2 * Copyright © 2009 Red Hat, Inc.
3 * Copyright © 2011 Google, Inc.
5 * This is part of HarfBuzz, a text shaping library.
7 * Permission is hereby granted, without written agreement and without
8 * license or royalty fees, to use, copy, modify, and distribute this
9 * software and its documentation for any purpose, provided that the
10 * above copyright notice and the following two paragraphs appear in
11 * all copies of this software.
13 * IN NO EVENT SHALL THE COPYRIGHT HOLDER BE LIABLE TO ANY PARTY FOR
14 * DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES
15 * ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS DOCUMENTATION, EVEN
16 * IF THE COPYRIGHT HOLDER HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH
19 * THE COPYRIGHT HOLDER SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING,
20 * BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
21 * FITNESS FOR A PARTICULAR PURPOSE. THE SOFTWARE PROVIDED HEREUNDER IS
22 * ON AN "AS IS" BASIS, AND THE COPYRIGHT HOLDER HAS NO OBLIGATION TO
23 * PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
25 * Red Hat Author(s): Behdad Esfahbod
26 * Google Author(s): Behdad Esfahbod
29 #include "hb-private.hh"
39 hb_ot_old_tag_from_script (hb_script_t script
)
41 switch ((hb_tag_t
) script
) {
42 case HB_SCRIPT_INVALID
: return HB_OT_TAG_DEFAULT_SCRIPT
;
44 /* KATAKANA and HIRAGANA both map to 'kana' */
45 case HB_SCRIPT_HIRAGANA
: return HB_TAG('k','a','n','a');
47 /* Spaces at the end are preserved, unlike ISO 15924 */
48 case HB_SCRIPT_LAO
: return HB_TAG('l','a','o',' ');
49 case HB_SCRIPT_YI
: return HB_TAG('y','i',' ',' ');
50 /* Unicode-5.0 additions */
51 case HB_SCRIPT_NKO
: return HB_TAG('n','k','o',' ');
52 /* Unicode-5.1 additions */
53 case HB_SCRIPT_VAI
: return HB_TAG('v','a','i',' ');
54 /* Unicode-5.2 additions */
55 /* Unicode-6.0 additions */
58 /* Else, just change first char to lowercase and return */
59 return ((hb_tag_t
) script
) | 0x20000000;
63 hb_ot_old_tag_to_script (hb_tag_t tag
)
65 if (unlikely (tag
== HB_OT_TAG_DEFAULT_SCRIPT
))
66 return HB_SCRIPT_INVALID
;
68 /* This side of the conversion is fully algorithmic. */
70 /* Any spaces at the end of the tag are replaced by repeating the last
71 * letter. Eg 'nko ' -> 'Nkoo' */
72 if (unlikely ((tag
& 0x0000FF00) == 0x00002000))
73 tag
|= (tag
>> 8) & 0x0000FF00; /* Copy second letter to third */
74 if (unlikely ((tag
& 0x000000FF) == 0x00000020))
75 tag
|= (tag
>> 8) & 0x000000FF; /* Copy third letter to fourth */
77 /* Change first char to uppercase and return */
78 return (hb_script_t
) (tag
& ~0x20000000);
82 hb_ot_new_tag_from_script (hb_script_t script
)
84 switch ((hb_tag_t
) script
) {
85 case HB_SCRIPT_BENGALI
: return HB_TAG('b','n','g','2');
86 case HB_SCRIPT_DEVANAGARI
: return HB_TAG('d','e','v','2');
87 case HB_SCRIPT_GUJARATI
: return HB_TAG('g','j','r','2');
88 case HB_SCRIPT_GURMUKHI
: return HB_TAG('g','u','r','2');
89 case HB_SCRIPT_KANNADA
: return HB_TAG('k','n','d','2');
90 case HB_SCRIPT_MALAYALAM
: return HB_TAG('m','l','m','2');
91 case HB_SCRIPT_ORIYA
: return HB_TAG('o','r','y','2');
92 case HB_SCRIPT_TAMIL
: return HB_TAG('t','m','l','2');
93 case HB_SCRIPT_TELUGU
: return HB_TAG('t','e','l','2');
96 return HB_OT_TAG_DEFAULT_SCRIPT
;
100 hb_ot_new_tag_to_script (hb_tag_t tag
)
103 case HB_TAG('b','n','g','2'): return HB_SCRIPT_BENGALI
;
104 case HB_TAG('d','e','v','2'): return HB_SCRIPT_DEVANAGARI
;
105 case HB_TAG('g','j','r','2'): return HB_SCRIPT_GUJARATI
;
106 case HB_TAG('g','u','r','2'): return HB_SCRIPT_GURMUKHI
;
107 case HB_TAG('k','n','d','2'): return HB_SCRIPT_KANNADA
;
108 case HB_TAG('m','l','m','2'): return HB_SCRIPT_MALAYALAM
;
109 case HB_TAG('o','r','y','2'): return HB_SCRIPT_ORIYA
;
110 case HB_TAG('t','m','l','2'): return HB_SCRIPT_TAMIL
;
111 case HB_TAG('t','e','l','2'): return HB_SCRIPT_TELUGU
;
114 return HB_SCRIPT_UNKNOWN
;
119 * https://www.microsoft.com/typography/otspec/scripttags.htm
120 * https://www.microsoft.com/typography/otspec160/scripttagsProposed.htm
122 * Most of the script tags are the same as the ISO 15924 tag but lowercased.
123 * So we just do that, and handle the exceptional cases in a switch.
