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 #include <string.h>
32
33
34
35 /* hb_script_t */
36
37 static hb_tag_t
38 hb_ot_old_tag_from_script (hb_script_t script)
39 {
40 /* This seems to be accurate as of end of 2012. */
41
42 switch ((hb_tag_t) script) {
43 case HB_SCRIPT_INVALID: return HB_OT_TAG_DEFAULT_SCRIPT;
44
45 /* KATAKANA and HIRAGANA both map to 'kana' */
46 case HB_SCRIPT_HIRAGANA: return HB_TAG('k','a','n','a');
47
48 /* Spaces at the end are preserved, unlike ISO 15924 */
49 case HB_SCRIPT_LAO: return HB_TAG('l','a','o',' ');
50 case HB_SCRIPT_YI: return HB_TAG('y','i',' ',' ');
51 /* Unicode-5.0 additions */
52 case HB_SCRIPT_NKO: return HB_TAG('n','k','o',' ');
53 /* Unicode-5.1 additions */
54 case HB_SCRIPT_VAI: return HB_TAG('v','a','i',' ');
55 /* Unicode-5.2 additions */
56 /* Unicode-6.0 additions */
57 }
58
59 /* Else, just change first char to lowercase and return */
60 return ((hb_tag_t) script) | 0x20000000u;
61 }
62
63 static hb_script_t
64 hb_ot_old_tag_to_script (hb_tag_t tag)
65 {
66 if (unlikely (tag == HB_OT_TAG_DEFAULT_SCRIPT))
67 return HB_SCRIPT_INVALID;
68
69 /* This side of the conversion is fully algorithmic. */
70
71 /* Any spaces at the end of the tag are replaced by repeating the last
72 * letter. Eg 'nko ' -> 'Nkoo' */
73 if (unlikely ((tag & 0x0000FF00u) == 0x00002000u))
74 tag |= (tag >> 8) & 0x0000FF00u; /* Copy second letter to third */
75 if (unlikely ((tag & 0x000000FFu) == 0x00000020u))
76 tag |= (tag >> 8) & 0x000000FFu; /* Copy third letter to fourth */
77
78 /* Change first char to uppercase and return */
79 return (hb_script_t) (tag & ~0x20000000u);
80 }
81
82 static hb_tag_t
83 hb_ot_new_tag_from_script (hb_script_t script)
84 {
85 switch ((hb_tag_t) script) {
86 case HB_SCRIPT_BENGALI: return HB_TAG('b','n','g','2');
87 case HB_SCRIPT_DEVANAGARI: return HB_TAG('d','e','v','2');
88 case HB_SCRIPT_GUJARATI: return HB_TAG('g','j','r','2');
89 case HB_SCRIPT_GURMUKHI: return HB_TAG('g','u','r','2');
90 case HB_SCRIPT_KANNADA: return HB_TAG('k','n','d','2');
91 case HB_SCRIPT_MALAYALAM: return HB_TAG('m','l','m','2');
92 case HB_SCRIPT_ORIYA: return HB_TAG('o','r','y','2');
93 case HB_SCRIPT_TAMIL: return HB_TAG('t','m','l','2');
94 case HB_SCRIPT_TELUGU: return HB_TAG('t','e','l','2');
95 case HB_SCRIPT_MYANMAR: return HB_TAG('m','y','m','2');
96 }
97
98 return HB_OT_TAG_DEFAULT_SCRIPT;
99 }
100
101 static hb_script_t
102 hb_ot_new_tag_to_script (hb_tag_t tag)
103 {
104 switch (tag) {
105 case HB_TAG('b','n','g','2'): return HB_SCRIPT_BENGALI;
106 case HB_TAG('d','e','v','2'): return HB_SCRIPT_DEVANAGARI;
107 case HB_TAG('g','j','r','2'): return HB_SCRIPT_GUJARATI;
108 case HB_TAG('g','u','r','2'): return HB_SCRIPT_GURMUKHI;
109 case HB_TAG('k','n','d','2'): return HB_SCRIPT_KANNADA;
110 case HB_TAG('m','l','m','2'): return HB_SCRIPT_MALAYALAM;
111 case HB_TAG('o','r','y','2'): return HB_SCRIPT_ORIYA;
112 case HB_TAG('t','m','l','2'): return HB_SCRIPT_TAMIL;
113 case HB_TAG('t','e','l','2'): return HB_SCRIPT_TELUGU;
114 case HB_TAG('m','y','m','2'): return HB_SCRIPT_MYANMAR;
115 }
116
117 return HB_SCRIPT_UNKNOWN;
118 }
119
120 /*
121 * Complete list at:
122 * https://www.microsoft.com/typography/otspec/scripttags.htm
123 * https://www.microsoft.com/typography/otspec160/scripttagsProposed.htm
124 *
125 * Most of the script tags are the same as the ISO 15924 tag but lowercased.
126 * So we just do that, and handle the exceptional cases in a switch.
127 */
128
129 void
130 hb_ot_tags_from_script (hb_script_t script,
131 hb_tag_t *script_tag_1,
132 hb_tag_t *script_tag_2)
133 {
134 hb_tag_t new_tag;
135
136 *script_tag_2 = HB_OT_TAG_DEFAULT_SCRIPT;
137 *script_tag_1 = hb_ot_old_tag_from_script (script);
138
139 new_tag = hb_ot_new_tag_from_script (script);
140 if (unlikely (new_tag != HB_OT_TAG_DEFAULT_SCRIPT)) {
141 *script_tag_2 = *script_tag_1;
142 *script_tag_1 = new_tag;
143 }
144 }
145
146 hb_script_t
147 hb_ot_tag_to_script (hb_tag_t tag)
148 {
149 if (unlikely ((tag & 0x000000FFu) == '2'))
150 return hb_ot_new_tag_to_script (tag);
151
152 return hb_ot_old_tag_to_script (tag);
153 }
154
155
156 /* hb_language_t */
157
158 typedef struct {
159 char language[4];
160 hb_tag_t tag;
161 } LangTag;
162
163 /*
164 * Complete list at:
165 * http://www.microsoft.com/typography/otspec/languagetags.htm
166 *
167 * Generated by intersecting the OpenType language tag list from
168 * Draft OpenType 1.5 spec, with with the ISO 639-3 codes from
169 * 2008-08-04, matching on name, and finally adjusted manually.
170 *
171 * Updated on 2012-12-07 with more research into remaining codes.
172 *
173 * Updated on 2013-11-23 based on usage in SIL and Microsoft fonts,
174 * the new proposal from Microsoft, and latest ISO 639-3 names.
175 *
176 * Some items still missing. Those are commented out at the end.
177 * Keep sorted for bsearch.
