]>
Commit | Line | Data |
---|---|---|
f3c0d7a5 A |
1 | # © 2016 and later: Unicode, Inc. and others. |
2 | # License & terms of use: http://www.unicode.org/copyright.html#License | |
3 | # | |
2ca993e8 | 4 | # File: kk_kk_Latn_BGN.txt |
f3c0d7a5 | 5 | # Generated from CLDR |
729e4ab9 | 6 | # |
2ca993e8 A |
7 | |
8 | # | |
9 | ######################################################################## | |
10 | # BGN/PCGN 1979 System | |
11 | # | |
12 | # The BGN/PCGN system for Kazakh Cyrillic was designed for use in | |
13 | # romanizing names written in the Kazakh Cyrillic alphabet. | |
14 | # The Kazakh Cyrillic alphabet contains nine letters not present | |
15 | # in the Russian alphabet: Әә, Ғғ, Ққ, Ңң, Өө, Ұұ, Үү, Һһ, and Іі. | |
16 | # | |
17 | # The Kazakh Cyrillic Alphabet as defined by the BGN (Page 47): | |
18 | # | |
19 | # АӘБВГҒДЕЁЖЗИЙКҚЛМНҢОӨПРСТУҰҮФХҺЦЧШЩЪЫІЬЭЮЯ | |
20 | # аәбвгғдеёжзийкқлмнңоөпрстуұүфхһцчшщъыіьэюя | |
21 | # | |
22 | # Originally prepared by Michael Everson <everson@evertype.com> | |
23 | ######################################################################## | |
24 | # | |
25 | # MINIMAL FILTER: KazakhCyrl-Latin | |
26 | # | |
729e4ab9 A |
27 | :: [АӘБВГҒДЕЁЖЗИЙКҚЛМНҢОӨПРСТУҰҮФХҺЦЧШЩЪЫІЬЭЮЯаәбвгғдеёжзийкқлмнңоөпрстуұүфхһцчшщъыіьэюя] ; |
28 | :: NFD (NFC) ; | |
2ca993e8 A |
29 | # |
30 | # | |
31 | ######################################################################## | |
32 | # | |
33 | ######################################################################## | |
34 | # | |
35 | # Define All Transformation Variables | |
36 | # | |
37 | ######################################################################## | |
38 | # | |
51004dcb | 39 | $prime = ʹ ; |
729e4ab9 A |
40 | $doublePrime = ʺ ; |
41 | $upperConsonants = [БВГҒДЖЗЙКҚЛМНҢПРСТФХҺЦЧШЩЪЬ] ; | |
42 | $lowerConsonants = [бвгғджзйкқлмнңпрстфхһцчшщъь] ; | |
43 | $consonants = [$upperConsonants $lowerConsonants] ; | |
44 | $upperVowels = [АӘЕЁИОӨУҰҮЫІЭЮЯ] ; | |
45 | $lowerVowels = [аәеёиоөуұүыіэюя] ; | |
46 | $vowels = [$upperVowels $lowerVowels] ; | |
47 | $lower = [$lowerConsonants $lowerVowels] ; | |
2ca993e8 A |
48 | # |
49 | # | |
50 | # Use this $wordBoundary until bug 2034 is fixed in ICU: | |
51 | # http://bugs.icu-project.org/cgi-bin/icu-bugs/transliterate?id=2034;expression=boundary;user=guest | |
52 | # | |
51004dcb | 53 | $wordBoundary = [^[:L:][:M:][:N:]] ; |
2ca993e8 A |
54 | # |
55 | # | |
56 | ######################################################################## | |
57 | # | |
58 | ######################################################################## | |
59 | # | |
60 | # Start of Alphabetic Transformations | |
61 | # | |
62 | ######################################################################## | |
63 | # | |
729e4ab9 A |
64 | А → A ; # CYRILLIC CAPITAL LETTER A |
65 | а → a ; # CYRILLIC SMALL LETTER A | |
66 | Ә → Ä ; # CYRILLIC CAPITAL LETTER SCHWA | |
67 | ә → ä ; # CYRILLIC SMALL LETTER SCHWA | |
68 | Б → B ; # CYRILLIC CAPITAL LETTER BE | |
69 | б → b ; # CYRILLIC SMALL LETTER BE | |
70 | В → V ; # CYRILLIC CAPITAL LETTER VE | |
71 | в → v ; # CYRILLIC SMALL LETTER VE | |
2ca993e8 A |
72 | # |
73 | # | |
74 | ######################################################################## | |
75 | # | |
76 | # BGN Page 48 Rule 1 | |
77 | # | |
78 | # The character sequences гһ, зһ, кһ, нг, сһ and цһ may be romanized | |
79 | # g·h, z·h, k·h, n·g, s·h and ts·h in order to differentiate those | |
80 | # romanizations from the digraphs gh, zh, kh, ng, sh, and the letter | |
