« back to all changes in this revision
Viewing changes to src/pinyin/pinyin_data.cpp
- browse files at revision 1
- compare with another revision
- download diff
- download tarball
- view history from revision 1
- Committer: Bazaar Package Importer
- Author(s): Zhengpeng Hou
- Date: 2010-09-06 12:23:46 UTC
- Revision ID: james.westby@ubuntu.com-20100906122346-yamofztk2j5p85fs
Tags: upstream-2.0.2
Import upstream version 2.0.2
- files added:
- build
- data
- doc
- python
- python/importer
- python/importer/pinyin_data.py
- raw
- src
- src/config
- src/ime-core
- src/lexicon
- src/pinyin
- src/slm
- src/slm/getWordFreq
- src/slm/ids2ngram
- src/slm/mmseg
- src/slm/slmbuild
- src/slm/slminfo
- src/slm/slmprune
- src/slm/slmseg
- src/slm/thread
- src/slm/tools
- src/slm/tslmendian
- src/slm/tslminfo
- src/slm/tslmpack
- swap
added
removed
src/pinyin/pinyin_data.cpp
1
/*
2
* DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS HEADER.
3
*
4
* Copyright (c) 2007 Sun Microsystems, Inc. All Rights Reserved.
5
*
6
* The contents of this file are subject to the terms of either the GNU Lesser
7
* General Public License Version 2.1 only ("LGPL") or the Common Development and
8
* Distribution License ("CDDL")(collectively, the "License"). You may not use this
9
* file except in compliance with the License. You can obtain a copy of the CDDL at
10
* http://www.opensource.org/licenses/cddl1.php and a copy of the LGPLv2.1 at
11
* http://www.opensource.org/licenses/lgpl-license.php. See the License for the
12
* specific language governing permissions and limitations under the License. When
13
* distributing the software, include this License Header Notice in each file and
14
* include the full text of the License in the License file as well as the
15
* following notice:
16
*
17
* NOTICE PURSUANT TO SECTION 9 OF THE COMMON DEVELOPMENT AND DISTRIBUTION LICENSE
18
* (CDDL)
19
* For Covered Software in this distribution, this License shall be governed by the
20
* laws of the State of California (excluding conflict-of-law provisions).
21
* Any litigation relating to this License shall be subject to the jurisdiction of
22
* the Federal Courts of the Northern District of California and the state courts
23
* of the State of California, with venue lying in Santa Clara County, California.
24
*
25
* Contributor(s):
26
*
27
* If you wish your version of this file to be governed by only the CDDL or only
28
* the LGPL Version 2.1, indicate your decision by adding "[Contributor]" elects to
29
* include this software in this distribution under the [CDDL or LGPL Version 2.1]
30
* license." If you don't indicate a single choice of license, a recipient has the
31
* option to distribute your version of this file under either the CDDL or the LGPL
32
* Version 2.1, or to extend the choice of license to its licensees as provided
33
* above. However, if you add LGPL Version 2.1 code and therefore, elected the LGPL
34
* Version 2 license, then the option applies only if the new code is made subject
35
* to such option by the copyright holder.