127 hb_ot_tags_from_script (hb_script_t script
,
128 hb_tag_t
*script_tag_1
,
129 hb_tag_t
*script_tag_2
)
133 *script_tag_2
= HB_OT_TAG_DEFAULT_SCRIPT
;
134 *script_tag_1
= hb_ot_old_tag_from_script (script
);
136 new_tag
= hb_ot_new_tag_from_script (script
);
137 if (unlikely (new_tag
!= HB_OT_TAG_DEFAULT_SCRIPT
)) {
138 *script_tag_2
= *script_tag_1
;
139 *script_tag_1
= new_tag
;
144 hb_ot_tag_to_script (hb_tag_t tag
)
146 if (unlikely ((tag
& 0x000000FF) == '2'))
147 return hb_ot_new_tag_to_script (tag
);
149 return hb_ot_old_tag_to_script (tag
);
162 * http://www.microsoft.com/typography/otspec/languagetags.htm
164 * Generated by intersecting the OpenType language tag list from
165 * Draft OpenType 1.5 spec, with with the ISO 639-3 codes from
166 * 2008/08/04, matching on name, and finally adjusted manually.
168 * Many items still missing. Those are commented out at the end.
169 * Keep sorted for bsearch.
172 static const LangTag ot_languages
[] = {
173 {"aa", HB_TAG('A','F','R',' ')}, /* Afar */
174 {"ab", HB_TAG('A','B','K',' ')}, /* Abkhazian */
175 {"abq", HB_TAG('A','B','A',' ')}, /* Abaza */
176 {"ady", HB_TAG('A','D','Y',' ')}, /* Adyghe */
177 {"af", HB_TAG('A','F','K',' ')}, /* Afrikaans */
178 {"aiw", HB_TAG('A','R','I',' ')}, /* Aari */
179 {"am", HB_TAG('A','M','H',' ')}, /* Amharic */
180 {"ar", HB_TAG('A','R','A',' ')}, /* Arabic */
181 {"arn", HB_TAG('M','A','P',' ')}, /* Mapudungun */
182 {"as", HB_TAG('A','S','M',' ')}, /* Assamese */
183 {"av", HB_TAG('A','V','R',' ')}, /* Avaric */
184 {"awa", HB_TAG('A','W','A',' ')}, /* Awadhi */
185 {"ay", HB_TAG('A','Y','M',' ')}, /* Aymara */
186 {"az", HB_TAG('A','Z','E',' ')}, /* Azerbaijani */
187 {"ba", HB_TAG('B','S','H',' ')}, /* Bashkir */
188 {"bal", HB_TAG('B','L','I',' ')}, /* Baluchi */
189 {"bcq", HB_TAG('B','C','H',' ')}, /* Bench */
190 {"bem", HB_TAG('B','E','M',' ')}, /* Bemba (Zambia) */
191 {"bfq", HB_TAG('B','A','D',' ')}, /* Badaga */
192 {"bft", HB_TAG('B','L','T',' ')}, /* Balti */
193 {"bg", HB_TAG('B','G','R',' ')}, /* Bulgarian */
194 {"bhb", HB_TAG('B','H','I',' ')}, /* Bhili */
195 {"bho", HB_TAG('B','H','O',' ')}, /* Bhojpuri */
196 {"bik", HB_TAG('B','I','K',' ')}, /* Bikol */
197 {"bin", HB_TAG('E','D','O',' ')}, /* Bini */
198 {"bm", HB_TAG('B','M','B',' ')}, /* Bambara */
199 {"bn", HB_TAG('B','E','N',' ')}, /* Bengali */
200 {"bo", HB_TAG('T','I','B',' ')}, /* Tibetan */
201 {"br", HB_TAG('B','R','E',' ')}, /* Breton */
202 {"brh", HB_TAG('B','R','H',' ')}, /* Brahui */
203 {"bs", HB_TAG('B','O','S',' ')}, /* Bosnian */
204 {"btb", HB_TAG('B','T','I',' ')}, /* Beti (Cameroon) */
205 {"ca", HB_TAG('C','A','T',' ')}, /* Catalan */
206 {"ce", HB_TAG('C','H','E',' ')}, /* Chechen */
207 {"ceb", HB_TAG('C','E','B',' ')}, /* Cebuano */
208 {"chp", HB_TAG('C','H','P',' ')}, /* Chipewyan */
209 {"chr", HB_TAG('C','H','R',' ')}, /* Cherokee */
210 {"cop", HB_TAG('C','O','P',' ')}, /* Coptic */
211 {"cr", HB_TAG('C','R','E',' ')}, /* Cree */
212 {"crh", HB_TAG('C','R','T',' ')}, /* Crimean Tatar */
213 {"crm", HB_TAG('M','C','R',' ')}, /* Moose Cree */
214 {"crx", HB_TAG('C','R','R',' ')}, /* Carrier */
215 {"cs", HB_TAG('C','S','Y',' ')}, /* Czech */
216 {"cu", HB_TAG('C','S','L',' ')}, /* Church Slavic */
217 {"cv", HB_TAG('C','H','U',' ')}, /* Chuvash */
218 {"cwd", HB_TAG('D','C','R',' ')}, /* Woods Cree */
219 {"cy", HB_TAG('W','E','L',' ')}, /* Welsh */
220 {"da", HB_TAG('D','A','N',' ')}, /* Danish */
221 {"dap", HB_TAG('N','I','S',' ')}, /* Nisi (India) */
222 {"dar", HB_TAG('D','A','R',' ')}, /* Dargwa */