178 *
179 * Updated as of 2015-05-06: OT1.7 on MS website has some newer
180 * items that we don't have here, eg. Zazaki. This is the new
181 * items in OpenType 1.7 (red items), most of which we have:
182 * http://www.microsoft.com/typography/otspec170/languagetags.htm
183 */
184
185 static const LangTag ot_languages[] = {
186 {"aa", HB_TAG('A','F','R',' ')}, /* Afar */
187 {"ab", HB_TAG('A','B','K',' ')}, /* Abkhazian */
188 {"abq", HB_TAG('A','B','A',' ')}, /* Abaza */
189 {"acf", HB_TAG('F','A','N',' ')}, /* French Antillean */
190 {"ach", HB_TAG('A','C','H',' ')}, /* Acoli */
191 {"acr", HB_TAG('A','C','R',' ')}, /* Achi */
192 {"ada", HB_TAG('D','N','G',' ')}, /* Dangme */
193 {"ady", HB_TAG('A','D','Y',' ')}, /* Adyghe */
194 {"af", HB_TAG('A','F','K',' ')}, /* Afrikaans */
195 {"ahg", HB_TAG('A','G','W',' ')}, /* Agaw */
196 {"aii", HB_TAG('S','W','A',' ')}, /* Swadaya Aramaic */
197 {"aio", HB_TAG('A','I','O',' ')}, /* Aiton */
198 {"aiw", HB_TAG('A','R','I',' ')}, /* Aari */
199 {"ak", HB_TAG('T','W','I',' ')}, /* Akan [macrolanguage] */
200 {"aka", HB_TAG('A','K','A',' ')}, /* Akan */
201 {"alt", HB_TAG('A','L','T',' ')}, /* [Southern] Altai */
202 {"am", HB_TAG('A','M','H',' ')}, /* Amharic */
203 {"amf", HB_TAG('H','B','N',' ')}, /* Hammer-Banna */
204 {"an", HB_TAG('A','R','G',' ')}, /* Aragonese */
205 {"ang", HB_TAG('A','N','G',' ')}, /* Old English (ca. 450-1100) */
206 {"ar", HB_TAG('A','R','A',' ')}, /* Arabic [macrolanguage] */
207 {"arb", HB_TAG('A','R','A',' ')}, /* Standard Arabic */
208 {"arn", HB_TAG('M','A','P',' ')}, /* Mapudungun */
209 {"ary", HB_TAG('M','O','R',' ')}, /* Moroccan Arabic */
210 {"as", HB_TAG('A','S','M',' ')}, /* Assamese */
211 {"ast", HB_TAG('A','S','T',' ')}, /* Asturian/Asturleonese/Bable/Leonese */
212 {"ath", HB_TAG('A','T','H',' ')}, /* Athapaskan [family] */
213 {"atj", HB_TAG('R','C','R',' ')}, /* R-Cree */
214 {"atv", HB_TAG('A','L','T',' ')}, /* [Northern] Altai */
215 {"av", HB_TAG('A','V','R',' ')}, /* Avaric */
216 {"awa", HB_TAG('A','W','A',' ')}, /* Awadhi */
217 {"ay", HB_TAG('A','Y','M',' ')}, /* Aymara [macrolanguage] */
218 {"az", HB_TAG('A','Z','E',' ')}, /* Azerbaijani [macrolanguage] */
219 {"azb", HB_TAG('A','Z','B',' ')}, /* South Azerbaijani */
220 {"azj", HB_TAG('A','Z','E',' ')}, /* North Azerbaijani */
221 {"ba", HB_TAG('B','S','H',' ')}, /* Bashkir */
222 {"bad", HB_TAG('B','A','D','0')}, /* Banda */
223 {"bai", HB_TAG('B','M','L',' ')}, /* Bamileke [family] */
224 {"bal", HB_TAG('B','L','I',' ')}, /* Baluchi [macrolangauge] */
225 {"ban", HB_TAG('B','A','N',' ')}, /* Balinese */
226 {"bar", HB_TAG('B','A','R',' ')}, /* Bavarian */
227 {"bbc", HB_TAG('B','B','C',' ')}, /* Batak Toba */
228 {"bci", HB_TAG('B','A','U',' ')}, /* Baoulé */
229 {"bcl", HB_TAG('B','I','K',' ')}, /* Central Bikol */
230 {"bcq", HB_TAG('B','C','H',' ')}, /* Bench */
231 {"bdy", HB_TAG('B','D','Y',' ')}, /* Bandjalang */
232 {"be", HB_TAG('B','E','L',' ')}, /* Belarusian */
233 {"bem", HB_TAG('B','E','M',' ')}, /* Bemba (Zambia) */
234 {"ber", HB_TAG('B','E','R',' ')}, /* Berber [family] */
235 {"bfq", HB_TAG('B','A','D',' ')}, /* Badaga */
236 {"bft", HB_TAG('B','L','T',' ')}, /* Balti */
237 {"bfu", HB_TAG('L','A','H',' ')}, /* Lahuli */
238 {"bfy", HB_TAG('B','A','G',' ')}, /* Baghelkhandi */
239 {"bg", HB_TAG('B','G','R',' ')}, /* Bulgarian */
240 {"bgc", HB_TAG('B','G','C',' ')}, /* Haryanvi */
241 {"bgq", HB_TAG('B','G','Q',' ')}, /* Bagri */
242 {"bhb", HB_TAG('B','H','I',' ')}, /* Bhili */
243 {"bhk", HB_TAG('B','I','K',' ')}, /* Albay Bicolano (retired code) */
244 {"bho", HB_TAG('B','H','O',' ')}, /* Bhojpuri */
245 {"bi", HB_TAG('B','I','S',' ')}, /* Bislama */
246 {"bik", HB_TAG('B','I','K',' ')}, /* Bikol [macrolanguage] */
247 {"bin", HB_TAG('E','D','O',' ')}, /* Bini */
248 {"bjj", HB_TAG('B','J','J',' ')}, /* Kanauji */
249 {"bjt", HB_TAG('B','L','N',' ')}, /* Balanta-Ganja */
250 {"bla", HB_TAG('B','K','F',' ')}, /* Blackfoot */
251 {"ble", HB_TAG('B','L','N',' ')}, /* Balanta-Kentohe */
252 {"blk", HB_TAG('B','L','K',' ')}, /* Pa'O/Pa'o Karen */
253 {"bln", HB_TAG('B','I','K',' ')}, /* Southern Catanduanes Bikol */
254 {"bm", HB_TAG('B','M','B',' ')}, /* Bambara */
255 {"bn", HB_TAG('B','E','N',' ')}, /* Bengali */
256 {"bo", HB_TAG('T','I','B',' ')}, /* Tibetan */
257 {"bpy", HB_TAG('B','P','Y',' ')}, /* Bishnupriya */
258 {"bqi", HB_TAG('L','R','C',' ')}, /* Bakhtiari */
259 {"br", HB_TAG('B','R','E',' ')}, /* Breton */
260 {"bra", HB_TAG('B','R','I',' ')}, /* Braj Bhasha */
261 {"brh", HB_TAG('B','R','H',' ')}, /* Brahui */
262 {"brx", HB_TAG('B','R','X',' ')}, /* Bodo (India) */
263 {"bs", HB_TAG('B','O','S',' ')}, /* Bosnian */
264 {"btb", HB_TAG('B','T','I',' ')}, /* Beti (Cameroon) */
265 {"bto", HB_TAG('B','I','K',' ')}, /* Rinconada Bikol */
266 {"bts", HB_TAG('B','T','S',' ')}, /* Batak Simalungun */
267 {"bug", HB_TAG('B','U','G',' ')}, /* Buginese */
268 {"bxr", HB_TAG('R','B','U',' ')}, /* Russian Buriat */
269 {"byn", HB_TAG('B','I','L',' ')}, /* Bilen */
