line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package Lingua::ZH::PinyinConvert::ID; |
2
|
|
|
|
|
|
|
|
3
|
1
|
|
|
1
|
|
21102
|
use 5.010001; |
|
1
|
|
|
|
|
4
|
|
|
1
|
|
|
|
|
43
|
|
4
|
1
|
|
|
1
|
|
6
|
use strict; |
|
1
|
|
|
|
|
1
|
|
|
1
|
|
|
|
|
40
|
|
5
|
1
|
|
|
1
|
|
6
|
use warnings; |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
5464
|
|
6
|
|
|
|
|
|
|
|
7
|
|
|
|
|
|
|
our $VERSION = '0.04'; # VERSION |
8
|
|
|
|
|
|
|
|
9
|
|
|
|
|
|
|
# notes: |
10
|
|
|
|
|
|
|
|
11
|
|
|
|
|
|
|
# * for 'w' as alternative spelling to 'u'/'o', only those occuring |
12
|
|
|
|
|
|
|
# after vowel is listed (e.g. taw, but not khwai), except kwa/kwe/kwi |
13
|
|
|
|
|
|
|
# and hwaX (commonly encountered) |
14
|
|
|
|
|
|
|
|
15
|
|
|
|
|
|
|
my %hy2id = ( |
16
|
|
|
|
|
|
|
a => 'a', |
17
|
|
|
|
|
|
|
ai => 'ai', |
18
|
|
|
|
|
|
|
an => 'an', |
19
|
|
|
|
|
|
|
ang => 'ang', |
20
|
|
|
|
|
|
|
ao => ['au', 'ao', 'aw'], |
21
|
|
|
|
|
|
|
|
22
|
|
|
|
|
|
|
ba => 'pa', |
23
|
|
|
|
|
|
|
bai => 'pai', |
24
|
|
|
|
|
|
|
ban => 'pan', |
25
|
|
|
|
|
|
|
bang => 'pang', |
26
|
|
|
|
|
|
|
bao => ['pau', 'pao', 'paw'], |
27
|
|
|
|
|
|
|
bei => 'pei', |
28
|
|
|
|
|
|
|
ben => 'pen', |
29
|
|
|
|
|
|
|
beng => 'peng', |
30
|
|
|
|
|
|
|
bi => 'pi', |
31
|
|
|
|
|
|
|
bian => 'pien', |
32
|
|
|
|
|
|
|
biao => ['piau', 'piao', 'piaw'], |
33
|
|
|
|
|
|
|
bie => 'pie', |
34
|
|
|
|
|
|
|
bin => 'pin', |
35
|
|
|
|
|
|
|
bing => 'ping', |
36
|
|
|
|
|
|
|
bo => ['po', 'puo'], #pwo |
37
|
|
|
|
|
|
|
bu => 'pu', |
38
|
|
|
|
|
|
|
|
39
|
|
|
|
|
|
|
ca => 'cha', |
40
|
|
|
|
|
|
|
cai => 'chai', |
41
|
|
|
|
|
|
|
can => 'chan', |
42
|
|
|
|
|
|
|
cang => 'chang', |
43
|
|
|
|
|
|
|
cao => ['chau', 'chao', 'chaw'], |
44
|
|
|
|
|
|
|
ce => 'che', |
45
|
|
|
|
|
|
|
cen => 'chen', |
46
|
|
|
|
|
|
|
ceng => 'cheng', |
47
|
|
|
|
|
|
|
cha => 'cha', |
48
|
|
|
|
|
|
|
chai => 'chai', |
49
|
|
|
|
|
|
|
chan => 'chan', |
50
|
|
|
|
|
|
|
chang => 'chang', |
51
|
|
|
|
|
|
|
chao => ['chau', 'chao', 'chaw'], |
52
|
|
|
|
|
|
|
che => 'che', |
53
|
|
|
|
|
|
|
chen => 'chen', |
54
|
|
|
|
|
|
|
cheng => 'cheng', |
55
|
|
|
|
|
|
|
chi => 'che', |
56
|
|
|
|
|
|
|
chong => 'chung', |
57
|
|
|
|
|
|
|
chou => ['chou', 'chow', 'cheu'], # chEw |
58
|
|
|
|
|
|
|
chu => 'chu', |
59
|
|
|
|
|
|
|
chuai => ['chuai'],# 'chwai'], |
60
|
|
|
|
|
|
|
chuan => ['chuan'],# 'chwan'], |
61
|
|
|
|
|
|
|
chuang => ['chuang'],# 'chwang'], |
62
|
|
|
|
|
|
|
chui => ['chuei'],# 'chwei'], |
63
|
|
|
|
|
|
|
chun => ['chuen'],# 'chwen'], |
64
|
|
|
|
|
|
|
chuo => ['chuo'],# 'chwo'], |
65
|
|
|
|
|
|
|
ci => 'che', |
66
|
|
|
|
|
|
|
cong => 'chung', |
67
|
|
|
|
|
|
|
cou => ['chou', 'chow', 'chew'], # chEw |
68
|
|
|
|
|
|
|
cu => 'chu', |
69
|
|
|
|
|
|
|
cuan => ['chuan'],# 'chwan'], |
70
|
|
|
|
|
|
|
cui => ['chuei'],# 'chwei'], |
71
|
|
|
|
|
|
|
cun => ['chuen'],# 'chwen'], |
72
|
|
|
|
|
|
|
cuo => ['chuo'],# 'chwo'], |
73
|
|
|
|
|
|
|
|
74
|
|
|
|
|
|
|
da => 'ta', |
75
|
|
|
|
|
|
|
dai => 'tai', |
76
|
|
|
|
|
|
|
dan => 'tan', |
77
|
|
|
|
|
|
|
dang => 'tang', |
78
|
|
|
|
|
|
|
dao => ['tau', 'tao', 'taw'], |
79
|
|
|
|
|
|
|
de => 'te', |
80
|
|
|
|
|
|
|
dei => 'tei', |
81
|
|
|
|
|
|
|
deng => 'teng', |
82
|
|
|
|
|
|
|
di => 'ti', |
83
|
|
|
|
|
|
|
dian => 'tien', |
84
|
|
|
|
|
|
|
diao => ['tiau', 'tiao', 'tiaw'], |
85
|
|
|
|
|
|
|
die => 'tie', |
86
|
|
|
|
|
|
|
ding => 'ting', |
87
|
|
|
|
|
|
|
diu => 'tiu', |
88
|
|
|
|
|
|
|
dong => 'tung', |
89
|
|
|
|
|
|
|
dou => ['tou', 'tow', 'teu', 'tew'], |
90
|
|
|
|
|
|
|
du => 'tu', |
91
|
|
|
|
|
|
|
duan => ['tuan'],# 'twan'], |
92
|
|
|
|
|
|
|
dui => ['tuei'],# 'twei'], |
93
|
|
|
|
|
|
|
dun => ['tuen'],# 'twen'], |
94
|
|
|
|
|
|
|
duo => 'tuo', # two |
95
|
|
|
|
|
|
|
|
96
|
|
|
|
|
|
|
e => 'e', |
97
|
|
|
|
|
|
|
en => 'en', |
98
|
|
|
|
|
|
|
er => ['er', 'el'], |
99
|
|
|
|
|
|
|
|
100
|
|
|
|
|
|
|
fa => 'fa', |
101
|
|
|
|
|
|
|
fan => 'fan', |
102
|
|
|
|
|
|
|
fang => 'fang', |
103
|
|
|
|
|
|
|
fei => 'fei', |
104
|
|
|
|
|
|
|
fen => 'fen', |
105
|
|
|
|
|
|
|
feng => 'feng', |
106
|
|
|
|
|
|
|
fo => ['fo', 'fuo'], # fwo |
107
|
|
|
|
|
|
|
fou => ['fou', 'fow'], |
108
|
|
|
|
|
|
|
fu => 'fu', |
109
|
|
|
|
|
|
|
|
110
|
|
|
|
|
|
|
ga => 'ka', |
111
|
|
|
|
|
|
|
gai => 'kai', |
112
|
|
|
|
|
|
|
gan => 'kan', |
113
|
|
|
|
|
|
|
gang => 'kang', |
114
|
|
|
|
|
|
|
gao => ['kau', 'kao', 'kaw'], |
115
|
|
|
|
|
|
|
ge => 'ke', |
116
|
|
|
|
|
|
|
gei => 'kei', |
117
|
|
|
|
|
|
|
gen => 'ken', |
118
|
|
|
|
|
|
|
geng => 'keng', |
119
|
|
|
|
|
|
|
gong => 'kung', |
120
|
|
|
|
|
|
|
gou => ['kou', 'kow', 'keu'], # kEw |
121
|
|
|
|
|
|
|
gu => 'ku', |
122
|
|
|
|
|
|
|
gua => ['kua', 'kwa'], |
123
|
|
|
|
|
|
|
guai => ['kuai', 'kwai'], |
124
|
|
|
|
|
|
|
guan => ['kuan', 'kwan'], |
125
|
|
|
|
|
|
|
guang => ['kuang', 'kwang'], |
126
|
|
|
|
|
|
|
gui => ['kuei', 'kwei'], |
127
|
|
|
|
|
|
|
gun => ['kuen'],# 'kwen'], |
128
|
|
|
|
|
|
|
guo => ['kuo'],# 'kwo'], |
129
|
|
|
|
|
|
|
|
130
|
|
|
|
|
|
|
ha => 'ha', |
131
|
|
|
|
|
|
|
hai => 'hai', |
132
|
|
|
|
|
|
|
han => 'han', |
133
|
|
|
|
|
|
|
hang => 'hang', |
134
|
|
|
|
|
|
|
hao => ['hau', 'hao', 'haw'], |
135
|
|
|
|
|
|
|
he => 'he', |
136
|
|
|
|
|
|
|
hei => 'hei', |
137
|
|
|
|
|
|
|
hen => 'hen', |
138
|
|
|
|
|
|
|
heng => 'heng', |
139
|
|
|
|
|
|
|
hong => 'hung', |
140
|
|
|
|
|
|
|
hou => ['hou', 'how', 'heu'], # hEw |
141
|
|
|
|
|
|
|
hu => 'hu', |
142
|
|
|
|
|
|
|
hua => ['hua', 'hwa'], |
143
|
|
|
|
|
|
|
huai => ['huai', 'hwai'], |
144
|
|
|
|
|
|
|
huan => ['huan', 'hwan'], |
145
|
|
|
|
|
|
|
huang => ['huang', 'hwang'], |
146
|
|
|
|
|
|
|
hui => ['huei', 'hwei'], |
147
|
|
|
|
|
|
|
hun => ['huen'],# 'hwen'], |
148
|
|
|
|
|
|
|
huo => 'huo', # hwo |
149
|
|
|
|
|
|
|
|
150
|
|
|
|
|
|
|
ji => 'ci', |
151
|
|
|
|
|
|
|
jia => 'cia', |
152
|
|
|
|
|
|
|
jian => 'cien', |
153
|
|
|
|
|
|
|
jiang => 'ciang', |
154
|
|
|
|
|
|
|
jiao => ['ciau', 'ciao', 'ciaw'], |
155
|
|
|
|
|
|
|
jie => 'cie', |
156
|
|
|
|
|
|
|
jin => 'cin', |
157
|
|
|
|
|
|
|
jing => 'cing', |
158
|
|
|
|
|
|
|
jiong => 'ciung', |
159
|
|
|
|
|
|
|
jiu => 'ciu', |
160
|
|
|
|
|
|
|
ju => 'cu', #? |
161
|
|
|
|
|
|
|
juan => 'cien', #? |
162
|
|
|
|
|
|
|
jue => 'cue', #? |
163
|
|
|
|
|
|
|
jun => ['cun', 'cin'], #? |
164
|
|
|
|
|
|
|
|
165
|
|
|
|
|
|
|
