1 /*
2 * Copyright © 2009 Red Hat, Inc.
3 * Copyright © 2011 Google, Inc.
4 *
5 * This is part of HarfBuzz, a text shaping library.
6 *
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.
12 *
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
17 * DAMAGE.
18 *
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.
24 *
25 * Red Hat Author(s): Behdad Esfahbod
26 * Google Author(s): Behdad Esfahbod, Roozbeh Pournader
27 */
28
29 #include "hb-private.hh"
30
31
32 /* hb_script_t */
33
34 static hb_tag_t
hb_ot_old_tag_from_script(hb_script_t script)35 hb_ot_old_tag_from_script (hb_script_t script)
36 {
37 /* This seems to be accurate as of end of 2012. */
38
39 switch ((hb_tag_t) script) {
40 case HB_SCRIPT_INVALID: return HB_OT_TAG_DEFAULT_SCRIPT;
41
42 /* KATAKANA and HIRAGANA both map to 'kana' */
43 case HB_SCRIPT_HIRAGANA: return HB_TAG('k','a','n','a');
44
45 /* Spaces at the end are preserved, unlike ISO 15924 */
46 case HB_SCRIPT_LAO: return HB_TAG('l','a','o',' ');
47 case HB_SCRIPT_YI: return HB_TAG('y','i',' ',' ');
48 /* Unicode-5.0 additions */
49 case HB_SCRIPT_NKO: return HB_TAG('n','k','o',' ');
50 /* Unicode-5.1 additions */
51 case HB_SCRIPT_VAI: return HB_TAG('v','a','i',' ');
52 /* Unicode-5.2 additions */
53 /* Unicode-6.0 additions */
54 }
55
56 /* Else, just change first char to lowercase and return */
57 return ((hb_tag_t) script) | 0x20000000u;
58 }
59
60 static hb_script_t
hb_ot_old_tag_to_script(hb_tag_t tag)61 hb_ot_old_tag_to_script (hb_tag_t tag)
62 {
63 if (unlikely (tag == HB_OT_TAG_DEFAULT_SCRIPT))
64 return HB_SCRIPT_INVALID;
65
66 /* This side of the conversion is fully algorithmic. */
67
68 /* Any spaces at the end of the tag are replaced by repeating the last
69 * letter. Eg 'nko ' -> 'Nkoo' */
70 if (unlikely ((tag & 0x0000FF00u) == 0x00002000u))
71 tag |= (tag >> 8) & 0x0000FF00u; /* Copy second letter to third */
72 if (unlikely ((tag & 0x000000FFu) == 0x00000020u))
73 tag |= (tag >> 8) & 0x000000FFu; /* Copy third letter to fourth */
74
75 /* Change first char to uppercase and return */
76 return (hb_script_t) (tag & ~0x20000000u);
77 }
78
79 static hb_tag_t
hb_ot_new_tag_from_script(hb_script_t script)80 hb_ot_new_tag_from_script (hb_script_t script)
81 {
82 switch ((hb_tag_t) script) {
83 case HB_SCRIPT_BENGALI: return HB_TAG('b','n','g','2');
84 case HB_SCRIPT_DEVANAGARI: return HB_TAG('d','e','v','2');
85 case HB_SCRIPT_GUJARATI: return HB_TAG('g','j','r','2');
86 case HB_SCRIPT_GURMUKHI: return HB_TAG('g','u','r','2');
87 case HB_SCRIPT_KANNADA: return HB_TAG('k','n','d','2');
88 case HB_SCRIPT_MALAYALAM: return HB_TAG('m','l','m','2');
89 case HB_SCRIPT_ORIYA: return HB_TAG('o','r','y','2');
90 case HB_SCRIPT_TAMIL: return HB_TAG('t','m','l','2');
91 case HB_SCRIPT_TELUGU: return HB_TAG('t','e','l','2');
92 case HB_SCRIPT_MYANMAR: return HB_TAG('m','y','m','2');
93 }
94
95 return HB_OT_TAG_DEFAULT_SCRIPT;
96 }
97
98 static hb_script_t
hb_ot_new_tag_to_script(hb_tag_t tag)99 hb_ot_new_tag_to_script (hb_tag_t tag)
100 {
101 switch (tag) {
102 case HB_TAG('b','n','g','2'): return HB_SCRIPT_BENGALI;
103 case HB_TAG('d','e','v','2'): return HB_SCRIPT_DEVANAGARI;
104 case HB_TAG('g','j','r','2'): return HB_SCRIPT_GUJARATI;
105 case HB_TAG('g','u','r','2'): return HB_SCRIPT_GURMUKHI;
106 case HB_TAG('k','n','d','2'): return HB_SCRIPT_KANNADA;
107 case HB_TAG('m','l','m','2'): return HB_SCRIPT_MALAYALAM;
108 case HB_TAG('o','r','y','2'): return HB_SCRIPT_ORIYA;
109 case HB_TAG('t','m','l','2'): return HB_SCRIPT_TAMIL;
110 case HB_TAG('t','e','l','2'): return HB_SCRIPT_TELUGU;
111 case HB_TAG('m','y','m','2'): return HB_SCRIPT_MYANMAR;
112 }
113
114 return HB_SCRIPT_UNKNOWN;
115 }
116
117 /*
118 * Complete list at:
119 * https://www.microsoft.com/typography/otspec/scripttags.htm
120 * https://www.microsoft.com/typography/otspec160/scripttagsProposed.htm
121 *
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.
124 */
125
126 void
hb_ot_tags_from_script(hb_script_t script,hb_tag_t * script_tag_1,hb_tag_t * script_tag_2)127 hb_ot_tags_from_script (hb_script_t script,
128 hb_tag_t *script_tag_1,
129 hb_tag_t *script_tag_2)
130 {
131 hb_tag_t new_tag;
132
133 *script_tag_2 = HB_OT_TAG_DEFAULT_SCRIPT;
134 *script_tag_1 = hb_ot_old_tag_from_script (script);
135
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;
140 }
141 }
142
143 hb_script_t
hb_ot_tag_to_script(hb_tag_t tag)144 hb_ot_tag_to_script (hb_tag_t tag)
145 {
146 if (unlikely ((tag & 0x000000FFu) == '2'))
147 return hb_ot_new_tag_to_script (tag);
148
149 return hb_ot_old_tag_to_script (tag);
150 }
151
152
153 /* hb_language_t */
154
155 typedef struct {
156 char language[4];
157 hb_tag_t tag;
158 } LangTag;
159
160 /*
161 * Complete list at:
162 * http://www.microsoft.com/typography/otspec/languagetags.htm
163 *
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.
167 *
168 * Updated on 2012-12-07 with more research into remaining codes.
169 *
170 * Updated on 2013-11-23 based on usage in SIL and Microsoft fonts,
171 * the new proposal from Microsoft, and latest ISO 639-3 names.
172 *
173 * Some items still missing. Those are commented out at the end.
174 * Keep sorted for bsearch.