81 | # sequence tsh, which are used to render the characters г, ж, х, ң, ш, | |
82 | # and the character sequence тш. | |
83 | # | |
84 | ######################################################################## | |
85 | # | |
729e4ab9 A |
86 | ГҺ → G·H ; # CYRILLIC CAPITAL LETTER GHE |
87 | Гһ → G·h ; # CYRILLIC CAPITAL LETTER GHE | |
88 | гһ → g·h ; # CYRILLIC SMALL LETTER GHE | |
89 | Г → G ; # CYRILLIC CAPITAL LETTER GHE | |
90 | г → g ; # CYRILLIC SMALL LETTER GHE | |
2ca993e8 A |
91 | # |
92 | # | |
93 | ######################################################################## | |
94 | # | |
95 | # End Rule 1 | |
96 | # | |
97 | ######################################################################## | |
98 | # | |
729e4ab9 A |
99 | Ғ} $lower → Gh ; # CYRILLIC CAPITAL LETTER GHE WITH STROKE |
100 | Ғ → GH ; # CYRILLIC CAPITAL LETTER GHE WITH STROKE | |
101 | ғ → gh ; # CYRILLIC SMALL LETTER GHE WITH STROKE | |
102 | Д → D ; # CYRILLIC CAPITAL LETTER DE | |
103 | д → d ; # CYRILLIC SMALL LETTER DE | |
104 | Е → E ; # CYRILLIC CAPITAL LETTER IE | |
105 | е → e ; # CYRILLIC SMALL LETTER IE | |
106 | Ё} $lower → Yo ; # CYRILLIC CAPITAL LETTER IO | |
107 | Ё → YO ; # CYRILLIC CAPITAL LETTER IO | |
108 | ё → yo ; # CYRILLIC SMALL LETTER IO | |
109 | Ж} $lower → Zh ; # CYRILLIC CAPITAL LETTER ZHE | |
110 | Ж → ZH ; # CYRILLIC CAPITAL LETTER ZHE | |
111 | ж → zh ; # CYRILLIC SMALL LETTER ZHE | |
2ca993e8 A |
112 | # |
113 | # | |
114 | ######################################################################## | |
115 | # | |
116 | # BGN Page 48 Rule 1 | |
117 | # | |
118 | # зһ becomes z·h | |
119 | # | |
120 | ######################################################################## | |
121 | # | |
729e4ab9 A |
122 | ЗҺ → Z·H ; # CYRILLIC CAPITAL LETTER ZE |
123 | Зһ → Z·h ; # CYRILLIC CAPITAL LETTER ZE | |
124 | зһ → z·h ; # CYRILLIC SMALL LETTER ZE | |
125 | З → Z ; # CYRILLIC CAPITAL LETTER ZE | |
126 | з → z ; # CYRILLIC SMALL LETTER ZE | |
2ca993e8 A |
127 | # |
128 | # | |
129 | ######################################################################## | |
130 | # | |
131 | # End Rule 1 | |
132 | # | |
133 | ######################################################################## | |
134 | # | |
729e4ab9 A |
135 | И → Ī ; # CYRILLIC CAPITAL LETTER I |
136 | и → ī ; # CYRILLIC SMALL LETTER I | |
137 | Й → Y ; # CYRILLIC CAPITAL LETTER I | |
138 | й → y ; # CYRILLIC SMALL LETTER I | |
2ca993e8 A |
139 | # |
140 | # | |
141 | ######################################################################## | |
142 | # | |
143 | # BGN Page 48 Rule 1 | |
144 | # | |
145 | # кһ becomes k·h | |
146 | # | |
147 | ######################################################################## | |
148 | # | |
729e4ab9 A |
149 | КҺ → K·H ; # CYRILLIC CAPITAL LETTER KA |
150 | Кһ → K·h ; # CYRILLIC CAPITAL LETTER KA | |
151 | кһ → k·h ; # CYRILLIC SMALL LETTER KA | |
152 | К → K ; # CYRILLIC CAPITAL LETTER KA | |
153 | к → k ; # CYRILLIC SMALL LETTER KA | |
2ca993e8 A |
154 | # |
155 | # | |
156 | ######################################################################## | |
157 | # | |
158 | # End Rule 1 | |
159 | # | |
160 | ######################################################################## | |
161 | # | |
729e4ab9 A |
162 | Қ → Q ; # CYRILLIC CAPITAL LETTER KA WITH DESCENDER |
163 | қ → q ; # CYRILLIC SMALL LETTER KA WITH DESCENDER | |
164 | Л → L ; # CYRILLIC CAPITAL LETTER EL | |
165 | л → l ; # CYRILLIC SMALL LETTER EL | |
166 | М → M ; # CYRILLIC CAPITAL LETTER EM | |
167 | м → m ; # CYRILLIC SMALL LETTER EM | |
2ca993e8 A |
168 | # |
169 | # | |