223 {"de", HB_TAG('D','E','U',' ')}, /* German */
224 {"din", HB_TAG('D','N','K',' ')}, /* Dinka */
225 {"dng", HB_TAG('D','U','N',' ')}, /* Dungan */
226 {"doi", HB_TAG('D','G','R',' ')}, /* Dogri */
227 {"dsb", HB_TAG('L','S','B',' ')}, /* Lower Sorbian */
228 {"dv", HB_TAG('D','I','V',' ')}, /* Dhivehi */
229 {"dz", HB_TAG('D','Z','N',' ')}, /* Dzongkha */
230 {"ee", HB_TAG('E','W','E',' ')}, /* Ewe */
231 {"efi", HB_TAG('E','F','I',' ')}, /* Efik */
232 {"el", HB_TAG('E','L','L',' ')}, /* Modern Greek (1453-) */
233 {"en", HB_TAG('E','N','G',' ')}, /* English */
234 {"eo", HB_TAG('N','T','O',' ')}, /* Esperanto */
235 {"eot", HB_TAG('B','T','I',' ')}, /* Beti (Côte d'Ivoire) */
236 {"es", HB_TAG('E','S','P',' ')}, /* Spanish */
237 {"et", HB_TAG('E','T','I',' ')}, /* Estonian */
238 {"eu", HB_TAG('E','U','Q',' ')}, /* Basque */
239 {"eve", HB_TAG('E','V','N',' ')}, /* Even */
240 {"evn", HB_TAG('E','V','K',' ')}, /* Evenki */
241 {"fa", HB_TAG('F','A','R',' ')}, /* Persian */
242 {"ff", HB_TAG('F','U','L',' ')}, /* Fulah */
243 {"fi", HB_TAG('F','I','N',' ')}, /* Finnish */
244 {"fil", HB_TAG('P','I','L',' ')}, /* Filipino */
245 {"fj", HB_TAG('F','J','I',' ')}, /* Fijian */
246 {"fo", HB_TAG('F','O','S',' ')}, /* Faroese */
247 {"fon", HB_TAG('F','O','N',' ')}, /* Fon */
248 {"fr", HB_TAG('F','R','A',' ')}, /* French */
249 {"fur", HB_TAG('F','R','L',' ')}, /* Friulian */
250 {"fy", HB_TAG('F','R','I',' ')}, /* Western Frisian */
251 {"ga", HB_TAG('I','R','I',' ')}, /* Irish */
252 {"gaa", HB_TAG('G','A','D',' ')}, /* Ga */
253 {"gag", HB_TAG('G','A','G',' ')}, /* Gagauz */
254 {"gbm", HB_TAG('G','A','W',' ')}, /* Garhwali */
255 {"gd", HB_TAG('G','A','E',' ')}, /* Scottish Gaelic */
256 {"gl", HB_TAG('G','A','L',' ')}, /* Galician */
257 {"gld", HB_TAG('N','A','N',' ')}, /* Nanai */
258 {"gn", HB_TAG('G','U','A',' ')}, /* Guarani */
259 {"gon", HB_TAG('G','O','N',' ')}, /* Gondi */
260 {"grt", HB_TAG('G','R','O',' ')}, /* Garo */
261 {"gu", HB_TAG('G','U','J',' ')}, /* Gujarati */
262 {"guk", HB_TAG('G','M','Z',' ')}, /* Gumuz */
263 {"gv", HB_TAG('M','N','X',' ')}, /* Manx Gaelic */
264 {"ha", HB_TAG('H','A','U',' ')}, /* Hausa */
265 {"har", HB_TAG('H','R','I',' ')}, /* Harari */
266 {"he", HB_TAG('I','W','R',' ')}, /* Hebrew */
267 {"hi", HB_TAG('H','I','N',' ')}, /* Hindi */
268 {"hil", HB_TAG('H','I','L',' ')}, /* Hiligaynon */
269 {"hoc", HB_TAG('H','O',' ',' ')}, /* Ho */
270 {"hr", HB_TAG('H','R','V',' ')}, /* Croatian */
271 {"hsb", HB_TAG('U','S','B',' ')}, /* Upper Sorbian */
272 {"ht", HB_TAG('H','A','I',' ')}, /* Haitian */
273 {"hu", HB_TAG('H','U','N',' ')}, /* Hungarian */
274 {"hy", HB_TAG('H','Y','E',' ')}, /* Armenian */
275 {"id", HB_TAG('I','N','D',' ')}, /* Indonesian */
276 {"ig", HB_TAG('I','B','O',' ')}, /* Igbo */
277 {"igb", HB_TAG('E','B','I',' ')}, /* Ebira */
278 {"inh", HB_TAG('I','N','G',' ')}, /* Ingush */
279 {"is", HB_TAG('I','S','L',' ')}, /* Icelandic */
280 {"it", HB_TAG('I','T','A',' ')}, /* Italian */
281 {"iu", HB_TAG('I','N','U',' ')}, /* Inuktitut */
282 {"ja", HB_TAG('J','A','N',' ')}, /* Japanese */
283 {"jv", HB_TAG('J','A','V',' ')}, /* Javanese */
284 {"ka", HB_TAG('K','A','T',' ')}, /* Georgian */
285 {"kam", HB_TAG('K','M','B',' ')}, /* Kamba (Kenya) */
286 {"kbd", HB_TAG('K','A','B',' ')}, /* Kabardian */
287 {"kdr", HB_TAG('K','R','M',' ')}, /* Karaim */
288 {"kdt", HB_TAG('K','U','Y',' ')}, /* Kuy */
289 {"kfr", HB_TAG('K','A','C',' ')}, /* Kachchi */
290 {"kfy", HB_TAG('K','M','N',' ')}, /* Kumaoni */
291 {"kha", HB_TAG('K','S','I',' ')}, /* Khasi */