270 {"ca", HB_TAG('C','A','T',' ')}, /* Catalan */
271 {"cak", HB_TAG('C','A','K',' ')}, /* Kaqchikel */
272 {"cbk", HB_TAG('C','B','K',' ')}, /* Chavacano */
273 {"ce", HB_TAG('C','H','E',' ')}, /* Chechen */
274 {"ceb", HB_TAG('C','E','B',' ')}, /* Cebuano */
275 {"cgg", HB_TAG('C','G','G',' ')}, /* Chiga */
276 {"ch", HB_TAG('C','H','A',' ')}, /* Chamorro */
277 {"chk", HB_TAG('C','H','K','0')}, /* Chuukese */
278 {"cho", HB_TAG('C','H','O',' ')}, /* Choctaw */
279 {"chp", HB_TAG('C','H','P',' ')}, /* Chipewyan */
280 {"chr", HB_TAG('C','H','R',' ')}, /* Cherokee */
281 {"chy", HB_TAG('C','H','Y',' ')}, /* Cheyenne */
282 {"ckb", HB_TAG('K','U','R',' ')}, /* Central Kurdish (Sorani) */
283 {"ckt", HB_TAG('C','H','K',' ')}, /* Chukchi */
284 {"cop", HB_TAG('C','O','P',' ')}, /* Coptic */
285 {"cpp", HB_TAG('C','P','P',' ')}, /* Creoles */
286 {"cr", HB_TAG('C','R','E',' ')}, /* Cree */
287 {"cre", HB_TAG('Y','C','R',' ')}, /* Y-Cree */
288 {"crh", HB_TAG('C','R','T',' ')}, /* Crimean Tatar */
289 {"crj", HB_TAG('E','C','R',' ')}, /* [Southern] East Cree */
290 {"crk", HB_TAG('W','C','R',' ')}, /* West-Cree */
291 {"crl", HB_TAG('E','C','R',' ')}, /* [Northern] East Cree */
292 {"crm", HB_TAG('M','C','R',' ')}, /* Moose Cree */
293 {"crx", HB_TAG('C','R','R',' ')}, /* Carrier */
294 {"cs", HB_TAG('C','S','Y',' ')}, /* Czech */
295 {"csb", HB_TAG('C','S','B',' ')}, /* Kashubian */
296 {"ctg", HB_TAG('C','T','G',' ')}, /* Chittagonian */
297 {"cts", HB_TAG('B','I','K',' ')}, /* Northern Catanduanes Bikol */
298 {"cu", HB_TAG('C','S','L',' ')}, /* Church Slavic */
299 {"cuk", HB_TAG('C','U','K',' ')}, /* San Blas Kuna */
300 {"cv", HB_TAG('C','H','U',' ')}, /* Chuvash */
301 {"cwd", HB_TAG('D','C','R',' ')}, /* Woods Cree */
302 {"cy", HB_TAG('W','E','L',' ')}, /* Welsh */
303 {"da", HB_TAG('D','A','N',' ')}, /* Danish */
304 {"dap", HB_TAG('N','I','S',' ')}, /* Nisi (India) */
305 {"dar", HB_TAG('D','A','R',' ')}, /* Dargwa */
306 {"dax", HB_TAG('D','A','X',' ')}, /* Dayi */
307 {"de", HB_TAG('D','E','U',' ')}, /* German */
308 {"dgo", HB_TAG('D','G','O',' ')}, /* Dogri */
309 {"dhd", HB_TAG('M','A','W',' ')}, /* Dhundari */
310 {"dhg", HB_TAG('D','H','G',' ')}, /* Dhangu */
311 {"din", HB_TAG('D','N','K',' ')}, /* Dinka [macrolanguage] */
312 {"diq", HB_TAG('D','I','Q',' ')}, /* Dimli */
313 {"dje", HB_TAG('D','J','R',' ')}, /* Zarma */
314 {"djr", HB_TAG('D','J','R','0')}, /* Djambarrpuyngu */
315 {"dng", HB_TAG('D','U','N',' ')}, /* Dungan */
316 {"dnj", HB_TAG('D','N','J',' ')}, /* Dan */
317 {"doi", HB_TAG('D','G','R',' ')}, /* Dogri [macrolanguage] */
318 {"dsb", HB_TAG('L','S','B',' ')}, /* Lower Sorbian */
319 {"duj", HB_TAG('D','U','J',' ')}, /* Dhuwal */
320 {"dv", HB_TAG('D','I','V',' ')}, /* Dhivehi/Divehi/Maldivian */
321 {"dyu", HB_TAG('J','U','L',' ')}, /* Jula */
322 {"dz", HB_TAG('D','Z','N',' ')}, /* Dzongkha */
323 {"ee", HB_TAG('E','W','E',' ')}, /* Ewe */
324 {"efi", HB_TAG('E','F','I',' ')}, /* Efik */
325 {"ekk", HB_TAG('E','T','I',' ')}, /* Standard Estonian */
326 {"el", HB_TAG('E','L','L',' ')}, /* Modern Greek (1453-) */
327 {"emk", HB_TAG('M','N','K',' ')}, /* Eastern Maninkakan */
328 {"en", HB_TAG('E','N','G',' ')}, /* English */
329 {"enf", HB_TAG('F','N','E',' ')}, /* Forest Nenets */
330 {"enh", HB_TAG('T','N','E',' ')}, /* Tundra Nenets */
331 {"eo", HB_TAG('N','T','O',' ')}, /* Esperanto */
332 {"eot", HB_TAG('B','T','I',' ')}, /* Beti (Côte d'Ivoire) */
333 {"es", HB_TAG('E','S','P',' ')}, /* Spanish */
334 {"esu", HB_TAG('E','S','U',' ')}, /* Central Yupik */
335 {"et", HB_TAG('E','T','I',' ')}, /* Estonian [macrolanguage] */
336 {"eu", HB_TAG('E','U','Q',' ')}, /* Basque */
337 {"eve", HB_TAG('E','V','N',' ')}, /* Even */
338 {"evn", HB_TAG('E','V','K',' ')}, /* Evenki */
339 {"fa", HB_TAG('F','A','R',' ')}, /* Persian [macrolanguage] */
340 {"fan", HB_TAG('F','A','N','0')}, /* Fang */
341 {"fat", HB_TAG('F','A','T',' ')}, /* Fanti */
342 {"ff", HB_TAG('F','U','L',' ')}, /* Fulah [macrolanguage] */
343 {"fi", HB_TAG('F','I','N',' ')}, /* Finnish */
344 {"fil", HB_TAG('P','I','L',' ')}, /* Filipino */
345 {"fj", HB_TAG('F','J','I',' ')}, /* Fijian */
346 {"flm", HB_TAG('H','A','L',' ')}, /* Halam */
347 {"fo", HB_TAG('F','O','S',' ')}, /* Faroese */
348 {"fon", HB_TAG('F','O','N',' ')}, /* Fon */
349 {"fr", HB_TAG('F','R','A',' ')}, /* French */
350 {"frc", HB_TAG('F','R','C',' ')}, /* Cajun French */
351 {"frp", HB_TAG('F','R','P',' ')}, /* Arpitan/Francoprovençal */
352 {"fuf", HB_TAG('F','T','A',' ')}, /* Futa */
353 {"fur", HB_TAG('F','R','L',' ')}, /* Friulian */
354 {"fuv", HB_TAG('F','U','V',' ')}, /* Nigerian Fulfulde */
355 {"fy", HB_TAG('F','R','I',' ')}, /* Western Frisian */
356 {"ga", HB_TAG('I','R','I',' ')}, /* Irish */
357 {"gaa", HB_TAG('G','A','D',' ')}, /* Ga */
358 {"gag", HB_TAG('G','A','G',' ')}, /* Gagauz */
359 {"gbm", HB_TAG('G','A','W',' ')}, /* Garhwali */
360 {"gd", HB_TAG('G','A','E',' ')}, /* Scottish Gaelic */
361 {"gez", HB_TAG('G','E','Z',' ')}, /* Ge'ez */