175 *
176 * Updated as of 2015-05-06: OT1.7 on MS website has some newer
177 * items that we don't have here, eg. Zazaki. This is the new
178 * items in OpenType 1.7 (red items), most of which we have:
179 * http://www.microsoft.com/typography/otspec170/languagetags.htm
180 */
181
182 static const LangTag ot_languages[] = {
183 {"aa", HB_TAG('A','F','R',' ')}, /* Afar */
184 {"ab", HB_TAG('A','B','K',' ')}, /* Abkhazian */
185 {"abq", HB_TAG('A','B','A',' ')}, /* Abaza */
186 {"acf", HB_TAG('F','A','N',' ')}, /* French Antillean */
187 {"ach", HB_TAG('A','C','H',' ')}, /* Acoli */
188 {"acr", HB_TAG('A','C','R',' ')}, /* Achi */
189 {"ada", HB_TAG('D','N','G',' ')}, /* Dangme */
190 {"ady", HB_TAG('A','D','Y',' ')}, /* Adyghe */
191 {"af", HB_TAG('A','F','K',' ')}, /* Afrikaans */
192 {"ahg", HB_TAG('A','G','W',' ')}, /* Agaw */
193 {"aii", HB_TAG('S','W','A',' ')}, /* Swadaya Aramaic */
194 {"aio", HB_TAG('A','I','O',' ')}, /* Aiton */
195 {"aiw", HB_TAG('A','R','I',' ')}, /* Aari */
196 {"ak", HB_TAG('T','W','I',' ')}, /* Akan [macrolanguage] */
197 {"aka", HB_TAG('A','K','A',' ')}, /* Akan */
198 {"alt", HB_TAG('A','L','T',' ')}, /* [Southern] Altai */
199 {"am", HB_TAG('A','M','H',' ')}, /* Amharic */
200 {"amf", HB_TAG('H','B','N',' ')}, /* Hammer-Banna */
201 {"amw", HB_TAG('S','Y','R',' ')}, /* Western Neo-Aramaic */
202 {"an", HB_TAG('A','R','G',' ')}, /* Aragonese */
203 {"ang", HB_TAG('A','N','G',' ')}, /* Old English (ca. 450-1100) */
204 {"ar", HB_TAG('A','R','A',' ')}, /* Arabic [macrolanguage] */
205 {"arb", HB_TAG('A','R','A',' ')}, /* Standard Arabic */
206 {"arn", HB_TAG('M','A','P',' ')}, /* Mapudungun */
207 {"ary", HB_TAG('M','O','R',' ')}, /* Moroccan Arabic */
208 {"as", HB_TAG('A','S','M',' ')}, /* Assamese */
209 {"ast", HB_TAG('A','S','T',' ')}, /* Asturian/Asturleonese/Bable/Leonese */
210 {"ath", HB_TAG('A','T','H',' ')}, /* Athapaskan [family] */
211 {"atj", HB_TAG('R','C','R',' ')}, /* R-Cree */
212 {"atv", HB_TAG('A','L','T',' ')}, /* [Northern] Altai */
213 {"av", HB_TAG('A','V','R',' ')}, /* Avaric */
214 {"awa", HB_TAG('A','W','A',' ')}, /* Awadhi */
215 {"ay", HB_TAG('A','Y','M',' ')}, /* Aymara [macrolanguage] */
216 {"az", HB_TAG('A','Z','E',' ')}, /* Azerbaijani [macrolanguage] */
217 {"azb", HB_TAG('A','Z','B',' ')}, /* South Azerbaijani */
218 {"azj", HB_TAG('A','Z','E',' ')}, /* North Azerbaijani */
219 {"ba", HB_TAG('B','S','H',' ')}, /* Bashkir */
220 {"bad", HB_TAG('B','A','D','0')}, /* Banda */
221 {"bai", HB_TAG('B','M','L',' ')}, /* Bamileke [family] */
222 {"bal", HB_TAG('B','L','I',' ')}, /* Baluchi [macrolangauge] */
223 {"ban", HB_TAG('B','A','N',' ')}, /* Balinese */
224 {"bar", HB_TAG('B','A','R',' ')}, /* Bavarian */
225 {"bbc", HB_TAG('B','B','C',' ')}, /* Batak Toba */
226 {"bci", HB_TAG('B','A','U',' ')}, /* Baoulé */
227 {"bcl", HB_TAG('B','I','K',' ')}, /* Central Bikol */
228 {"bcq", HB_TAG('B','C','H',' ')}, /* Bench */
229 {"bdy", HB_TAG('B','D','Y',' ')}, /* Bandjalang */
230 {"be", HB_TAG('B','E','L',' ')}, /* Belarusian */
231 {"bem", HB_TAG('B','E','M',' ')}, /* Bemba (Zambia) */
232 {"ber", HB_TAG('B','E','R',' ')}, /* Berber [family] */
233 {"bfq", HB_TAG('B','A','D',' ')}, /* Badaga */
234 {"bft", HB_TAG('B','L','T',' ')}, /* Balti */
235 {"bfu", HB_TAG('L','A','H',' ')}, /* Lahuli */
236 {"bfy", HB_TAG('B','A','G',' ')}, /* Baghelkhandi */
237 {"bg", HB_TAG('B','G','R',' ')}, /* Bulgarian */
238 {"bgc", HB_TAG('B','G','C',' ')}, /* Haryanvi */
239 {"bgq", HB_TAG('B','G','Q',' ')}, /* Bagri */
240 {"bgr", HB_TAG('Q','I','N',' ')}, /* Bawm Chin */
241 {"bhb", HB_TAG('B','H','I',' ')}, /* Bhili */
242 {"bhk", HB_TAG('B','I','K',' ')}, /* Albay Bicolano (retired code) */
243 {"bho", HB_TAG('B','H','O',' ')}, /* Bhojpuri */
244 {"bi", HB_TAG('B','I','S',' ')}, /* Bislama */
245 {"bik", HB_TAG('B','I','K',' ')}, /* Bikol [macrolanguage] */
246 {"bin", HB_TAG('E','D','O',' ')}, /* Bini */
247 {"bjj", HB_TAG('B','J','J',' ')}, /* Kanauji */
248 {"bjt", HB_TAG('B','L','N',' ')}, /* Balanta-Ganja */
249 {"bla", HB_TAG('B','K','F',' ')}, /* Blackfoot */
250 {"ble", HB_TAG('B','L','N',' ')}, /* Balanta-Kentohe */
251 {"blk", HB_TAG('B','L','K',' ')}, /* Pa'O/Pa'o Karen */
252 {"bln", HB_TAG('B','I','K',' ')}, /* Southern Catanduanes Bikol */
253 {"bm", HB_TAG('B','M','B',' ')}, /* Bambara */
254 {"bn", HB_TAG('B','E','N',' ')}, /* Bengali */
255 {"bo", HB_TAG('T','I','B',' ')}, /* Tibetan */
256 {"bpy", HB_TAG('B','P','Y',' ')}, /* Bishnupriya */
257 {"bqi", HB_TAG('L','R','C',' ')}, /* Bakhtiari */
258 {"br", HB_TAG('B','R','E',' ')}, /* Breton */
259 {"bra", HB_TAG('B','R','I',' ')}, /* Braj Bhasha */
260 {"brh", HB_TAG('B','R','H',' ')}, /* Brahui */
261 {"brx", HB_TAG('B','R','X',' ')}, /* Bodo (India) */
262 {"bs", HB_TAG('B','O','S',' ')}, /* Bosnian */
263 {"btb", HB_TAG('B','T','I',' ')}, /* Beti (Cameroon) */
264 {"bto", HB_TAG('B','I','K',' ')}, /* Rinconada Bikol */
265 {"bts", HB_TAG('B','T','S',' ')}, /* Batak Simalungun */
266 {"bug", HB_TAG('B','U','G',' ')}, /* Buginese */
267 {"bxr", HB_TAG('R','B','U',' ')}, /* Russian Buriat */
268 {"byn", HB_TAG('B','I','L',' ')}, /* Bilen */
269 {"ca", HB_TAG('C','A','T',' ')}, /* Catalan */
270 {"cak", HB_TAG('C','A','K',' ')}, /* Kaqchikel */
271 {"cbk", HB_TAG('C','B','K',' ')}, /* Chavacano */
272 {"cbl", HB_TAG('Q','I','N',' ')}, /* Bualkhaw Chin */
273 {"ce", HB_TAG('C','H','E',' ')}, /* Chechen */
274 {"ceb", HB_TAG('C','E','B',' ')}, /* Cebuano */
275 {"cfm", HB_TAG('H','A','L',' ')}, /* Halam/Falam Chin */
276 {"cgg", HB_TAG('C','G','G',' ')}, /* Chiga */
277 {"ch", HB_TAG('C','H','A',' ')}, /* Chamorro */
278 {"chk", HB_TAG('C','H','K','0')}, /* Chuukese */
279 {"cho", HB_TAG('C','H','O',' ')}, /* Choctaw */
280 {"chp", HB_TAG('C','H','P',' ')}, /* Chipewyan */
281 {"chr", HB_TAG('C','H','R',' ')}, /* Cherokee */
282 {"chy", HB_TAG('C','H','Y',' ')}, /* Cheyenne */
283 {"cja", HB_TAG('C','J','A',' ')}, /* Western Cham */
284 {"cjm", HB_TAG('C','J','M',' ')}, /* Eastern Cham */
285 {"cka", HB_TAG('Q','I','N',' ')}, /* Khumi Awa Chin */
286 {"ckb", HB_TAG('K','U','R',' ')}, /* Central Kurdish (Sorani) */
287 {"ckt", HB_TAG('C','H','K',' ')}, /* Chukchi */
288 {"cld", HB_TAG('S','Y','R',' ')}, /* Chaldean Neo-Aramaic */
289 {"cmr", HB_TAG('Q','I','N',' ')}, /* Mro-Khimi Chin */
290 {"cnb", HB_TAG('Q','I','N',' ')}, /* Chinbon Chin */