292 {"khw", HB_TAG('K','H','W',' ')}, /* Khowar */
293 {"ki", HB_TAG('K','I','K',' ')}, /* Kikuyu */
294 {"kk", HB_TAG('K','A','Z',' ')}, /* Kazakh */
295 {"kl", HB_TAG('G','R','N',' ')}, /* Kalaallisut */
296 {"kln", HB_TAG('K','A','L',' ')}, /* Kalenjin */
297 {"km", HB_TAG('K','H','M',' ')}, /* Central Khmer */
298 {"kmw", HB_TAG('K','M','O',' ')}, /* Komo (Democratic Republic of Congo) */
299 {"kn", HB_TAG('K','A','N',' ')}, /* Kannada */
300 {"ko", HB_TAG('K','O','R',' ')}, /* Korean */
301 {"koi", HB_TAG('K','O','P',' ')}, /* Komi-Permyak */
302 {"kok", HB_TAG('K','O','K',' ')}, /* Konkani */
303 {"kpe", HB_TAG('K','P','L',' ')}, /* Kpelle */
304 {"kpv", HB_TAG('K','O','Z',' ')}, /* Komi-Zyrian */
305 {"kpy", HB_TAG('K','Y','K',' ')}, /* Koryak */
306 {"kqy", HB_TAG('K','R','T',' ')}, /* Koorete */
307 {"kr", HB_TAG('K','N','R',' ')}, /* Kanuri */
308 {"kri", HB_TAG('K','R','I',' ')}, /* Krio */
309 {"krl", HB_TAG('K','R','L',' ')}, /* Karelian */
310 {"kru", HB_TAG('K','U','U',' ')}, /* Kurukh */
311 {"ks", HB_TAG('K','S','H',' ')}, /* Kashmiri */
312 {"ku", HB_TAG('K','U','R',' ')}, /* Kurdish */
313 {"kum", HB_TAG('K','U','M',' ')}, /* Kumyk */
314 {"kvd", HB_TAG('K','U','I',' ')}, /* Kui (Indonesia) */
315 {"kxu", HB_TAG('K','U','I',' ')}, /* Kui (India) */
316 {"ky", HB_TAG('K','I','R',' ')}, /* Kirghiz */
317 {"la", HB_TAG('L','A','T',' ')}, /* Latin */
318 {"lad", HB_TAG('J','U','D',' ')}, /* Ladino */
319 {"lb", HB_TAG('L','T','Z',' ')}, /* Luxembourgish */
320 {"lbe", HB_TAG('L','A','K',' ')}, /* Lak */
321 {"lbj", HB_TAG('L','D','K',' ')}, /* Ladakhi */
322 {"lif", HB_TAG('L','M','B',' ')}, /* Limbu */
323 {"lld", HB_TAG('L','A','D',' ')}, /* Ladin */
324 {"ln", HB_TAG('L','I','N',' ')}, /* Lingala */
325 {"lo", HB_TAG('L','A','O',' ')}, /* Lao */
326 {"lt", HB_TAG('L','T','H',' ')}, /* Lithuanian */
327 {"luo", HB_TAG('L','U','O',' ')}, /* Luo (Kenya and Tanzania) */
328 {"luw", HB_TAG('L','U','O',' ')}, /* Luo (Cameroon) */
329 {"lv", HB_TAG('L','V','I',' ')}, /* Latvian */
330 {"lzz", HB_TAG('L','A','Z',' ')}, /* Laz */
331 {"mai", HB_TAG('M','T','H',' ')}, /* Maithili */
332 {"mdc", HB_TAG('M','L','E',' ')}, /* Male (Papua New Guinea) */
333 {"mdf", HB_TAG('M','O','K',' ')}, /* Moksha */
334 {"mdy", HB_TAG('M','L','E',' ')}, /* Male (Ethiopia) */
335 {"men", HB_TAG('M','D','E',' ')}, /* Mende (Sierra Leone) */
336 {"mg", HB_TAG('M','L','G',' ')}, /* Malagasy */
337 {"mi", HB_TAG('M','R','I',' ')}, /* Maori */
338 {"mk", HB_TAG('M','K','D',' ')}, /* Macedonian */
339 {"ml", HB_TAG('M','L','R',' ')}, /* Malayalam */
340 {"mn", HB_TAG('M','N','G',' ')}, /* Mongolian */
341 {"mnc", HB_TAG('M','C','H',' ')}, /* Manchu */
342 {"mni", HB_TAG('M','N','I',' ')}, /* Manipuri */
343 {"mnk", HB_TAG('M','N','D',' ')}, /* Mandinka */
344 {"mns", HB_TAG('M','A','N',' ')}, /* Mansi */
345 {"mnw", HB_TAG('M','O','N',' ')}, /* Mon */
346 {"mo", HB_TAG('M','O','L',' ')}, /* Moldavian */
347 {"moh", HB_TAG('M','O','H',' ')}, /* Mohawk */
348 {"mpe", HB_TAG('M','A','J',' ')}, /* Majang */
349 {"mr", HB_TAG('M','A','R',' ')}, /* Marathi */
350 {"ms", HB_TAG('M','L','Y',' ')}, /* Malay */
351 {"mt", HB_TAG('M','T','S',' ')}, /* Maltese */
352 {"mwr", HB_TAG('M','A','W',' ')}, /* Marwari */
353 {"my", HB_TAG('B','R','M',' ')}, /* Burmese */
354 {"mym", HB_TAG('M','E','N',' ')}, /* Me'en */
355 {"myv", HB_TAG('E','R','Z',' ')}, /* Erzya */
356 {"nb", HB_TAG('N','O','R',' ')}, /* Norwegian Bokmål */
357 {"nco", HB_TAG('S','I','B',' ')}, /* Sibe */
358 {"ne", HB_TAG('N','E','P',' ')}, /* Nepali */
359 {"new", HB_TAG('N','E','W',' ')}, /* Newari */