362 {"ggo", HB_TAG('G','O','N',' ')}, /* Southern Gondi */
363 {"gih", HB_TAG('G','I','H',' ')}, /* Githabul */
364 {"gil", HB_TAG('G','I','L','0')}, /* Kiribati (Gilbertese) */
365 {"gkp", HB_TAG('G','K','P',' ')}, /* Kpelle (Guinea) */
366 {"gl", HB_TAG('G','A','L',' ')}, /* Galician */
367 {"gld", HB_TAG('N','A','N',' ')}, /* Nanai */
368 {"gle", HB_TAG('I','R','T',' ')}, /* Irish Traditional */
369 {"glk", HB_TAG('G','L','K',' ')}, /* Gilaki */
370 {"gn", HB_TAG('G','U','A',' ')}, /* Guarani [macrolanguage] */
371 {"gnn", HB_TAG('G','N','N',' ')}, /* Gumatj */
372 {"gno", HB_TAG('G','O','N',' ')}, /* Northern Gondi */
373 {"gog", HB_TAG('G','O','G',' ')}, /* Gogo */
374 {"gon", HB_TAG('G','O','N',' ')}, /* Gondi [macrolanguage] */
375 {"grt", HB_TAG('G','R','O',' ')}, /* Garo */
376 {"gru", HB_TAG('S','O','G',' ')}, /* Sodo Gurage */
377 {"gsw", HB_TAG('A','L','S',' ')}, /* Alsatian */
378 {"gu", HB_TAG('G','U','J',' ')}, /* Gujarati */
379 {"guc", HB_TAG('G','U','C',' ')}, /* Wayuu */
380 {"guf", HB_TAG('G','U','F',' ')}, /* Gupapuyngu */
381 {"guk", HB_TAG('G','M','Z',' ')}, /* Gumuz */
382 /*{"guk", HB_TAG('G','U','K',' ')},*/ /* Gumuz (in SIL fonts) */
383 {"guz", HB_TAG('G','U','Z',' ')}, /* Ekegusii/Gusii */
384 {"gv", HB_TAG('M','N','X',' ')}, /* Manx */
385 {"ha", HB_TAG('H','A','U',' ')}, /* Hausa */
386 {"har", HB_TAG('H','R','I',' ')}, /* Harari */
387 {"haw", HB_TAG('H','A','W',' ')}, /* Hawaiian */
388 {"hay", HB_TAG('H','A','Y',' ')}, /* Haya */
389 {"haz", HB_TAG('H','A','Z',' ')}, /* Hazaragi */
390 {"he", HB_TAG('I','W','R',' ')}, /* Hebrew */
391 {"hi", HB_TAG('H','I','N',' ')}, /* Hindi */
392 {"hil", HB_TAG('H','I','L',' ')}, /* Hiligaynon */
393 {"hmn", HB_TAG('H','M','N',' ')}, /* Hmong */
394 {"hnd", HB_TAG('H','N','D',' ')}, /* [Southern] Hindko */
395 {"hne", HB_TAG('C','H','H',' ')}, /* Chattisgarhi */
396 {"hno", HB_TAG('H','N','D',' ')}, /* [Northern] Hindko */
397 {"ho", HB_TAG('H','M','O',' ')}, /* Hiri Motu */
398 {"hoc", HB_TAG('H','O',' ',' ')}, /* Ho */
399 {"hoj", HB_TAG('H','A','R',' ')}, /* Harauti */
400 {"hr", HB_TAG('H','R','V',' ')}, /* Croatian */
401 {"hsb", HB_TAG('U','S','B',' ')}, /* Upper Sorbian */
402 {"ht", HB_TAG('H','A','I',' ')}, /* Haitian/Haitian Creole */
403 {"hu", HB_TAG('H','U','N',' ')}, /* Hungarian */
404 {"hy", HB_TAG('H','Y','E',' ')}, /* Armenian */
405 {"hz", HB_TAG('H','E','R',' ')}, /* Herero */
406 {"ia", HB_TAG('I','N','A',' ')}, /* Interlingua (International Auxiliary Language Association) */
407 {"iba", HB_TAG('I','B','A',' ')}, /* Iban */
408 {"ibb", HB_TAG('I','B','B',' ')}, /* Ibibio */
409 {"id", HB_TAG('I','N','D',' ')}, /* Indonesian */
410 {"ie", HB_TAG('I','L','E',' ')}, /* Interlingue/Occidental */
411 {"ig", HB_TAG('I','B','O',' ')}, /* Igbo */
412 {"igb", HB_TAG('E','B','I',' ')}, /* Ebira */
413 {"ii", HB_TAG('Y','I','M',' ')}, /* Yi Modern */
414 {"ijc", HB_TAG('I','J','O',' ')}, /* Izon */
415 {"ijo", HB_TAG('I','J','O',' ')}, /* Ijo [family] */
416 {"ik", HB_TAG('I','P','K',' ')}, /* Inupiaq [macrolanguage] */
417 {"ilo", HB_TAG('I','L','O',' ')}, /* Ilokano */
418 {"inh", HB_TAG('I','N','G',' ')}, /* Ingush */
419 {"io", HB_TAG('I','D','O',' ')}, /* Ido */
420 {"is", HB_TAG('I','S','L',' ')}, /* Icelandic */
421 {"it", HB_TAG('I','T','A',' ')}, /* Italian */
422 {"iu", HB_TAG('I','N','U',' ')}, /* Inuktitut [macrolanguage] */
423 {"ja", HB_TAG('J','A','N',' ')}, /* Japanese */
424 {"jam", HB_TAG('J','A','M',' ')}, /* Jamaican Creole English */
425 {"jbo", HB_TAG('J','B','O',' ')}, /* Lojban */
426 {"jv", HB_TAG('J','A','V',' ')}, /* Javanese */
427 {"ka", HB_TAG('K','A','T',' ')}, /* Georgian */
428 {"kaa", HB_TAG('K','R','K',' ')}, /* Karakalpak */
429 {"kab", HB_TAG('K','A','B','0')}, /* Kabyle */
430 {"kam", HB_TAG('K','M','B',' ')}, /* Kamba (Kenya) */
431 {"kar", HB_TAG('K','R','N',' ')}, /* Karen [family] */
432 {"kat", HB_TAG('K','G','E',' ')}, /* Khutsuri Georgian */
433 {"kbd", HB_TAG('K','A','B',' ')}, /* Kabardian */
434 {"kde", HB_TAG('K','D','E',' ')}, /* Makonde */
435 {"kdr", HB_TAG('K','R','M',' ')}, /* Karaim */
436 {"kdt", HB_TAG('K','U','Y',' ')}, /* Kuy */
437 {"kea", HB_TAG('K','E','A',' ')}, /* Kabuverdianu (Crioulo) */
438 {"kek", HB_TAG('K','E','K',' ')}, /* Kekchi */
439 {"kex", HB_TAG('K','K','N',' ')}, /* Kokni */
440 {"kfa", HB_TAG('K','O','D',' ')}, /* Kodagu */
441 {"kfr", HB_TAG('K','A','C',' ')}, /* Kachchi */
442 {"kfx", HB_TAG('K','U','L',' ')}, /* Kulvi */
443 {"kfy", HB_TAG('K','M','N',' ')}, /* Kumaoni */
444 {"kg", HB_TAG('K','O','N',' ')}, /* Kongo [macrolanguage] */
445 {"kha", HB_TAG('K','S','I',' ')}, /* Khasi */
446 {"khb", HB_TAG('X','B','D',' ')}, /* Lü */
447 {"kht", HB_TAG('K','H','N',' ')}, /* Khamti (Microsoft fonts) */
448 /*{"kht", HB_TAG('K','H','T',' ')},*/ /* Khamti (OpenType spec and SIL fonts) */
449 {"khw", HB_TAG('K','H','W',' ')}, /* Khowar */
450 {"ki", HB_TAG('K','I','K',' ')}, /* Gikuyu/Kikuyu */
451 {"kiu", HB_TAG('K','I','U',' ')}, /* Kirmanjki */
452 {"kj", HB_TAG('K','U','A',' ')}, /* Kuanyama/Kwanyama */