291 {"cnh", HB_TAG('Q','I','N',' ')}, /* Hakha Chin */
292 {"cnk", HB_TAG('Q','I','N',' ')}, /* Khumi Chin */
293 {"cnw", HB_TAG('Q','I','N',' ')}, /* Ngawn Chin */
294 {"cop", HB_TAG('C','O','P',' ')}, /* Coptic */
295 {"cpp", HB_TAG('C','P','P',' ')}, /* Creoles */
296 {"cr", HB_TAG('C','R','E',' ')}, /* Cree */
297 {"cre", HB_TAG('Y','C','R',' ')}, /* Y-Cree */
298 {"crh", HB_TAG('C','R','T',' ')}, /* Crimean Tatar */
299 {"crj", HB_TAG('E','C','R',' ')}, /* [Southern] East Cree */
300 {"crk", HB_TAG('W','C','R',' ')}, /* West-Cree */
301 {"crl", HB_TAG('E','C','R',' ')}, /* [Northern] East Cree */
302 {"crm", HB_TAG('M','C','R',' ')}, /* Moose Cree */
303 {"crx", HB_TAG('C','R','R',' ')}, /* Carrier */
304 {"cs", HB_TAG('C','S','Y',' ')}, /* Czech */
305 {"csb", HB_TAG('C','S','B',' ')}, /* Kashubian */
306 {"csh", HB_TAG('Q','I','N',' ')}, /* Asho Chin */
307 {"csy", HB_TAG('Q','I','N',' ')}, /* Siyin Chin */
308 {"ctd", HB_TAG('Q','I','N',' ')}, /* Tedim Chin */
309 {"ctg", HB_TAG('C','T','G',' ')}, /* Chittagonian */
310 {"cts", HB_TAG('B','I','K',' ')}, /* Northern Catanduanes Bikol */
311 {"cu", HB_TAG('C','S','L',' ')}, /* Church Slavic */
312 {"cuk", HB_TAG('C','U','K',' ')}, /* San Blas Kuna */
313 {"cv", HB_TAG('C','H','U',' ')}, /* Chuvash */
314 {"cwd", HB_TAG('D','C','R',' ')}, /* Woods Cree */
315 {"cy", HB_TAG('W','E','L',' ')}, /* Welsh */
316 {"czt", HB_TAG('Q','I','N',' ')}, /* Zotung Chin */
317 {"da", HB_TAG('D','A','N',' ')}, /* Danish */
318 {"dao", HB_TAG('Q','I','N',' ')}, /* Daai Chin */
319 {"dap", HB_TAG('N','I','S',' ')}, /* Nisi (India) */
320 {"dar", HB_TAG('D','A','R',' ')}, /* Dargwa */
321 {"dax", HB_TAG('D','A','X',' ')}, /* Dayi */
322 {"de", HB_TAG('D','E','U',' ')}, /* German */
323 {"dgo", HB_TAG('D','G','O',' ')}, /* Dogri */
324 {"dhd", HB_TAG('M','A','W',' ')}, /* Dhundari */
325 {"dhg", HB_TAG('D','H','G',' ')}, /* Dhangu */
326 {"din", HB_TAG('D','N','K',' ')}, /* Dinka [macrolanguage] */
327 {"diq", HB_TAG('D','I','Q',' ')}, /* Dimli */
328 {"dje", HB_TAG('D','J','R',' ')}, /* Zarma */
329 {"djr", HB_TAG('D','J','R','0')}, /* Djambarrpuyngu */
330 {"dng", HB_TAG('D','U','N',' ')}, /* Dungan */
331 {"dnj", HB_TAG('D','N','J',' ')}, /* Dan */
332 {"doi", HB_TAG('D','G','R',' ')}, /* Dogri [macrolanguage] */
333 {"dsb", HB_TAG('L','S','B',' ')}, /* Lower Sorbian */
334 {"duj", HB_TAG('D','U','J',' ')}, /* Dhuwal */
335 {"dv", HB_TAG('D','I','V',' ')}, /* Dhivehi/Divehi/Maldivian */
336 {"dyu", HB_TAG('J','U','L',' ')}, /* Jula */
337 {"dz", HB_TAG('D','Z','N',' ')}, /* Dzongkha */
338 {"ee", HB_TAG('E','W','E',' ')}, /* Ewe */
339 {"efi", HB_TAG('E','F','I',' ')}, /* Efik */
340 {"ekk", HB_TAG('E','T','I',' ')}, /* Standard Estonian */
341 {"el", HB_TAG('E','L','L',' ')}, /* Modern Greek (1453-) */
342 {"emk", HB_TAG('M','N','K',' ')}, /* Eastern Maninkakan */
343 {"en", HB_TAG('E','N','G',' ')}, /* English */
344 {"enf", HB_TAG('F','N','E',' ')}, /* Forest Nenets */
345 {"enh", HB_TAG('T','N','E',' ')}, /* Tundra Nenets */
346 {"eo", HB_TAG('N','T','O',' ')}, /* Esperanto */
347 {"eot", HB_TAG('B','T','I',' ')}, /* Beti (Côte d'Ivoire) */
348 {"es", HB_TAG('E','S','P',' ')}, /* Spanish */
349 {"esu", HB_TAG('E','S','U',' ')}, /* Central Yupik */
350 {"et", HB_TAG('E','T','I',' ')}, /* Estonian [macrolanguage] */
351 {"eu", HB_TAG('E','U','Q',' ')}, /* Basque */
352 {"eve", HB_TAG('E','V','N',' ')}, /* Even */
353 {"evn", HB_TAG('E','V','K',' ')}, /* Evenki */
354 {"fa", HB_TAG('F','A','R',' ')}, /* Persian [macrolanguage] */
355 {"fan", HB_TAG('F','A','N','0')}, /* Fang */
356 {"fat", HB_TAG('F','A','T',' ')}, /* Fanti */
357 {"ff", HB_TAG('F','U','L',' ')}, /* Fulah [macrolanguage] */
358 {"fi", HB_TAG('F','I','N',' ')}, /* Finnish */
359 {"fil", HB_TAG('P','I','L',' ')}, /* Filipino */
360 {"fj", HB_TAG('F','J','I',' ')}, /* Fijian */
361 {"flm", HB_TAG('H','A','L',' ')}, /* Halam/Falam Chin [retired ISO639 code] */
362 {"fo", HB_TAG('F','O','S',' ')}, /* Faroese */
363 {"fon", HB_TAG('F','O','N',' ')}, /* Fon */
364 {"fr", HB_TAG('F','R','A',' ')}, /* French */
365 {"frc", HB_TAG('F','R','C',' ')}, /* Cajun French */
366 {"frp", HB_TAG('F','R','P',' ')}, /* Arpitan/Francoprovençal */
367 {"fuf", HB_TAG('F','T','A',' ')}, /* Futa */
368 {"fur", HB_TAG('F','R','L',' ')}, /* Friulian */
369 {"fuv", HB_TAG('F','U','V',' ')}, /* Nigerian Fulfulde */
370 {"fy", HB_TAG('F','R','I',' ')}, /* Western Frisian */
371 {"ga", HB_TAG('I','R','I',' ')}, /* Irish */
372 {"gaa", HB_TAG('G','A','D',' ')}, /* Ga */
373 {"gag", HB_TAG('G','A','G',' ')}, /* Gagauz */
374 {"gbm", HB_TAG('G','A','W',' ')}, /* Garhwali */
375 {"gd", HB_TAG('G','A','E',' ')}, /* Scottish Gaelic */
376 {"gez", HB_TAG('G','E','Z',' ')}, /* Ge'ez */
377 {"ggo", HB_TAG('G','O','N',' ')}, /* Southern Gondi */
378 {"gih", HB_TAG('G','I','H',' ')}, /* Githabul */
379 {"gil", HB_TAG('G','I','L','0')}, /* Kiribati (Gilbertese) */
380 {"gkp", HB_TAG('G','K','P',' ')}, /* Kpelle (Guinea) */
381 {"gl", HB_TAG('G','A','L',' ')}, /* Galician */
382 {"gld", HB_TAG('N','A','N',' ')}, /* Nanai */
383 {"gle", HB_TAG('I','R','T',' ')}, /* Irish Traditional */
384 {"glk", HB_TAG('G','L','K',' ')}, /* Gilaki */
385 {"gn", HB_TAG('G','U','A',' ')}, /* Guarani [macrolanguage] */
386 {"gnn", HB_TAG('G','N','N',' ')}, /* Gumatj */
387 {"gno", HB_TAG('G','O','N',' ')}, /* Northern Gondi */
388 {"gog", HB_TAG('G','O','G',' ')}, /* Gogo */
389 {"gon", HB_TAG('G','O','N',' ')}, /* Gondi [macrolanguage] */
390 {"grt", HB_TAG('G','R','O',' ')}, /* Garo */
391 {"gru", HB_TAG('S','O','G',' ')}, /* Sodo Gurage */
392 {"gsw", HB_TAG('A','L','S',' ')}, /* Alsatian */
393 {"gu", HB_TAG('G','U','J',' ')}, /* Gujarati */
394 {"guc", HB_TAG('G','U','C',' ')}, /* Wayuu */
395 {"guf", HB_TAG('G','U','F',' ')}, /* Gupapuyngu */
396 {"guk", HB_TAG('G','M','Z',' ')}, /* Gumuz */
397 /*{"guk", HB_TAG('G','U','K',' ')},*/ /* Gumuz (in SIL fonts) */
398 {"guz", HB_TAG('G','U','Z',' ')}, /* Ekegusii/Gusii */
399 {"gv", HB_TAG('M','N','X',' ')}, /* Manx */
400 {"ha", HB_TAG('H','A','U',' ')}, /* Hausa */
401 {"har", HB_TAG('H','R','I',' ')}, /* Harari */
402 {"haw", HB_TAG('H','A','W',' ')}, /* Hawaiian */
403 {"hay", HB_TAG('H','A','Y',' ')}, /* Haya */
404 {"haz", HB_TAG('H','A','Z',' ')}, /* Hazaragi */
405 {"he", HB_TAG('I','W','R',' ')}, /* Hebrew */
406 {"hi", HB_TAG('H','I','N',' ')}, /* Hindi */
407 {"hil", HB_TAG('H','I','L',' ')}, /* Hiligaynon */
