RFC3491 and RFC3454.
|
|
|
2786 *
|
|
|
2787 * @param array $input Unicode Characters
|
|
|
2788 *
|
|
|
2789 * @return string Unicode Characters, Nameprep'd
|
|
|
2790 * @throws Exception
|
|
|
2791 * @access private
|
|
|
2792 */
|
|
|
2793 private function _nameprep($input)
|
|
|
2794 {
|
|
|
2795 $output = array();
|
|
|
2796
|
|
|
2797 // Walking through the input array, performing the required steps on each of
|
|
|
2798 // the input chars and putting the result into the output array
|
|
|
2799 // While mapping required chars we apply the canonical ordering
|
|
|
2800
|
|
|
2801 foreach ($input as $v) {
|
|
|
2802 // Map to nothing == skip that code point
|
|
|
2803 if (in_array($v, self::$_np_map_nothing)) {
|
|
|
2804 continue;
|
|
|
2805 }
|
|
|
2806
|
|
|
2807 // Try to find prohibited input
|
|
|
2808 if (in_array($v, self::$_np_prohibit) || in_array($v, self::$_general_prohibited)) {
|
|
|
2809 throw new Net_IDNA2_Exception_Nameprep('Prohibited input U+' . sprintf('%08X', $v));
|
|
|
2810 }
|
|
|
2811
|
|
|
2812 foreach (self::$_np_prohibit_ranges as $range) {
|
|
|
2813 if ($range[0] <= $v && $v <= $range[1]) {
|
|
|
2814 throw new Net_IDNA2_Exception_Nameprep('Prohibited input U+' . sprintf('%08X', $v));
|
|
|
2815 }
|
|
|
2816 }
|
|
|
2817
|
|
|
2818 // Hangul syllable decomposition
|
|
|
2819 if (0xAC00 <= $v && $v <= 0xD7AF) {
|
|
|
2820 foreach ($this->_hangulDecompose($v) as $out) {
|
|
|
2821 $output[] = $out;
|
|
|
2822 }
|
|
|
2823 } else if (($this->_version == '2003') && isset(self::$_np_replacemaps[$v])) {
|
|
|
2824 // There's a decomposition mapping for that code point
|
|
|
2825 // Decompositions only in version 2003 (original) of IDNA
|
|
|
2826 foreach ($this->_applyCannonicalOrdering(self::$_np_replacemaps[$v]) as $out) {
|
|
|
2827 $output[] = $out;
|
|
|
2828 }
|
|
|
2829 } else {
|
|
|
2830 $output[] = $v;
|
|
|
2831 }
|
|
|
2832 }
|
|
|
2833
|
|
|
2834 // Combine code points
|
|
|
2835
|
|
|
2836 $last_class = 0;
|
|
|
2837 $last_starter = 0;
|
|
|
2838 $out_len = count($output);
|
|
|
2839
|
|
|
2840 for ($i = 0; $i < $out_len; ++$i) {
|
|
|
2841 $class = $this->_getCombiningClass($output[$i]);
|
|
|
2842
|
|
|
2843 if ((!$last_class || $last_class != $class) && $class) {
|
|
|
2844 // Try to match
|
|
|
2845 $seq_len = $i - $last_starter;
|
|
|
2846 $out = $this->_combine(array_slice($output, $last_starter, $seq_len));
|
|
|
2847
|
|
|
2848 // On match: Replace the last starter with the composed character and remove
|
|
|
2849 // the now redundant non-starter(s)
|
|
|
2850 if ($out) {
|
|
|
2851 $output[$last_starter] = $out;
|
|
|
2852
|
|
|
2853 if (count($out) != $seq_len) {
|
|
|
2854 for ($j = $i + 1; $j < $out_len; ++$j) {
|
|
|
2855 $output[$j - 1] = $output[$j];
|
|
|
2856 }
|
|
|
2857
|
|
|
2858 unset($output[$out_len]);
|
|
|
2859 }
|
|
|
2860
|
|
|
2861 // Rewind the for loop by one, since there can be more possible compositions
|
|
|
2862 $i--;
|
|
|
2863 $out_len--;
|
|
|
2864 $last_class = ($i == $last_starter)? 0 : $this->_getCombiningClass($output[$i - 1]);
|
|
|
2865
|
|
|
2866 continue;
|
|
|
2867 }
|
|
|
2868 }
|
|
|
2869
|
|
|
2870 // The current class is 0
|
|
|
2871 if (!$class) {
|
|
|
2872 $last_starter = $i;
|
|
|
2873 }
|
|
|
2874
|
|
|
2875 $last_class = $class;
|
|
|
2876 }
|
|
|
2877
|
|
|
2878 return $output;
|
|
|
2879 }
|
|
|
2880
|
|
|
2881 /**
|
|
|
2882 * Decomposes a Hangul syllable
|
|
|
2883 * (see http://www.unicode.org/unicode/reports/tr15/#Hangul).