453 {"kjd", HB_TAG('K','J','D',' ')}, /* Southern Kiwai */
454 {"kjh", HB_TAG('K','H','A',' ')}, /* Khakass */
455 {"kjp", HB_TAG('K','J','P',' ')}, /* Pwo Eastern Karen */
456 {"kk", HB_TAG('K','A','Z',' ')}, /* Kazakh */
457 {"kl", HB_TAG('G','R','N',' ')}, /* Kalaallisut */
458 {"kln", HB_TAG('K','A','L',' ')}, /* Kalenjin */
459 {"km", HB_TAG('K','H','M',' ')}, /* Central Khmer */
460 {"kmb", HB_TAG('M','B','N',' ')}, /* Kimbundu */
461 {"kmw", HB_TAG('K','M','O',' ')}, /* Komo (Democratic Republic of Congo) */
462 {"kn", HB_TAG('K','A','N',' ')}, /* Kannada */
463 {"knn", HB_TAG('K','O','K',' ')}, /* Konkani */
464 {"ko", HB_TAG('K','O','R',' ')}, /* Korean */
465 {"koi", HB_TAG('K','O','P',' ')}, /* Komi-Permyak */
466 {"kok", HB_TAG('K','O','K',' ')}, /* Konkani [macrolanguage] */
467 {"kon", HB_TAG('K','O','N','0')}, /* Kongo */
468 {"kos", HB_TAG('K','O','S',' ')}, /* Kosraean */
469 {"kpe", HB_TAG('K','P','L',' ')}, /* Kpelle [macrolanguage] */
470 {"kpv", HB_TAG('K','O','Z',' ')}, /* Komi-Zyrian */
471 {"kpy", HB_TAG('K','Y','K',' ')}, /* Koryak */
472 {"kqy", HB_TAG('K','R','T',' ')}, /* Koorete */
473 {"kr", HB_TAG('K','N','R',' ')}, /* Kanuri [macrolanguage] */
474 {"kri", HB_TAG('K','R','I',' ')}, /* Krio */
475 {"krl", HB_TAG('K','R','L',' ')}, /* Karelian */
476 {"kru", HB_TAG('K','U','U',' ')}, /* Kurukh */
477 {"ks", HB_TAG('K','S','H',' ')}, /* Kashmiri */
478 {"ksh", HB_TAG('K','S','H','0')}, /* Ripuarian, Kölsch */
479 /*{"ksw", HB_TAG('K','R','N',' ')},*/ /* S'gaw Karen (Microsoft fonts?) */
480 {"ksw", HB_TAG('K','S','W',' ')}, /* S'gaw Karen (OpenType spec and SIL fonts) */
481 {"ktb", HB_TAG('K','E','B',' ')}, /* Kebena */
482 {"ktu", HB_TAG('K','O','N',' ')}, /* Kikongo */
483 {"ku", HB_TAG('K','U','R',' ')}, /* Kurdish [macrolanguage] */
484 {"kum", HB_TAG('K','U','M',' ')}, /* Kumyk */
485 {"kv", HB_TAG('K','O','M',' ')}, /* Komi [macrolanguage] */
486 {"kvd", HB_TAG('K','U','I',' ')}, /* Kui (Indonesia) */
487 {"kw", HB_TAG('C','O','R',' ')}, /* Cornish */
488 {"kxc", HB_TAG('K','M','S',' ')}, /* Komso */
489 {"kxu", HB_TAG('K','U','I',' ')}, /* Kui (India) */
490 {"ky", HB_TAG('K','I','R',' ')}, /* Kirghiz/Kyrgyz */
491 {"kyu", HB_TAG('K','Y','U',' ')}, /* Western Kayah */
492 {"la", HB_TAG('L','A','T',' ')}, /* Latin */
493 {"lad", HB_TAG('J','U','D',' ')}, /* Ladino */
494 {"lb", HB_TAG('L','T','Z',' ')}, /* Luxembourgish */
495 {"lbe", HB_TAG('L','A','K',' ')}, /* Lak */
496 {"lbj", HB_TAG('L','D','K',' ')}, /* Ladakhi */
497 {"lez", HB_TAG('L','E','Z',' ')}, /* Lezgi */
498 {"lg", HB_TAG('L','U','G',' ')}, /* Ganda */
499 {"li", HB_TAG('L','I','M',' ')}, /* Limburgan/Limburger/Limburgish */
500 {"lif", HB_TAG('L','M','B',' ')}, /* Limbu */
501 {"lij", HB_TAG('L','I','J',' ')}, /* Ligurian */
502 {"lis", HB_TAG('L','I','S',' ')}, /* Lisu */
503 {"ljp", HB_TAG('L','J','P',' ')}, /* Lampung Api */
504 {"lki", HB_TAG('L','K','I',' ')}, /* Laki */
505 {"lld", HB_TAG('L','A','D',' ')}, /* Ladin */
506 {"lmn", HB_TAG('L','A','M',' ')}, /* Lambani */
507 {"lmo", HB_TAG('L','M','O',' ')}, /* Lombard */
508 {"ln", HB_TAG('L','I','N',' ')}, /* Lingala */
509 {"lo", HB_TAG('L','A','O',' ')}, /* Lao */
510 {"lom", HB_TAG('L','O','M',' ')}, /* Loma */
511 {"lrc", HB_TAG('L','R','C',' ')}, /* Northern Luri */
512 {"lt", HB_TAG('L','T','H',' ')}, /* Lithuanian */
513 {"lu", HB_TAG('L','U','B',' ')}, /* Luba-Katanga */
514 {"lua", HB_TAG('L','U','B',' ')}, /* Luba-Kasai */
515 {"luo", HB_TAG('L','U','O',' ')}, /* Luo (Kenya and Tanzania) */
516 {"lus", HB_TAG('M','I','Z',' ')}, /* Mizo */
517 {"luy", HB_TAG('L','U','H',' ')}, /* Luyia/Oluluyia [macrolanguage] */
518 {"luz", HB_TAG('L','R','C',' ')}, /* Southern Luri */
519 {"lv", HB_TAG('L','V','I',' ')}, /* Latvian */
520 {"lzz", HB_TAG('L','A','Z',' ')}, /* Laz */
521 {"mad", HB_TAG('M','A','D',' ')}, /* Madurese */
522 {"mag", HB_TAG('M','A','G',' ')}, /* Magahi */
523 {"mai", HB_TAG('M','T','H',' ')}, /* Maithili */
524 {"mak", HB_TAG('M','K','R',' ')}, /* Makasar */
525 {"mal", HB_TAG('M','A','L',' ')}, /* Malayalam */
526 {"mam", HB_TAG('M','A','M',' ')}, /* Mam */
527 {"man", HB_TAG('M','N','K',' ')}, /* Manding/Mandingo [macrolanguage] */
528 {"mdc", HB_TAG('M','L','E',' ')}, /* Male (Papua New Guinea) */
529 {"mdf", HB_TAG('M','O','K',' ')}, /* Moksha */
530 {"mdr", HB_TAG('M','D','R',' ')}, /* Mandar */
531 {"mdy", HB_TAG('M','L','E',' ')}, /* Male (Ethiopia) */
532 {"men", HB_TAG('M','D','E',' ')}, /* Mende (Sierra Leone) */
533 {"mer", HB_TAG('M','E','R',' ')}, /* Meru */
534 {"mfe", HB_TAG('M','F','E',' ')}, /* Morisyen */
535 {"mg", HB_TAG('M','L','G',' ')}, /* Malagasy [macrolanguage] */
536 {"mh", HB_TAG('M','A','H',' ')}, /* Marshallese */
537 {"mhr", HB_TAG('L','M','A',' ')}, /* Low Mari */
538 {"mi", HB_TAG('M','R','I',' ')}, /* Maori */
539 {"min", HB_TAG('M','I','N',' ')}, /* Minangkabau */
540 {"mk", HB_TAG('M','K','D',' ')}, /* Macedonian */
541 {"mku", HB_TAG('M','N','K',' ')}, /* Konyanka Maninka */