408 {"hlt", HB_TAG('Q','I','N',' ')}, /* Matu Chin */
409 {"hmn", HB_TAG('H','M','N',' ')}, /* Hmong */
410 {"hnd", HB_TAG('H','N','D',' ')}, /* [Southern] Hindko */
411 {"hne", HB_TAG('C','H','H',' ')}, /* Chattisgarhi */
412 {"hno", HB_TAG('H','N','D',' ')}, /* [Northern] Hindko */
413 {"ho", HB_TAG('H','M','O',' ')}, /* Hiri Motu */
414 {"hoc", HB_TAG('H','O',' ',' ')}, /* Ho */
415 {"hoj", HB_TAG('H','A','R',' ')}, /* Harauti */
416 {"hr", HB_TAG('H','R','V',' ')}, /* Croatian */
417 {"hsb", HB_TAG('U','S','B',' ')}, /* Upper Sorbian */
418 {"ht", HB_TAG('H','A','I',' ')}, /* Haitian/Haitian Creole */
419 {"hu", HB_TAG('H','U','N',' ')}, /* Hungarian */
420 {"hy", HB_TAG('H','Y','E',' ')}, /* Armenian */
421 {"hz", HB_TAG('H','E','R',' ')}, /* Herero */
422 {"ia", HB_TAG('I','N','A',' ')}, /* Interlingua (International Auxiliary Language Association) */
423 {"iba", HB_TAG('I','B','A',' ')}, /* Iban */
424 {"ibb", HB_TAG('I','B','B',' ')}, /* Ibibio */
425 {"id", HB_TAG('I','N','D',' ')}, /* Indonesian */
426 {"ie", HB_TAG('I','L','E',' ')}, /* Interlingue/Occidental */
427 {"ig", HB_TAG('I','B','O',' ')}, /* Igbo */
428 {"igb", HB_TAG('E','B','I',' ')}, /* Ebira */
429 {"ii", HB_TAG('Y','I','M',' ')}, /* Yi Modern */
430 {"ijc", HB_TAG('I','J','O',' ')}, /* Izon */
431 {"ijo", HB_TAG('I','J','O',' ')}, /* Ijo [family] */
432 {"ik", HB_TAG('I','P','K',' ')}, /* Inupiaq [macrolanguage] */
433 {"ilo", HB_TAG('I','L','O',' ')}, /* Ilokano */
434 {"inh", HB_TAG('I','N','G',' ')}, /* Ingush */
435 {"io", HB_TAG('I','D','O',' ')}, /* Ido */
436 {"is", HB_TAG('I','S','L',' ')}, /* Icelandic */
437 {"it", HB_TAG('I','T','A',' ')}, /* Italian */
438 {"iu", HB_TAG('I','N','U',' ')}, /* Inuktitut [macrolanguage] */
439 {"ja", HB_TAG('J','A','N',' ')}, /* Japanese */
440 {"jam", HB_TAG('J','A','M',' ')}, /* Jamaican Creole English */
441 {"jbo", HB_TAG('J','B','O',' ')}, /* Lojban */
442 {"jv", HB_TAG('J','A','V',' ')}, /* Javanese */
443 {"ka", HB_TAG('K','A','T',' ')}, /* Georgian */
444 {"kaa", HB_TAG('K','R','K',' ')}, /* Karakalpak */
445 {"kab", HB_TAG('K','A','B','0')}, /* Kabyle */
446 {"kam", HB_TAG('K','M','B',' ')}, /* Kamba (Kenya) */
447 {"kar", HB_TAG('K','R','N',' ')}, /* Karen [family] */
448 {"kat", HB_TAG('K','G','E',' ')}, /* Khutsuri Georgian */
449 {"kbd", HB_TAG('K','A','B',' ')}, /* Kabardian */
450 {"kde", HB_TAG('K','D','E',' ')}, /* Makonde */
451 {"kdr", HB_TAG('K','R','M',' ')}, /* Karaim */
452 {"kdt", HB_TAG('K','U','Y',' ')}, /* Kuy */
453 {"kea", HB_TAG('K','E','A',' ')}, /* Kabuverdianu (Crioulo) */
454 {"kek", HB_TAG('K','E','K',' ')}, /* Kekchi */
455 {"kex", HB_TAG('K','K','N',' ')}, /* Kokni */
456 {"kfa", HB_TAG('K','O','D',' ')}, /* Kodagu */
457 {"kfr", HB_TAG('K','A','C',' ')}, /* Kachchi */
458 {"kfx", HB_TAG('K','U','L',' ')}, /* Kulvi */
459 {"kfy", HB_TAG('K','M','N',' ')}, /* Kumaoni */
460 {"kg", HB_TAG('K','O','N',' ')}, /* Kongo [macrolanguage] */
461 {"kha", HB_TAG('K','S','I',' ')}, /* Khasi */
462 {"khb", HB_TAG('X','B','D',' ')}, /* Lü */
463 {"kht", HB_TAG('K','H','N',' ')}, /* Khamti (Microsoft fonts) */
464 /*{"kht", HB_TAG('K','H','T',' ')},*/ /* Khamti (OpenType spec and SIL fonts) */
465 {"khw", HB_TAG('K','H','W',' ')}, /* Khowar */
466 {"ki", HB_TAG('K','I','K',' ')}, /* Gikuyu/Kikuyu */
467 {"kiu", HB_TAG('K','I','U',' ')}, /* Kirmanjki */
468 {"kj", HB_TAG('K','U','A',' ')}, /* Kuanyama/Kwanyama */
469 {"kjd", HB_TAG('K','J','D',' ')}, /* Southern Kiwai */
470 {"kjh", HB_TAG('K','H','A',' ')}, /* Khakass */
471 {"kjp", HB_TAG('K','J','P',' ')}, /* Pwo Eastern Karen */
472 {"kk", HB_TAG('K','A','Z',' ')}, /* Kazakh */
473 {"kl", HB_TAG('G','R','N',' ')}, /* Kalaallisut */
474 {"kln", HB_TAG('K','A','L',' ')}, /* Kalenjin */
475 {"km", HB_TAG('K','H','M',' ')}, /* Central Khmer */
476 {"kmb", HB_TAG('M','B','N',' ')}, /* Kimbundu */
477 {"kmw", HB_TAG('K','M','O',' ')}, /* Komo (Democratic Republic of Congo) */
478 {"kn", HB_TAG('K','A','N',' ')}, /* Kannada */
479 {"knn", HB_TAG('K','O','K',' ')}, /* Konkani */
480 {"ko", HB_TAG('K','O','R',' ')}, /* Korean */
481 {"koi", HB_TAG('K','O','P',' ')}, /* Komi-Permyak */
482 {"kok", HB_TAG('K','O','K',' ')}, /* Konkani [macrolanguage] */
483 {"kon", HB_TAG('K','O','N','0')}, /* Kongo */
484 {"kos", HB_TAG('K','O','S',' ')}, /* Kosraean */
485 {"kpe", HB_TAG('K','P','L',' ')}, /* Kpelle [macrolanguage] */
486 {"kpv", HB_TAG('K','O','Z',' ')}, /* Komi-Zyrian */
487 {"kpy", HB_TAG('K','Y','K',' ')}, /* Koryak */
488 {"kqy", HB_TAG('K','R','T',' ')}, /* Koorete */
489 {"kr", HB_TAG('K','N','R',' ')}, /* Kanuri [macrolanguage] */
490 {"kri", HB_TAG('K','R','I',' ')}, /* Krio */
491 {"krl", HB_TAG('K','R','L',' ')}, /* Karelian */
492 {"kru", HB_TAG('K','U','U',' ')}, /* Kurukh */
493 {"ks", HB_TAG('K','S','H',' ')}, /* Kashmiri */
494 {"ksh", HB_TAG('K','S','H','0')}, /* Ripuarian, Kölsch */
495 /*{"ksw", HB_TAG('K','R','N',' ')},*/ /* S'gaw Karen (Microsoft fonts?) */
496 {"ksw", HB_TAG('K','S','W',' ')}, /* S'gaw Karen (OpenType spec and SIL fonts) */
497 {"ktb", HB_TAG('K','E','B',' ')}, /* Kebena */
498 {"ktu", HB_TAG('K','O','N',' ')}, /* Kikongo */
499 {"ku", HB_TAG('K','U','R',' ')}, /* Kurdish [macrolanguage] */
500 {"kum", HB_TAG('K','U','M',' ')}, /* Kumyk */
501 {"kv", HB_TAG('K','O','M',' ')}, /* Komi [macrolanguage] */
502 {"kvd", HB_TAG('K','U','I',' ')}, /* Kui (Indonesia) */
503 {"kw", HB_TAG('C','O','R',' ')}, /* Cornish */
504 {"kxc", HB_TAG('K','M','S',' ')}, /* Komso */
505 {"kxu", HB_TAG('K','U','I',' ')}, /* Kui (India) */
506 {"ky", HB_TAG('K','I','R',' ')}, /* Kirghiz/Kyrgyz */
507 {"kyu", HB_TAG('K','Y','U',' ')}, /* Western Kayah */
508 {"la", HB_TAG('L','A','T',' ')}, /* Latin */
509 {"lad", HB_TAG('J','U','D',' ')}, /* Ladino */
510 {"lb", HB_TAG('L','T','Z',' ')}, /* Luxembourgish */
511 {"lbe", HB_TAG('L','A','K',' ')}, /* Lak */
512 {"lbj", HB_TAG('L','D','K',' ')}, /* Ladakhi */
513 {"lez", HB_TAG('L','E','Z',' ')}, /* Lezgi */
514 {"lg", HB_TAG('L','U','G',' ')}, /* Ganda */
515 {"li", HB_TAG('L','I','M',' ')}, /* Limburgan/Limburger/Limburgish */
516 {"lif", HB_TAG('L','M','B',' ')}, /* Limbu */
517 {"lij", HB_TAG('L','I','J',' ')}, /* Ligurian */
518 {"lis", HB_TAG('L','I','S',' ')}, /* Lisu */
519 {"ljp", HB_TAG('L','J','P',' ')}, /* Lampung Api */
520 {"lki", HB_TAG('L','K','I',' ')}, /* Laki */
521 {"lld", HB_TAG('L','A','D',' ')}, /* Ladin */