360 {"ng", HB_TAG('N','D','G',' ')}, /* Ndonga */
361 {"ngl", HB_TAG('L','M','W',' ')}, /* Lomwe */
362 {"niu", HB_TAG('N','I','U',' ')}, /* Niuean */
363 {"niv", HB_TAG('G','I','L',' ')}, /* Gilyak */
364 {"nl", HB_TAG('N','L','D',' ')}, /* Dutch */
365 {"nn", HB_TAG('N','Y','N',' ')}, /* Norwegian Nynorsk */
366 {"no", HB_TAG('N','O','R',' ')}, /* Norwegian (deprecated) */
367 {"nog", HB_TAG('N','O','G',' ')}, /* Nogai */
368 {"nqo", HB_TAG('N','K','O',' ')}, /* N'Ko */
369 {"nsk", HB_TAG('N','A','S',' ')}, /* Naskapi */
370 {"ny", HB_TAG('C','H','I',' ')}, /* Nyanja */
371 {"oc", HB_TAG('O','C','I',' ')}, /* Occitan (post 1500) */
372 {"oj", HB_TAG('O','J','B',' ')}, /* Ojibwa */
373 {"om", HB_TAG('O','R','O',' ')}, /* Oromo */
374 {"or", HB_TAG('O','R','I',' ')}, /* Oriya */
375 {"os", HB_TAG('O','S','S',' ')}, /* Ossetian */
376 {"pa", HB_TAG('P','A','N',' ')}, /* Panjabi */
377 {"pi", HB_TAG('P','A','L',' ')}, /* Pali */
378 {"pl", HB_TAG('P','L','K',' ')}, /* Polish */
379 {"plp", HB_TAG('P','A','P',' ')}, /* Palpa */
380 {"prs", HB_TAG('D','R','I',' ')}, /* Dari */
381 {"ps", HB_TAG('P','A','S',' ')}, /* Pushto */
382 {"pt", HB_TAG('P','T','G',' ')}, /* Portuguese */
383 {"raj", HB_TAG('R','A','J',' ')}, /* Rajasthani */
384 {"ria", HB_TAG('R','I','A',' ')}, /* Riang (India) */
385 {"ril", HB_TAG('R','I','A',' ')}, /* Riang (Myanmar) */
386 {"ro", HB_TAG('R','O','M',' ')}, /* Romanian */
387 {"rom", HB_TAG('R','O','Y',' ')}, /* Romany */
388 {"ru", HB_TAG('R','U','S',' ')}, /* Russian */
389 {"rue", HB_TAG('R','S','Y',' ')}, /* Rusyn */
390 {"sa", HB_TAG('S','A','N',' ')}, /* Sanskrit */
391 {"sah", HB_TAG('Y','A','K',' ')}, /* Yakut */
392 {"sat", HB_TAG('S','A','T',' ')}, /* Santali */
393 {"sck", HB_TAG('S','A','D',' ')}, /* Sadri */
394 {"sd", HB_TAG('S','N','D',' ')}, /* Sindhi */
395 {"se", HB_TAG('N','S','M',' ')}, /* Northern Sami */
396 {"seh", HB_TAG('S','N','A',' ')}, /* Sena */
397 {"sel", HB_TAG('S','E','L',' ')}, /* Selkup */
398 {"sg", HB_TAG('S','G','O',' ')}, /* Sango */
399 {"shn", HB_TAG('S','H','N',' ')}, /* Shan */
400 {"si", HB_TAG('S','N','H',' ')}, /* Sinhala */
401 {"sid", HB_TAG('S','I','D',' ')}, /* Sidamo */
402 {"sjd", HB_TAG('K','S','M',' ')}, /* Kildin Sami */
403 {"sk", HB_TAG('S','K','Y',' ')}, /* Slovak */
404 {"skr", HB_TAG('S','R','K',' ')}, /* Seraiki */
405 {"sl", HB_TAG('S','L','V',' ')}, /* Slovenian */
406 {"sm", HB_TAG('S','M','O',' ')}, /* Samoan */
407 {"sma", HB_TAG('S','S','M',' ')}, /* Southern Sami */
408 {"smj", HB_TAG('L','S','M',' ')}, /* Lule Sami */
409 {"smn", HB_TAG('I','S','M',' ')}, /* Inari Sami */
410 {"sms", HB_TAG('S','K','S',' ')}, /* Skolt Sami */
411 {"snk", HB_TAG('S','N','K',' ')}, /* Soninke */
412 {"so", HB_TAG('S','M','L',' ')}, /* Somali */
413 {"sq", HB_TAG('S','Q','I',' ')}, /* Albanian */
414 {"sr", HB_TAG('S','R','B',' ')}, /* Serbian */
415 {"srr", HB_TAG('S','R','R',' ')}, /* Serer */
416 {"suq", HB_TAG('S','U','R',' ')}, /* Suri */
417 {"sv", HB_TAG('S','V','E',' ')}, /* Swedish */
418 {"sva", HB_TAG('S','V','A',' ')}, /* Svan */
419 {"sw", HB_TAG('S','W','K',' ')}, /* Swahili */
420 {"swb", HB_TAG('C','M','R',' ')}, /* Comorian */
421 {"syr", HB_TAG('S','Y','R',' ')}, /* Syriac */
422 {"ta", HB_TAG('T','A','M',' ')}, /* Tamil */
423 {"tcy", HB_TAG('T','U','L',' ')}, /* Tulu */
424 {"te", HB_TAG('T','E','L',' ')}, /* Telugu */
425 {"tg", HB_TAG('T','A','J',' ')}, /* Tajik */
426 {"th", HB_TAG('T','H','A',' ')}, /* Thai */
427 {"ti", HB_TAG('T','G','Y',' ')}, /* Tigrinya */
428 {"tig", HB_TAG('T','G','R',' ')}, /* Tigre */