ka => 'kha', |
166
|
|
|
|
|
|
|
kai => 'khai', |
167
|
|
|
|
|
|
|
kan => 'khan', |
168
|
|
|
|
|
|
|
kang => 'khang', |
169
|
|
|
|
|
|
|
kao => ['khau', 'khao', 'khaw'], |
170
|
|
|
|
|
|
|
ke => 'khe', |
171
|
|
|
|
|
|
|
ken => 'khen', |
172
|
|
|
|
|
|
|
keng => 'kheng', |
173
|
|
|
|
|
|
|
kong => 'khung', |
174
|
|
|
|
|
|
|
kou => ['khou', 'khow', 'kheu'], # khEw |
175
|
|
|
|
|
|
|
ku => 'khu', |
176
|
|
|
|
|
|
|
kua => ['khua'],# 'khwa'], |
177
|
|
|
|
|
|
|
kuai => ['khuai'],# 'khwai'], |
178
|
|
|
|
|
|
|
kuan => ['khuan'],# 'khwan'], |
179
|
|
|
|
|
|
|
kuang => ['khuang'],# 'khwang'], |
180
|
|
|
|
|
|
|
kui => ['khuei'],# 'khwei'], |
181
|
|
|
|
|
|
|
kun => 'khuen', # khwen |
182
|
|
|
|
|
|
|
kuo => 'khuo', # khwo |
183
|
|
|
|
|
|
|
|
184
|
|
|
|
|
|
|
la => 'la', |
185
|
|
|
|
|
|
|
lai => 'lai', |
186
|
|
|
|
|
|
|
lan => 'lan', |
187
|
|
|
|
|
|
|
lang => 'lang', |
188
|
|
|
|
|
|
|
lao => ['lau', 'lao'], |
189
|
|
|
|
|
|
|
le => 'le', |
190
|
|
|
|
|
|
|
lei => 'lei', |
191
|
|
|
|
|
|
|
leng => 'leng', |
192
|
|
|
|
|
|
|
li => 'li', |
193
|
|
|
|
|
|
|
lia => 'lia', |
194
|
|
|
|
|
|
|
lian => 'lien', |
195
|
|
|
|
|
|
|
liang => 'liang', |
196
|
|
|
|
|
|
|
liao => ['liau', 'liao', 'liaw'], |
197
|
|
|
|
|
|
|
lie => 'lie', |
198
|
|
|
|
|
|
|
lin => 'lin', |
199
|
|
|
|
|
|
|
ling => 'ling', |
200
|
|
|
|
|
|
|
liu => 'liu', |
201
|
|
|
|
|
|
|
long => 'lung', |
202
|
|
|
|
|
|
|
lou => ['lou', 'low', 'leu'],#, 'lEw'] |
203
|
|
|
|
|
|
|
lu => 'lu', |
204
|
|
|
|
|
|
|
lv => ['li'], # lu |
205
|
|
|
|
|
|
|
luan => ['luan'], # lwan |
206
|
|
|
|
|
|
|
lve => ['lie'], #'lue'], # lwe |
207
|
|
|
|
|
|
|
lun => ['luen'], # lwen |
208
|
|
|
|
|
|
|
luo => ['luo'], # lwo |
209
|
|
|
|
|
|
|
|
210
|
|
|
|
|
|
|
ma => 'ma', |
211
|
|
|
|
|
|
|
mai => 'mai', |
212
|
|
|
|
|
|
|
man => 'man', |
213
|
|
|
|
|
|
|
mang => 'mang', |
214
|
|
|
|
|
|
|
mao => ['mau', 'mao', 'maw'], |
215
|
|
|
|
|
|
|
me => 'me', |
216
|
|
|
|
|
|
|
mei => 'mei', |
217
|
|
|
|
|
|
|
men => 'men', |
218
|
|
|
|
|
|
|
meng => 'meng', |
219
|
|
|
|
|
|
|
mi => 'mi', |
220
|
|
|
|
|
|
|
mian => 'mien', |
221
|
|
|
|
|
|
|
miao => ['miau', 'miao', 'miaw'], |
222
|
|
|
|
|
|
|
mie => 'mie', |
223
|
|
|
|
|
|
|
min => 'min', |
224
|
|
|
|
|
|
|
ming => 'ming', |
225
|
|
|
|
|
|
|
miu => 'miu', |
226
|
|
|
|
|
|
|
mo => ['mo', 'muo'], # mwo |
227
|
|
|
|
|
|
|
mou => ['mou', 'mow'], |
228
|
|
|
|
|
|
|
mu => 'mu', |
229
|
|
|
|
|
|
|
|
230
|
|
|
|
|
|
|
na => 'na', |
231
|
|
|
|
|
|
|
nai => 'nai', |
232
|
|
|
|
|
|
|
nan => 'nan', |
233
|
|
|
|
|
|
|
nang => 'nang', |
234
|
|
|
|
|
|
|
nao => ['nau', 'nao'], |
235
|
|
|
|
|
|
|
ne => 'ne', |
236
|
|
|
|
|
|
|
nei => 'nei', |
237
|
|
|
|
|
|
|
nen => 'nen', |
238
|
|
|
|
|
|
|
neng => 'neng', |
239
|
|
|
|
|
|
|
ni => 'ni', |
240
|
|
|
|
|
|
|
nian => 'nien', |
241
|
|
|
|
|
|
|
niang => 'niang', |
242
|
|
|
|
|
|
|
niao => ['niau', 'niao'], |
243
|
|
|
|
|
|
|
nie => 'nie', |
244
|
|
|
|
|
|
|
nin => 'nin', |
245
|
|
|
|
|
|
|
ning => 'ning', |
246
|
|
|
|
|
|
|
niu => 'niu', |
247
|
|
|
|
|
|
|
nong => 'nung', |
248
|
|
|
|
|
|
|
nou => ['nou'], # now |
249
|
|
|
|
|
|
|
nu => 'nu', |
250
|
|
|
|
|
|
|
nv => ['ni'], # nu |
251
|
|
|
|
|
|
|
nuan => 'nuan', |
252
|
|
|
|
|
|
|
nve => 'nie', #nue? |
253
|
|
|
|
|
|
|
nuo => 'nuo', |
254
|
|
|
|
|
|
|
|
255
|
|
|
|
|
|
|
o => 'o', |
256
|
|
|
|
|
|
|
ou => ['ou', 'ow'], |
257
|
|
|
|
|
|
|
|
258
|
|
|
|
|
|
|
pa => 'pha', |
259
|
|
|
|
|
|
|
pai => 'phai', |
260
|
|
|
|
|
|
|
pan => 'phan', |
261
|
|
|
|
|
|
|
pang => 'phang', |
262
|
|
|
|
|
|
|
pao => ['phau', 'phao', 'phaw'], |
263
|
|
|
|
|
|
|
pei => 'phei', |
264
|
|
|
|
|
|
|
pen => 'phen', |
265
|
|
|
|
|
|
|
peng => 'pheng', |
266
|
|
|
|
|
|
|
pi => 'phi', |
267
|
|
|
|
|
|
|
pian => 'phien', |
268
|
|
|
|
|
|
|
piao => ['phiau', 'phiao', 'phiaw'], |
269
|
|
|
|
|
|
|
pie => 'phie', |
270
|
|
|
|
|
|
|
pin => 'phin', |
271
|
|
|
|
|
|
|
ping => 'phing', |
272
|
|
|
|
|
|
|
po => ['pho', 'phuo'], |
273
|
|
|
|
|
|
|
pou => ['phou', 'phow'], |
274
|
|
|
|
|
|
|
pu => 'phu', |
275
|
|
|
|
|
|
|
|
276
|
|
|
|
|
|
|
qi => 'chi', |
277
|
|
|
|
|
|
|
qia => 'chia', |
278
|
|
|
|
|
|
|
qian => 'chien', |
279
|
|
|
|
|
|
|
qiang => 'chiang', |
280
|
|
|
|
|
|
|
qiao => ['chiau', 'chiao', 'chiaw'], |
281
|
|
|
|
|
|
|
qie => 'chie', |
282
|
|
|
|
|
|
|
qin => 'chin', |
283
|
|
|
|
|
|
|
qing => 'ching', |
284
|
|
|
|
|
|
|
qiong => 'chiung', |
285
|
|
|
|
|
|
|
qiu => 'chiu', |
286
|
|
|
|
|
|
|
qu => 'chi', |
287
|
|
|
|
|
|
|
quan => 'chuen', #? |
288
|
|
|
|
|
|
|
que => 'chue', |
289
|
|
|
|
|
|
|
qun => 'chuen', |
290
|
|
|
|
|
|
|
|
291
|
|
|
|
|
|
|
ran => ['ran', 'jan'], |
292
|
|
|
|
|
|
|
rang => ['rang', 'jang'], |
293
|
|
|
|
|
|
|
rao => ['rau', 'rao', 'raw', 'jau', 'jao', 'jaw'], |
294
|
|
|
|
|
|
|
re => ['re', 'je'], |
295
|
|
|
|
|
|
|
ren => ['ren', 'jen'], |
296
|
|
|
|
|
|
|
reng => ['reng', 'jeng'], |
297
|
|
|
|
|
|
|
ri => ['re', 'je'], |
298
|
|
|
|
|
|
|
rong => ['rung', 'jung'], |
299
|
|
|
|
|
|
|
rou => ['rou', 'row', 'jou', 'jow', 'reu', 'jeu'], # rEw |
300
|
|
|
|
|
|
|
ru => ['ru', 'ju'], |
301
|
|
|
|
|
|
|
ruan => ['ruan', 'juan'], |
302
|
|
|
|
|
|
|
rui => ['ruei', 'juei'], |
303
|
|
|
|
|
|
|
run => ['ruen', 'juen'], |
304
|
|
|
|
|
|
|
ruo => ['ruo', 'juo'], |
305
|
|
|
|
|
|
|
|
306
|
|
|
|
|
|
|
sa => 'sa', |
307
|
|
|
|
|
|
|
sai => 'sai', |
308
|
|
|
|
|
|
|
san => 'san', |
309
|
|
|
|
|
|
|
sang => 'sang', |
310
|
|
|
|
|
|
|
sao => ['sau', 'sao', 'saw'], |
311
|
|
|
|
|
|
|
se => 'se', |
312
|
|
|
|
|
|
|
sen => 'sen', |
313
|
|
|
|
|
|
|
seng => 'seng', |
314
|
|
|
|
|
|
|
sha => 'sha', |
315
|
|
|
|
|
|
|
shai => 'shai', |
316
|
|
|
|
|
|
|
shan => 'shan', |
317
|
|
|
|
|
|
|
shang => 'shang', |
318
|
|
|
|
|
|
|
shao => ['shau', 'shao', 'shaw'], |
319
|
|
|
|
|
|
|
she => 'she', |
320
|
|
|
|
|
|
|
shei => 'shei', |
321
|
|
|
|
|
|
|
shen => 'shen', |
322
|
|
|
|
|
|
|
sheng => 'sheng', |
323
|
|
|
|
|
|
|
shi => 'she', |
324
|
|
|
|
|
|
|
shou => ['shou', 'sheu'],# 'shEw', 'show'], |
325
|
|
|
|
|
|
|
shu => 'shu', |
326
|
|
|
|
|
|
|
shua => 'shua', |
327
|
|
|
|
|
|
|
shuai => 'shuai', |
328
|
|
|
|
|
|
|
shuan => 'shuan', |
329
|
|
|
|
|
|
|
shuang => 'shuang', |
330
|
|
|
|
|
|
|
shui => 'shuei', |
331
|
|
|
|
|
|
|
shun => 'shuen', |
332
|
|
|
|
|
|
|
shuo => 'shuo', |
333
|
|
|
|
|
|
|
si => 'se', |
334
|
|
|
|
|