|
|
|
2884 *
|
|
|
2885 * @param integer $char 32bit UCS4 code point
|
|
|
2886 *
|
|
|
2887 * @return array Either Hangul Syllable decomposed or original 32bit
|
|
|
2888 * value as one value array
|
|
|
2889 * @access private
|
|
|
2890 */
|
|
|
2891 private function _hangulDecompose($char)
|
|
|
2892 {
|
|
|
2893 $sindex = $char - $this->_sbase;
|
|
|
2894
|
|
|
2895 if ($sindex < 0 || $sindex >= $this->_scount) {
|
|
|
2896 return array($char);
|
|
|
2897 }
|
|
|
2898
|
|
|
2899 $result = array();
|
|
|
2900 $T = $this->_tbase + $sindex % $this->_tcount;
|
|
|
2901 $result[] = (int)($this->_lbase + $sindex / $this->_ncount);
|
|
|
2902 $result[] = (int)($this->_vbase + ($sindex % $this->_ncount) / $this->_tcount);
|
|
|
2903
|
|
|
2904 if ($T != $this->_tbase) {
|
|
|
2905 $result[] = $T;
|
|
|
2906 }
|
|
|
2907
|
|
|
2908 return $result;
|
|
|
2909 }
|
|
|
2910
|
|
|
2911 /**
|
|
|
2912 * Ccomposes a Hangul syllable
|
|
|
2913 * (see http://www.unicode.org/unicode/reports/tr15/#Hangul).
|
|
|
2914 *
|
|
|
2915 * @param array $input Decomposed UCS4 sequence
|
|
|
2916 *
|
|
|
2917 * @return array UCS4 sequence with syllables composed
|
|
|
2918 * @access private
|
|
|
2919 */
|
|
|
2920 private function _hangulCompose($input)
|
|
|
2921 {
|
|
|
2922 $inp_len = count($input);
|
|
|
2923
|
|
|
2924 if (!$inp_len) {
|
|
|
2925 return array();
|
|
|
2926 }
|
|
|
2927
|
|
|
2928 $result = array();
|
|
|
2929 $last = $input[0];
|
|
|
2930 $result[] = $last; // copy first char from input to output
|
|
|
2931
|
|
|
2932 for ($i = 1; $i < $inp_len; ++$i) {
|
|
|
2933 $char = $input[$i];
|
|
|
2934
|
|
|
2935 // Find out, wether two current characters from L and V
|
|
|
2936 $lindex = $last - $this->_lbase;
|
|
|
2937
|
|
|
2938 if (0 <= $lindex && $lindex < $this->_lcount) {
|
|
|
2939 $vindex = $char - $this->_vbase;
|
|
|
2940
|
|
|
2941 if (0 <= $vindex && $vindex < $this->_vcount) {
|
|
|
2942 // create syllable of form LV
|
|
|
2943 $last = ($this->_sbase + ($lindex * $this->_vcount + $vindex) * $this->_tcount);
|
|
|
2944 $out_off = count($result) - 1;
|
|
|
2945 $result[$out_off] = $last; // reset last
|
|
|
2946
|
|
|
2947 // discard char
|
|
|
2948 continue;
|
|
|
2949 }
|
|
|
2950 }
|
|
|
2951
|
|
|
2952 // Find out, wether two current characters are LV and T
|
|
|
2953 $sindex = $last - $this->_sbase;
|
|
|
2954
|
|
|
2955 if (0 <= $sindex && $sindex < $this->_scount && ($sindex % $this->_tcount) == 0) {
|
|
|
2956 $tindex = $char - $this->_tbase;
|
|
|
2957
|
|
|
2958 if (0 <= $tindex && $tindex <= $this->_tcount) {
|
|
|
2959 // create syllable of form LVT
|
|
|
2960 $last += $tindex;
|
|
|
2961 $out_off = count($result) - 1;
|
|
|
2962 $result[$out_off] = $last; // reset last
|
|
|
2963
|
|
|
2964 // discard char
|
|
|
2965 continue;
|
|
|
2966 }
|
|
|
2967 }
|
|
|
2968
|
|
|
2969 // if neither case was true, just add the character
|
|
|
2970 $last = $char;
|
|
|
2971 $result[] = $char;
|
|
|
2972 }
|
|
|
2973
|
|
|
2974 return $result;
|
|
|
2975 }
|
|
|
2976
|
|
|
2977 /**
|
|
|
2978 * Returns the combining class of a certain wide char.