170 | ######################################################################## | |
171 | # | |
172 | # BGN Page 48 Rule 1 | |
173 | # | |
174 | # нг becomes n·g | |
175 | # | |
176 | ######################################################################## | |
177 | # | |
729e4ab9 A |
178 | НГ → N·G ; # CYRILLIC CAPITAL LETTER EN |
179 | Нг → N·g ; # CYRILLIC CAPITAL LETTER EN | |
180 | нг → n·g ; # CYRILLIC SMALL LETTER EN | |
181 | Н → N ; # CYRILLIC CAPITAL LETTER EN | |
182 | н → n ; # CYRILLIC SMALL LETTER EN | |
2ca993e8 A |
183 | # |
184 | # | |
185 | ######################################################################## | |
186 | # | |
187 | # End Rule 1 | |
188 | # | |
189 | ######################################################################## | |
190 | # | |
729e4ab9 A |
191 | Ң} $lower → Ng ; # CYRILLIC CAPITAL LETTER EN WITH DESCENDER |
192 | Ң → NG ; # CYRILLIC CAPITAL LETTER EN WITH DESCENDER | |
193 | ң → ng ; # CYRILLIC SMALL LETTER EN WITH DESCENDER | |
194 | О → O ; # CYRILLIC CAPITAL LETTER O | |
195 | о → o ; # CYRILLIC SMALL LETTER O | |
196 | Ө → Ö ; # CYRILLIC CAPITAL LETTER BARRED O | |
197 | ө → ö ; # CYRILLIC SMALL LETTER BARRED O | |
198 | П → P ; # CYRILLIC CAPITAL LETTER PE | |
199 | п → p ; # CYRILLIC SMALL LETTER PE | |
200 | Р → R ; # CYRILLIC CAPITAL LETTER ER | |
201 | р → r ; # CYRILLIC SMALL LETTER ER | |
2ca993e8 A |
202 | # |
203 | # | |
204 | ######################################################################## | |
205 | # | |
206 | # BGN Page 48 Rule 1 | |
207 | # | |
208 | # сһ becomes s·h | |
209 | # | |
210 | ######################################################################## | |
211 | # | |
729e4ab9 A |
212 | СҺ → S·H ; # CYRILLIC CAPITAL LETTER ES |
213 | Сһ → S·h ; # CYRILLIC CAPITAL LETTER ES | |
214 | сһ → s·h ; # CYRILLIC SMALL LETTER ES | |
215 | С → S ; # CYRILLIC CAPITAL LETTER ES | |
216 | с → s ; # CYRILLIC SMALL LETTER ES | |
2ca993e8 A |
217 | # |
218 | # | |
219 | ######################################################################## | |
220 | # | |
221 | # End Rule 1 | |
222 | # | |
223 | ######################################################################## | |
224 | # | |
729e4ab9 A |
225 | Т → T ; # CYRILLIC CAPITAL LETTER TE |
226 | т → t ; # CYRILLIC SMALL LETTER TE | |
227 | У → Ū ; # CYRILLIC CAPITAL LETTER U | |
228 | у → ū ; # CYRILLIC SMALL LETTER U | |
229 | Ұ → U ; # CYRILLIC CAPITAL LETTER STRAIGHT U WITH STROKE | |
230 | ұ → u ; # CYRILLIC SMALL LETTER STRAIGHT U WITH STROKE | |
231 | Ү → Ü ; # CYRILLIC CAPITAL LETTER STRAIGHT U | |
232 | ү → ü ; # CYRILLIC SMALL LETTER STRAIGHT U | |
233 | Ф → F ; # CYRILLIC CAPITAL LETTER EF | |
234 | ф → f ; # CYRILLIC SMALL LETTER EF | |
235 | Х} $lower → Kh ; # CYRILLIC CAPITAL LETTER HA | |
236 | Х → KH ; # CYRILLIC CAPITAL LETTER HA | |
237 | х → kh ; # CYRILLIC SMALL LETTER HA | |
238 | Һ → H ; # CYRILLIC CAPITAL LETTER SHHA | |
239 | һ → h ; # CYRILLIC SMALL LETTER SHHA | |
2ca993e8 A |
240 | # |
241 | # | |
242 | ######################################################################## | |
243 | # | |
244 | # BGN Page 48 Rule 1 | |
245 | # | |
246 | # цһ becomes ts·h | |
247 | # | |
248 | ######################################################################## | |
249 | # | |
729e4ab9 A |
250 | ЦҺ → TS·H ; # CYRILLIC CAPITAL LETTER GHE |
251 | Цһ → Ts·h ; # CYRILLIC CAPITAL LETTER GHE | |
252 | цһ → ts·h ; # CYRILLIC SMALL LETTER GHE | |
253 | Ц} $lower → Ts ; # CYRILLIC CAPITAL LETTER TSE | |
254 | Ц → TS ; # CYRILLIC CAPITAL LETTER TSE | |
255 | ц → ts ; # CYRILLIC SMALL LETTER TSE | |