|
|
song => 'sung', |
335
|
|
|
|
|
|
|
sou => ['sou', 'sow', 'seu', 'sew'], |
336
|
|
|
|
|
|
|
su => 'su', |
337
|
|
|
|
|
|
|
suan => 'suan', |
338
|
|
|
|
|
|
|
sui => 'suei', |
339
|
|
|
|
|
|
|
sun => 'suen', |
340
|
|
|
|
|
|
|
suo => 'suo', |
341
|
|
|
|
|
|
|
|
342
|
|
|
|
|
|
|
ta => 'tha', |
343
|
|
|
|
|
|
|
tai => 'thai', |
344
|
|
|
|
|
|
|
tan => 'than', |
345
|
|
|
|
|
|
|
tang => 'thang', |
346
|
|
|
|
|
|
|
tao => ['thau', 'thao', 'thaw'], |
347
|
|
|
|
|
|
|
te => 'the', |
348
|
|
|
|
|
|
|
teng => 'theng', |
349
|
|
|
|
|
|
|
ti => 'thi', |
350
|
|
|
|
|
|
|
tian => 'thien', |
351
|
|
|
|
|
|
|
tiao => ['thiau', 'thiao', 'thiaw'], |
352
|
|
|
|
|
|
|
tie => 'thie', |
353
|
|
|
|
|
|
|
ting => 'thing', |
354
|
|
|
|
|
|
|
tong => 'thung', |
355
|
|
|
|
|
|
|
tou => ['thou', 'thow', 'theu', 'thew'], |
356
|
|
|
|
|
|
|
tu => 'thu', |
357
|
|
|
|
|
|
|
tuan => 'thuan', |
358
|
|
|
|
|
|
|
tui => 'thuei', |
359
|
|
|
|
|
|
|
tun => 'thuen', |
360
|
|
|
|
|
|
|
tuo => 'thuo', |
361
|
|
|
|
|
|
|
|
362
|
|
|
|
|
|
|
wa => 'wa', |
363
|
|
|
|
|
|
|
wai => 'wai', |
364
|
|
|
|
|
|
|
wan => 'wan', |
365
|
|
|
|
|
|
|
wang => 'wang', |
366
|
|
|
|
|
|
|
wei => 'wei', |
367
|
|
|
|
|
|
|
wen => 'wen', |
368
|
|
|
|
|
|
|
weng => 'weng', |
369
|
|
|
|
|
|
|
wo => 'wo', |
370
|
|
|
|
|
|
|
wu => 'wu', |
371
|
|
|
|
|
|
|
|
372
|
|
|
|
|
|
|
xi => 'si', |
373
|
|
|
|
|
|
|
xia => 'sia', |
374
|
|
|
|
|
|
|
xian => 'sien', |
375
|
|
|
|
|
|
|
xiang => 'siang', |
376
|
|
|
|
|
|
|
xiao => ['siau', 'siao', 'siau'], |
377
|
|
|
|
|
|
|
xie => 'sie', |
378
|
|
|
|
|
|
|
xin => 'sin', |
379
|
|
|
|
|
|
|
xing => 'sing', |
380
|
|
|
|
|
|
|
xiong => 'siung', |
381
|
|
|
|
|
|
|
xiu => 'siu', |
382
|
|
|
|
|
|
|
xu => ['si', 'syu'], |
383
|
|
|
|
|
|
|
xuan => ['suan', 'swan'], |
384
|
|
|
|
|
|
|
xue => ['sie'],# 'sue'], |
385
|
|
|
|
|
|
|
xun => 'suen', |
386
|
|
|
|
|
|
|
|
387
|
|
|
|
|
|
|
ya => 'ya', |
388
|
|
|
|
|
|
|
yai => 'yai', |
389
|
|
|
|
|
|
|
yan => 'yen', |
390
|
|
|
|
|
|
|
yang => 'yang', |
391
|
|
|
|
|
|
|
yao => ['yau', 'yao', 'yaw'], |
392
|
|
|
|
|
|
|
ye => 'ye', |
393
|
|
|
|
|
|
|
yi => ['i', 'yi'], |
394
|
|
|
|
|
|
|
yin => ['in', 'yin'], |
395
|
|
|
|
|
|
|
ying => ['ing', 'ying'], |
396
|
|
|
|
|
|
|
yong => 'yung', |
397
|
|
|
|
|
|
|
you => ['yu', 'you'],# 'yow'], |
398
|
|
|
|
|
|
|
yu => ['yi', 'yu'], |
399
|
|
|
|
|
|
|
yuan => 'yuen', |
400
|
|
|
|
|
|
|
yue => ['yue', 'ye'], |
401
|
|
|
|
|
|
|
yun => ['yun', 'yin'], |
402
|
|
|
|
|
|
|
|
403
|
|
|
|
|
|
|
za => 'ca', |
404
|
|
|
|
|
|
|
zai => 'cai', |
405
|
|
|
|
|
|
|
zan => 'can', |
406
|
|
|
|
|
|
|
zang => 'cang', |
407
|
|
|
|
|
|
|
zao => ['cau', 'cao', 'caw'], |
408
|
|
|
|
|
|
|
ze => 'ce', |
409
|
|
|
|
|
|
|
zei => 'cei', |
410
|
|
|
|
|
|
|
zen => 'cen', |
411
|
|
|
|
|
|
|
zeng => 'ceng', |
412
|
|
|
|
|
|
|
zha => 'ca', |
413
|
|
|
|
|
|
|
zhai => 'cai', |
414
|
|
|
|
|
|
|
zhan => 'can', |
415
|
|
|
|
|
|
|
zhang => 'cang', |
416
|
|
|
|
|
|
|
zhao => ['cau', 'cao', 'caw'], |
417
|
|
|
|
|
|
|
zhe => 'ce', |
418
|
|
|
|
|
|
|
zhei => 'cei', |
419
|
|
|
|
|
|
|
zhen => 'cen', |
420
|
|
|
|
|
|
|
zheng => 'ceng', |
421
|
|
|
|
|
|
|
zhi => 'ce', |
422
|
|
|
|
|
|
|
zhong => 'cung', |
423
|
|
|
|
|
|
|
zhou => ['cou', 'ceu', 'cew'], #cow |
424
|
|
|
|
|
|
|
zhu => 'cu', |
425
|
|
|
|
|
|
|
zhua => 'cua', |
426
|
|
|
|
|
|
|
zhuai => 'cuai', |
427
|
|
|
|
|
|
|
zhuan => 'cuan', |
428
|
|
|
|
|
|
|
zhuang => 'cuang', |
429
|
|
|
|
|
|
|
zhui => 'cuei', |
430
|
|
|
|
|
|
|
zhun => 'cuen', |
431
|
|
|
|
|
|
|
zhuo => 'cuo', |
432
|
|
|
|
|
|
|
zi => 'ce', |
433
|
|
|
|
|
|
|
zong => 'cung', |
434
|
|
|
|
|
|
|
zou => ['cou', 'ceu', 'cew'], #cow |
435
|
|
|
|
|
|
|
zu => 'cu', |
436
|
|
|
|
|
|
|
zuan => 'cuan', |
437
|
|
|
|
|
|
|
zui => 'cuei', |
438
|
|
|
|
|
|
|
zun => 'cuen', |
439
|
|
|
|
|
|
|
zuo => 'cuo', |
440
|
|
|
|
|
|
|
); |
441
|
|
|
|
|
|
|
|
442
|
|
|
|
|
|
|
my %jy2id = ( |
443
|
|
|
|
|
|
|
aa => 'a', |
444
|
|
|
|
|
|
|
aai => 'ai', |
445
|
|
|
|
|
|
|
aak => ['ak', 'ngak'], |
446
|
|
|
|
|
|
|
aam => 'am', |
447
|
|
|
|
|
|
|
aan => ['an', 'ngan'], |
448
|
|
|
|
|
|
|
aang => ['ang', 'ngang'], |
449
|
|
|
|
|
|
|
aap => ['ap', 'ngap'], |
450
|
|
|
|
|
|
|
aat => ['at', 'ngat'], |
451
|
|
|
|
|
|
|
aau => ['au', 'ngau'], |
452
|
|
|
|
|
|
|
ai => ['ai', 'ngai'], |
453
|
|
|
|
|
|
|
ak => ['ak', 'ngak'], |
454
|
|
|
|
|
|
|
am => 'am', |
455
|
|
|
|
|
|
|
ang => ['ang', 'ngang'], |
456
|
|
|
|
|
|
|
ap => 'ap', |
457
|
|
|
|
|
|
|
au => 'au', |
458
|
|
|
|
|
|
|
|
459
|
|
|
|
|
|
|
baa => 'pa', |
460
|
|
|
|
|
|
|
baai => 'pai', |
461
|
|
|
|
|
|
|
baak => 'pak', |
462
|
|
|
|
|
|
|
baan => 'pan', |
463
|
|
|
|
|
|
|
baang => 'pang', |
464
|
|
|
|
|
|
|
baat => 'pat', |
465
|
|
|
|
|
|
|
baau => 'pau', |
466
|
|
|
|
|
|
|
bai => 'pai', |
467
|
|
|
|
|
|
|
bak => 'pak', |
468
|
|
|
|
|
|
|
bam => 'pam', |
469
|
|
|
|
|
|
|
ban => 'pan', |
470
|
|
|
|
|
|
|
bang => 'pang', |
471
|
|
|
|
|
|
|
bat => 'pat', |
472
|
|
|
|
|
|
|
bau => 'pau', |
473
|
|
|
|
|
|
|
be => 'pe', |
474
|
|
|
|
|
|
|
bei => 'pei', |
475
|
|
|
|
|
|
|
bek => 'pek', |
476
|
|
|
|
|
|
|
beng => 'peng', |
477
|
|
|
|
|
|
|
bik => 'pik', |
478
|
|
|
|
|
|
|
bin => 'pin', |
479
|
|
|
|
|
|
|
bing => 'ping', |
480
|
|
|
|
|
|
|
bit => 'pit', |
481
|
|
|
|
|
|
|
biu => 'piu', |
482
|
|
|
|
|
|
|
bo => 'po', |
483
|
|
|
|
|
|
|
bok => 'pok', |
484
|
|
|
|
|
|
|
bong => 'pong', |
485
|
|
|
|
|
|
|
bou => 'pou', # pow |
486
|
|
|
|
|
|
|
bui => 'pui', |
487
|
|
|
|
|
|
|
buk => 'puk', |
488
|
|
|
|
|
|
|
bun => 'pun', |
489
|
|
|
|
|
|
|
bung => 'pung', |
490
|
|
|
|
|
|
|
but => 'put', |
491
|
|
|
|
|
|
|
|
492
|
|
|
|
|
|
|
caa => 'cha', |
493
|
|
|
|
|
|
|
caai => 'chai', |
494
|
|
|
|
|
|
|
caak => 'chak', |
495
|
|
|
|
|
|
|
caam => 'cham', |
496
|
|
|
|
|
|
|
caan => 'chan', |
497
|
|
|
|
|
|
|
caang => 'chang', |
498
|
|
|
|
|
|
|
caap => 'chap', |
499
|
|
|
|
|
|
|
caat => 'chat', |
500
|
|
|
|
|
|
|
caau => 'chau', |
501
|
|
|
|
|
|
|
cai => 'chai', |
502
|
|
|
|
|
|
|
cak => 'chak', |
503
|
|
|
|
|
|
|
cam => 'cham', |
504
|
|
|
|
|
|
|
can => 'chan', |
505
|
|
|
|
|
|
|
cang => 'chang', |
506
|
|
|
|
|
|
|
cap => 'chap', |
507
|
|
|
|
|
|
|
cat => 'chat', |
508
|
|
|
|
|
|
|
cau => 'chau', |
509
|
|
|
|
|
|
|
ce => 'che', |
510
|
|
|
|
|
|
|
cek => 'chek', |
511
|
|
|
|
|
|
|
ceng => 'cheng', |
512
|
|
|
|
|
|
|