429 {"tk", HB_TAG('T','K','M',' ')}, /* Turkmen */
430 {"tn", HB_TAG('T','N','A',' ')}, /* Tswana */
431 {"tnz", HB_TAG('T','N','G',' ')}, /* Tonga (Thailand) */
432 {"to", HB_TAG('T','N','G',' ')}, /* Tonga (Tonga Islands) */
433 {"tog", HB_TAG('T','N','G',' ')}, /* Tonga (Nyasa) */
434 {"toi", HB_TAG('T','N','G',' ')}, /* Tonga (Zambia) */
435 {"tr", HB_TAG('T','R','K',' ')}, /* Turkish */
436 {"ts", HB_TAG('T','S','G',' ')}, /* Tsonga */
437 {"tt", HB_TAG('T','A','T',' ')}, /* Tatar */
438 {"tw", HB_TAG('T','W','I',' ')}, /* Twi */
439 {"ty", HB_TAG('T','H','T',' ')}, /* Tahitian */
440 {"udm", HB_TAG('U','D','M',' ')}, /* Udmurt */
441 {"ug", HB_TAG('U','Y','G',' ')}, /* Uighur */
442 {"uk", HB_TAG('U','K','R',' ')}, /* Ukrainian */
443 {"unr", HB_TAG('M','U','N',' ')}, /* Mundari */
444 {"ur", HB_TAG('U','R','D',' ')}, /* Urdu */
445 {"uz", HB_TAG('U','Z','B',' ')}, /* Uzbek */
446 {"ve", HB_TAG('V','E','N',' ')}, /* Venda */
447 {"vi", HB_TAG('V','I','T',' ')}, /* Vietnamese */
448 {"wbm", HB_TAG('W','A',' ',' ')}, /* Wa */
449 {"wbr", HB_TAG('W','A','G',' ')}, /* Wagdi */
450 {"wo", HB_TAG('W','L','F',' ')}, /* Wolof */
451 {"xal", HB_TAG('K','L','M',' ')}, /* Kalmyk */
452 {"xh", HB_TAG('X','H','S',' ')}, /* Xhosa */
453 {"xom", HB_TAG('K','M','O',' ')}, /* Komo (Sudan) */
454 {"xsl", HB_TAG('S','S','L',' ')}, /* South Slavey */
455 {"yi", HB_TAG('J','I','I',' ')}, /* Yiddish */
456 {"yo", HB_TAG('Y','B','A',' ')}, /* Yoruba */
457 {"yso", HB_TAG('N','I','S',' ')}, /* Nisi (China) */
458 {"zne", HB_TAG('Z','N','D',' ')}, /* Zande */
459 {"zu", HB_TAG('Z','U','L',' ')} /* Zulu */
461 /* I couldn't find the language id for these */
463 /*{"??", HB_TAG('A','G','W',' ')},*/ /* Agaw */
464 /*{"??", HB_TAG('A','L','S',' ')},*/ /* Alsatian */
465 /*{"??", HB_TAG('A','L','T',' ')},*/ /* Altai */
466 /*{"??", HB_TAG('A','R','K',' ')},*/ /* Arakanese */
467 /*{"??", HB_TAG('A','T','H',' ')},*/ /* Athapaskan */
468 /*{"??", HB_TAG('B','A','G',' ')},*/ /* Baghelkhandi */
469 /*{"??", HB_TAG('B','A','L',' ')},*/ /* Balkar */
470 /*{"??", HB_TAG('B','A','U',' ')},*/ /* Baule */
471 /*{"??", HB_TAG('B','B','R',' ')},*/ /* Berber */
472 /*{"??", HB_TAG('B','C','R',' ')},*/ /* Bible Cree */
473 /*{"??", HB_TAG('B','E','L',' ')},*/ /* Belarussian */
474 /*{"??", HB_TAG('B','I','L',' ')},*/ /* Bilen */
475 /*{"??", HB_TAG('B','K','F',' ')},*/ /* Blackfoot */
476 /*{"??", HB_TAG('B','L','N',' ')},*/ /* Balante */
477 /*{"??", HB_TAG('B','M','L',' ')},*/ /* Bamileke */
478 /*{"??", HB_TAG('B','R','I',' ')},*/ /* Braj Bhasha */
479 /*{"??", HB_TAG('C','H','G',' ')},*/ /* Chaha Gurage */
480 /*{"??", HB_TAG('C','H','H',' ')},*/ /* Chattisgarhi */
481 /*{"??", HB_TAG('C','H','K',' ')},*/ /* Chukchi */
482 /*{"??", HB_TAG('D','J','R',' ')},*/ /* Djerma */
483 /*{"??", HB_TAG('D','N','G',' ')},*/ /* Dangme */
484 /*{"??", HB_TAG('E','C','R',' ')},*/ /* Eastern Cree */
485 /*{"??", HB_TAG('F','A','N',' ')},*/ /* French Antillean */
486 /*{"??", HB_TAG('F','L','E',' ')},*/ /* Flemish */
487 /*{"??", HB_TAG('F','N','E',' ')},*/ /* Forest Nenets */
488 /*{"??", HB_TAG('F','T','A',' ')},*/ /* Futa */
489 /*{"??", HB_TAG('G','A','R',' ')},*/ /* Garshuni */
490 /*{"??", HB_TAG('G','E','Z',' ')},*/ /* Ge'ez */
491 /*{"??", HB_TAG('H','A','L',' ')},*/ /* Halam */
492 /*{"??", HB_TAG('H','A','R',' ')},*/ /* Harauti */
493 /*{"??", HB_TAG('H','A','W',' ')},*/ /* Hawaiin */
494 /*{"??", HB_TAG('H','B','N',' ')},*/ /* Hammer-Banna */
495 /*{"??", HB_TAG('H','M','A',' ')},*/ /* High Mari */
496 /*{"??", HB_TAG('H','N','D',' ')},*/ /* Hindko */