|
|
|
2979 *
|
|
|
2980 * @param integer $char Wide char to check (32bit integer)
|
|
|
2981 *
|
|
|
2982 * @return integer Combining class if found, else 0
|
|
|
2983 * @access private
|
|
|
2984 */
|
|
|
2985 private function _getCombiningClass($char)
|
|
|
2986 {
|
|
|
2987 return isset(self::$_np_norm_combcls[$char])? self::$_np_norm_combcls[$char] : 0;
|
|
|
2988 }
|
|
|
2989
|
|
|
2990 /**
|
|
|
2991 * Apllies the canonical ordering of a decomposed UCS4 sequence.
|
|
|
2992 *
|
|
|
2993 * @param array $input Decomposed UCS4 sequence
|
|
|
2994 *
|
|
|
2995 * @return array Ordered USC4 sequence
|
|
|
2996 * @access private
|
|
|
2997 */
|
|
|
2998 private function _applyCannonicalOrdering($input)
|
|
|
2999 {
|
|
|
3000 $swap = true;
|
|
|
3001 $size = count($input);
|
|
|
3002
|
|
|
3003 while ($swap) {
|
|
|
3004 $swap = false;
|
|
|
3005 $last = $this->_getCombiningClass($input[0]);
|
|
|
3006
|
|
|
3007 for ($i = 0; $i < $size - 1; ++$i) {
|
|
|
3008 $next = $this->_getCombiningClass($input[$i + 1]);
|
|
|
3009
|
|
|
3010 if ($next != 0 && $last > $next) {
|
|
|
3011 // Move item leftward until it fits
|
|
|
3012 for ($j = $i + 1; $j > 0; --$j) {
|
|
|
3013 if ($this->_getCombiningClass($input[$j - 1]) <= $next) {
|
|
|
3014 break;
|
|
|
3015 }
|
|
|
3016
|
|
|
3017 $t = $input[$j];
|
|
|
3018 $input[$j] = $input[$j - 1];
|
|
|
3019 $input[$j - 1] = $t;
|
|
|
3020 $swap = 1;
|
|
|
3021 }
|
|
|
3022
|
|
|
3023 // Reentering the loop looking at the old character again
|
|
|
3024 $next = $last;
|
|
|
3025 }
|
|
|
3026
|
|
|
3027 $last = $next;
|
|
|
3028 }
|
|
|
3029 }
|
|
|
3030
|
|
|
3031 return $input;
|
|
|
3032 }
|
|
|
3033
|
|
|
3034 /**
|
|
|
3035 * Do composition of a sequence of starter and non-starter.
|
|
|
3036 *
|
|
|
3037 * @param array $input UCS4 Decomposed sequence
|
|
|
3038 *
|
|
|
3039 * @return array Ordered USC4 sequence
|
|
|
3040 * @access private
|
|
|
3041 */
|
|
|
3042 private function _combine($input)
|
|
|
3043 {
|
|
|
3044 $inp_len = count($input);
|
|
|
3045
|
|
|
3046 // Is it a Hangul syllable?