36
*/
37
38
#include <stdio.h>
39
#include <stdlib.h>
40
#include <string.h>
41
#include "pinyin_data.h"
42
43
static const char *initials[] = {"", "b", "p", "m", "f", "d", "t", "n", "l", "g", "k", "h", "j", "q", "x", "zh", "ch", "sh", "r", "z", "c", "s", "y", "w", };
44
static const unsigned num_initials = sizeof(initials)/sizeof(*initials);
45
46
static const char *finals[] = {"", "a", "o", "e", "ai", "ei", "ao", "ou", "an", "en", "ang", "eng", "er", "i", "ia", "ie", "iao", "iu", "ian", "in", "iang", "ing", "u", "ua", "uo", "uai", "ui", "uan", "un", "uang", "ong", "v", "ue", "iong", };
47
static const unsigned num_finals = sizeof(finals)/sizeof(*finals);
48
49
static const char *fuzzy_finals[] = {"ia", "iao", "ian", "iang", "ie", "ua", "uai", "uan", "uang", "ue"};
50
static const unsigned num_fuzzy_finals = sizeof(fuzzy_finals)/sizeof(*fuzzy_finals);
51
52
static const unsigned fuzzy_finals_map[] = {
53
0x0e, 0x10, 1, /* ia -> a len 1 */
54
0x10, 0x60, 2, /* iao -> ao len 2 */
55
0x12, 0x80, 2, /* ian -> an len 2 */
56
0x14, 0xa0, 3, /* iang -> ang len 3 */
57
0x0f, 0x30, 1, /* ie -> e len 1 */
58
0x17, 0x10, 1, /* ua -> a len 1 */
59
0x19, 0x40, 2, /* uai -> ai len 2 */
60
0x1b, 0x80, 2, /* uan -> an len 2 */
61
0x1d, 0xa0, 3, /* uang -> ang len 3 */
62
0x20, 0x30, 1, /* ue -> e len 1 */
63
};
64
65
static const unsigned fuzzy_pre_syllables [] = {
66
0x0d0e0, 'n', 0x0d120, /* qian */
67
0x09080, 'g', 0x090a0, /* gang */
68
0x080e0, 'n', 0x08120, /* lian */
69
0x15090, 'g', 0x150b0, /* seng */
70
0x04010, 'n', 0x04080, /* fan */
71
0x10030, 'n', 0x10090, /* chen */
72
0x050e0, 'n', 0x05120, /* dian */
73
0x15160, 'n', 0x151c0, /* sun */
74
0x07080, 'g', 0x070a0, /* nang */
75
0x0a160, 'n', 0x0a1c0, /* kun */
76
0x05030, 'n', 0x05090, /* den */
77
0x07090, 'g', 0x070b0, /* neng */
78
0x03030, 'n', 0x03090, /* men */
79
0x09090, 'g', 0x090b0, /* geng */
80
0x10080, 'g', 0x100a0, /* chang */
81
0x0f010, 'n', 0x0f080, /* zhan */
82
0x14010, 'n', 0x14080, /* can */
83
0x07130, 'g', 0x07150, /* ning */
84
0x17080, 'g', 0x170a0, /* wang */
85
0x01090, 'g', 0x010b0, /* beng */
86
0x0f1b0, 'g', 0x0f1d0, /* zhuang */
87
0x06010, 'n', 0x06080, /* tan */
88
0x00090, 'g', 0x000b0, /* eng */
89
0x0f080, 'g', 0x0f0a0, /* zhang */
90
0x02130, 'g', 0x02150, /* ping */
91
0x08010, 'n', 0x08080, /* lan */
92
0x0e160, 'n', 0x0e1c0, /* xun */
93
0x03010, 'n', 0x03080, /* man */
94
0x0c120, 'g', 0x0c140, /* jiang */
95
0x0a1b0, 'g', 0x0a1d0, /* kuang */
96
0x01130, 'g', 0x01150, /* bing */
97
0x13010, 'n', 0x13080, /* zan */
98
0x13030, 'n', 0x13090, /* zen */
99
0x02080, 'g', 0x020a0, /* pang */
100
0x0c0d0, 'n', 0x0c130, /* jin */
101
0x14030, 'n', 0x14090, /* cen */
102
0x05010, 'n', 0x05080, /* dan */
103
0x0f030, 'n', 0x0f090, /* zhen */
104
0x01080, 'g', 0x010a0, /* bang */
105
0x17090, 'g', 0x170b0, /* weng */
106