ceoi => ['cheui', 'cheoi' , |
513
|
|
|
|
|
|
|
'cheuy', 'cheoy' ], |
514
|
|
|
|
|
|
|
ceon => ['cheun', 'cheon' ], |
515
|
|
|
|
|
|
|
ceot => ['cheut', 'cheot' ], |
516
|
|
|
|
|
|
|
ci => 'chi', |
517
|
|
|
|
|
|
|
cik => 'chik', |
518
|
|
|
|
|
|
|
cim => 'chim', |
519
|
|
|
|
|
|
|
cin => 'chin', |
520
|
|
|
|
|
|
|
cing => 'ching', |
521
|
|
|
|
|
|
|
cip => 'chip', |
522
|
|
|
|
|
|
|
cit => 'chit', |
523
|
|
|
|
|
|
|
ciu => 'chiu', |
524
|
|
|
|
|
|
|
co => 'cho', |
525
|
|
|
|
|
|
|
coek => ['cheuk', 'choek'], |
526
|
|
|
|
|
|
|
coeng => ['cheung', 'choeng'], |
527
|
|
|
|
|
|
|
coi => ['choi', 'choy'], |
528
|
|
|
|
|
|
|
cok => 'chok', |
529
|
|
|
|
|
|
|
cong => 'chong', |
530
|
|
|
|
|
|
|
cou => ['chou', 'chow'], |
531
|
|
|
|
|
|
|
cuk => 'chuk', |
532
|
|
|
|
|
|
|
cung => 'chung', |
533
|
|
|
|
|
|
|
cyu => ['chyu', 'chiu'], |
534
|
|
|
|
|
|
|
cyun => ['chyun', 'chiun'], |
535
|
|
|
|
|
|
|
cyut => ['chyut', 'chiut'], |
536
|
|
|
|
|
|
|
|
537
|
|
|
|
|
|
|
daa => 'ta', |
538
|
|
|
|
|
|
|
daai => 'tai', |
539
|
|
|
|
|
|
|
daak => 'tak', |
540
|
|
|
|
|
|
|
daam => 'tam', |
541
|
|
|
|
|
|
|
daan => 'tan', |
542
|
|
|
|
|
|
|
daap => 'tap', |
543
|
|
|
|
|
|
|
daat => 'tat', |
544
|
|
|
|
|
|
|
dai => 'tai', |
545
|
|
|
|
|
|
|
dak => 'tak', |
546
|
|
|
|
|
|
|
dam => 'tam', |
547
|
|
|
|
|
|
|
dan => 'tan', |
548
|
|
|
|
|
|
|
dang => 'tang', |
549
|
|
|
|
|
|
|
dap => 'tap', |
550
|
|
|
|
|
|
|
dat => 'tat', |
551
|
|
|
|
|
|
|
dau => 'tau', |
552
|
|
|
|
|
|
|
de => 'te', |
553
|
|
|
|
|
|
|
dei => 'tei', |
554
|
|
|
|
|
|
|
dek => 'tek', |
555
|
|
|
|
|
|
|
deng => 'teng', |
556
|
|
|
|
|
|
|
deoi => ['teui', 'teoi', |
557
|
|
|
|
|
|
|
'teuy', 'teoy'], |
558
|
|
|
|
|
|
|
deon => ['teun', 'teon'], |
559
|
|
|
|
|
|
|
deot => ['teut', 'teot'], |
560
|
|
|
|
|
|
|
deu => ['teu','tew'], |
561
|
|
|
|
|
|
|
dik => 'tik', |
562
|
|
|
|
|
|
|
dim => 'tim', |
563
|
|
|
|
|
|
|
din => 'tin', |
564
|
|
|
|
|
|
|
ding => 'ting', |
565
|
|
|
|
|
|
|
dip => 'tip', |
566
|
|
|
|
|
|
|
dit => 'tit', |
567
|
|
|
|
|
|
|
diu => 'tiu', |
568
|
|
|
|
|
|
|
do => 'to', |
569
|
|
|
|
|
|
|
doe => ['teu','toe'], |
570
|
|
|
|
|
|
|
doek => ['teuk', 'toek'], |
571
|
|
|
|
|
|
|
doi => ['toi', 'toy'], |
572
|
|
|
|
|
|
|
dok => 'tok', |
573
|
|
|
|
|
|
|
dong => 'tong', |
574
|
|
|
|
|
|
|
dou => ['tou', 'tow'], |
575
|
|
|
|
|
|
|
duk => 'tuk', |
576
|
|
|
|
|
|
|
dung => 'tung', |
577
|
|
|
|
|
|
|
dyun => ['tyun', 'tiun'], |
578
|
|
|
|
|
|
|
dyut => ['tyut', 'tiut'], |
579
|
|
|
|
|
|
|
|
580
|
|
|
|
|
|
|
e => 'e', |
581
|
|
|
|
|
|
|
ei => 'ei', |
582
|
|
|
|
|
|
|
|
583
|
|
|
|
|
|
|
faa => 'fa', |
584
|
|
|
|
|
|
|
faai => 'fai', |
585
|
|
|
|
|
|
|
faak => 'fak', |
586
|
|
|
|
|
|
|
faan => 'fan', |
587
|
|
|
|
|
|
|
faat => 'fat', |
588
|
|
|
|
|
|
|
fai => 'fai', |
589
|
|
|
|
|
|
|
fan => 'fan', |
590
|
|
|
|
|
|
|
fang => 'fang', |
591
|
|
|
|
|
|
|
fat => 'fat', |
592
|
|
|
|
|
|
|
fau => 'fau', |
593
|
|
|
|
|
|
|
fe => 'fe', |
594
|
|
|
|
|
|
|
fei => 'fei', |
595
|
|
|
|
|
|
|
fo => 'fo', |
596
|
|
|
|
|
|
|
fok => 'fok', |
597
|
|
|
|
|
|
|
fong => 'fong', |
598
|
|
|
|
|
|
|
fu => 'fu', |
599
|
|
|
|
|
|
|
fui => 'fui', |
600
|
|
|
|
|
|
|
fuk => 'fuk', |
601
|
|
|
|
|
|
|
fun => 'fun', |
602
|
|
|
|
|
|
|
fung => 'fung', |
603
|
|
|
|
|
|
|
fut => 'fut', |
604
|
|
|
|
|
|
|
|
605
|
|
|
|
|
|
|
# i think g is also a common indo transliteration for |
606
|
|
|
|
|
|
|
# non-mandarin, for instance in names, e.g. kwik kian gee. but |
607
|
|
|
|
|
|
|
# i'll default to k. |
608
|
|
|
|
|
|
|
gaa => ['ka', 'ga'], |
609
|
|
|
|
|
|
|
gaai => ['kai', 'gai'], |
610
|
|
|
|
|
|
|
gaak => ['kak', 'gak'], |
611
|
|
|
|
|
|
|
gaam => ['kam', 'gam'], |
612
|
|
|
|
|
|
|
gaan => ['kan', 'gan'], |
613
|
|
|
|
|
|
|
gaang => ['kang', 'gang'], |
614
|
|
|
|
|
|
|
gaap => ['kap', 'gap'], |
615
|
|
|
|
|
|
|
gaat => ['kat', 'gat'], |
616
|
|
|
|
|
|
|
gaau => ['kau', 'gau'], |
617
|
|
|
|
|
|
|
gai => ['kai', 'gai'], |
618
|
|
|
|
|
|
|
gam => ['kam','gam'], |
619
|
|
|
|
|
|
|
gan => ['kan','gan'], |
620
|
|
|
|
|
|
|
gang => ['kang','gang'], |
621
|
|
|
|
|
|
|
gap => ['kap','gap'], |
622
|
|
|
|
|
|
|
gat => ['kat','gat'], |
623
|
|
|
|
|
|
|
gau => ['kau','gau'], |
624
|
|
|
|
|
|
|
ge => ['ke','ge'], |
625
|
|
|
|
|
|
|
gei => ['kei','gei'], |
626
|
|
|
|
|
|
|
geng => ['keng','geng'], |
627
|
|
|
|
|
|
|
geoi => ['keui', 'geui', 'keoi', 'geoi', |
628
|
|
|
|
|
|
|
'keuy', 'geuy', 'keoy', 'geoy'], |
629
|
|
|
|
|
|
|
gep => ['kep','gep'], |
630
|
|
|
|
|
|
|
gik => ['kik','gik'], |
631
|
|
|
|
|
|
|
gim => ['kim','gim'], |
632
|
|
|
|
|
|
|
gin => ['kin','gin'], |
633
|
|
|
|
|
|
|
ging => ['king','ging'], |
634
|
|
|
|
|
|
|
gip => ['kip','gip'], |
635
|
|
|
|
|
|
|
git => ['kit','git'], |
636
|
|
|
|
|
|
|
giu => ['kiu','giu'], |
637
|
|
|
|
|
|
|
go => ['ko','go'], |
638
|
|
|
|
|
|
|
goe => ['keu','geu','koe','goe'], |
639
|
|
|
|
|
|
|
goek => ['keuk','geuk','koek','goek'], |
640
|
|
|
|
|
|
|
goeng => ['keung','geung','koeng','goeng'], |
641
|
|
|
|
|
|
|
goi => ['koi','goi', |
642
|
|
|
|
|
|
|
'koy','goy'], |
643
|
|
|
|
|
|
|
gok => ['kok','gok'], |
644
|
|
|
|
|
|
|
gon => ['kon','gon'], |
645
|
|
|
|
|
|
|
gong => ['kong','gong'], |
646
|
|
|
|
|
|
|
got => ['kot','got'], |
647
|
|
|
|
|
|
|
gou => ['kou', 'kow', 'gou', 'gow'], |
648
|
|
|
|
|
|
|
gu => ['ku','gu'], |
649
|
|
|
|
|
|
|
gui => ['kui','gui'], |
650
|
|
|
|
|
|
|
guk => ['kuk','guk'], |
651
|
|
|
|
|
|
|
gun => ['kun','gun'], |
652
|
|
|
|
|
|
|
gung => ['kung','gung'], |
653
|
|
|
|
|
|
|
gwaa => ['kwa','gwa', 'kua','gua'], |
654
|
|
|
|
|
|
|
gwaai => ['kwai','gwai','kuai','guai'], |
655
|
|
|
|
|
|
|
gwaak => ['kwak','gwak'], |
656
|
|
|
|
|
|
|
gwaan => ['kwan','gwan'], |
657
|
|
|
|
|
|
|
gwaang => ['kwang','gwang'], |
658
|
|
|
|
|
|
|
gwaat => ['kwat','gwat'], |
659
|
|
|
|
|
|
|
gwai => ['kwai','gwai'], |
660
|
|
|
|
|
|
|
gwan => ['kwan','gwan'], |
661
|
|
|
|
|
|
|
gwang => ['kwang','gwang'], |
662
|
|
|
|
|
|
|
gwat => ['kwat','gwat'], |
663
|
|
|
|
|
|
|
gwik => ['kwik','gwik'], |
664
|
|
|
|
|
|
|
gwing => ['kwing','gwing'], |
665
|
|
|
|
|
|
|
gwo => ['kwo','gwo'], |
666
|
|
|
|
|
|
|
gwok => ['kwok','gwok'], |
667
|
|
|
|
|
|
|
gwong => ['kwong','gwong'], |
668
|
|
|
|
|
|
|
gyun => ['kyun','gyun','kiun','giun'], |
669
|
|
|
|
|
|
|
gyut => ['kyut','gyut','kiut','kyut'], |
670
|
|
|
|
|
|
|
|
671
|