542 {"mkw", HB_TAG('M','K','W',' ')}, /* Kituba (Congo) */
543 {"ml", HB_TAG('M','L','R',' ')}, /* Malayalam */
544 {"mlq", HB_TAG('M','N','K',' ')}, /* Western Maninkakan */
545 {"mn", HB_TAG('M','N','G',' ')}, /* Mongolian [macrolanguage] */
546 {"mnc", HB_TAG('M','C','H',' ')}, /* Manchu */
547 {"mni", HB_TAG('M','N','I',' ')}, /* Manipuri */
548 {"mnk", HB_TAG('M','N','D',' ')}, /* Mandinka */
549 {"mns", HB_TAG('M','A','N',' ')}, /* Mansi */
550 {"mnw", HB_TAG('M','O','N',' ')}, /* Mon */
551 {"mo", HB_TAG('M','O','L',' ')}, /* Moldavian */
552 {"moh", HB_TAG('M','O','H',' ')}, /* Mohawk */
553 {"mos", HB_TAG('M','O','S',' ')}, /* Mossi */
554 {"mpe", HB_TAG('M','A','J',' ')}, /* Majang */
555 {"mr", HB_TAG('M','A','R',' ')}, /* Marathi */
556 {"mrj", HB_TAG('H','M','A',' ')}, /* High Mari */
557 {"ms", HB_TAG('M','L','Y',' ')}, /* Malay [macrolanguage] */
558 {"msc", HB_TAG('M','N','K',' ')}, /* Sankaran Maninka */
559 {"mt", HB_TAG('M','T','S',' ')}, /* Maltese */
560 {"mtr", HB_TAG('M','A','W',' ')}, /* Mewari */
561 {"mus", HB_TAG('M','U','S',' ')}, /* Creek */
562 {"mve", HB_TAG('M','A','W',' ')}, /* Marwari (Pakistan) */
563 {"mwk", HB_TAG('M','N','K',' ')}, /* Kita Maninkakan */
564 {"mwl", HB_TAG('M','W','L',' ')}, /* Mirandese */
565 {"mwr", HB_TAG('M','A','W',' ')}, /* Marwari [macrolanguage] */
566 {"mww", HB_TAG('M','W','W',' ')}, /* Hmong Daw */
567 {"my", HB_TAG('B','R','M',' ')}, /* Burmese */
568 {"mym", HB_TAG('M','E','N',' ')}, /* Me'en */
569 {"myn", HB_TAG('M','Y','N',' ')}, /* Mayan */
570 {"myq", HB_TAG('M','N','K',' ')}, /* Forest Maninka (retired code) */
571 {"myv", HB_TAG('E','R','Z',' ')}, /* Erzya */
572 {"mzn", HB_TAG('M','Z','N',' ')}, /* Mazanderani */
573 {"na", HB_TAG('N','A','U',' ')}, /* Nauru */
574 {"nag", HB_TAG('N','A','G',' ')}, /* Naga-Assamese */
575 {"nah", HB_TAG('N','A','H',' ')}, /* Nahuatl [family] */
576 {"nap", HB_TAG('N','A','P',' ')}, /* Neapolitan */
577 {"nb", HB_TAG('N','O','R',' ')}, /* Norwegian Bokmål */
578 {"nco", HB_TAG('S','I','B',' ')}, /* Sibe */
579 {"nd", HB_TAG('N','D','B',' ')}, /* [North] Ndebele */
580 {"ndc", HB_TAG('N','D','C',' ')}, /* Ndau */
581 {"nds", HB_TAG('N','D','S',' ')}, /* Low German/Low Saxon */
582 {"ne", HB_TAG('N','E','P',' ')}, /* Nepali */
583 {"new", HB_TAG('N','E','W',' ')}, /* Newari */
584 {"ng", HB_TAG('N','D','G',' ')}, /* Ndonga */
585 {"nga", HB_TAG('N','G','A',' ')}, /* Ngabaka */
586 {"ngl", HB_TAG('L','M','W',' ')}, /* Lomwe */
587 {"ngo", HB_TAG('S','X','T',' ')}, /* Sutu */
588 {"niu", HB_TAG('N','I','U',' ')}, /* Niuean */
589 {"niv", HB_TAG('G','I','L',' ')}, /* Gilyak */
590 {"nl", HB_TAG('N','L','D',' ')}, /* Dutch */
591 {"nn", HB_TAG('N','Y','N',' ')}, /* Norwegian Nynorsk */
592 {"no", HB_TAG('N','O','R',' ')}, /* Norwegian [macrolanguage] */
593 {"nod", HB_TAG('N','T','A',' ')}, /* Northern Thai */
594 {"noe", HB_TAG('N','O','E',' ')}, /* Nimadi */
595 {"nog", HB_TAG('N','O','G',' ')}, /* Nogai */
596 {"nov", HB_TAG('N','O','V',' ')}, /* Novial */
597 {"nqo", HB_TAG('N','K','O',' ')}, /* N'Ko */
598 {"nr", HB_TAG('N','D','B',' ')}, /* [South] Ndebele */
599 {"nsk", HB_TAG('N','A','S',' ')}, /* Naskapi */
600 {"nso", HB_TAG('S','O','T',' ')}, /* [Northern] Sotho */
601 {"nv", HB_TAG('N','A','V',' ')}, /* Navajo */
602 {"ny", HB_TAG('C','H','I',' ')}, /* Chewa/Chichwa/Nyanja */
603 {"nym", HB_TAG('N','Y','M',' ')}, /* Nyamwezi */
604 {"nyn", HB_TAG('N','K','L',' ')}, /* Nyankole */
605 {"oc", HB_TAG('O','C','I',' ')}, /* Occitan (post 1500) */
606 {"oj", HB_TAG('O','J','B',' ')}, /* Ojibwa [macrolanguage] */
607 {"ojs", HB_TAG('O','C','R',' ')}, /* Oji-Cree */
608 {"okm", HB_TAG('K','O','H',' ')}, /* Korean Old Hangul */
609 {"om", HB_TAG('O','R','O',' ')}, /* Oromo [macrolanguage] */
610 {"or", HB_TAG('O','R','I',' ')}, /* Oriya */
611 {"os", HB_TAG('O','S','S',' ')}, /* Ossetian */
612 {"pa", HB_TAG('P','A','N',' ')}, /* Panjabi */
613 {"pag", HB_TAG('P','A','G',' ')}, /* Pangasinan */
614 {"pam", HB_TAG('P','A','M',' ')}, /* Kapampangan/Pampanga */
615 {"pap", HB_TAG('P','A','P','0')}, /* Papiamento */
616 {"pau", HB_TAG('P','A','U',' ')}, /* Palauan */
617 {"pcc", HB_TAG('P','C','C',' ')}, /* Bouyei */
618 {"pcd", HB_TAG('P','C','D',' ')}, /* Picard */
619 {"pce", HB_TAG('P','L','G',' ')}, /* [Ruching] Palaung */
620 {"pdc", HB_TAG('P','D','C',' ')}, /* Pennsylvania German */
621 {"pes", HB_TAG('F','A','R',' ')}, /* Iranian Persian */
622 {"phk", HB_TAG('P','H','K',' ')}, /* Phake */
623 {"pi", HB_TAG('P','A','L',' ')}, /* Pali */
624 {"pih", HB_TAG('P','I','H',' ')}, /* Pitcairn-Norfolk */
625 {"pl", HB_TAG('P','L','K',' ')}, /* Polish */
626 {"pll", HB_TAG('P','L','G',' ')}, /* [Shwe] Palaung */
627 {"plp", HB_TAG('P','A','P',' ')}, /* Palpa */
628 {"pms", HB_TAG('P','M','S',' ')}, /* Piemontese */
629 {"pnb", HB_TAG('P','N','B',' ')}, /* Western Panjabi */
630 {"poh", HB_TAG('P','O','H',' ')}, /* Pocomchi */
631 {"pon", HB_TAG('P','O','N',' ')}, /* Pohnpeian */
632 {"prs", HB_TAG('D','R','I',' ')}, /* Afghan Persian/Dari */