0x00030, 'n', 0x00090, /* en */
107
0x0a080, 'g', 0x0a0a0, /* kang */
108
0x09160, 'n', 0x091c0, /* gun */
109
0x00030, 'r', 0x000c0, /* er */
110
0x0a090, 'g', 0x0a0b0, /* keng */
111
0x15080, 'g', 0x150a0, /* sang */
112
0x12030, 'n', 0x12090, /* ren */
113
0x11160, 'n', 0x111c0, /* shun */
114
0x0d160, 'n', 0x0d1c0, /* qun */
115
0x16160, 'n', 0x161c0, /* yun */
116
0x0e120, 'g', 0x0e140, /* xiang */
117
0x12080, 'g', 0x120a0, /* rang */
118
0x09170, 'n', 0x091b0, /* guan */
119
0x16130, 'g', 0x16150, /* ying */
120
0x0a170, 'n', 0x0a1b0, /* kuan */
121
0x10010, 'n', 0x10080, /* chan */
122
0x160d0, 'n', 0x16130, /* yin */
123
0x0e0d0, 'n', 0x0e130, /* xin */
124
0x07120, 'g', 0x07140, /* niang */
125
0x0b160, 'n', 0x0b1c0, /* hun */
126
0x11170, 'n', 0x111b0, /* shuan */
127
0x05080, 'g', 0x050a0, /* dang */
128
0x00080, 'g', 0x000a0, /* ang */
129
0x15010, 'n', 0x15080, /* san */
130
0x12090, 'g', 0x120b0, /* reng */
131
0x03130, 'g', 0x03150, /* ming */
132
0x030d0, 'n', 0x03130, /* min */
133
0x07030, 'n', 0x07090, /* nen */
134
0x0a010, 'n', 0x0a080, /* kan */
135
0x16080, 'g', 0x160a0, /* yang */
136
0x05090, 'g', 0x050b0, /* deng */
137
0x101b0, 'g', 0x101d0, /* chuang */
138
0x04090, 'g', 0x040b0, /* feng */
139
0x03090, 'g', 0x030b0, /* meng */
140
0x10090, 'g', 0x100b0, /* cheng */
141
0x09030, 'n', 0x09090, /* gen */
142
0x01010, 'n', 0x01080, /* ban */
143
0x07160, 'n', 0x071c0, /* nun */
144
0x15030, 'n', 0x15090, /* sen */
145
0x04080, 'g', 0x040a0, /* fang */
146
0x08160, 'n', 0x081c0, /* lun */
147
0x0a030, 'n', 0x0a090, /* ken */
148
0x0b1b0, 'g', 0x0b1d0, /* huang */
149
0x03080, 'g', 0x030a0, /* mang */
150
0x06160, 'n', 0x061c0, /* tun */
151
0x0d0d0, 'n', 0x0d130, /* qin */
152
0x02090, 'g', 0x020b0, /* peng */
153
0x05160, 'n', 0x051c0, /* dun */
154
0x10160, 'n', 0x101c0, /* chun */
155
0x09010, 'n', 0x09080, /* gan */
156
0x13090, 'g', 0x130b0, /* zeng */
157
0x06080, 'g', 0x060a0, /* tang */
158
0x14080, 'g', 0x140a0, /* cang */
159
0x0b090, 'g', 0x0b0b0, /* heng */
160
0x0e0e0, 'n', 0x0e120, /* xian */
161
0x0f160, 'n', 0x0f1c0, /* zhun */
162
0x111b0, 'g', 0x111d0, /* shuang */
163
0x11010, 'n', 0x11080, /* shan */
164
0x02010, 'n', 0x02080, /* pan */
165
0x070d0, 'n', 0x07130, /* nin */
166
0x0b080, 'g', 0x0b0a0, /* hang */
167
0x0f170, 'n', 0x0f1b0, /* zhuan */
168
0x080d0, 'n', 0x08130, /* lin */
169
0x091b0, 'g', 0x091d0, /* guang */
170
0x0b010, 'n', 0x0b080, /* han */
171
0x14160, 'n', 0x141c0, /* cun */
172
0x010d0, 'n', 0x01130, /* bin */
173
0x11030, 'n', 0x11090, /* shen */
174
0x0e130, 'g', 0x0e150, /* xing */
175
0x0d120, 'g', 0x0d140, /* qiang */
176
0x12160, 'n', 0x121c0, /* run */
177
0x11090, 'g', 0x110b0, /* sheng */
178
0x10170, 'n', 0x101b0, /* chuan */
179
0x0d130, 'g', 0x0d150, /* qing */
180
0x0c0e0, 'n', 0x0c120, /* jian */
181
0x17010, 'n', 0x17080, /* wan */