|
|
|
3047 if (1 != $inp_len) {
|
|
|
3048 $hangul = $this->_hangulCompose($input);
|
|
|
3049
|
|
|
3050 // This place is probably wrong
|
|
|
3051 if (count($hangul) != $inp_len) {
|
|
|
3052 return $hangul;
|
|
|
3053 }
|
|
|
3054 }
|
|
|
3055
|
|
|
3056 foreach (self::$_np_replacemaps as $np_src => $np_target) {
|
|
|
3057 if ($np_target[0] != $input[0]) {
|
|
|
3058 continue;
|
|
|
3059 }
|
|
|
3060
|
|
|
3061 if (count($np_target) != $inp_len) {
|
|
|
3062 continue;
|
|
|
3063 }
|
|
|
3064
|
|
|
3065 $hit = false;
|
|
|
3066
|
|
|
3067 foreach ($input as $k2 => $v2) {
|
|
|
3068 if ($v2 == $np_target[$k2]) {
|
|
|
3069 $hit = true;
|
|
|
3070 } else {
|
|
|
3071 $hit = false;
|
|
|
3072 break;
|
|
|
3073 }
|
|
|
3074 }
|
|
|
3075
|
|
|
3076 if ($hit) {
|
|
|
3077 return $np_src;
|
|
|
3078 }
|
|
|
3079 }
|
|
|
3080
|
|
|
3081 return false;
|
|
|
3082 }
|
|
|
3083
|
|
|
3084 /**
|
|
|
3085 * This converts an UTF-8 encoded string to its UCS-4 (array) representation
|
|
|
3086 * By talking about UCS-4 we mean arrays of 32bit integers representing
|
|
|
3087 * each of the "chars". This is due to PHP not being able to handle strings with
|
|
|
3088 * bit depth different from 8. This applies to the reverse method _ucs4_to_utf8(), too.
|
|
|
3089 * The following UTF-8 encodings are supported:
|
|
|
3090 *
|
|
|
3091 * bytes bits representation
|
|
|
3092 * 1 7 0xxxxxxx
|
|
|
3093 * 2 11 110xxxxx 10xxxxxx
|
|
|
3094 * 3 16 1110xxxx 10xxxxxx 10xxxxxx
|
|
|
3095 * 4 21 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
|
|
|
3096 * 5 26 111110xx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx
|
|
|
3097 * 6 31 1111110x 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx
|
|
|
3098 *
|
|
|
3099 * Each x represents a bit that can be used to store character data.
|
|
|
3100 *
|
|
|
3101 * @param string $input utf8-encoded string
|
|
|
3102 *
|
|
|
3103 * @return array ucs4-encoded array
|
|
|
3104 * @throws Exception
|
|
|
3105 * @access private
|
|
|
3106 */
|
|
|
3107 private function _utf8_to_ucs4($input)
|
|
|
3108 {
|
|
|
3109 $output = array();
|
|
|
3110 $out_len = 0;
|
|
|
3111 $inp_len = self::_byteLength($input, '8bit');
|
|
|
3112 $mode = 'next';
|
|
|
3113 $test = 'none';
|
|
|
3114 for ($k = 0; $k < $inp_len; ++$k) {
|
|
|
3115 $v = ord($input{$k}); // Extract byte from input string
|
|
|
3116
|
|
|
3117 if ($v < 128) { // We found an ASCII char - put into string as is
|
|
|
3118 $output[$out_len] = $v;
|
|
|
3119 ++$out_len;
|
|
|
3120 if ('add' == $mode) {
|
|
|
3121 throw new UnexpectedValueException('Conversion from UTF-8 to UCS-4 failed: malformed input at byte '.$k);
|
|
|
3122 }
|
|
|
3123 continue;
|
|
|
3124 }
|
|
|
3125 if ('next' == $mode) { // Try to find the next start byte; determine the width of the Unicode char
|
|
|
3126 $start_byte = $v;
|
|
|
3127 $mode = 'add';
|
|
|