522 {"lmn", HB_TAG('L','A','M',' ')}, /* Lambani */
523 {"lmo", HB_TAG('L','M','O',' ')}, /* Lombard */
524 {"ln", HB_TAG('L','I','N',' ')}, /* Lingala */
525 {"lo", HB_TAG('L','A','O',' ')}, /* Lao */
526 {"lom", HB_TAG('L','O','M',' ')}, /* Loma */
527 {"lrc", HB_TAG('L','R','C',' ')}, /* Northern Luri */
528 {"lt", HB_TAG('L','T','H',' ')}, /* Lithuanian */
529 {"lu", HB_TAG('L','U','B',' ')}, /* Luba-Katanga */
530 {"lua", HB_TAG('L','U','B',' ')}, /* Luba-Kasai */
531 {"luo", HB_TAG('L','U','O',' ')}, /* Luo (Kenya and Tanzania) */
532 {"lus", HB_TAG('M','I','Z',' ')}, /* Mizo */
533 {"luy", HB_TAG('L','U','H',' ')}, /* Luyia/Oluluyia [macrolanguage] */
534 {"luz", HB_TAG('L','R','C',' ')}, /* Southern Luri */
535 {"lv", HB_TAG('L','V','I',' ')}, /* Latvian */
536 {"lzz", HB_TAG('L','A','Z',' ')}, /* Laz */
537 {"mad", HB_TAG('M','A','D',' ')}, /* Madurese */
538 {"mag", HB_TAG('M','A','G',' ')}, /* Magahi */
539 {"mai", HB_TAG('M','T','H',' ')}, /* Maithili */
540 {"mak", HB_TAG('M','K','R',' ')}, /* Makasar */
541 {"mal", HB_TAG('M','A','L',' ')}, /* Malayalam */
542 {"mam", HB_TAG('M','A','M',' ')}, /* Mam */
543 {"man", HB_TAG('M','N','K',' ')}, /* Manding/Mandingo [macrolanguage] */
544 {"mdc", HB_TAG('M','L','E',' ')}, /* Male (Papua New Guinea) */
545 {"mdf", HB_TAG('M','O','K',' ')}, /* Moksha */
546 {"mdr", HB_TAG('M','D','R',' ')}, /* Mandar */
547 {"mdy", HB_TAG('M','L','E',' ')}, /* Male (Ethiopia) */
548 {"men", HB_TAG('M','D','E',' ')}, /* Mende (Sierra Leone) */
549 {"mer", HB_TAG('M','E','R',' ')}, /* Meru */
550 {"mfe", HB_TAG('M','F','E',' ')}, /* Morisyen */
551 {"mg", HB_TAG('M','L','G',' ')}, /* Malagasy [macrolanguage] */
552 {"mh", HB_TAG('M','A','H',' ')}, /* Marshallese */
553 {"mhr", HB_TAG('L','M','A',' ')}, /* Low Mari */
554 {"mi", HB_TAG('M','R','I',' ')}, /* Maori */
555 {"min", HB_TAG('M','I','N',' ')}, /* Minangkabau */
556 {"mk", HB_TAG('M','K','D',' ')}, /* Macedonian */
557 {"mku", HB_TAG('M','N','K',' ')}, /* Konyanka Maninka */
558 {"mkw", HB_TAG('M','K','W',' ')}, /* Kituba (Congo) */
559 {"ml", HB_TAG('M','L','R',' ')}, /* Malayalam */
560 {"mlq", HB_TAG('M','N','K',' ')}, /* Western Maninkakan */
561 {"mn", HB_TAG('M','N','G',' ')}, /* Mongolian [macrolanguage] */
562 {"mnc", HB_TAG('M','C','H',' ')}, /* Manchu */
563 {"mni", HB_TAG('M','N','I',' ')}, /* Manipuri */
564 {"mnk", HB_TAG('M','N','D',' ')}, /* Mandinka */
565 {"mns", HB_TAG('M','A','N',' ')}, /* Mansi */
566 {"mnw", HB_TAG('M','O','N',' ')}, /* Mon */
567 {"mo", HB_TAG('M','O','L',' ')}, /* Moldavian */
568 {"moh", HB_TAG('M','O','H',' ')}, /* Mohawk */
569 {"mos", HB_TAG('M','O','S',' ')}, /* Mossi */
570 {"mpe", HB_TAG('M','A','J',' ')}, /* Majang */
571 {"mr", HB_TAG('M','A','R',' ')}, /* Marathi */
572 {"mrh", HB_TAG('Q','I','N',' ')}, /* Mara Chin */
573 {"mrj", HB_TAG('H','M','A',' ')}, /* High Mari */
574 {"ms", HB_TAG('M','L','Y',' ')}, /* Malay [macrolanguage] */
575 {"msc", HB_TAG('M','N','K',' ')}, /* Sankaran Maninka */
576 {"mt", HB_TAG('M','T','S',' ')}, /* Maltese */
577 {"mtr", HB_TAG('M','A','W',' ')}, /* Mewari */
578 {"mus", HB_TAG('M','U','S',' ')}, /* Creek */
579 {"mve", HB_TAG('M','A','W',' ')}, /* Marwari (Pakistan) */
580 {"mwk", HB_TAG('M','N','K',' ')}, /* Kita Maninkakan */
581 {"mwl", HB_TAG('M','W','L',' ')}, /* Mirandese */
582 {"mwr", HB_TAG('M','A','W',' ')}, /* Marwari [macrolanguage] */
583 {"mww", HB_TAG('M','W','W',' ')}, /* Hmong Daw */
584 {"my", HB_TAG('B','R','M',' ')}, /* Burmese */
585 {"mym", HB_TAG('M','E','N',' ')}, /* Me'en */
586 {"myn", HB_TAG('M','Y','N',' ')}, /* Mayan */
587 {"myq", HB_TAG('M','N','K',' ')}, /* Forest Maninka (retired code) */
588 {"myv", HB_TAG('E','R','Z',' ')}, /* Erzya */
589 {"mzn", HB_TAG('M','Z','N',' ')}, /* Mazanderani */
590 {"na", HB_TAG('N','A','U',' ')}, /* Nauru */
591 {"nag", HB_TAG('N','A','G',' ')}, /* Naga-Assamese */
592 {"nah", HB_TAG('N','A','H',' ')}, /* Nahuatl [family] */
593 {"nap", HB_TAG('N','A','P',' ')}, /* Neapolitan */
594 {"nb", HB_TAG('N','O','R',' ')}, /* Norwegian Bokmål */
595 {"nco", HB_TAG('S','I','B',' ')}, /* Sibe */
596 {"nd", HB_TAG('N','D','B',' ')}, /* [North] Ndebele */
597 {"ndc", HB_TAG('N','D','C',' ')}, /* Ndau */
598 {"nds", HB_TAG('N','D','S',' ')}, /* Low German/Low Saxon */
599 {"ne", HB_TAG('N','E','P',' ')}, /* Nepali */
600 {"new", HB_TAG('N','E','W',' ')}, /* Newari */
601 {"ng", HB_TAG('N','D','G',' ')}, /* Ndonga */
602 {"nga", HB_TAG('N','G','A',' ')}, /* Ngabaka */
603 {"ngl", HB_TAG('L','M','W',' ')}, /* Lomwe */
604 {"ngo", HB_TAG('S','X','T',' ')}, /* Sutu */
605 {"niu", HB_TAG('N','I','U',' ')}, /* Niuean */
606 {"niv", HB_TAG('G','I','L',' ')}, /* Gilyak */
607 {"nl", HB_TAG('N','L','D',' ')}, /* Dutch */
608 {"nn", HB_TAG('N','Y','N',' ')}, /* Norwegian Nynorsk */
609 {"no", HB_TAG('N','O','R',' ')}, /* Norwegian [macrolanguage] */
610 {"nod", HB_TAG('N','T','A',' ')}, /* Northern Thai */
611 {"noe", HB_TAG('N','O','E',' ')}, /* Nimadi */
612 {"nog", HB_TAG('N','O','G',' ')}, /* Nogai */
613 {"nov", HB_TAG('N','O','V',' ')}, /* Novial */
614 {"nqo", HB_TAG('N','K','O',' ')}, /* N'Ko */
615 {"nr", HB_TAG('N','D','B',' ')}, /* [South] Ndebele */
616 {"nsk", HB_TAG('N','A','S',' ')}, /* Naskapi */
617 {"nso", HB_TAG('S','O','T',' ')}, /* [Northern] Sotho */
618 {"nv", HB_TAG('N','A','V',' ')}, /* Navajo */
619 {"ny", HB_TAG('C','H','I',' ')}, /* Chewa/Chichwa/Nyanja */
620 {"nym", HB_TAG('N','Y','M',' ')}, /* Nyamwezi */
621 {"nyn", HB_TAG('N','K','L',' ')}, /* Nyankole */
622 {"oc", HB_TAG('O','C','I',' ')}, /* Occitan (post 1500) */
623 {"oj", HB_TAG('O','J','B',' ')}, /* Ojibwa [macrolanguage] */
624 {"ojs", HB_TAG('O','C','R',' ')}, /* Oji-Cree */
625 {"okm", HB_TAG('K','O','H',' ')}, /* Korean Old Hangul */
626 {"om", HB_TAG('O','R','O',' ')}, /* Oromo [macrolanguage] */
627 {"or", HB_TAG('O','R','I',' ')}, /* Oriya */
628 {"os", HB_TAG('O','S','S',' ')}, /* Ossetian */
629 {"pa", HB_TAG('P','A','N',' ')}, /* Panjabi */
630 {"pag", HB_TAG('P','A','G',' ')}, /* Pangasinan */
631 {"pam", HB_TAG('P','A','M',' ')}, /* Kapampangan/Pampanga */
632 {"pap", HB_TAG('P','A','P','0')}, /* Papiamento */
633 {"pau", HB_TAG('P','A','U',' ')}, /* Palauan */
634 {"pcc", HB_TAG('P','C','C',' ')}, /* Bouyei */
635 {"pcd", HB_TAG('P','C','D',' ')}, /* Picard */
636 {"pce", HB_TAG('P','L','G',' ')}, /* [Ruching] Palaung */