182
0x0c130, 'g', 0x0c150, /* jing */
183
0x16010, 'n', 0x16080, /* yan */
184
0x08120, 'g', 0x08140, /* liang */
185
0x0b170, 'n', 0x0b1b0, /* huan */
186
0x0b030, 'n', 0x0b090, /* hen */
187
0x11080, 'g', 0x110a0, /* shang */
188
0x0c160, 'n', 0x0c1c0, /* jun */
189
0x08130, 'g', 0x08150, /* ling */
190
0x14090, 'g', 0x140b0, /* ceng */
191
0x020d0, 'n', 0x02130, /* pin */
192
0x00010, 'n', 0x00080, /* an */
193
0x13080, 'g', 0x130a0, /* zang */
194
0x07010, 'n', 0x07080, /* nan */
195
0x0f090, 'g', 0x0f0b0, /* zheng */
196
0x13160, 'n', 0x131c0, /* zun */
197
0x08080, 'g', 0x080a0, /* lang */
198
0x0,
199
};
200
201
static const unsigned fuzzy_pro_syllables [] = {
202
0x09030, 'g', 0x00030, /* ge */
203
0x090a0, 'g', 0x000a0, /* gang */
204
0x09010, 'g', 0x00010, /* ga */
205
0x12070, 'r', 0x00070, /* rou */
206
0x07050, 'n', 0x00050, /* nei */
207
0x070a0, 'n', 0x000a0, /* nang */
208
0x070b0, 'n', 0x000b0, /* neng */
209
0x090b0, 'g', 0x000b0, /* geng */
210
0x07070, 'n', 0x00070, /* nou */
211
0x12030, 'r', 0x00030, /* re */
212
0x12090, 'r', 0x00090, /* ren */
213
0x09070, 'g', 0x00070, /* gou */
214
0x120a0, 'r', 0x000a0, /* rang */
215
0x120b0, 'r', 0x000b0, /* reng */
216
0x12080, 'r', 0x00080, /* ran */
217
0x12060, 'r', 0x00060, /* rao */
218
0x07090, 'n', 0x00090, /* nen */
219
0x09050, 'g', 0x00050, /* gei */
220
0x09090, 'g', 0x00090, /* gen */
221
0x09060, 'g', 0x00060, /* gao */
222
0x09080, 'g', 0x00080, /* gan */
223
0x09040, 'g', 0x00040, /* gai */
224
0x07060, 'n', 0x00060, /* nao */
225
0x07010, 'n', 0x00010, /* na */
226
0x07040, 'n', 0x00040, /* nai */
227
0x07080, 'n', 0x00080, /* nan */
228
0x07030, 'n', 0x00030, /* ne */
229
0x0,
230
};
231
232
static const char * fuzzy_pairs[] = {
233
"z", "zh",
234
"c", "ch",
235
"s", "sh",
236
"an", "ang",
237
"on", "ong",
238
"en", "eng",
239
"in", "ing",
240
"eng", "ong",
241
"ian", "iang",
242
"uan", "uang",
243
"l", "n",
244
"f", "h",
245
"r", "l",
246
"k", "g",
247
};
248
static const unsigned num_fuzzy_pairs = sizeof(fuzzy_pairs)/sizeof(*fuzzy_pairs)/2;
249
250
static const char * auto_correction_pairs[] = {
251
"ign", "ing",
252
"img", "ing",
253
"uei", "ui",
254
"uen", "un",
255
"iou", "iu",
256
};
257
static const unsigned num_auto_correction_pairs = sizeof(auto_correction_pairs)/sizeof(*auto_correction_pairs)/2;
258
259
static const TPyTabEntry
260
pinyin_table[] = {
261
{"a", 0x00010},
262
{"ai", 0x00040},
263
{"an", 0x00080},
264
{"ang", 0x000a0},
265
{"ao", 0x00060},
266
{"b", 0x01000},
267
{"ba", 0x01010},
268
{"bai", 0x01040},
269
{"ban", 0x01080},
270
{"bang", 0x010a0},
271
{"bao", 0x01060},
272
{"bei", 0x01050},
273
{"ben", 0x01090},
274
{"beng", 0x010b0},
275
{"bi", 0x010d0},
276
{"bian", 0x01120},
277
{"biao", 0x01100},
278
{"bie", 0x010f0},
279
{"bin", 0x01130},
280
{"bing", 0x01150},
281
{"bo", 0x01020},
282