|
|
|
|
|
|
haa => 'ha', |
672
|
|
|
|
|
|
|
haai => 'hai', |
673
|
|
|
|
|
|
|
haak => 'hak', |
674
|
|
|
|
|
|
|
haam => 'ham', |
675
|
|
|
|
|
|
|
haan => 'han', |
676
|
|
|
|
|
|
|
haang => 'hang', |
677
|
|
|
|
|
|
|
haap => 'hap', |
678
|
|
|
|
|
|
|
haau => 'hau', |
679
|
|
|
|
|
|
|
hai => 'hai', |
680
|
|
|
|
|
|
|
hak => 'hak', |
681
|
|
|
|
|
|
|
ham => 'ham', |
682
|
|
|
|
|
|
|
han => 'han', |
683
|
|
|
|
|
|
|
hang => 'hang', |
684
|
|
|
|
|
|
|
hap => 'hap', |
685
|
|
|
|
|
|
|
hat => 'hat', |
686
|
|
|
|
|
|
|
hau => 'hau', |
687
|
|
|
|
|
|
|
hei => 'hei', |
688
|
|
|
|
|
|
|
hek => 'hek', |
689
|
|
|
|
|
|
|
heng => 'heng', |
690
|
|
|
|
|
|
|
heoi => ['heoi', |
691
|
|
|
|
|
|
|
'heoy'], |
692
|
|
|
|
|
|
|
hik => 'hik', |
693
|
|
|
|
|
|
|
him => 'him', |
694
|
|
|
|
|
|
|
hin => 'hin', |
695
|
|
|
|
|
|
|
hing => 'hing', |
696
|
|
|
|
|
|
|
hip => 'hip', |
697
|
|
|
|
|
|
|
hit => 'hit', |
698
|
|
|
|
|
|
|
hiu => 'hiu', |
699
|
|
|
|
|
|
|
hm => 'hm', |
700
|
|
|
|
|
|
|
hng => 'hng', |
701
|
|
|
|
|
|
|
ho => 'ho', |
702
|
|
|
|
|
|
|
hoe => ['heu','hoe'], |
703
|
|
|
|
|
|
|
hoeng => ['heung','hoeng'], |
704
|
|
|
|
|
|
|
hoi => ['hoi','hoy'], |
705
|
|
|
|
|
|
|
hok => 'hok', |
706
|
|
|
|
|
|
|
hon => 'hon', |
707
|
|
|
|
|
|
|
hong => 'hong', |
708
|
|
|
|
|
|
|
hot => 'hot', |
709
|
|
|
|
|
|
|
hou => ['hou','how'], |
710
|
|
|
|
|
|
|
huk => 'huk', |
711
|
|
|
|
|
|
|
hung => 'hung', |
712
|
|
|
|
|
|
|
hyun => ['hiun','hyun'], |
713
|
|
|
|
|
|
|
hyut => ['hyut','hiut'], |
714
|
|
|
|
|
|
|
|
715
|
|
|
|
|
|
|
jaa => 'ya', |
716
|
|
|
|
|
|
|
jaai => 'yai', |
717
|
|
|
|
|
|
|
jaak => 'yak', |
718
|
|
|
|
|
|
|
jai => 'yai', |
719
|
|
|
|
|
|
|
jam => 'yam', |
720
|
|
|
|
|
|
|
jan => 'yan', |
721
|
|
|
|
|
|
|
jap => 'yap', |
722
|
|
|
|
|
|
|
jat => 'yat', |
723
|
|
|
|
|
|
|
jau => 'yau', |
724
|
|
|
|
|
|
|
je => 'ye', |
725
|
|
|
|
|
|
|
jeng => 'yeng', |
726
|
|
|
|
|
|
|
jeoi => ['yeui','yeoi', |
727
|
|
|
|
|
|
|
'yeuy','yeoy'], |
728
|
|
|
|
|
|
|
jeon => ['yeun','yeon'], |
729
|
|
|
|
|
|
|
ji => 'yi', |
730
|
|
|
|
|
|
|
jik => 'yik', |
731
|
|
|
|
|
|
|
jim => 'yim', |
732
|
|
|
|
|
|
|
jin => 'yin', |
733
|
|
|
|
|
|
|
jing => 'ying', |
734
|
|
|
|
|
|
|
jip => 'yip', |
735
|
|
|
|
|
|
|
jit => 'yit', |
736
|
|
|
|
|
|
|
jiu => 'yiu', |
737
|
|
|
|
|
|
|
jo => 'yo', |
738
|
|
|
|
|
|
|
joek => ['yeuk','yoek'], |
739
|
|
|
|
|
|
|
joeng => ['yeung','yoeng'], |
740
|
|
|
|
|
|
|
juk => 'yuk', |
741
|
|
|
|
|
|
|
jung => 'yung', |
742
|
|
|
|
|
|
|
jyu => ['yiu', 'yu'], |
743
|
|
|
|
|
|
|
jyun => ['yiun', 'yun'], |
744
|
|
|
|
|
|
|
jyut => ['yiut', 'yut'], |
745
|
|
|
|
|
|
|
|
746
|
|
|
|
|
|
|
kaa => 'kha', |
747
|
|
|
|
|
|
|
kaai => 'khai', |
748
|
|
|
|
|
|
|
kaak => 'khak', |
749
|
|
|
|
|
|
|
kaat => 'khat', |
750
|
|
|
|
|
|
|
kaau => 'khau', |
751
|
|
|
|
|
|
|
kai => 'khai', |
752
|
|
|
|
|
|
|
kam => 'kham', |
753
|
|
|
|
|
|
|
kan => 'khan', |
754
|
|
|
|
|
|
|
kang => 'khang', |
755
|
|
|
|
|
|
|
kap => 'khap', |
756
|
|
|
|
|
|
|
kat => 'khat', |
757
|
|
|
|
|
|
|
kau => 'khau', |
758
|
|
|
|
|
|
|
ke => 'khe', |
759
|
|
|
|
|
|
|
kei => 'khei', |
760
|
|
|
|
|
|
|
kek => 'khek', |
761
|
|
|
|
|
|
|
keoi => ['kheui', 'kheoi', |
762
|
|
|
|
|
|
|
'kheuy', 'kheoy'], |
763
|
|
|
|
|
|
|
kik => 'khik', |
764
|
|
|
|
|
|
|
kim => 'khim', |
765
|
|
|
|
|
|
|
kin => 'khin', |
766
|
|
|
|
|
|
|
king => 'khing', |
767
|
|
|
|
|
|
|
kit => 'khit', |
768
|
|
|
|
|
|
|
kiu => 'khiu', |
769
|
|
|
|
|
|
|
ko => 'kho', |
770
|
|
|
|
|
|
|
koe => ['kheu','khoe'], |
771
|
|
|
|
|
|
|
koek => ['kheuk','khoek'], |
772
|
|
|
|
|
|
|
koeng => ['kheung','khoeng'], |
773
|
|
|
|
|
|
|
koi => ['khoi','khoy'], |
774
|
|
|
|
|
|
|
kok => 'khok', |
775
|
|
|
|
|
|
|
kong => 'khong', |
776
|
|
|
|
|
|
|
ku => 'khu', |
777
|
|
|
|
|
|
|
kui => 'khui', |
778
|
|
|
|
|
|
|
kuk => 'khuk', |
779
|
|
|
|
|
|
|
kung => 'khung', |
780
|
|
|
|
|
|
|
kut => 'khut', |
781
|
|
|
|
|
|
|
kwaa => 'khwa', |
782
|
|
|
|
|
|
|
kwaai => 'khwai', |
783
|
|
|
|
|
|
|
kwaang => 'khwang', |
784
|
|
|
|
|
|
|
kwai => 'khwai', |
785
|
|
|
|
|
|
|
kwan => 'khwan', |
786
|
|
|
|
|
|
|
kwik => 'khwik', |
787
|
|
|
|
|
|
|
kwok => 'khwok', |
788
|
|
|
|
|
|
|
kwong => 'khwong', |
789
|
|
|
|
|
|
|
kyun => ['khiun','khyun'], |
790
|
|
|
|
|
|
|
kyut => ['khiut','khyut'], |
791
|
|
|
|
|
|
|
|
792
|
|
|
|
|
|
|
laa => 'la', |
793
|
|
|
|
|
|
|
laai => 'lai', |
794
|
|
|
|
|
|
|
laak => 'lak', |
795
|
|
|
|
|
|
|
laam => 'lam', |
796
|
|
|
|
|
|
|
laan => 'lan', |
797
|
|
|
|
|
|
|
laang => 'lang', |
798
|
|
|
|
|
|
|
laap => 'lap', |
799
|
|
|
|
|
|
|
laat => 'lat', |
800
|
|
|
|
|
|
|
laau => 'lau', |
801
|
|
|
|
|
|
|
lai => 'lai', |
802
|
|
|
|
|
|
|
lak => 'lak', |
803
|
|
|
|
|
|
|
lam => 'lam', |
804
|
|
|
|
|
|
|
lang => 'lang', |
805
|
|
|
|
|
|
|
lap => 'lap', |
806
|
|
|
|
|
|
|
lat => 'lat', |
807
|
|
|
|
|
|
|
lau => 'lau', |
808
|
|
|
|
|
|
|
le => 'le', |
809
|
|
|
|
|
|
|
lei => 'lei', |
810
|
|
|
|
|
|
|
lek => 'lek', |
811
|
|
|
|
|
|
|
lem => 'lem', |
812
|
|
|
|
|
|
|
leng => 'leng', |
813
|
|
|
|
|
|
|
leoi => ['leui','leoi', |
814
|
|
|
|
|
|
|
'leuy','leoy'], |
815
|
|
|
|
|
|
|
leon => ['leun','leon'], |
816
|
|
|
|
|
|
|
leot => ['leut','leot'], |
817
|
|
|
|
|
|
|
li => 'li', |
818
|
|
|
|
|
|
|
lik => 'lik', |
819
|
|
|
|
|
|
|
lim => 'lim', |
820
|
|
|
|
|
|
|
lin => 'lin', |
821
|
|
|
|
|
|
|
ling => 'ling', |
822
|
|
|
|
|
|
|
lip => 'lip', |
823
|
|
|
|
|
|
|
lit => 'lit', |
824
|
|
|
|
|
|
|
liu => 'liu', |
825
|
|
|
|
|
|
|
lo => 'lo', |
826
|
|
|
|
|
|
|
loek => ['leuk','loek'], |
827
|
|
|
|
|
|
|
loeng => ['leung','loeng'], |
828
|
|
|
|
|
|
|
loi => ['loi','loy'], |
829
|
|
|
|
|
|
|
lok => 'lok', |
830
|
|
|
|
|
|
|
long => 'long', |
831
|
|
|
|
|
|
|
lou => ['lou','low'], |
832
|
|
|
|
|
|
|
luk => 'luk', |
833
|
|
|
|
|
|
|
lung => 'lung', |
834
|
|
|
|
|
|
|
lyun => ['liun','lyun'], |
835
|
|
|
|
|
|
|
lyut => ['liut','lyut'], |
836
|
|
|
|
|
|
|
|
837
|
|
|
|
|
|
|
m => 'm', |
838
|
|
|
|
|
|
|
maa => 'ma', |
839
|
|
|
|
|
|
|
maai => 'mai', |
840
|
|
|
|
|
|
|
maak => 'mak', |
841
|
|
|
|
|
|
|
maan => 'man', |
842
|
|
|
|
|
|
|
maang => 'mang', |
843
|
|
|
|
|
|
|
maat => 'mat', |
844
|
|
|
|
|
|
|
maau => 'mau', |
845
|
|
|
|
|
|
|
mai => 'mai', |
846
|
|
|
|
|
|
|
mak => 'mak', |
847
|
|
|
|
|
|
|
man => 'man', |
848
|
|
|
|
|
|
|
mang => 'mang', |
849
|
|