497 /*{"??", HB_TAG('I','J','O',' ')},*/ /* Ijo */
498 /*{"??", HB_TAG('I','L','O',' ')},*/ /* Ilokano */
499 /*{"??", HB_TAG('I','R','T',' ')},*/ /* Irish Traditional */
500 /*{"??", HB_TAG('J','U','L',' ')},*/ /* Jula */
501 /*{"??", HB_TAG('K','A','R',' ')},*/ /* Karachay */
502 /*{"??", HB_TAG('K','E','B',' ')},*/ /* Kebena */
503 /*{"??", HB_TAG('K','G','E',' ')},*/ /* Khutsuri Georgian */
504 /*{"??", HB_TAG('K','H','A',' ')},*/ /* Khakass */
505 /*{"??", HB_TAG('K','H','K',' ')},*/ /* Khanty-Kazim */
506 /*{"??", HB_TAG('K','H','S',' ')},*/ /* Khanty-Shurishkar */
507 /*{"??", HB_TAG('K','H','V',' ')},*/ /* Khanty-Vakhi */
508 /*{"??", HB_TAG('K','I','S',' ')},*/ /* Kisii */
509 /*{"??", HB_TAG('K','K','N',' ')},*/ /* Kokni */
510 /*{"??", HB_TAG('K','M','S',' ')},*/ /* Komso */
511 /*{"??", HB_TAG('K','O','D',' ')},*/ /* Kodagu */
512 /*{"??", HB_TAG('K','O','H',' ')},*/ /* Korean Old Hangul */
513 /*{"??", HB_TAG('K','O','N',' ')},*/ /* Kikongo */
514 /*{"??", HB_TAG('K','R','K',' ')},*/ /* Karakalpak */
515 /*{"??", HB_TAG('K','R','N',' ')},*/ /* Karen */
516 /*{"??", HB_TAG('K','U','L',' ')},*/ /* Kulvi */
517 /*{"??", HB_TAG('L','A','H',' ')},*/ /* Lahuli */
518 /*{"??", HB_TAG('L','A','M',' ')},*/ /* Lambani */
519 /*{"??", HB_TAG('L','C','R',' ')},*/ /* L-Cree */
520 /*{"??", HB_TAG('L','E','Z',' ')},*/ /* Lezgi */
521 /*{"??", HB_TAG('L','M','A',' ')},*/ /* Low Mari */
522 /*{"??", HB_TAG('L','U','B',' ')},*/ /* Luba */
523 /*{"??", HB_TAG('L','U','G',' ')},*/ /* Luganda */
524 /*{"??", HB_TAG('L','U','H',' ')},*/ /* Luhya */
525 /*{"??", HB_TAG('M','A','K',' ')},*/ /* Makua */
526 /*{"??", HB_TAG('M','A','L',' ')},*/ /* Malayalam Traditional */
527 /*{"??", HB_TAG('M','B','N',' ')},*/ /* Mbundu */
528 /*{"??", HB_TAG('M','I','Z',' ')},*/ /* Mizo */
529 /*{"??", HB_TAG('M','L','N',' ')},*/ /* Malinke */
530 /*{"??", HB_TAG('M','N','K',' ')},*/ /* Maninka */
531 /*{"??", HB_TAG('M','O','R',' ')},*/ /* Moroccan */
532 /*{"??", HB_TAG('N','A','G',' ')},*/ /* Naga-Assamese */
533 /*{"??", HB_TAG('N','C','R',' ')},*/ /* N-Cree */
534 /*{"??", HB_TAG('N','D','B',' ')},*/ /* Ndebele */
535 /*{"??", HB_TAG('N','G','R',' ')},*/ /* Nagari */
536 /*{"??", HB_TAG('N','H','C',' ')},*/ /* Norway House Cree */
537 /*{"??", HB_TAG('N','K','L',' ')},*/ /* Nkole */
538 /*{"??", HB_TAG('N','T','A',' ')},*/ /* Northern Tai */
539 /*{"??", HB_TAG('O','C','R',' ')},*/ /* Oji-Cree */
540 /*{"??", HB_TAG('P','A','A',' ')},*/ /* Palestinian Aramaic */
541 /*{"??", HB_TAG('P','G','R',' ')},*/ /* Polytonic Greek */
542 /*{"??", HB_TAG('P','L','G',' ')},*/ /* Palaung */
543 /*{"??", HB_TAG('Q','I','N',' ')},*/ /* Chin */
544 /*{"??", HB_TAG('R','B','U',' ')},*/ /* Russian Buriat */
545 /*{"??", HB_TAG('R','C','R',' ')},*/ /* R-Cree */
546 /*{"??", HB_TAG('R','M','S',' ')},*/ /* Rhaeto-Romanic */
547 /*{"??", HB_TAG('R','U','A',' ')},*/ /* Ruanda */
548 /*{"??", HB_TAG('S','A','Y',' ')},*/ /* Sayisi */
549 /*{"??", HB_TAG('S','E','K',' ')},*/ /* Sekota */
550 /*{"??", HB_TAG('S','I','G',' ')},*/ /* Silte Gurage */
551 /*{"??", HB_TAG('S','L','A',' ')},*/ /* Slavey */
552 /*{"??", HB_TAG('S','O','G',' ')},*/ /* Sodo Gurage */
553 /*{"??", HB_TAG('S','O','T',' ')},*/ /* Sotho */
554 /*{"??", HB_TAG('S','W','A',' ')},*/ /* Swadaya Aramaic */
555 /*{"??", HB_TAG('S','W','Z',' ')},*/ /* Swazi */
556 /*{"??", HB_TAG('S','X','T',' ')},*/ /* Sutu */
557 /*{"??", HB_TAG('T','A','B',' ')},*/ /* Tabasaran */
558 /*{"??", HB_TAG('T','C','R',' ')},*/ /* TH-Cree */
559 /*{"??", HB_TAG('T','G','N',' ')},*/ /* Tongan */