{"bu", 0x01160},
283
{"c", 0x14000},
284
{"ca", 0x14010},
285
{"cai", 0x14040},
286
{"can", 0x14080},
287
{"cang", 0x140a0},
288
{"cao", 0x14060},
289
{"ce", 0x14030},
290
{"cei", 0x14050},
291
{"cen", 0x14090},
292
{"ceng", 0x140b0},
293
{"ch", 0x10000},
294
{"cha", 0x10010},
295
{"chai", 0x10040},
296
{"chan", 0x10080},
297
{"chang", 0x100a0},
298
{"chao", 0x10060},
299
{"che", 0x10030},
300
{"chen", 0x10090},
301
{"cheng", 0x100b0},
302
{"chi", 0x100d0},
303
{"chong", 0x101e0},
304
{"chou", 0x10070},
305
{"chu", 0x10160},
306
{"chua", 0x10170},
307
{"chuai", 0x10190},
308
{"chuan", 0x101b0},
309
{"chuang", 0x101d0},
310
{"chui", 0x101a0},
311
{"chun", 0x101c0},
312
{"chuo", 0x10180},
313
{"ci", 0x140d0},
314
{"cong", 0x141e0},
315
{"cou", 0x14070},
316
{"cu", 0x14160},
317
{"cuan", 0x141b0},
318
{"cui", 0x141a0},
319
{"cun", 0x141c0},
320
{"cuo", 0x14180},
321
{"d", 0x05000},
322
{"da", 0x05010},
323
{"dai", 0x05040},
324
{"dan", 0x05080},
325
{"dang", 0x050a0},
326
{"dao", 0x05060},
327
{"de", 0x05030},
328
{"dei", 0x05050},
329
{"den", 0x05090},
330
{"deng", 0x050b0},
331
{"di", 0x050d0},
332
{"dia", 0x050e0},
333
{"dian", 0x05120},
334
{"diao", 0x05100},
335
{"die", 0x050f0},
336
{"ding", 0x05150},
337
{"diu", 0x05110},
338
{"dong", 0x051e0},
339
{"dou", 0x05070},
340
{"du", 0x05160},
341
{"duan", 0x051b0},
342
{"dui", 0x051a0},
343
{"dun", 0x051c0},
344
{"duo", 0x05180},
345
{"e", 0x00030},
346
{"ei", 0x00050},
347
{"en", 0x00090},
348
{"eng", 0x000b0},
349
{"er", 0x000c0},
350
{"f", 0x04000},
351
{"fa", 0x04010},
352
{"fan", 0x04080},
353
{"fang", 0x040a0},
354
{"fei", 0x04050},
355
{"fen", 0x04090},
356
{"feng", 0x040b0},
357
{"fiao", 0x04100},
358
{"fo", 0x04020},
359
{"fou", 0x04070},
360
{"fu", 0x04160},
361
{"g", 0x09000},
362
{"ga", 0x09010},
363
{"gai", 0x09040},
364
{"gan", 0x09080},
365
{"gang", 0x090a0},
366
{"gao", 0x09060},
367
{"ge", 0x09030},
368
{"gei", 0x09050},
369
{"gen", 0x09090},
370
{"geng", 0x090b0},
371
{"gong", 0x091e0},
372
{"gou", 0x09070},
373
{"gu", 0x09160},
374
{"gua", 0x09170},
375
{"guai", 0x09190},
376
{"guan", 0x091b0},
377
{"guang", 0x091d0},
378
{"gui", 0x091a0},
379
{"gun", 0x091c0},
380
{"guo", 0x09180},
381
{"h", 0x0b000},
382
{"ha", 0x0b010},
383
{"hai", 0x0b040},
384
{"han", 0x0b080},
385
{"hang", 0x0b0a0},
386
{"hao", 0x0b060},
387
{"he", 0x0b030},
388
{"hei", 0x0b050},
389
{"hen", 0x0b090},
390
{"heng", 0x0b0b0},
391
{"hong", 0x0b1e0},
392
{"hou", 0x0b070},
393
{"hu", 0x0b160},
394
{"hua", 0x0b170},
395
{"huai", 0x0b190},
396
{"huan", 0x0b1b0},
397
{"huang", 0x0b1d0},
398
{"hui", 0x0b1a0},
399
{"hun", 0x0b1c0},
400
{"huo", 0x0b180},
401
{"j", 0x0c000},
402
{"ji", 0x0c0d0},
403
{"jia", 0x0c0e0},
404
{"jian", 0x0c120},
405
{"jiang", 0x0c140},
406
{"jiao", 0x0c100},
407
{"jie", 0x0c0f0},
408
{"jin", 0x0c130},
409
{"jing", 0x0c150},
410
{"jiong", 0x0c210},