637 {"pck", HB_TAG('Q','I','N',' ')}, /* Paite Chin */
638 {"pdc", HB_TAG('P','D','C',' ')}, /* Pennsylvania German */
639 {"pes", HB_TAG('F','A','R',' ')}, /* Iranian Persian */
640 {"phk", HB_TAG('P','H','K',' ')}, /* Phake */
641 {"pi", HB_TAG('P','A','L',' ')}, /* Pali */
642 {"pih", HB_TAG('P','I','H',' ')}, /* Pitcairn-Norfolk */
643 {"pl", HB_TAG('P','L','K',' ')}, /* Polish */
644 {"pll", HB_TAG('P','L','G',' ')}, /* [Shwe] Palaung */
645 {"plp", HB_TAG('P','A','P',' ')}, /* Palpa */
646 {"pms", HB_TAG('P','M','S',' ')}, /* Piemontese */
647 {"pnb", HB_TAG('P','N','B',' ')}, /* Western Panjabi */
648 {"poh", HB_TAG('P','O','H',' ')}, /* Pocomchi */
649 {"pon", HB_TAG('P','O','N',' ')}, /* Pohnpeian */
650 {"prs", HB_TAG('D','R','I',' ')}, /* Afghan Persian/Dari */
651 {"ps", HB_TAG('P','A','S',' ')}, /* Pashto/Pushto [macrolanguage] */
652 {"pt", HB_TAG('P','T','G',' ')}, /* Portuguese */
653 {"pwo", HB_TAG('P','W','O',' ')}, /* Pwo Western Karen */
654 {"qu", HB_TAG('Q','U','Z',' ')}, /* Quechua [macrolanguage] */
655 {"quc", HB_TAG('Q','U','C',' ')}, /* K'iche'/Quiché */
656 {"quh", HB_TAG('Q','U','H',' ')}, /* Quechua (Bolivia) */
657 {"quz", HB_TAG('Q','U','Z',' ')}, /* Cusco Quechua */
658 {"qvi", HB_TAG('Q','V','I',' ')}, /* Quechua (Ecuador) */
659 {"qwh", HB_TAG('Q','W','H',' ')}, /* Quechua (Peru) */
660 {"raj", HB_TAG('R','A','J',' ')}, /* Rajasthani [macrolanguage] */
661 {"rar", HB_TAG('R','A','R',' ')}, /* Rarotongan */
662 {"rbb", HB_TAG('P','L','G',' ')}, /* Rumai Palaung */
663 {"rej", HB_TAG('R','E','J',' ')}, /* Rejang */
664 {"ria", HB_TAG('R','I','A',' ')}, /* Riang (India) */
665 {"rif", HB_TAG('R','I','F',' ')}, /* Tarifit */
666 {"ril", HB_TAG('R','I','A',' ')}, /* Riang (Myanmar) */
667 {"rit", HB_TAG('R','I','T',' ')}, /* Ritarungo */
668 {"rki", HB_TAG('A','R','K',' ')}, /* Rakhine */
669 {"rkw", HB_TAG('R','K','W',' ')}, /* Arakwal */
670 {"rm", HB_TAG('R','M','S',' ')}, /* Romansh */
671 {"rmy", HB_TAG('R','M','Y',' ')}, /* Vlax Romani */
672 {"rn", HB_TAG('R','U','N',' ')}, /* Rundi */
673 {"ro", HB_TAG('R','O','M',' ')}, /* Romanian */
674 {"rom", HB_TAG('R','O','Y',' ')}, /* Romany [macrolanguage] */
675 {"rtm", HB_TAG('R','T','M',' ')}, /* Rotuman */
676 {"ru", HB_TAG('R','U','S',' ')}, /* Russian */
677 {"rue", HB_TAG('R','S','Y',' ')}, /* Rusyn */
678 {"rup", HB_TAG('R','U','P',' ')}, /* Aromanian/Arumanian/Macedo-Romanian */
679 {"rw", HB_TAG('R','U','A',' ')}, /* Kinyarwanda */
680 {"rwr", HB_TAG('M','A','W',' ')}, /* Marwari (India) */
681 {"sa", HB_TAG('S','A','N',' ')}, /* Sanskrit */
682 {"sah", HB_TAG('Y','A','K',' ')}, /* Yakut */
683 {"sam", HB_TAG('P','A','A',' ')}, /* Palestinian Aramaic */
684 {"sas", HB_TAG('S','A','S',' ')}, /* Sasak */
685 {"sat", HB_TAG('S','A','T',' ')}, /* Santali */
686 {"sc", HB_TAG('S','R','D',' ')}, /* Sardinian [macrolanguage] */
687 {"sck", HB_TAG('S','A','D',' ')}, /* Sadri */
688 {"scn", HB_TAG('S','C','N',' ')}, /* Sicilian */
689 {"sco", HB_TAG('S','C','O',' ')}, /* Scots */
690 {"scs", HB_TAG('S','L','A',' ')}, /* [North] Slavey */
691 {"sd", HB_TAG('S','N','D',' ')}, /* Sindhi */
692 {"se", HB_TAG('N','S','M',' ')}, /* Northern Sami */
693 {"seh", HB_TAG('S','N','A',' ')}, /* Sena */
694 {"sel", HB_TAG('S','E','L',' ')}, /* Selkup */
695 {"sez", HB_TAG('Q','I','N',' ')}, /* Senthang Chin */
696 {"sg", HB_TAG('S','G','O',' ')}, /* Sango */
697 {"sga", HB_TAG('S','G','A',' ')}, /* Old Irish (to 900) */
698 {"sgs", HB_TAG('S','G','S',' ')}, /* Samogitian */
699 {"sgw", HB_TAG('C','H','G',' ')}, /* Sebat Bet Gurage */
700 /*{"sgw", HB_TAG('S','G','W',' ')},*/ /* Sebat Bet Gurage (in SIL fonts) */
701 {"shi", HB_TAG('S','H','I',' ')}, /* Tachelhit */
702 {"shn", HB_TAG('S','H','N',' ')}, /* Shan */
703 {"si", HB_TAG('S','N','H',' ')}, /* Sinhala */
704 {"sid", HB_TAG('S','I','D',' ')}, /* Sidamo */
705 {"sjd", HB_TAG('K','S','M',' ')}, /* Kildin Sami */
706 {"sk", HB_TAG('S','K','Y',' ')}, /* Slovak */
707 {"skr", HB_TAG('S','R','K',' ')}, /* Seraiki */
708 {"sl", HB_TAG('S','L','V',' ')}, /* Slovenian */
709 {"sm", HB_TAG('S','M','O',' ')}, /* Samoan */
710 {"sma", HB_TAG('S','S','M',' ')}, /* Southern Sami */
711 {"smj", HB_TAG('L','S','M',' ')}, /* Lule Sami */
712 {"smn", HB_TAG('I','S','M',' ')}, /* Inari Sami */
713 {"sms", HB_TAG('S','K','S',' ')}, /* Skolt Sami */
714 {"sn", HB_TAG('S','N','A','0')}, /* Shona */
715 {"snk", HB_TAG('S','N','K',' ')}, /* Soninke */
716 {"so", HB_TAG('S','M','L',' ')}, /* Somali */
717 {"sop", HB_TAG('S','O','P',' ')}, /* Songe */
718 {"sq", HB_TAG('S','Q','I',' ')}, /* Albanian [macrolanguage] */
719 {"sr", HB_TAG('S','R','B',' ')}, /* Serbian */
720 {"srr", HB_TAG('S','R','R',' ')}, /* Serer */
721 {"ss", HB_TAG('S','W','Z',' ')}, /* Swati */
722 {"st", HB_TAG('S','O','T',' ')}, /* [Southern] Sotho */
723 {"stq", HB_TAG('S','T','Q',' ')}, /* Saterfriesisch */
724 {"stv", HB_TAG('S','I','G',' ')}, /* Silt'e */
725 {"su", HB_TAG('S','U','N',' ')}, /* Sundanese */
726 {"suk", HB_TAG('S','U','K',' ')}, /* Sukama */
727 {"suq", HB_TAG('S','U','R',' ')}, /* Suri */
728 {"sv", HB_TAG('S','V','E',' ')}, /* Swedish */
729 {"sva", HB_TAG('S','V','A',' ')}, /* Svan */
730 {"sw", HB_TAG('S','W','K',' ')}, /* Swahili [macrolanguage] */
731 {"swb", HB_TAG('C','M','R',' ')}, /* Comorian */
732 {"swh", HB_TAG('S','W','K',' ')}, /* Kiswahili/Swahili */
733 {"swv", HB_TAG('M','A','W',' ')}, /* Shekhawati */
734 {"sxu", HB_TAG('S','X','U',' ')}, /* Upper Saxon */
735 {"syc", HB_TAG('S','Y','R',' ')}, /* Classical Syriac */
736 {"syl", HB_TAG('S','Y','L',' ')}, /* Sylheti */
737 {"syr", HB_TAG('S','Y','R',' ')}, /* Syriac [macrolanguage] */
738 {"szl", HB_TAG('S','Z','L',' ')}, /* Silesian */
739 {"ta", HB_TAG('T','A','M',' ')}, /* Tamil */
740 {"tab", HB_TAG('T','A','B',' ')}, /* Tabasaran */
741 {"tcp", HB_TAG('Q','I','N',' ')}, /* Tawr Chin */
742 {"tcy", HB_TAG('T','U','L',' ')}, /* Tulu */
743 {"tcz", HB_TAG('Q','I','N',' ')}, /* Thado Chin */
744 {"tdd", HB_TAG('T','D','D',' ')}, /* Tai Nüa */
745 {"te", HB_TAG('T','E','L',' ')}, /* Telugu */
746 {"tem", HB_TAG('T','M','N',' ')}, /* Temne */
747 {"tet", HB_TAG('T','E','T',' ')}, /* Tetum */
748 {"tg", HB_TAG('T','A','J',' ')}, /* Tajik */
749 {"th", HB_TAG('T','H','A',' ')}, /* Thai */
750 {"ti", HB_TAG('T','G','Y',' ')}, /* Tigrinya */
751 {"tig", HB_TAG('T','G','R',' ')}, /* Tigre */
752 {"tiv", HB_TAG('T','I','V',' ')}, /* Tiv */