633 {"ps", HB_TAG('P','A','S',' ')}, /* Pashto/Pushto [macrolanguage] */
634 {"pt", HB_TAG('P','T','G',' ')}, /* Portuguese */
635 {"pwo", HB_TAG('P','W','O',' ')}, /* Pwo Western Karen */
636 {"qu", HB_TAG('Q','U','Z',' ')}, /* Quechua [macrolanguage] */
637 {"quc", HB_TAG('Q','U','C',' ')}, /* K'iche'/Quiché */
638 {"quh", HB_TAG('Q','U','H',' ')}, /* Quechua (Bolivia) */
639 {"quz", HB_TAG('Q','U','Z',' ')}, /* Cusco Quechua */
640 {"qvi", HB_TAG('Q','V','I',' ')}, /* Quechua (Ecuador) */
641 {"qwh", HB_TAG('Q','W','H',' ')}, /* Quechua (Peru) */
642 {"raj", HB_TAG('R','A','J',' ')}, /* Rajasthani [macrolanguage] */
643 {"rar", HB_TAG('R','A','R',' ')}, /* Rarotongan */
644 {"rbb", HB_TAG('P','L','G',' ')}, /* Rumai Palaung */
645 {"rej", HB_TAG('R','E','J',' ')}, /* Rejang */
646 {"ria", HB_TAG('R','I','A',' ')}, /* Riang (India) */
647 {"rif", HB_TAG('R','I','F',' ')}, /* Tarifit */
648 {"ril", HB_TAG('R','I','A',' ')}, /* Riang (Myanmar) */
649 {"rit", HB_TAG('R','I','T',' ')}, /* Ritarungo */
650 {"rki", HB_TAG('A','R','K',' ')}, /* Rakhine */
651 {"rkw", HB_TAG('R','K','W',' ')}, /* Arakwal */
652 {"rm", HB_TAG('R','M','S',' ')}, /* Romansh */
653 {"rmy", HB_TAG('R','M','Y',' ')}, /* Vlax Romani */
654 {"rn", HB_TAG('R','U','N',' ')}, /* Rundi */
655 {"ro", HB_TAG('R','O','M',' ')}, /* Romanian */
656 {"rom", HB_TAG('R','O','Y',' ')}, /* Romany [macrolanguage] */
657 {"rtm", HB_TAG('R','T','M',' ')}, /* Rotuman */
658 {"ru", HB_TAG('R','U','S',' ')}, /* Russian */
659 {"rue", HB_TAG('R','S','Y',' ')}, /* Rusyn */
660 {"rup", HB_TAG('R','U','P',' ')}, /* Aromanian/Arumanian/Macedo-Romanian */
661 {"rw", HB_TAG('R','U','A',' ')}, /* Kinyarwanda */
662 {"rwr", HB_TAG('M','A','W',' ')}, /* Marwari (India) */
663 {"sa", HB_TAG('S','A','N',' ')}, /* Sanskrit */
664 {"sah", HB_TAG('Y','A','K',' ')}, /* Yakut */
665 {"sam", HB_TAG('P','A','A',' ')}, /* Palestinian Aramaic */
666 {"sas", HB_TAG('S','A','S',' ')}, /* Sasak */
667 {"sat", HB_TAG('S','A','T',' ')}, /* Santali */
668 {"sc", HB_TAG('S','R','D',' ')}, /* Sardinian [macrolanguage] */
669 {"sck", HB_TAG('S','A','D',' ')}, /* Sadri */
670 {"scn", HB_TAG('S','C','N',' ')}, /* Sicilian */
671 {"sco", HB_TAG('S','C','O',' ')}, /* Scots */
672 {"scs", HB_TAG('S','L','A',' ')}, /* [North] Slavey */
673 {"sd", HB_TAG('S','N','D',' ')}, /* Sindhi */
674 {"se", HB_TAG('N','S','M',' ')}, /* Northern Sami */
675 {"seh", HB_TAG('S','N','A',' ')}, /* Sena */
676 {"sel", HB_TAG('S','E','L',' ')}, /* Selkup */
677 {"sg", HB_TAG('S','G','O',' ')}, /* Sango */
678 {"sga", HB_TAG('S','G','A',' ')}, /* Old Irish (to 900) */
679 {"sgs", HB_TAG('S','G','S',' ')}, /* Samogitian */
680 {"sgw", HB_TAG('C','H','G',' ')}, /* Sebat Bet Gurage */
681 /*{"sgw", HB_TAG('S','G','W',' ')},*/ /* Sebat Bet Gurage (in SIL fonts) */
682 {"shi", HB_TAG('S','H','I',' ')}, /* Tachelhit */
683 {"shn", HB_TAG('S','H','N',' ')}, /* Shan */
684 {"si", HB_TAG('S','N','H',' ')}, /* Sinhala */
685 {"sid", HB_TAG('S','I','D',' ')}, /* Sidamo */
686 {"sjd", HB_TAG('K','S','M',' ')}, /* Kildin Sami */
687 {"sk", HB_TAG('S','K','Y',' ')}, /* Slovak */
688 {"skr", HB_TAG('S','R','K',' ')}, /* Seraiki */
689 {"sl", HB_TAG('S','L','V',' ')}, /* Slovenian */
690 {"sm", HB_TAG('S','M','O',' ')}, /* Samoan */
691 {"sma", HB_TAG('S','S','M',' ')}, /* Southern Sami */
692 {"smj", HB_TAG('L','S','M',' ')}, /* Lule Sami */
693 {"smn", HB_TAG('I','S','M',' ')}, /* Inari Sami */
694 {"sms", HB_TAG('S','K','S',' ')}, /* Skolt Sami */
695 {"sn", HB_TAG('S','N','A','0')}, /* Shona */
696 {"snk", HB_TAG('S','N','K',' ')}, /* Soninke */
697 {"so", HB_TAG('S','M','L',' ')}, /* Somali */
698 {"sop", HB_TAG('S','O','P',' ')}, /* Songe */
699 {"sq", HB_TAG('S','Q','I',' ')}, /* Albanian [macrolanguage] */
700 {"sr", HB_TAG('S','R','B',' ')}, /* Serbian */
701 {"srr", HB_TAG('S','R','R',' ')}, /* Serer */
702 {"ss", HB_TAG('S','W','Z',' ')}, /* Swati */
703 {"st", HB_TAG('S','O','T',' ')}, /* [Southern] Sotho */
704 {"stq", HB_TAG('S','T','Q',' ')}, /* Saterfriesisch */
705 {"stv", HB_TAG('S','I','G',' ')}, /* Silt'e */
706 {"su", HB_TAG('S','U','N',' ')}, /* Sundanese */
707 {"suk", HB_TAG('S','U','K',' ')}, /* Sukama */
708 {"suq", HB_TAG('S','U','R',' ')}, /* Suri */
709 {"sv", HB_TAG('S','V','E',' ')}, /* Swedish */
710 {"sva", HB_TAG('S','V','A',' ')}, /* Svan */
711 {"sw", HB_TAG('S','W','K',' ')}, /* Swahili [macrolanguage] */
712 {"swb", HB_TAG('C','M','R',' ')}, /* Comorian */
713 {"swh", HB_TAG('S','W','K',' ')}, /* Kiswahili/Swahili */
714 {"swv", HB_TAG('M','A','W',' ')}, /* Shekhawati */
715 {"sxu", HB_TAG('S','X','U',' ')}, /* Upper Saxon */
716 {"syl", HB_TAG('S','Y','L',' ')}, /* Sylheti */
717 {"syr", HB_TAG('S','Y','R',' ')}, /* Syriac [macrolanguage] */
718 {"szl", HB_TAG('S','Z','L',' ')}, /* Silesian */
719 {"ta", HB_TAG('T','A','M',' ')}, /* Tamil */
720 {"tab", HB_TAG('T','A','B',' ')}, /* Tabasaran */
721 {"tcy", HB_TAG('T','U','L',' ')}, /* Tulu */
722 {"tdd", HB_TAG('T','D','D',' ')}, /* Tai Nüa */
723 {"te", HB_TAG('T','E','L',' ')}, /* Telugu */