3128 $test = 'range';
|
|
|
3129 if ($v >> 5 == 6) { // &110xxxxx 10xxxxx
|
|
|
3130 $next_byte = 0; // Tells, how many times subsequent bitmasks must rotate 6bits to the left
|
|
|
3131 $v = ($v - 192) << 6;
|
|
|
3132 } elseif ($v >> 4 == 14) { // &1110xxxx 10xxxxxx 10xxxxxx
|
|
|
3133 $next_byte = 1;
|
|
|
3134 $v = ($v - 224) << 12;
|
|
|
3135 } elseif ($v >> 3 == 30) { // &11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
|
|
|
3136 $next_byte = 2;
|
|
|
3137 $v = ($v - 240) << 18;
|
|
|
3138 } elseif ($v >> 2 == 62) { // &111110xx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx
|
|
|
3139 $next_byte = 3;
|
|
|
3140 $v = ($v - 248) << 24;
|
|
|
3141 } elseif ($v >> 1 == 126) { // &1111110x 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx
|
|
|
3142 $next_byte = 4;
|
|
|
3143 $v = ($v - 252) << 30;
|
|
|
3144 } else {
|
|
|
3145 throw new UnexpectedValueException('This might be UTF-8, but I don\'t understand it at byte '.$k);
|
|
|
3146 }
|
|
|
3147 if ('add' == $mode) {
|
|
|
3148 $output[$out_len] = (int) $v;
|
|
|
3149 ++$out_len;
|
|
|
3150 continue;
|
|
|
3151 }
|
|
|
3152 }
|
|
|
3153 if ('add' == $mode) {
|
|
|
3154 if (!$this->_allow_overlong && $test == 'range') {
|
|
|
3155 $test = 'none';
|
|
|
3156 if (($v < 0xA0 && $start_byte == 0xE0) || ($v < 0x90 && $start_byte == 0xF0) || ($v > 0x8F && $start_byte == 0xF4)) {
|
|
|
3157 throw new OutOfRangeException('Bogus UTF-8 character detected (out of legal range) at byte '.$k);
|
|
|
3158 }
|
|
|
3159 }
|
|
|
3160 if ($v >> 6 == 2) { // Bit mask must be 10xxxxxx
|
|
|
3161 $v = ($v - 128) << ($next_byte * 6);
|
|
|
3162 $output[($out_len - 1)] += $v;
|
|
|
3163 --$next_byte;
|
|
|
3164 } else {
|
|
|
3165 throw new UnexpectedValueException('Conversion from UTF-8 to UCS-4 failed: malformed input at byte '.$k);
|
|
|
3166 }
|
|
|
3167 if ($next_byte < 0) {
|
|
|
3168 $mode = 'next';
|
|
|
3169 }
|
|
|
3170 }
|
|
|
3171 } // for
|
|
|
3172 return $output;
|
|
|
3173 }
|
|
|
3174
|
|
|
3175 /**
|
|
|
3176 * Convert UCS-4 array into UTF-8 string
|
|
|
3177 *
|
|
|
3178 * @param array $input ucs4-encoded array
|
|
|
3179 *
|
|
|
3180 * @return string utf8-encoded string
|
|
|
3181 * @throws Exception
|
|
|
3182 * @access private
|
|
|
3183 */
|
|
|
3184 private function _ucs4_to_utf8($input)
|
|
|
3185 {
|
|
|
3186 $output = '';
|
|
|
3187
|
|
|
3188 foreach ($input as $v) {
|
|
|
3189 // $v = ord($v);
|
|
|
3190
|
|
|
3191 if ($v < 128) {
|
|
|
3192 // 7bit are transferred literally
|
|
|
3193 $output .= chr($v);
|
|
|
3194 } else if ($v < 1 << 11) {
|
|
|
3195 // 2 bytes
|
|
|
3196 $output .= chr(192 + ($v >> 6))
|
|
|
3197 . chr(128 + ($v & 63));
|
|
|
3198 } else if ($v < 1 << 16) {
|
|
|
3199 // 3 bytes
|
|
|
3200 $output .= chr(224 + ($v >> 12))
|
|
|
3201 . chr(128 + (($v >> 6) & 63))
|
|
|
3202 . chr(128 + ($v & 63));
|
|
|