411
{"jiu", 0x0c110},
412
{"ju", 0x0c160},
413
{"juan", 0x0c1b0},
414
{"jue", 0x0c200},
415
{"jun", 0x0c1c0},
416
{"k", 0x0a000},
417
{"ka", 0x0a010},
418
{"kai", 0x0a040},
419
{"kan", 0x0a080},
420
{"kang", 0x0a0a0},
421
{"kao", 0x0a060},
422
{"ke", 0x0a030},
423
{"kei", 0x0a050},
424
{"ken", 0x0a090},
425
{"keng", 0x0a0b0},
426
{"kong", 0x0a1e0},
427
{"kou", 0x0a070},
428
{"ku", 0x0a160},
429
{"kua", 0x0a170},
430
{"kuai", 0x0a190},
431
{"kuan", 0x0a1b0},
432
{"kuang", 0x0a1d0},
433
{"kui", 0x0a1a0},
434
{"kun", 0x0a1c0},
435
{"kuo", 0x0a180},
436
{"l", 0x08000},
437
{"la", 0x08010},
438
{"lai", 0x08040},
439
{"lan", 0x08080},
440
{"lang", 0x080a0},
441
{"lao", 0x08060},
442
{"le", 0x08030},
443
{"lei", 0x08050},
444
{"leng", 0x080b0},
445
{"li", 0x080d0},
446
{"lia", 0x080e0},
447
{"lian", 0x08120},
448
{"liang", 0x08140},
449
{"liao", 0x08100},
450
{"lie", 0x080f0},
451
{"lin", 0x08130},
452
{"ling", 0x08150},
453
{"liu", 0x08110},
454
{"lo", 0x08020},
455
{"long", 0x081e0},
456
{"lou", 0x08070},
457
{"lu", 0x08160},
458
{"luan", 0x081b0},
459
{"lue", 0x08200},
460
{"lun", 0x081c0},
461
{"luo", 0x08180},
462
{"lv", 0x081f0},
463
{"m", 0x03000},
464
{"ma", 0x03010},
465
{"mai", 0x03040},
466
{"man", 0x03080},
467
{"mang", 0x030a0},
468
{"mao", 0x03060},
469
{"me", 0x03030},
470
{"mei", 0x03050},
471
{"men", 0x03090},
472
{"meng", 0x030b0},
473
{"mi", 0x030d0},
474
{"mian", 0x03120},
475
{"miao", 0x03100},
476
{"mie", 0x030f0},
477
{"min", 0x03130},
478
{"ming", 0x03150},
479
{"miu", 0x03110},
480
{"mo", 0x03020},
481
{"mou", 0x03070},
482
{"mu", 0x03160},
483
{"n", 0x07000},
484
{"na", 0x07010},
485
{"nai", 0x07040},
486
{"nan", 0x07080},
487
{"nang", 0x070a0},
488
{"nao", 0x07060},
489
{"ne", 0x07030},
490
{"nei", 0x07050},
491
{"nen", 0x07090},
492
{"neng", 0x070b0},
493
{"ni", 0x070d0},
494
{"nian", 0x07120},
495
{"niang", 0x07140},
496
{"niao", 0x07100},
497
{"nie", 0x070f0},
498
{"nin", 0x07130},
499
{"ning", 0x07150},
500
{"niu", 0x07110},
501
{"nong", 0x071e0},
502
{"nou", 0x07070},
503
{"nu", 0x07160},
504
{"nuan", 0x071b0},
505
{"nue", 0x07200},
506
{"nun", 0x071c0},
507
{"nuo", 0x07180},
508
{"nv", 0x071f0},
509
{"o", 0x00020},
510
{"ou", 0x00070},
511
{"p", 0x02000},
512
{"pa", 0x02010},
513
{"pai", 0x02040},
514
{"pan", 0x02080},
515
{"pang", 0x020a0},
516
{"pao", 0x02060},
517
{"pei", 0x02050},
518
{"pen", 0x02090},
519
{"peng", 0x020b0},
520
{"pi", 0x020d0},
521
{"pian", 0x02120},
522
{"piao", 0x02100},
523
{"pie", 0x020f0},
524
{"pin", 0x02130},
525
{"ping", 0x02150},
526
{"po", 0x02020},
527
{"pou", 0x02070},
528
{"pu", 0x02160},
529
{"q", 0x0d000},
530
{"qi", 0x0d0d0},
531
{"qia", 0x0d0e0},
532
{"qian", 0x0d120},
533
{"qiang", 0x0d140},
534
{"qiao", 0x0d100},
535
{"qie", 0x0d0f0},
536
{"qin", 0x0d130},
537
{"qing", 0x0d150},
538
{"qiong", 0x0d210},
539
{"qiu", 0x0d110},
540