753 {"tk", HB_TAG('T','K','M',' ')}, /* Turkmen */
754 {"tl", HB_TAG('T','G','L',' ')}, /* Tagalog */
755 {"tmh", HB_TAG('T','M','H',' ')}, /* Tamashek */
756 {"tn", HB_TAG('T','N','A',' ')}, /* Tswana */
757 {"to", HB_TAG('T','G','N',' ')}, /* Tonga (Tonga Islands) */
758 {"tod", HB_TAG('T','O','D','0')}, /* Toma */
759 {"toi", HB_TAG('T','N','G',' ')}, /* Tonga */
760 {"tpi", HB_TAG('T','P','I',' ')}, /* Tok Pisin */
761 {"tr", HB_TAG('T','R','K',' ')}, /* Turkish */
762 {"tru", HB_TAG('T','U','A',' ')}, /* Turoyo Aramaic */
763 {"ts", HB_TAG('T','S','G',' ')}, /* Tsonga */
764 {"tt", HB_TAG('T','A','T',' ')}, /* Tatar */
765 {"tum", HB_TAG('T','U','M',' ')}, /* Tumbuka */
766 {"tvl", HB_TAG('T','V','L',' ')}, /* Tuvalu */
767 {"tw", HB_TAG('T','W','I',' ')}, /* Twi */
768 {"ty", HB_TAG('T','H','T',' ')}, /* Tahitian */
769 {"tyv", HB_TAG('T','U','V',' ')}, /* Tuvin */
770 {"tyz", HB_TAG('T','Y','Z',' ')}, /* Tày */
771 {"tzm", HB_TAG('T','Z','M',' ')}, /* Central Atlas Tamazight */
772 {"tzo", HB_TAG('T','Z','O',' ')}, /* Tzotzil */
773 {"udm", HB_TAG('U','D','M',' ')}, /* Udmurt */
774 {"ug", HB_TAG('U','Y','G',' ')}, /* Uighur */
775 {"uk", HB_TAG('U','K','R',' ')}, /* Ukrainian */
776 {"umb", HB_TAG('U','M','B',' ')}, /* Umbundu */
777 {"unr", HB_TAG('M','U','N',' ')}, /* Mundari */
778 {"ur", HB_TAG('U','R','D',' ')}, /* Urdu */
779 {"uz", HB_TAG('U','Z','B',' ')}, /* Uzbek [macrolanguage] */
780 {"uzn", HB_TAG('U','Z','B',' ')}, /* Northern Uzbek */
781 {"uzs", HB_TAG('U','Z','B',' ')}, /* Southern Uzbek */
782 {"ve", HB_TAG('V','E','N',' ')}, /* Venda */
783 {"vec", HB_TAG('V','E','C',' ')}, /* Venetian */
784 {"vi", HB_TAG('V','I','T',' ')}, /* Vietnamese */
785 {"vls", HB_TAG('F','L','E',' ')}, /* Vlaams */
786 {"vmw", HB_TAG('M','A','K',' ')}, /* Makhuwa */
787 {"vo", HB_TAG('V','O','L',' ')}, /* Volapük */
788 {"vro", HB_TAG('V','R','O',' ')}, /* Võro */
789 {"wa", HB_TAG('W','L','N',' ')}, /* Walloon */
790 {"war", HB_TAG('W','A','R',' ')}, /* Waray (Philippines) */
791 {"wbm", HB_TAG('W','A',' ',' ')}, /* Wa */
792 {"wbr", HB_TAG('W','A','G',' ')}, /* Wagdi */
793 {"wle", HB_TAG('S','I','G',' ')}, /* Wolane */
794 {"wo", HB_TAG('W','L','F',' ')}, /* Wolof */
795 {"wry", HB_TAG('M','A','W',' ')}, /* Merwari */
796 {"wtm", HB_TAG('W','T','M',' ')}, /* Mewati */
797 {"xal", HB_TAG('K','L','M',' ')}, /* Kalmyk */
798 {"xan", HB_TAG('S','E','K',' ')}, /* Sekota */
799 {"xh", HB_TAG('X','H','S',' ')}, /* Xhosa */
800 {"xjb", HB_TAG('X','J','B',' ')}, /* Minjangbal */
801 {"xog", HB_TAG('X','O','G',' ')}, /* Soga */
802 {"xom", HB_TAG('K','M','O',' ')}, /* Komo (Sudan) */
803 {"xpe", HB_TAG('X','P','E',' ')}, /* Kpelle (Liberia) */
804 {"xsl", HB_TAG('S','S','L',' ')}, /* South Slavey */
805 {"xst", HB_TAG('S','I','G',' ')}, /* Silt'e (retired code) */
806 {"xwo", HB_TAG('T','O','D',' ')}, /* Written Oirat (Todo) */
807 {"yao", HB_TAG('Y','A','O',' ')}, /* Yao */
808 {"yap", HB_TAG('Y','A','P',' ')}, /* Yapese */
809 {"yi", HB_TAG('J','I','I',' ')}, /* Yiddish [macrolanguage] */
810 {"yo", HB_TAG('Y','B','A',' ')}, /* Yoruba */
811 {"yos", HB_TAG('Q','I','N',' ')}, /* Yos, deprecated by IANA in favor of Zou [zom] */
812 {"yso", HB_TAG('N','I','S',' ')}, /* Nisi (China) */
813 {"za", HB_TAG('Z','H','A',' ')}, /* Chuang/Zhuang [macrolanguage] */
814 {"zea", HB_TAG('Z','E','A',' ')}, /* Zeeuws */
815 {"zgh", HB_TAG('Z','G','H',' ')}, /* Standard Morrocan Tamazigh */
816 {"zne", HB_TAG('Z','N','D',' ')}, /* Zande */
817 {"zom", HB_TAG('Q','I','N',' ')}, /* Zou */
818 {"zu", HB_TAG('Z','U','L',' ')}, /* Zulu */
819 {"zum", HB_TAG('L','R','C',' ')}, /* Kumzari */
820 {"zza", HB_TAG('Z','Z','A',' ')}, /* Zazaki */
821
822 /* The corresponding languages IDs for the following IDs are unclear,
823 * overlap, or are architecturally weird. Needs more research. */
824
825 /*{"chp", HB_TAG('S','A','Y',' ')},*/ /* Sayisi */
826 /*{"cwd", HB_TAG('T','C','R',' ')},*/ /* TH-Cree */
827 /*{"emk", HB_TAG('E','M','K',' ')},*/ /* Eastern Maninkakan */
828 /*{"krc", HB_TAG('B','A','L',' ')},*/ /* Balkar */
829 /*{"??", HB_TAG('B','C','R',' ')},*/ /* Bible Cree */
830 /*{"zh?", HB_TAG('C','H','N',' ')},*/ /* Chinese (seen in Microsoft fonts) */
831 /*{"ar-Syrc?", HB_TAG('G','A','R',' ')},*/ /* Garshuni */
832 /*{"hy?", HB_TAG('H','Y','E','0')},*/ /* Armenian East (ISO 639-3 hye according to Microsoft, but that’s equivalent to ISO 639-1 hy) */
833 /*{"ga-Latg?/" HB_TAG('I','R','T',' ')},*/ /* Irish Traditional */
834 /*{"krc", HB_TAG('K','A','R',' ')},*/ /* Karachay */
835 /*{"ka-Geok?", HB_TAG('K','G','E',' ')},*/ /* Khutsuri Georgian */
836 /*{"kca", HB_TAG('K','H','K',' ')},*/ /* Khanty-Kazim */
837 /*{"kca", HB_TAG('K','H','S',' ')},*/ /* Khanty-Shurishkar */
838 /*{"kca", HB_TAG('K','H','V',' ')},*/ /* Khanty-Vakhi */
839 /*{"kqs, kss", HB_TAG('K','I','S',' ')},*/ /* Kisii */
840 /*{"lua", HB_TAG('L','U','A',' ')},*/ /* Luba-Lulua */
841 /*{"mlq", HB_TAG('M','L','N',' ')},*/ /* Malinke */
842 /*{"nso", HB_TAG('N','S','O',' ')},*/ /* Sotho, Northern */
843 /*{"??", HB_TAG('M','A','L',' ')},*/ /* Malayalam Traditional */
844 /*{"csw", HB_TAG('N','C','R',' ')},*/ /* N-Cree */
845 /*{"csw", HB_TAG('N','H','C',' ')},*/ /* Norway House Cree */
846 /*{"el-polyton", HB_TAG('P','G','R',' ')},*/ /* Polytonic Greek */
847 /*{"bgr, cnh, cnw, czt, sez, tcp, csy, ctd, flm, pck, tcz, zom, cmr, dao, hlt, cka, cnk, mrh, mwg, cbl, cnb, csh", HB_TAG('Q','I','N',' ')},*/ /* Chin */
848 /*{"??", HB_TAG('Y','I','C',' ')},*/ /* Yi Classic */
849 /*{"zh-Latn-pinyin", HB_TAG('Z','H','P',' ')},*/ /* Chinese Phonetic */
850 };
851
852 typedef struct {
853 char language[11];
854 hb_tag_t tag;
855 } LangTagLong;
856 static const LangTagLong ot_languages_zh[] = {
857 /* Store longest-first, if one is a prefix of another. */
858 {"zh-cn", HB_TAG('Z','H','S',' ')}, /* Chinese (China) */
859 {"zh-hk", HB_TAG('Z','H','H',' ')}, /* Chinese (Hong Kong) */
860 {"zh-mo", HB_TAG('Z','H','H',' ')}, /* Chinese (Macao) */
861 {"zh-sg", HB_TAG('Z','H','S',' ')}, /* Chinese (Singapore) */
862 {"zh-tw", HB_TAG('Z','H','T',' ')}, /* Chinese (Taiwan) */
863 {"zh-hans", HB_TAG('Z','H','S',' ')}, /* Chinese (Simplified) */
864 {"zh-hant-hk",HB_TAG('Z','H','H',' ')}, /* Chinese (Hong Kong) */
865 {"zh-hant-mo",HB_TAG('Z','H','H',' ')}, /* Chinese (Macao) */
866 {"zh-hant", HB_TAG('Z','H','T',' ')}, /* Chinese (Traditional) */
867 };
868
869 static int