3203 } else if ($v < 1 << 21) {
|
|
|
3204 // 4 bytes
|
|
|
3205 $output .= chr(240 + ($v >> 18))
|
|
|
3206 . chr(128 + (($v >> 12) & 63))
|
|
|
3207 . chr(128 + (($v >> 6) & 63))
|
|
|
3208 . chr(128 + ($v & 63));
|
|
|
3209 } else if ($v < 1 << 26) {
|
|
|
3210 // 5 bytes
|
|
|
3211 $output .= chr(248 + ($v >> 24))
|
|
|
3212 . chr(128 + (($v >> 18) & 63))
|
|
|
3213 . chr(128 + (($v >> 12) & 63))
|
|
|
3214 . chr(128 + (($v >> 6) & 63))
|
|
|
3215 . chr(128 + ($v & 63));
|
|
|
3216 } else if ($v < 1 << 31) {
|
|
|
3217 // 6 bytes
|
|
|
3218 $output .= chr(252 + ($v >> 30))
|
|
|
3219 . chr(128 + (($v >> 24) & 63))
|
|
|
3220 . chr(128 + (($v >> 18) & 63))
|
|
|
3221 . chr(128 + (($v >> 12) & 63))
|
|
|
3222 . chr(128 + (($v >> 6) & 63))
|
|
|
3223 . chr(128 + ($v & 63));
|
|
|
3224 } else {
|
|
|
3225 throw new UnexpectedValueException('Conversion from UCS-4 to UTF-8 failed: malformed input');
|
|
|
3226 }
|
|
|
3227 }
|
|
|
3228
|
|
|
3229 return $output;
|
|
|
3230 }
|
|
|
3231
|
|
|
3232 /**
|
|
|
3233 * Convert UCS-4 array into UCS-4 string
|
|
|
3234 *
|
|
|
3235 * @param array $input ucs4-encoded array
|
|
|
3236 *
|
|
|
3237 * @return string ucs4-encoded string
|
|
|
3238 * @throws Exception
|
|
|
3239 * @access private
|
|
|
3240 */
|
|
|
3241 private function _ucs4_to_ucs4_string($input)
|
|
|
3242 {
|
|
|
3243 $output = '';
|
|
|
3244 // Take array values and split output to 4 bytes per value
|
|
|
3245 // The bit mask is 255, which reads &11111111
|
|
|
3246 foreach ($input as $v) {
|
|
|
3247 $output .= ($v & (255 << 24) >> 24) . ($v & (255 << 16) >> 16) . ($v & (255 << 8) >> 8) . ($v & 255);
|
|
|
3248 }
|
|
|
3249 return $output;
|
|
|
3250 }
|
|
|
3251
|
|
|
3252 /**
|
|
|
3253 * Convert UCS-4 string into UCS-4 array
|
|
|
3254 *
|
|
|
3255 * @param string $input ucs4-encoded string
|
|
|
3256 *
|
|
|
3257 * @return array ucs4-encoded array
|
|
|
3258 * @throws InvalidArgumentException
|
|
|
3259 * @access private
|
|
|
3260 */
|
|
|
3261 private function _ucs4_string_to_ucs4($input)
|
|
|
3262 {
|
|
|
3263 $output = array();
|
|
|
3264
|
|
|
3265 $inp_len = self::_byteLength($input);
|
|
|
3266 // Input length must be dividable by 4
|
|
|
3267 if ($inp_len % 4) {
|
|
|
3268 throw new InvalidArgumentException('Input UCS4 string is broken');
|
|
|
3269 }
|
|
|
3270
|
|
|
3271 // Empty input - return empty output
|
|
|
3272 if (!$inp_len) {
|
|
|
3273 return $output;
|
|
|
3274 }
|
|
|
3275
|
|
|
3276 for ($i = 0, $out_len = -1; $i < $inp_len; ++$i) {
|
|
|
3277 // Increment output position every 4 input bytes
|
|
|
3278 if (!$i % 4) {
|
|
|
3279 $out_len++;
|
|
|
3280 $output[$out_len] = 0;
|
|
|
3281 }
|
|
|
3282 $output[$out_len] += ord($input{$i}) << (8 * (3 - ($i % 4) ) );
|
|
|
3283 }
|
|
|
3284 return $output;
|
|
|
3285 }
|
|
|
3286
|
|
|
3287 /**
|
|
|
3288 * Echo hex representation of UCS4 sequence.