{"qu", 0x0d160},
541
{"quan", 0x0d1b0},
542
{"que", 0x0d200},
543
{"qun", 0x0d1c0},
544
{"r", 0x12000},
545
{"ran", 0x12080},
546
{"rang", 0x120a0},
547
{"rao", 0x12060},
548
{"re", 0x12030},
549
{"ren", 0x12090},
550
{"reng", 0x120b0},
551
{"ri", 0x120d0},
552
{"rong", 0x121e0},
553
{"rou", 0x12070},
554
{"ru", 0x12160},
555
{"ruan", 0x121b0},
556
{"rui", 0x121a0},
557
{"run", 0x121c0},
558
{"ruo", 0x12180},
559
{"s", 0x15000},
560
{"sa", 0x15010},
561
{"sai", 0x15040},
562
{"san", 0x15080},
563
{"sang", 0x150a0},
564
{"sao", 0x15060},
565
{"se", 0x15030},
566
{"sen", 0x15090},
567
{"seng", 0x150b0},
568
{"sh", 0x11000},
569
{"sha", 0x11010},
570
{"shai", 0x11040},
571
{"shan", 0x11080},
572
{"shang", 0x110a0},
573
{"shao", 0x11060},
574
{"she", 0x11030},
575
{"shei", 0x11050},
576
{"shen", 0x11090},
577
{"sheng", 0x110b0},
578
{"shi", 0x110d0},
579
{"shou", 0x11070},
580
{"shu", 0x11160},
581
{"shua", 0x11170},
582
{"shuai", 0x11190},
583
{"shuan", 0x111b0},
584
{"shuang", 0x111d0},
585
{"shui", 0x111a0},
586
{"shun", 0x111c0},
587
{"shuo", 0x11180},
588
{"si", 0x150d0},
589
{"song", 0x151e0},
590
{"sou", 0x15070},
591
{"su", 0x15160},
592
{"suan", 0x151b0},
593
{"sui", 0x151a0},
594
{"sun", 0x151c0},
595
{"suo", 0x15180},
596
{"t", 0x06000},
597
{"ta", 0x06010},
598
{"tai", 0x06040},
599
{"tan", 0x06080},
600
{"tang", 0x060a0},
601
{"tao", 0x06060},
602
{"te", 0x06030},
603
{"tei", 0x06050},
604
{"teng", 0x060b0},
605
{"ti", 0x060d0},
606
{"tian", 0x06120},
607
{"tiao", 0x06100},
608
{"tie", 0x060f0},
609
{"ting", 0x06150},
610
{"tong", 0x061e0},
611
{"tou", 0x06070},
612
{"tu", 0x06160},
613
{"tuan", 0x061b0},
614
{"tui", 0x061a0},
615
{"tun", 0x061c0},
616
{"tuo", 0x06180},
617
{"w", 0x17000},
618
{"wa", 0x17010},
619
{"wai", 0x17040},
620
{"wan", 0x17080},
621
{"wang", 0x170a0},
622
{"wei", 0x17050},
623
{"wen", 0x17090},
624
{"weng", 0x170b0},
625
{"wo", 0x17020},
626
{"wu", 0x17160},
627
{"x", 0x0e000},
628
{"xi", 0x0e0d0},
629
{"xia", 0x0e0e0},
630
{"xian", 0x0e120},
631
{"xiang", 0x0e140},
632
{"xiao", 0x0e100},
633
{"xie", 0x0e0f0},
634
{"xin", 0x0e130},
635
{"xing", 0x0e150},
636
{"xiong", 0x0e210},
637
{"xiu", 0x0e110},
638
{"xu", 0x0e160},
639
{"xuan", 0x0e1b0},
640
{"xue", 0x0e200},
641
{"xun", 0x0e1c0},
642
{"y", 0x16000},
643
{"ya", 0x16010},
644
{"yai", 0x16040},
645
{"yan", 0x16080},
646
{"yang", 0x160a0},
647
{"yao", 0x16060},
648
{"ye", 0x16030},
649
{"yi", 0x160d0},
650
{"yin", 0x16130},
651
{"ying", 0x16150},
652
{"yo", 0x16020},
653
{"yong", 0x161e0},
654
{"you", 0x16070},
655
{"yu", 0x16160},
656
{"yuan", 0x161b0},
657
{"yue", 0x16200},
658
{"yun", 0x161c0},
659
{"z", 0x13000},
660
{"za", 0x13010},
661
{"zai", 0x13040},
662
{"zan", 0x13080},
663
{"zang", 0x130a0},
664
{"zao", 0x13060},
665
{"ze", 0x13030},
666
{"zei", 0x13050},
667
{"zen", 0x13090},
668
{"zeng", 0x130b0},
669
{"zh", 0x0f000},