|
|
|
|
|
mat => 'mat', |
850
|
|
|
|
|
|
|
mau => 'mau', |
851
|
|
|
|
|
|
|
me => 'me', |
852
|
|
|
|
|
|
|
mei => 'mei', |
853
|
|
|
|
|
|
|
meng => 'meng', |
854
|
|
|
|
|
|
|
mi => 'mi', |
855
|
|
|
|
|
|
|
mik => 'mik', |
856
|
|
|
|
|
|
|
min => 'min', |
857
|
|
|
|
|
|
|
ming => 'ming', |
858
|
|
|
|
|
|
|
mit => 'mit', |
859
|
|
|
|
|
|
|
miu => 'miu', |
860
|
|
|
|
|
|
|
mo => 'mo', |
861
|
|
|
|
|
|
|
mok => 'mok', |
862
|
|
|
|
|
|
|
mong => 'mong', |
863
|
|
|
|
|
|
|
mou => ['mou','mow'], |
864
|
|
|
|
|
|
|
mui => 'mui', |
865
|
|
|
|
|
|
|
muk => 'muk', |
866
|
|
|
|
|
|
|
mun => 'mun', |
867
|
|
|
|
|
|
|
mung => 'mung', |
868
|
|
|
|
|
|
|
mut => 'mut', |
869
|
|
|
|
|
|
|
|
870
|
|
|
|
|
|
|
naa => 'na', |
871
|
|
|
|
|
|
|
naai => 'nai', |
872
|
|
|
|
|
|
|
naam => 'nam', |
873
|
|
|
|
|
|
|
naan => 'nan', |
874
|
|
|
|
|
|
|
naap => 'nap', |
875
|
|
|
|
|
|
|
naat => 'nat', |
876
|
|
|
|
|
|
|
naau => 'nau', |
877
|
|
|
|
|
|
|
nai => 'nai', |
878
|
|
|
|
|
|
|
nam => 'nam', |
879
|
|
|
|
|
|
|
nan => 'nan', |
880
|
|
|
|
|
|
|
nang => 'nang', |
881
|
|
|
|
|
|
|
nap => 'nap', |
882
|
|
|
|
|
|
|
nat => 'nat', |
883
|
|
|
|
|
|
|
nau => 'nau', |
884
|
|
|
|
|
|
|
ne => 'ne', |
885
|
|
|
|
|
|
|
nei => 'nei', |
886
|
|
|
|
|
|
|
neoi => ['neui','neoi', |
887
|
|
|
|
|
|
|
'neuy','neoy'], |
888
|
|
|
|
|
|
|
neot => ['neut','neot'], |
889
|
|
|
|
|
|
|
ng => 'ng', |
890
|
|
|
|
|
|
|
|
891
|
|
|
|
|
|
|
ngaa => 'nga', |
892
|
|
|
|
|
|
|
ngaai => 'ngai', |
893
|
|
|
|
|
|
|
ngaak => 'ngak', |
894
|
|
|
|
|
|
|
ngaam => 'ngam', |
895
|
|
|
|
|
|
|
ngaan => 'ngan', |
896
|
|
|
|
|
|
|
ngaang => 'ngang', |
897
|
|
|
|
|
|
|
ngaap => 'ngap', |
898
|
|
|
|
|
|
|
ngaat => 'ngat', |
899
|
|
|
|
|
|
|
ngaau => 'ngau', |
900
|
|
|
|
|
|
|
ngai => 'ngai', |
901
|
|
|
|
|
|
|
ngak => 'ngak', |
902
|
|
|
|
|
|
|
ngam => 'ngam', |
903
|
|
|
|
|
|
|
ngan => 'ngan', |
904
|
|
|
|
|
|
|
ngang => 'ngang', |
905
|
|
|
|
|
|
|
ngap => 'ngap', |
906
|
|
|
|
|
|
|
ngat => 'ngat', |
907
|
|
|
|
|
|
|
ngau => 'ngau', |
908
|
|
|
|
|
|
|
ngit => 'ngit', |
909
|
|
|
|
|
|
|
ngo => 'ngo', |
910
|
|
|
|
|
|
|
ngoi => ['ngoi','ngoy'], |
911
|
|
|
|
|
|
|
ngok => 'ngok', |
912
|
|
|
|
|
|
|
ngon => 'ngon', |
913
|
|
|
|
|
|
|
ngong => 'ngong', |
914
|
|
|
|
|
|
|
ngou => ['ngou','ngow'], |
915
|
|
|
|
|
|
|
nguk => 'nguk', |
916
|
|
|
|
|
|
|
ngung => 'ngung', |
917
|
|
|
|
|
|
|
|
918
|
|
|
|
|
|
|
ni => 'ni', |
919
|
|
|
|
|
|
|
nik => 'nik', |
920
|
|
|
|
|
|
|
nim => 'nim', |
921
|
|
|
|
|
|
|
nin => 'nin', |
922
|
|
|
|
|
|
|
ning => 'ning', |
923
|
|
|
|
|
|
|
nip => 'nip', |
924
|
|
|
|
|
|
|
niu => 'niu', |
925
|
|
|
|
|
|
|
no => 'no', |
926
|
|
|
|
|
|
|
noeng => ['neung','noeng'], |
927
|
|
|
|
|
|
|
noi => ['noi','noy'], |
928
|
|
|
|
|
|
|
nok => 'nok', |
929
|
|
|
|
|
|
|
nong => 'nong', |
930
|
|
|
|
|
|
|
nou => ['nou','now'], |
931
|
|
|
|
|
|
|
nuk => 'nuk', |
932
|
|
|
|
|
|
|
nung => 'nung', |
933
|
|
|
|
|
|
|
nyun => ['niun','nyun'], |
934
|
|
|
|
|
|
|
|
935
|
|
|
|
|
|
|
o => 'o', |
936
|
|
|
|
|
|
|
oi => ['oi','oy'], |
937
|
|
|
|
|
|
|
ok => 'ok', |
938
|
|
|
|
|
|
|
on => 'on', |
939
|
|
|
|
|
|
|
ong => 'ong', |
940
|
|
|
|
|
|
|
ou => ['ou','ow'], |
941
|
|
|
|
|
|
|
|
942
|
|
|
|
|
|
|
paa => 'pha', |
943
|
|
|
|
|
|
|
paai => 'phai', |
944
|
|
|
|
|
|
|
paak => 'phak', |
945
|
|
|
|
|
|
|
paan => 'phan', |
946
|
|
|
|
|
|
|
paang => 'phang', |
947
|
|
|
|
|
|
|
paau => 'phau', |
948
|
|
|
|
|
|
|
pai => 'phai', |
949
|
|
|
|
|
|
|
pan => 'phan', |
950
|
|
|
|
|
|
|
pang => 'phang', |
951
|
|
|
|
|
|
|
pat => 'phat', |
952
|
|
|
|
|
|
|
pau => 'phau', |
953
|
|
|
|
|
|
|
pei => 'phei', |
954
|
|
|
|
|
|
|
pek => 'phek', |
955
|
|
|
|
|
|
|
peng => 'pheng', |
956
|
|
|
|
|
|
|
pik => 'phik', |
957
|
|
|
|
|
|
|
pin => 'phin', |
958
|
|
|
|
|
|
|
ping => 'phing', |
959
|
|
|
|
|
|
|
pit => 'phit', |
960
|
|
|
|
|
|
|
piu => 'phiu', |
961
|
|
|
|
|
|
|
po => 'pho', |
962
|
|
|
|
|
|
|
poi => 'phoi', |
963
|
|
|
|
|
|
|
pok => 'phok', |
964
|
|
|
|
|
|
|
pong => 'phong', |
965
|
|
|
|
|
|
|
pou => ['phou','phow'], |
966
|
|
|
|
|
|
|
pui => 'phui', |
967
|
|
|
|
|
|
|
puk => 'phuk', |
968
|
|
|
|
|
|
|
pun => 'phun', |
969
|
|
|
|
|
|
|
pung => 'phung', |
970
|
|
|
|
|
|
|
put => 'phut', |
971
|
|
|
|
|
|
|
|
972
|
|
|
|
|
|
|
saa => 'sa', |
973
|
|
|
|
|
|
|
saai => 'sai', |
974
|
|
|
|
|
|
|
saak => 'sak', |
975
|
|
|
|
|
|
|
saam => 'sam', |
976
|
|
|
|
|
|
|
saan => 'san', |
977
|
|
|
|
|
|
|
saang => 'sang', |
978
|
|
|
|
|
|
|
saap => 'sap', |
979
|
|
|
|
|
|
|
saat => 'sat', |
980
|
|
|
|
|
|
|
saau => 'sau', |
981
|
|
|
|
|
|
|
sai => 'sai', |
982
|
|
|
|
|
|
|
sak => 'sak', |
983
|
|
|
|
|
|
|
sam => 'sam', |
984
|
|
|
|
|
|
|
san => 'san', |
985
|
|
|
|
|
|
|
sang => 'sang', |
986
|
|
|
|
|
|
|
sap => 'sap', |
987
|
|
|
|
|
|
|
sat => 'sat', |
988
|
|
|
|
|
|
|
sau => 'sau', |
989
|
|
|
|
|
|
|
se => 'se', |
990
|
|
|
|
|
|
|
sei => 'sei', |
991
|
|
|
|
|
|
|
sek => 'sek', |
992
|
|
|
|
|
|
|
seng => 'seng', |
993
|
|
|
|
|
|
|
seoi => ['seui','seoi', |
994
|
|
|
|
|
|
|
'seuy','seoy'], |
995
|
|
|
|
|
|
|
seon => ['seun','seon'], |
996
|
|
|
|
|
|
|
seot => ['seut','seot'], |
997
|
|
|
|
|
|
|
si => 'si', |
998
|
|
|
|
|
|
|
sik => 'sik', |
999
|
|
|
|
|
|
|
sim => 'sim', |
1000
|
|
|
|
|
|
|
sin => 'sin', |
1001
|
|
|
|
|
|
|
sing => 'sing', |
1002
|
|
|
|
|
|
|
sip => 'sip', |
1003
|
|
|
|
|
|
|
sit => 'sit', |
1004
|
|
|
|
|
|
|
siu => 'siu', |
1005
|
|
|
|
|
|
|
so => 'so', |
1006
|
|
|
|
|
|
|
soek => ['seuk','soek'], |
1007
|
|
|
|
|
|
|
soeng => ['seung','soeng'], |
1008
|
|
|
|
|
|
|
soi => ['soi','soy'], |
1009
|
|
|
|
|
|
|
sok => 'sok', |
1010
|
|
|
|
|
|
|
song => 'song', |
1011
|
|
|
|
|
|
|
sou => ['sou','sow'], |
1012
|
|
|
|
|
|
|
suk => 'suk', |
1013
|
|
|
|
|
|
|
sung => 'sung', |
1014
|
|
|
|
|
|
|
syu => ['siu','syu'], |
1015
|
|
|
|
|
|
|
syun => ['siun','syun'], |
1016
|
|
|
|
|
|
|
syut => ['siut','syut'], |
1017
|
|
|
|
|
|
|
|
1018
|
|
|
|
|
|
|
taa => 'tha', |
1019
|
|
|
|
|
|
|
taai => 'thai', |
1020
|
|
|
|
|
|
|
taam => 'tham', |
1021
|
|
|
|
|
|
|
taan => 'than', |
1022
|
|
|
|
|
|
|
taap => 'thap', |
1023
|
|
|
|
|
|
|
taat => 'that', |
1024
|
|
|
|
|
|
|
tai => 'thai', |
1025
|
|
|
|
|
|
|
tam => 'tham', |
1026
|
|
|
|
|
|
|
tan => 'than', |
1027
|
|
|
|
|
|
|
tang => 'thang', |
1028
|
|
|
|
|
|
|
tau => 'thau', |
1029
|
|
|
|
|
|
|