lang_compare_first_component(const char * a,const char * b)870 lang_compare_first_component (const char *a,
871 const char *b)
872 {
873 unsigned int da, db;
874 const char *p;
875
876 p = strchr (a, '-');
877 da = p ? (unsigned int) (p - a) : strlen (a);
878
879 p = strchr (b, '-');
880 db = p ? (unsigned int) (p - b) : strlen (b);
881
882 return strncmp (a, b, MAX (da, db));
883 }
884
885 static hb_bool_t
lang_matches(const char * lang_str,const char * spec)886 lang_matches (const char *lang_str, const char *spec)
887 {
888 unsigned int len = strlen (spec);
889
890 return strncmp (lang_str, spec, len) == 0 &&
891 (lang_str[len] == '\0' || lang_str[len] == '-');
892 }
893
894 hb_tag_t
hb_ot_tag_from_language(hb_language_t language)895 hb_ot_tag_from_language (hb_language_t language)
896 {
897 const char *lang_str, *s;
898
899 if (language == HB_LANGUAGE_INVALID)
900 return HB_OT_TAG_DEFAULT_LANGUAGE;
901
902 lang_str = hb_language_to_string (language);
903
904 s = strstr (lang_str, "x-hbot");
905 if (s) {
906 char tag[4];
907 int i;
908 s += 6;
909 for (i = 0; i < 4 && ISALPHA (s[i]); i++)
910 tag[i] = TOUPPER (s[i]);
911 if (i) {
912 for (; i < 4; i++)
913 tag[i] = ' ';
914 return HB_TAG_CHAR4 (tag);
915 }
916 }
917
918 /*
919 * "fonipa" is a variant tag in BCP-47, meaning the International Phonetic Alphabet.
920 * It can be applied to any language.
921 */
922 if (strstr (lang_str, "-fonipa")) {
923 return HB_TAG('I','P','P','H'); /* Phonetic transcription—IPA conventions */
924 }
925
926 /*
927 * "fonnapa" is a variant tag in BCP-47, meaning the North American Phonetic Alphabet
928 * also known as Americanist Phonetic Notation. It can be applied to any language.
929 */
930 if (strstr (lang_str, "-fonnapa")) {
931 return HB_TAG('A','P','P','H'); /* Phonetic transcription—Americanist conventions */
932 }
933
934 /*
935 * "Syre" is a BCP-47 script tag, meaning the Estrangela variant of the Syriac script.
936 * It can be applied to any language.
937 */
938 if (strstr (lang_str, "-syre")) {
939 return HB_TAG('S','Y','R','E'); /* Estrangela Syriac */
940 }
941
942 /*
943 * "Syrj" is a BCP-47 script tag, meaning the Western variant of the Syriac script.
944 * It can be applied to any language.
945 */
946 if (strstr (lang_str, "-syrj")) {
947 return HB_TAG('S','Y','R','J'); /* Western Syriac */
948 }
949
950 /*
951 * "Syrn" is a BCP-47 script tag, meaning the Eastern variant of the Syriac script.
952 * It can be applied to any language.
953 */
954 if (strstr (lang_str, "-syrn")) {
955 return HB_TAG('S','Y','R','N'); /* Eastern Syriac */
956 }
957
958 /* Find a language matching in the first component */
959 {
960 const LangTag *lang_tag;
961 lang_tag = (LangTag *) bsearch (lang_str, ot_languages,
962 ARRAY_LENGTH (ot_languages), sizeof (LangTag),
963 (hb_compare_func_t) lang_compare_first_component);
964 if (lang_tag)
965 return lang_tag->tag;
966 }
967
968 /* Otherwise, check the Chinese ones */
969 if (0 == lang_compare_first_component (lang_str, "zh"))
970 {
971 unsigned int i;
972
973 for (i = 0; i < ARRAY_LENGTH (ot_languages_zh); i++)
974 {
975 const LangTagLong *lang_tag;
976 lang_tag = &ot_languages_zh[i];
977 if (lang_matches (lang_str, lang_tag->language))
978 return lang_tag->tag;
979 }
980
981 /* Otherwise just return 'ZHS ' */
982 return HB_TAG('Z','H','S',' ');
983 }
984
985 s = strchr (lang_str, '-');
986 if (!s)
987 s = lang_str + strlen (lang_str);
988 if (s - lang_str == 3) {
989 /* Assume it's ISO-639-3 and upper-case and use it. */
990 return hb_tag_from_string (lang_str, s - lang_str) & ~0x20202000u;
991 }
992
993 return HB_OT_TAG_DEFAULT_LANGUAGE;
994 }
995
996 /**
997 * hb_ot_tag_to_language:
998 *
999 *
1000 *
1001 * Return value: (transfer none):
1002 *
1003 * Since: 0.9.2
1004 **/
1005 hb_language_t
hb_ot_tag_to_language(hb_tag_t tag)1006 hb_ot_tag_to_language (hb_tag_t tag)
1007 {
1008 unsigned int i;
1009
1010 if (tag == HB_OT_TAG_DEFAULT_LANGUAGE)
1011 return NULL;
1012
1013 /* struct LangTag has only room for 3-letter language tags. */
1014 switch (tag) {
1015 case HB_TAG('A','P','P','H'): /* Phonetic transcription—Americanist conventions */
1016 return hb_language_from_string ("und-fonnapa", -1);
1017 case HB_TAG('I','P','P','H'): /* Phonetic transcription—IPA conventions */
1018 return hb_language_from_string ("und-fonipa", -1);
1019 case HB_TAG('S','Y','R',' '): /* Syriac [macrolanguage] */
1020 return hb_language_from_string ("syr", -1);
1021 case HB_TAG('S','Y','R','E'): /* Estrangela Syriac */
1022 return hb_language_from_string ("und-Syre", -1);
1023 case HB_TAG('S','Y','R','J'): /* Western Syriac */
1024 return hb_language_from_string ("und-Syrj", -1);
1025 case HB_TAG('S','Y','R','N'): /* Eastern Syriac */
1026 return hb_language_from_string ("und-Syrn", -1);
1027 }
1028
1029 for (i = 0; i < ARRAY_LENGTH (ot_languages); i++)
1030 if (ot_languages[i].tag == tag)
1031 return hb_language_from_string (ot_languages[i].language, -1);
1032
1033 /* If tag starts with ZH, it's Chinese */
1034 if ((tag & 0xFFFF0000u) == 0x5A480000u) {
1035 switch (tag) {
1036 case HB_TAG('Z','H','H',' '): return hb_language_from_string ("zh-hk", -1); /* Hong Kong */
1037 case HB_TAG('Z','H','S',' '): return hb_language_from_string ("zh-Hans", -1); /* Simplified */
1038 case HB_TAG('Z','H','T',' '): return hb_language_from_string ("zh-Hant", -1); /* Traditional */
1039 default: break; /* Fall through */
1040 }
1041 }
1042
1043 /* Else return a custom language in the form of "x-hbotABCD" */
1044 {
1045 unsigned char buf[11] = "x-hbot";
1046 buf[6] = tag >> 24;
1047 buf[7] = (tag >> 16) & 0xFF;
1048 buf[8] = (tag >> 8) & 0xFF;
1049 buf[9] = tag & 0xFF;
1050 if (buf[9] == 0x20)
1051 buf[9] = '\0';
1052 buf[10] = '\0';
1053 return hb_language_from_string ((char *) buf, -1);
1054 }
1055 }
1056
1057 #ifdef MAIN
1058 static inline void
test_langs_sorted(void)1059 test_langs_sorted (void)
1060 {
1061 for (unsigned int i = 1; i < ARRAY_LENGTH (ot_languages); i++)
1062 {
1063 int c = lang_compare_first_component (ot_languages[i-1].language, ot_languages[i].language);
1064 if (c >= 0)
1065 {
1066 fprintf (stderr, "ot_languages not sorted at index %d: %s %d %s\n",
1067 i, ot_languages[i-1].language, c, ot_languages[i].language);
1068 abort();
1069 }
1070 }
1071 }
1072
1073 int
main(void)1074 main (void)
1075 {
1076 test_langs_sorted ();
1077 return 0;
1078 }
1079
1080 #endif
1081