|
|
|
3289 *
|
|
|
3290 * @param array $input UCS4 sequence
|
|
|
3291 * @param boolean $include_bit Include bitmask in output
|
|
|
3292 *
|
|
|
3293 * @return void
|
|
|
3294 * @static
|
|
|
3295 * @access private
|
|
|
3296 */
|
|
|
3297 private static function _showHex($input, $include_bit = false)
|
|
|
3298 {
|
|
|
3299 foreach ($input as $k => $v) {
|
|
|
3300 echo '[', $k, '] => ', sprintf('%X', $v);
|
|
|
3301
|
|
|
3302 if ($include_bit) {
|
|
|
3303 echo ' (', Net_IDNA2::_showBitmask($v), ')';
|
|
|
3304 }
|
|
|
3305
|
|
|
3306 echo "\n";
|
|
|
3307 }
|
|
|
3308 }
|
|
|
3309
|
|
|
3310 /**
|
|
|
3311 * Gives you a bit representation of given Byte (8 bits), Word (16 bits) or DWord (32 bits)
|
|
|
3312 * Output width is automagically determined
|
|
|
3313 *
|
|
|
3314 * @param int $octet ...
|
|
|
3315 *
|
|
|
3316 * @return string Bitmask-representation
|
|
|
3317 * @static
|
|
|
3318 * @access private
|
|
|
3319 */
|
|
|
3320 private static function _showBitmask($octet)
|
|
|
3321 {
|
|
|
3322 if ($octet >= (1 << 16)) {
|
|
|
3323 $w = 31;
|
|
|
3324 } else if ($octet >= (1 << 8)) {
|
|
|
3325 $w = 15;
|
|
|
3326 } else {
|
|
|
3327 $w = 7;
|
|
|
3328 }
|
|
|
3329
|
|
|
3330 $return = '';
|
|
|
3331
|
|
|
3332 for ($i = $w; $i > -1; $i--) {
|
|
|
3333 $return .= ($octet & (1 << $i))? '1' : '0';
|
|
|
3334 }
|
|
|
3335
|
|
|
3336 return $return;
|
|
|
3337 }
|
|
|
3338
|
|
|
3339 /**
|
|
|
3340 * Gets the length of a string in bytes even if mbstring function
|
|
|
3341 * overloading is turned on
|
|
|
3342 *
|
|
|
3343 * @param string $string the string for which to get the length.
|
|
|
3344 *
|
|
|
3345 * @return integer the length of the string in bytes.
|
|
|
3346 *
|
|
|
3347 * @see Net_IDNA2::$_mb_string_overload
|
|
|
3348 */
|
|
|
3349 private static function _byteLength($string)
|
|
|
3350 {
|
|
|
3351 if (self::$_mb_string_overload) {
|
|
|
3352 return mb_strlen($string, '8bit');
|
|
|
3353 }
|
|
|
3354 return strlen((binary)$string);
|
|
|
3355 }
|
|
|
3356
|
|
|
3357 // }}}}
|
|
|
3358
|
|
|
3359 // {{{ factory
|
|
|
3360 /**
|
|
|
3361 * Attempts to return a concrete IDNA instance for either php4 or php5.
|
|
|
3362 *
|
|
|
3363 * @param array $params Set of paramaters
|
|
|
3364 *
|
|
|
3365 * @return Net_IDNA2
|
|
|
3366 * @access public
|
|
|
3367 */
|
|
|
3368 public static function getInstance($params = array())
|
|
|
3369 {
|
|
|
3370 return new Net_IDNA2($params);
|
|
|
3371 }
|
|
|
3372 // }}}
|
|
|
3373
|
|
|
3374 // {{{ singleton
|
|
|
3375 /**
|
|
|
3376 * Attempts to return a concrete IDNA instance for either php4 or php5,
|
|
|
3377 * only creating a new instance if no IDNA instance with the same
|
|
|
3378 * parameters currently exists.
|
|
|
3379 *
|
|
|
3380 * @param array $params Set of parameters
|
|
|
3381 *
|
|
|
3382 * @return object Net_IDNA2
|
|
|
3383 * @access public
|
|
|
3384 */
|
|
|
3385 public static function singleton($params = array())
|
|
|
3386 {
|
|
|
3387 static $instances;
|
|
|
3388 if (!isset($instances)) {
|
|
|
3389 $instances = array();
|
|
|
3390 }
|
|
|
3391
|
|
|
3392 $signature = serialize($params);
|
|
|
3393 if (!isset($instances[$signature])) {
|
|
|
3394 $instances[$signature] = Net_IDNA2::getInstance($params);
|
|
|
3395 }
|
|
|
3396
|
|
|
3397 return $instances[$signature];
|
|
|
3398 }
|
|
|
3399 // }}}
|
|
|
3400 }
|
|
|
3401
|
|
|
3402 ?>
|