2ca993e8 A |
256 | # |
257 | # | |
258 | ######################################################################## | |
259 | # | |
260 | # End Rule 1 | |
261 | # | |
262 | ######################################################################## | |
263 | # | |
729e4ab9 A |
264 | Ч} $lower → Ch ; # CYRILLIC CAPITAL LETTER CHE |
265 | Ч → CH ; # CYRILLIC CAPITAL LETTER CHE | |
266 | ч → ch ; # CYRILLIC SMALL LETTER CHE | |
2ca993e8 A |
267 | # |
268 | # | |
269 | ######################################################################## | |
270 | # | |
271 | # Implied rule from BGN Russian-Latin transliteration (Page 94 Note 3.6). | |
272 | # | |
273 | # шч becomes sh·ch | |
274 | # | |
275 | ######################################################################## | |
276 | # | |
729e4ab9 A |
277 | ШЧ → SH·CH ; # CYRILLIC CAPITAL LETTER SHA |
278 | Шч → Sh·ch ; # CYRILLIC CAPITAL LETTER SHA | |
279 | шч → sh·ch ; # CYRILLIC SMALL LETTER SHA | |
280 | Ш} $lower → Sh ; # CYRILLIC CAPITAL LETTER SHA | |
281 | Ш → SH ; # CYRILLIC CAPITAL LETTER SHA | |
282 | ш → sh ; # CYRILLIC SMALL LETTER SHA | |
283 | Щ} $lower → Shch ; # CYRILLIC CAPITAL LETTER SHCHA | |
284 | Щ → SHCH ; # CYRILLIC CAPITAL LETTER SHCHA | |
285 | щ → shch ; # CYRILLIC SMALL LETTER SHCHA | |
2ca993e8 A |
286 | # |
287 | # | |
288 | ######################################################################## | |
289 | # | |
290 | # End Implied rule | |
291 | # | |
292 | ######################################################################## | |
293 | # | |
729e4ab9 A |
294 | Ъ → $doublePrime ; # CYRILLIC CAPITAL LETTER HARD SIGN |
295 | ъ → $doublePrime ; # CYRILLIC SMALL LETTER HARD SIGN | |
2ca993e8 A |
296 | # |
297 | # | |
298 | ######################################################################## | |
299 | # | |
300 | # BGN Page 48 Note 2 | |
301 | # | |
302 | # The character Ыы may be romanized Ɨɨ instead of Yy, if so desired. | |
303 | # | |
304 | ######################################################################## | |
305 | # | |
729e4ab9 A |
306 | Ы → Y ; # CYRILLIC CAPITAL LETTER YERU |
307 | ы → y ; # CYRILLIC SMALL LETTER YERU | |
2ca993e8 A |
308 | # |
309 | # | |
310 | # Alternative rule to implement the option described here. To apply | |
311 | # uncomment the following by removing the '#' mark at the start of the | |
312 | # line and insert before the two rule lines above. | |
313 | # | |
314 | #Ы → Ɨ ; # CYRILLIC CAPITAL LETTER YERU | |
315 | #ы → ɨ ; # CYRILLIC SMALL LETTER YERU | |
316 | # | |
317 | ######################################################################## | |
318 | # | |
319 | # End BGN Page 48 Note 2 | |
320 | # | |
321 | ######################################################################## | |
322 | # | |
729e4ab9 A |
323 | І → I ; # CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I |
324 | і → i ; # CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I | |
325 | Ь → $prime ; # CYRILLIC CAPITAL LETTER SOFT SIGN | |
326 | ь → $prime ; # CYRILLIC SMALL LETTER SOFT SIGN | |
327 | Э → Ė ; # CYRILLIC CAPITAL LETTER E | |
328 | э → ė ; # CYRILLIC SMALL LETTER E | |
329 | Ю} $lower → Yu ; # CYRILLIC CAPITAL LETTER YU | |
330 | Ю → YU ; # CYRILLIC CAPITAL LETTER YU | |
331 | ю → yu ; # CYRILLIC SMALL LETTER YU | |
332 | Я} $lower → Ya ; # CYRILLIC CAPITAL LETTER YA | |
333 | Я → YA ; # CYRILLIC CAPITAL LETTER YA | |
334 | я → ya ; # CYRILLIC SMALL LETTER YA | |
2ca993e8 A |
335 | # |
336 | # | |
337 | ######################################################################## | |
338 |