tek => 'thek', |
1030
|
|
|
|
|
|
|
teng => 'theng', |
1031
|
|
|
|
|
|
|
teoi => ['theui','theoi', |
1032
|
|
|
|
|
|
|
'theuy','theoy'], |
1033
|
|
|
|
|
|
|
teon => ['theun','theon'], |
1034
|
|
|
|
|
|
|
tik => 'thik', |
1035
|
|
|
|
|
|
|
tim => 'thim', |
1036
|
|
|
|
|
|
|
tin => 'thin', |
1037
|
|
|
|
|
|
|
ting => 'thing', |
1038
|
|
|
|
|
|
|
tip => 'thip', |
1039
|
|
|
|
|
|
|
tit => 'thit', |
1040
|
|
|
|
|
|
|
tiu => 'thiu', |
1041
|
|
|
|
|
|
|
to => 'tho', |
1042
|
|
|
|
|
|
|
toe => ['theu','thoe'], |
1043
|
|
|
|
|
|
|
toi => ['thoi','thoy'], |
1044
|
|
|
|
|
|
|
tok => 'thok', |
1045
|
|
|
|
|
|
|
tong => 'thong', |
1046
|
|
|
|
|
|
|
tou => ['thou', 'thow'], |
1047
|
|
|
|
|
|
|
tuk => 'thuk', |
1048
|
|
|
|
|
|
|
tung => 'thung', |
1049
|
|
|
|
|
|
|
tyun => ['thiun','thyun'], |
1050
|
|
|
|
|
|
|
tyut => ['thiut','thyut'], |
1051
|
|
|
|
|
|
|
|
1052
|
|
|
|
|
|
|
uk => 'uk', |
1053
|
|
|
|
|
|
|
ung => 'ung', |
1054
|
|
|
|
|
|
|
|
1055
|
|
|
|
|
|
|
waa => 'wa', |
1056
|
|
|
|
|
|
|
waai => 'wai', |
1057
|
|
|
|
|
|
|
waak => 'wak', |
1058
|
|
|
|
|
|
|
waan => 'wan', |
1059
|
|
|
|
|
|
|
waang => 'wang', |
1060
|
|
|
|
|
|
|
waat => 'wat', |
1061
|
|
|
|
|
|
|
wai => 'wai', |
1062
|
|
|
|
|
|
|
wan => 'wan', |
1063
|
|
|
|
|
|
|
wang => 'wang', |
1064
|
|
|
|
|
|
|
wat => 'wat', |
1065
|
|
|
|
|
|
|
wik => 'wik', |
1066
|
|
|
|
|
|
|
wing => 'wing', |
1067
|
|
|
|
|
|
|
wo => 'wo', |
1068
|
|
|
|
|
|
|
wok => 'wok', |
1069
|
|
|
|
|
|
|
wong => 'wong', |
1070
|
|
|
|
|
|
|
wu => 'wu', |
1071
|
|
|
|
|
|
|
wui => 'wui', |
1072
|
|
|
|
|
|
|
wun => 'wun', |
1073
|
|
|
|
|
|
|
wut => 'wut', |
1074
|
|
|
|
|
|
|
|
1075
|
|
|
|
|
|
|
zaa => 'ca', |
1076
|
|
|
|
|
|
|
zaai => 'cai', |
1077
|
|
|
|
|
|
|
zaak => 'cak', |
1078
|
|
|
|
|
|
|
zaam => 'cam', |
1079
|
|
|
|
|
|
|
zaan => 'can', |
1080
|
|
|
|
|
|
|
zaang => 'cang', |
1081
|
|
|
|
|
|
|
zaap => 'cap', |
1082
|
|
|
|
|
|
|
zaat => 'cat', |
1083
|
|
|
|
|
|
|
zaau => 'cau', |
1084
|
|
|
|
|
|
|
zai => 'cai', |
1085
|
|
|
|
|
|
|
zak => 'cak', |
1086
|
|
|
|
|
|
|
zam => 'cam', |
1087
|
|
|
|
|
|
|
zan => 'can', |
1088
|
|
|
|
|
|
|
zang => 'cang', |
1089
|
|
|
|
|
|
|
zap => 'cap', |
1090
|
|
|
|
|
|
|
zat => 'cat', |
1091
|
|
|
|
|
|
|
zau => 'cau', |
1092
|
|
|
|
|
|
|
ze => 'ce', |
1093
|
|
|
|
|
|
|
zek => 'cek', |
1094
|
|
|
|
|
|
|
zeng => 'ceng', |
1095
|
|
|
|
|
|
|
zeoi => ['ceui','ceoi', |
1096
|
|
|
|
|
|
|
'ceuy','ceoy'], |
1097
|
|
|
|
|
|
|
zeon => ['ceun','ceon'], |
1098
|
|
|
|
|
|
|
zeot => ['ceut','ceot'], |
1099
|
|
|
|
|
|
|
zi => 'ci', |
1100
|
|
|
|
|
|
|
zik => 'cik', |
1101
|
|
|
|
|
|
|
zim => 'cim', |
1102
|
|
|
|
|
|
|
zin => 'cin', |
1103
|
|
|
|
|
|
|
zing => 'cing', |
1104
|
|
|
|
|
|
|
zip => 'cip', |
1105
|
|
|
|
|
|
|
zit => 'cit', |
1106
|
|
|
|
|
|
|
ziu => 'ciu', |
1107
|
|
|
|
|
|
|
zo => 'co', |
1108
|
|
|
|
|
|
|
zoek => ['ceuk','coek'], |
1109
|
|
|
|
|
|
|
zoeng => ['ceung','coeng'], |
1110
|
|
|
|
|
|
|
zoi => ['coi','coy'], |
1111
|
|
|
|
|
|
|
zok => 'cok', |
1112
|
|
|
|
|
|
|
zong => 'cong', |
1113
|
|
|
|
|
|
|
zou => ['cou','cow'], |
1114
|
|
|
|
|
|
|
zuk => 'cuk', |
1115
|
|
|
|
|
|
|
zung => 'cung', |
1116
|
|
|
|
|
|
|
zyu => ['ciu','cyu'], |
1117
|
|
|
|
|
|
|
zyun => ['ciun','cyun'], |
1118
|
|
|
|
|
|
|
zyut => ['ciut','cyut'], |
1119
|
|
|
|
|
|
|
); |
1120
|
|
|
|
|
|
|
|
1121
|
|
|
|
|
|
|
my %id2hy; |
1122
|
|
|
|
|
|
|
my %id2jy; |
1123
|
|
|
|
|
|
|
for (keys %hy2id) { |
1124
|
|
|
|
|
|
|
my $l = ref($hy2id{$_}) eq 'ARRAY' ? $hy2id{$_} : [$hy2id{$_}]; |
1125
|
|
|
|
|
|
|
for my $id (@$l) { |
1126
|
|
|
|
|
|
|
if (exists $id2hy{$id}) { |
1127
|
|
|
|
|
|
|
$id2hy{$id} = [$id2hy{$id}] unless ref($id2hy{$id}) eq 'ARRAY'; |
1128
|
|
|
|
|
|
|
push @{ $id2hy{$id} }, $_; |
1129
|
|
|
|
|
|
|
} else { |
1130
|
|
|
|
|
|
|
$id2hy{$id} = $_; |
1131
|
|
|
|
|
|
|
} |
1132
|
|
|
|
|
|
|
} |
1133
|
|
|
|
|
|
|
} |
1134
|
|
|
|
|
|
|
#use Data::Dumper; print Dumper \%id2hy; |
1135
|
|
|
|
|
|
|
for (keys %jy2id) { |
1136
|
|
|
|
|
|
|
my $l = ref($jy2id{$_}) eq 'ARRAY' ? $jy2id{$_} : [$jy2id{$_}]; |
1137
|
|
|
|
|
|
|
for my $id (@$l) { |
1138
|
|
|
|
|
|
|
if (exists $id2jy{$id}) { |
1139
|
|
|
|
|
|
|
$id2jy{$id} = [$id2jy{$id}] unless ref($id2jy{$id}) eq 'ARRAY'; |
1140
|
|
|
|
|
|
|
push @{ $id2jy{$id} }, $_; |
1141
|
|
|
|
|
|
|
} else { |
1142
|
|
|
|
|
|
|
$id2jy{$id} = $_; |
1143
|
|
|
|
|
|
|
} |
1144
|
|
|
|
|
|
|
} |
1145
|
|
|
|
|
|
|
} |
1146
|
|
|
|
|
|
|
#use Data::Dumper; print Dumper \%id2jy; |
1147
|
|
|
|
|
|
|
|
1148
|
|
|
|
|
|
|
my $hy_re = join("|", sort { length($b) <=> length($a) } keys %hy2id); $hy_re = qr/(?:$hy_re)/; |
1149
|
|
|
|
|
|
|
my $jy_re = join("|", sort { length($b) <=> length($a) } keys %jy2id); $jy_re = qr/(?:$jy_re)/; |
1150
|
|
|
|
|
|
|
my $idmand_re = join("|", sort { length($b) <=> length($a) } keys %id2hy); $idmand_re = qr/(?:$idmand_re)/; |
1151
|
|
|
|
|
|
|
my $idcant_re = join("|", sort { length($b) <=> length($a) } keys %id2jy); $idcant_re = qr/(?:$idcant_re)/; |
1152
|
|
|
|
|
|
|
my %all = (%hy2id, %jy2id, %id2hy, %id2jy); |
1153
|
|
|
|
|
|
|
my $all_re = join("|", sort { length($b) <=> length($a) } keys %all); $all_re = qr/(?:$all_re)/; |
1154
|
|
|
|
|
|
|
|
1155
|
|
|
|
|
|
|
sub new { |
1156
|
1
|
|
|
1
|
1
|
17
|
my ($class, %opts) = @_; |
1157
|
1
|
|
|
|
|
5
|
bless {}, $class; |
1158
|
|
|
|
|
|
|
} |
1159
|
|
|
|
|
|
|
|
1160
|
|
|
|
|
|
|
sub hanyu2id { |
1161
|
5
|
|
|
5
|
1
|
2133
|
my ($self, $text, $opts) = @_; |
1162
|
5
|
50
|
|
|
|
17
|
if (!defined($opts)) { $opts = {} } |
|
5
|
|
|
|
|
11
|
|
1163
|
|
|
|
|
|
|
my $sub1 = sub { |
1164
|
9
|
|
|
9
|
|
17
|
my $t = shift; |
1165
|
9
|
|
|
|
|
1011
|
$t =~ s/($hy_re)([12345]?)/ |
1166
|
13
|
100
|
|
|
|
82
|
(ref($hy2id{$1}) ? $hy2id{$1}[0] : $hy2id{$1}) . |
|
|
50
|
|
|
|
|
|
1167
|
|
|
|
|
|
|
($opts->{remove_tones} ? "" : $2)/eg; |
1168
|
9
|
|
|
|
|
119
|
$t; |
1169
|
5
|
|
|
|
|
24
|
}; |
1170
|
5
|
|
|
|
|
702
|
$text =~ s/\b((?:$hy_re[12345]?)+)\b/$sub1->($1)/eg; |
|
9
|
|
|
|
|
22
|
|
1171
|
5
|
|
|
|
|
86
|
$text; |
1172
|
|
|
|
|
|
|
} |
1173
|
|
|
|
|
|
|
|
1174
|
|
|
|
|
|
|
sub jyutping2id { |
1175
|
3
|
|
|
3
|
1
|
1384
|
my ($self, $text, $opts) = @_; |
1176
|
3
|
50
|
|
|
|
8
|
if (!defined($opts)) { $opts = {} } |
|
0
|
|
|
|
|
0
|
|
1177
|
|
|
|
|
|
|
my $sub1 = sub { |
1178
|
5
|
|
|
5
|
|
10
|
my $t = shift; |
1179
|
5
|
|
|
|
|
1268
|