560 /*{"??", HB_TAG('T','M','N',' ')},*/ /* Temne */
561 /*{"??", HB_TAG('T','N','E',' ')},*/ /* Tundra Nenets */
562 /*{"??", HB_TAG('T','O','D',' ')},*/ /* Todo */
563 /*{"??", HB_TAG('T','U','A',' ')},*/ /* Turoyo Aramaic */
564 /*{"??", HB_TAG('T','U','V',' ')},*/ /* Tuvin */
565 /*{"??", HB_TAG('W','C','R',' ')},*/ /* West-Cree */
566 /*{"??", HB_TAG('X','B','D',' ')},*/ /* Tai Lue */
567 /*{"??", HB_TAG('Y','C','R',' ')},*/ /* Y-Cree */
568 /*{"??", HB_TAG('Y','I','C',' ')},*/ /* Yi Classic */
569 /*{"??", HB_TAG('Y','I','M',' ')},*/ /* Yi Modern */
570 /*{"??", HB_TAG('Z','H','P',' ')},*/ /* Chinese Phonetic */
573 static const LangTag ot_languages_zh
[] = {
574 {"zh-cn", HB_TAG('Z','H','S',' ')}, /* Chinese (China) */
575 {"zh-hk", HB_TAG('Z','H','H',' ')}, /* Chinese (Hong Kong) */
576 {"zh-mo", HB_TAG('Z','H','T',' ')}, /* Chinese (Macao) */
577 {"zh-sg", HB_TAG('Z','H','S',' ')}, /* Chinese (Singapore) */
578 {"zh-tw", HB_TAG('Z','H','T',' ')} /* Chinese (Taiwan) */
582 lang_compare_first_component (const char *a
,
589 da
= p
? (unsigned int) (p
- a
) : strlen (a
);
592 db
= p
? (unsigned int) (p
- b
) : strlen (b
);
594 return strncmp (a
, b
, MAX (da
, db
));
598 lang_matches (const char *lang_str
, const char *spec
)
600 unsigned int len
= strlen (spec
);
602 return strncmp (lang_str
, spec
, len
) == 0 &&
603 (lang_str
[len
] == '\0' || lang_str
[len
] == '-');
607 hb_ot_tag_from_language (hb_language_t language
)
609 const char *lang_str
, *s
;
610 const LangTag
*lang_tag
;
612 if (language
== HB_LANGUAGE_INVALID
)
613 return HB_OT_TAG_DEFAULT_LANGUAGE
;
615 lang_str
= hb_language_to_string (language
);
617 s
= strstr (lang_str
, "x-hbot");
622 for (i
= 0; i
< 4 && ISALPHA (s
[i
]); i
++)
623 tag
[i
] = TOUPPER (s
[i
]);
627 return HB_TAG_CHAR4 (tag
);
631 /* Find a language matching in the first component */
632 lang_tag
= (LangTag
*) bsearch (lang_str
, ot_languages
,
633 ARRAY_LENGTH (ot_languages
), sizeof (LangTag
),
634 (hb_compare_func_t
) lang_compare_first_component
);
636 return lang_tag
->tag
;
638 /* Otherwise, check the Chinese ones */
639 if (0 == lang_compare_first_component (lang_str
, "zh"))
643 for (i
= 0; i
< ARRAY_LENGTH (ot_languages_zh
); i
++)
645 lang_tag
= &ot_languages_zh
[i
];
646 if (lang_matches (lang_tag
->language
, lang_str
))
647 return lang_tag
->tag
;
650 /* Otherwise just return 'ZHS ' */
651 return HB_TAG('Z','H','S',' ');
654 s
= strchr (lang_str
, '-');
656 s
= lang_str
+ strlen (lang_str
);
657 if (s
- lang_str
== 3) {
658 /* Assume it's ISO-639-3 and upper-case and use it. */
659 return hb_tag_from_string (lang_str
, s
- lang_str
) & ~0x20202000;
662 return HB_OT_TAG_DEFAULT_LANGUAGE
;
666 hb_ot_tag_to_language (hb_tag_t tag
)
670 if (tag
== HB_OT_TAG_DEFAULT_LANGUAGE
)
673 for (i
= 0; i
< ARRAY_LENGTH (ot_languages
); i
++)
674 if (ot_languages
[i
].tag
== tag
)
675 return hb_language_from_string (ot_languages
[i
].language
, -1);
677 /* If tag starts with ZH, it's Chinese */
678 if ((tag
& 0xFFFF0000) == 0x5A480000) {
680 case HB_TAG('Z','H','H',' '): return hb_language_from_string ("zh-hk", -1); /* Hong Kong */
682 /* Encode the tag... */
683 unsigned char buf
[14] = "zh-x-hbot";
685 buf
[10] = (tag
>> 16) & 0xFF;
686 buf
[11] = (tag
>> 8) & 0xFF;
687 buf
[12] = tag
& 0xFF;
691 return hb_language_from_string ((char *) buf
, -1);
696 /* Else return a custom language in the form of "x-hbotABCD" */
698 unsigned char buf
[11] = "x-hbot";
700 buf
[7] = (tag
>> 16) & 0xFF;
701 buf
[8] = (tag
>> 8) & 0xFF;
706 return hb_language_from_string ((char *) buf
, -1);