670
{"zha", 0x0f010},
671
{"zhai", 0x0f040},
672
{"zhan", 0x0f080},
673
{"zhang", 0x0f0a0},
674
{"zhao", 0x0f060},
675
{"zhe", 0x0f030},
676
{"zhei", 0x0f050},
677
{"zhen", 0x0f090},
678
{"zheng", 0x0f0b0},
679
{"zhi", 0x0f0d0},
680
{"zhong", 0x0f1e0},
681
{"zhou", 0x0f070},
682
{"zhu", 0x0f160},
683
{"zhua", 0x0f170},
684
{"zhuai", 0x0f190},
685
{"zhuan", 0x0f1b0},
686
{"zhuang", 0x0f1d0},
687
{"zhui", 0x0f1a0},
688
{"zhun", 0x0f1c0},
689
{"zhuo", 0x0f180},
690
{"zi", 0x130d0},
691
{"zong", 0x131e0},
692
{"zou", 0x13070},
693
{"zu", 0x13160},
694
{"zuan", 0x131b0},
695
{"zui", 0x131a0},
696
{"zun", 0x131c0},
697
{"zuo", 0x13180},
698
};
699
700
static int
701
pytab_entry_compare (const char *s, TPyTabEntry *v)
702
{return strcmp (s, v->pystr);}
703
704
TSyllable
705
CPinyinData::encodeSyllable (const char *pinyin)
706
{
707
typedef int (*bsearch_compare) (const void*, const void*);
708
TPyTabEntry *e = (TPyTabEntry*) bsearch (pinyin, pinyin_table,
709
sizeof(pinyin_table)/sizeof(pinyin_table[0]),
710
sizeof(pinyin_table[0]),
711
(bsearch_compare) pytab_entry_compare);
712
if (e)
713
return e->id;
714
715
return 0;
716
}
717
718
const char *
719
CPinyinData::decodeSyllable (TSyllable s, const char **i, const char **f)
720
{
721
if (i) *i = initials[s.initial];
722
if (f) *f = finals[s.final];
723
724
static char buf[128];
725
snprintf (buf, sizeof(buf), "%s%s", initials[s.initial], finals[s.final]);
726
727
typedef int (*bsearch_compare) (const void*, const void*);
728
TPyTabEntry *e = (TPyTabEntry*) bsearch (buf, pinyin_table,
729
sizeof(pinyin_table)/sizeof(pinyin_table[0]),
730
sizeof(pinyin_table[0]),
731
(bsearch_compare) pytab_entry_compare);
732
733
if (e)
734
return e->pystr;
735
736
return NULL;
737
}
738
739
const char **
740
CPinyinData::getAutoCorrectionPairs (unsigned &num)
741
{
742
num = num_auto_correction_pairs;
743
return auto_correction_pairs;
744
}
745
746
const char **
747
CPinyinData::getFuzzyPairs (unsigned &num)
748
{
749
num = num_fuzzy_pairs;
750
return fuzzy_pairs;
751
}
752
753
const char **
754
CPinyinData::getInitials (unsigned &num)
755
{
756
num = num_initials;
757
return initials;
758
}
759
760
const char **
761
CPinyinData::getFinals (unsigned &num)
762
{
763
num = num_finals;
764
return finals;
765
}
766
767
const TPyTabEntry *
768
CPinyinData::getPinyinTable(unsigned &num)
769
{
770
num = sizeof(pinyin_table) / sizeof(TPyTabEntry);
771
return pinyin_table;
772
}
773
774
const unsigned *
775
CPinyinData::getInnerFuzzyFinalMap (unsigned &num)
776
{
777
num = num_fuzzy_finals;
778
return fuzzy_finals_map;
779
}
780
781
void
782
CPinyinData::getFuzzyPreProSyllables (const unsigned **pre_syls, const unsigned **pro_syls)
783
{
784
*pre_syls = fuzzy_pre_syllables;
785
*pro_syls = fuzzy_pro_syllables;
786
}
Loggerhead is a web-based interface for Breezy
Version: 2.0.1