$t =~ s/($jy_re)([123456]?)/ |
1180
|
6
|
100
|
33
|
|
|
56
|
(ref($jy2id{$1}) ? $jy2id{$1}[0] : $jy2id{$1}) . |
|
|
50
|
|
|
|
|
|
1181
|
|
|
|
|
|
|
(!$opts->{remove_tones} && $2 ? $2 : "")/eg; |
1182
|
5
|
|
|
|
|
144
|
$t; |
1183
|
3
|
|
|
|
|
15
|
}; |
1184
|
3
|
|
|
|
|
626
|
$text =~ s/\b((?:$jy_re[123456]?)+)\b/$sub1->($1)/eg; |
|
5
|
|
|
|
|
10
|
|
1185
|
3
|
|
|
|
|
27
|
$text; |
1186
|
|
|
|
|
|
|
} |
1187
|
|
|
|
|
|
|
|
1188
|
|
|
|
|
|
|
sub id2hanyu { |
1189
|
6
|
|
|
6
|
1
|
2954
|
my ($self, $text, $opts) = @_; |
1190
|
6
|
100
|
|
|
|
19
|
if (!defined($opts)) { $opts = {} } |
|
5
|
|
|
|
|
9
|
|
1191
|
|
|
|
|
|
|
my $sub2 = sub { |
1192
|
9
|
|
|
9
|
|
17
|
my $t = shift; |
1193
|
9
|
100
|
|
|
|
19
|
if ($opts->{list_all}) { |
1194
|
2
|
|
|
|
|
936
|
$t =~ s/($idmand_re)([12345]?)/ |
1195
|
2
|
100
|
33
|
|
|
15
|
(ref($id2hy{$1}) ? "(".join("|", sort(@{ $id2hy{$1} })).")" : $id2hy{$1}) . |
|
1
|
50
|
|
|
|
18
|
|
1196
|
|
|
|
|
|
|
(!$opts->{remove_tones} && $2 ? $2 : "")/eg; |
1197
|
|
|
|
|
|
|
} else { |
1198
|
7
|
|
|
|
|
1511
|
$t =~ s/($idmand_re)([12345]?)/ |
1199
|
9
|
100
|
33
|
|
|
83
|
(ref($id2hy{$1}) ? die("ambiguous") : $id2hy{$1}) . |
|
|
100
|
|
|
|
|
|
1200
|
|
|
|
|
|
|
(!$opts->{remove_tones} && $2 ? $2 : "")/eg; |
1201
|
|
|
|
|
|
|
} |
1202
|
8
|
|
|
|
|
121
|
$t; |
1203
|
6
|
|
|
|
|
27
|
}; |
1204
|
6
|
|
|
|
|
7
|
eval { |
1205
|
6
|
|
|
|
|
604
|
$text =~ s/\b((?:$idmand_re[12345]?)+)\b/$sub2->($1)/eg; |
|
9
|
|
|
|
|
20
|
|
1206
|
|
|
|
|
|
|
}; |
1207
|
6
|
100
|
|
|
|
155
|
$@ ? undef : $text; |
1208
|
|
|
|
|
|
|
} |
1209
|
|
|
|
|
|
|
|
1210
|
|
|
|
|
|
|
sub id2jyutping { |
1211
|
1
|
|
|
1
|
1
|
464
|
my ($self, $text, $opts) = @_; |
1212
|
1
|
50
|
|
|
|
5
|
if (!defined($opts)) { $opts = {} } |
|
0
|
|
|
|
|
0
|
|
1213
|
|
|
|
|
|
|
my $sub2 = sub { |
1214
|
2
|
|
|
2
|
|
5
|
my $t = shift; |
1215
|
2
|
50
|
|
|
|
10
|
if ($opts->{list_all}) { |
1216
|
0
|
|
|
|
|
0
|
$t =~ s/($idcant_re)([123456]?)/ |
1217
|
0
|
0
|
|
|
|
0
|
(ref($id2jy{$1}) ? "(".join("|", sort(@{ $id2jy{$1} })).")" : $id2jy{$1}) . |
|
0
|
0
|
|
|
|
0
|
|
1218
|
|
|
|
|
|
|
($opts->{remove_tones} ? "" : $2)/eg; |
1219
|
|
|
|
|
|
|
} else { |
1220
|
2
|
|
|
|
|
1545
|
$t =~ s/($idcant_re)([123456]?)/ |
1221
|
2
|
50
|
|
|
|
19
|
(ref($id2jy{$1}) ? die("ambiguous") : $id2jy{$1}) . |
|
|
50
|
|
|
|
|
|
1222
|
|
|
|
|
|
|
($opts->{remove_tones} ? "" : $2)/eg; |
1223
|
|
|
|
|
|
|
} |
1224
|
2
|
|
|
|
|
128
|
$t; |
1225
|
1
|
|
|
|
|
5
|
}; |
1226
|
1
|
|
|
|
|
3
|
eval { |
1227
|
1
|
|
|
|
|
624
|
$text =~ s/\b((?:$idcant_re[123456]?)+)\b/$sub2->($1)/eg; |
|
2
|
|
|
|
|
5
|
|
1228
|
|
|
|
|
|
|
}; |
1229
|
1
|
50
|
|
|
|
15
|
$@ ? undef : $text; |
1230
|
|
|
|
|
|
|
} |
1231
|
|
|
|
|
|
|
|
1232
|
|
|
|
|
|
|
sub detect { |
1233
|
6
|
|
|
6
|
1
|
5006
|
my ($self, $text) = @_; |
1234
|
6
|
|
|
|
|
14
|
my $n_hy = 0; |
1235
|
6
|
|
|
|
|
10
|
my $n_jy = 0; |
1236
|
6
|
|
|
|
|
11
|
my $n_idmand = 0; |
1237
|
6
|
|
|
|
|
9
|
my $n_idcant = 0; |
1238
|
6
|
|
|
|
|
11
|
my $n_all = 0; |
1239
|
|
|
|
|
|
|
|
1240
|
|
|
|
|
|
|
my $sub_all = sub { |
1241
|
17
|
|
|
17
|
|
40
|
my $t = shift; |
1242
|
17
|
|
|
|
|
1361
|
$n_all++ while $t =~ /\b$all_re[123456]?\b/g; |
1243
|
17
|
|
|
|
|
94
|
$t; |
1244
|
6
|
|
|
|
|
30
|
}; |
1245
|
|
|
|
|
|
|
my $sub_hy = sub { |
1246
|
13
|
|
|
13
|
|
28
|
my $t = shift; |
1247
|
13
|
|
|
|
|
665
|
$n_hy++ while $t =~ /\b$hy_re[12345]?\b/g; |
1248
|
13
|
|
|
|
|
68
|
$t; |
1249
|
6
|
|
|
|
|
24
|
}; |
1250
|
|
|
|
|
|
|
my $sub_jy = sub { |
1251
|
14
|
|
|
14
|
|
34
|
my $t = shift; |
1252
|
14
|
|
|
|
|
878
|
$n_jy++ while $t =~ /\b$jy_re[123456]?\b/g; |
1253
|
|
|
|
|
|
|
#print "[jy:$1]" while $t =~ /\b($jy_re[123456]?)\b/g; |
1254
|
14
|
|
|
|
|
73
|
$t; |
1255
|
6
|
|
|
|
|
23
|
}; |
1256
|
|
|
|
|
|
|
my $sub_idmand = sub { |
1257
|
10
|
|
|
10
|
|
23
|
my $t = shift; |
1258
|
10
|
|
|
|
|
681
|
$n_idmand++ while $t =~ /\b$idmand_re[12345]?\b/g; |
1259
|
|
|
|
|
|
|
#print "[idmand:$1]" while $t =~ /\b($idmand_re[12345]?)\b/g; |
1260
|
10
|
|
|
|
|
51
|
$t; |
1261
|
6
|
|
|
|
|
24
|
}; |
1262
|
|
|
|
|
|
|
my $sub_idcant = sub { |
1263
|
10
|
|
|
10
|
|
23
|
my $t = shift; |
1264
|
10
|
|
|
|
|
892
|
$n_idcant++ while $t =~ /\b$idcant_re[123456]?\b/g; |
1265
|
10
|
|
|
|
|
70
|
$t; |
1266
|
6
|
|
|
|
|
21
|
}; |
1267
|
|
|
|
|
|
|
|
1268
|
6
|
|
|
|
|
9
|
my @res; |
1269
|
6
|
|
|
|
|
1216
|
$text =~ s/\b((?:$all_re[123456]?){1})\b/$sub_all->($1)/eg; |
|
17
|
|
|
|
|
39
|
|
1270
|
6
|
|
|
|
|
534
|
$text =~ s/\b((?:$hy_re[12345]?){1})\b/$sub_hy->($1)/eg; |
|
13
|
|
|
|
|
32
|
|
1271
|
6
|
|
|
|
|
803
|
$text =~ s/\b((?:$jy_re[123456]?){1})\b/$sub_jy->($1)/eg; |
|
14
|
|
|
|
|
32
|
|
1272
|
6
|
|
|
|
|
546
|
$text =~ s/\b((?:$idmand_re[12345]?){1})\b/$sub_idmand->($1)/eg; |
|
10
|
|
|
|
|
26
|
|
1273
|
6
|
|
|
|
|
859
|
$text =~ s/\b((?:$idcant_re[123456]?){1})\b/$sub_idcant->($1)/eg; |
|
10
|
|
|
|
|
26
|
|
1274
|
|
|
|
|
|
|
|
1275
|
|
|
|
|
|
|
#print "DEBUG: all=$n_all, hy=$n_hy, jy=$n_jy, idmand=$n_idmand, idcant=$n_idcant\n"; |
1276
|
|
|
|
|
|
|
|
1277
|
6
|
100
|
|
|
|
57
|
return @res unless $n_all; |
1278
|
5
|
100
|
|
|
|
23
|
push @res, "hanyu" if $n_hy/$n_all >= 0.9; |
1279
|
5
|
100
|
|
|
|
15
|
push @res, "jyutping" if $n_jy/$n_all >= 0.9; |
1280
|
5
|
100
|
|
|
|
15
|
push @res, "id-mandarin" if $n_idmand/$n_all >= 0.9; |
1281
|
5
|
100
|
|
|
|
21
|
push @res, "id-cantonese" if $n_idcant/$n_all >= 0.9; |
1282
|
5
|
|
|
|
|
95
|
@res; |
1283
|
|
|
|
|
|
|
} |
1284
|
|
|
|
|
|
|
|
1285
|
|
|
|
|
|
|
sub list_hanyu { |
1286
|
0
|
|
|
0
|
1
|
|
my ($self) = @_; |
1287
|
0
|
|
|
|
|
|
sort keys %hy2id; |
1288
|
|
|
|
|
|
|
} |
1289
|
|
|
|
|
|
|
|
1290
|
|
|
|
|
|
|
sub list_jyutping { |
1291
|
0
|
|
|
0
|
1
|
|
my ($self) = @_; |
1292
|
0
|
|
|
|
|
|
sort keys %jy2id; |
1293
|
|
|
|
|
|
|
} |
1294
|
|
|
|
|
|
|
|
1295
|
|
|
|
|
|
|
sub list_id_mandarin { |
1296
|
0
|
|
|
0
|
1
|
|
my ($self) = @_; |
1297
|
0
|
|
|
|
|
|
sort keys %id2hy; |
1298
|
|
|
|
|
|
|
} |
1299
|
|
|
|
|
|
|
|
1300
|
|
|
|
|
|
|
sub list_id_cantonese { |
1301
|
0
|
|
|
0
|
1
|
|
my ($self) = @_; |
1302
|
0
|
|
|
|
|
|
sort keys %id2jy; |
1303
|
|
|
|
|
|
|
} |
1304
|
|
|
|
|
|
|
|
1305
|
|
|
|
|
|
|
1; |
1306
|
|
|
|
|
|
|
# ABSTRACT: Convert between various Chinese pinyin system and Indonesian transliteration |
1307
|
|
|
|
|
|
|
|
1308
|
|
|
|
|
|
|
__END__ |