724 {"tem", HB_TAG('T','M','N',' ')}, /* Temne */
725 {"tet", HB_TAG('T','E','T',' ')}, /* Tetum */
726 {"tg", HB_TAG('T','A','J',' ')}, /* Tajik */
727 {"th", HB_TAG('T','H','A',' ')}, /* Thai */
728 {"ti", HB_TAG('T','G','Y',' ')}, /* Tigrinya */
729 {"tig", HB_TAG('T','G','R',' ')}, /* Tigre */
730 {"tiv", HB_TAG('T','I','V',' ')}, /* Tiv */
731 {"tk", HB_TAG('T','K','M',' ')}, /* Turkmen */
732 {"tl", HB_TAG('T','G','L',' ')}, /* Tagalog */
733 {"tmh", HB_TAG('T','M','H',' ')}, /* Tamashek */
734 {"tn", HB_TAG('T','N','A',' ')}, /* Tswana */
735 {"to", HB_TAG('T','G','N',' ')}, /* Tonga (Tonga Islands) */
736 {"tod", HB_TAG('T','O','D','0')}, /* Toma */
737 {"toi", HB_TAG('T','N','G',' ')}, /* Tonga */
738 {"tpi", HB_TAG('T','P','I',' ')}, /* Tok Pisin */
739 {"tr", HB_TAG('T','R','K',' ')}, /* Turkish */
740 {"tru", HB_TAG('T','U','A',' ')}, /* Turoyo Aramaic */
741 {"ts", HB_TAG('T','S','G',' ')}, /* Tsonga */
742 {"tt", HB_TAG('T','A','T',' ')}, /* Tatar */
743 {"tum", HB_TAG('T','U','M',' ')}, /* Tumbuka */
744 {"tvl", HB_TAG('T','V','L',' ')}, /* Tuvalu */
745 {"tw", HB_TAG('T','W','I',' ')}, /* Twi */
746 {"ty", HB_TAG('T','H','T',' ')}, /* Tahitian */
747 {"tyv", HB_TAG('T','U','V',' ')}, /* Tuvin */
748 {"tyz", HB_TAG('T','Y','Z',' ')}, /* Tày */
749 {"tzm", HB_TAG('T','Z','M',' ')}, /* Central Atlas Tamazight */
750 {"tzo", HB_TAG('T','Z','O',' ')}, /* Tzotzil */
751 {"udm", HB_TAG('U','D','M',' ')}, /* Udmurt */
752 {"ug", HB_TAG('U','Y','G',' ')}, /* Uighur */
753 {"uk", HB_TAG('U','K','R',' ')}, /* Ukrainian */
754 {"umb", HB_TAG('U','M','B',' ')}, /* Umbundu */
755 {"unr", HB_TAG('M','U','N',' ')}, /* Mundari */
756 {"ur", HB_TAG('U','R','D',' ')}, /* Urdu */
757 {"uz", HB_TAG('U','Z','B',' ')}, /* Uzbek [macrolanguage] */
758 {"uzn", HB_TAG('U','Z','B',' ')}, /* Northern Uzbek */
759 {"uzs", HB_TAG('U','Z','B',' ')}, /* Southern Uzbek */
760 {"ve", HB_TAG('V','E','N',' ')}, /* Venda */
761 {"vec", HB_TAG('V','E','C',' ')}, /* Venetian */
762 {"vi", HB_TAG('V','I','T',' ')}, /* Vietnamese */
763 {"vls", HB_TAG('F','L','E',' ')}, /* Vlaams */
764 {"vmw", HB_TAG('M','A','K',' ')}, /* Makhuwa */
765 {"vo", HB_TAG('V','O','L',' ')}, /* Volapük */
766 {"vro", HB_TAG('V','R','O',' ')}, /* Võro */
767 {"wa", HB_TAG('W','L','N',' ')}, /* Walloon */
768 {"war", HB_TAG('W','A','R',' ')}, /* Waray (Philippines) */
769 {"wbm", HB_TAG('W','A',' ',' ')}, /* Wa */
770 {"wbr", HB_TAG('W','A','G',' ')}, /* Wagdi */
771 {"wle", HB_TAG('S','I','G',' ')}, /* Wolane */
772 {"wo", HB_TAG('W','L','F',' ')}, /* Wolof */
773 {"wry", HB_TAG('M','A','W',' ')}, /* Merwari */
774 {"wtm", HB_TAG('W','T','M',' ')}, /* Mewati */
775 {"xal", HB_TAG('K','L','M',' ')}, /* Kalmyk */
776 {"xan", HB_TAG('S','E','K',' ')}, /* Sekota */
777 {"xh", HB_TAG('X','H','S',' ')}, /* Xhosa */
778 {"xjb", HB_TAG('X','J','B',' ')}, /* Minjangbal */
779 {"xog", HB_TAG('X','O','G',' ')}, /* Soga */
780 {"xom", HB_TAG('K','M','O',' ')}, /* Komo (Sudan) */
781 {"xpe", HB_TAG('X','P','E',' ')}, /* Kpelle (Liberia) */
782 {"xsl", HB_TAG('S','S','L',' ')}, /* South Slavey */
783 {"xst", HB_TAG('S','I','G',' ')}, /* Silt'e (retired code) */
784 {"xwo", HB_TAG('T','O','D',' ')}, /* Written Oirat (Todo) */
785 {"yao", HB_TAG('Y','A','O',' ')}, /* Yao */
786 {"yap", HB_TAG('Y','A','P',' ')}, /* Yapese */
787 {"yi", HB_TAG('J','I','I',' ')}, /* Yiddish [macrolanguage] */
788 {"yo", HB_TAG('Y','B','A',' ')}, /* Yoruba */
789 {"yso", HB_TAG('N','I','S',' ')}, /* Nisi (China) */
790 {"za", HB_TAG('Z','H','A',' ')}, /* Chuang/Zhuang [macrolanguage] */
791 {"zea", HB_TAG('Z','E','A',' ')}, /* Zeeuws */
792 {"zgh", HB_TAG('Z','G','H',' ')}, /* Standard Morrocan Tamazigh */
793 {"zne", HB_TAG('Z','N','D',' ')}, /* Zande */
794 {"zu", HB_TAG('Z','U','L',' ')}, /* Zulu */
795 {"zum", HB_TAG('L','R','C',' ')}, /* Kumzari */
796 {"zza", HB_TAG('Z','Z','A',' ')}, /* Zazaki */
797
798 /* The corresponding languages IDs for the following IDs are unclear,
799 * overlap, or are architecturally weird. Needs more research. */
800
801 /*{"chp", HB_TAG('S','A','Y',' ')},*/ /* Sayisi */
802 /*{"cwd", HB_TAG('T','C','R',' ')},*/ /* TH-Cree */
803 /*{"emk", HB_TAG('E','M','K',' ')},*/ /* Eastern Maninkakan */
804 /*{"krc", HB_TAG('B','A','L',' ')},*/ /* Balkar */
805 /*{"??", HB_TAG('B','C','R',' ')},*/ /* Bible Cree */
806 /*{"zh?", HB_TAG('C','H','N',' ')},*/ /* Chinese (seen in Microsoft fonts) */
807 /*{"ar-Syrc?", HB_TAG('G','A','R',' ')},*/ /* Garshuni */
808 /*{"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) */
809 /*{"ga-Latg?/" HB_TAG('I','R','T',' ')},*/ /* Irish Traditional */
810 /*{"krc", HB_TAG('K','A','R',' ')},*/ /* Karachay */
811 /*{"ka-Geok?", HB_TAG('K','G','E',' ')},*/ /* Khutsuri Georgian */
812 /*{"kca", HB_TAG('K','H','K',' ')},*/ /* Khanty-Kazim */
813 /*{"kca", HB_TAG('K','H','S',' ')},*/ /* Khanty-Shurishkar */
814 /*{"kca", HB_TAG('K','H','V',' ')},*/ /* Khanty-Vakhi */
815 /*{"kqs, kss", HB_TAG('K','I','S',' ')},*/ /* Kisii */
816 /*{"lua", HB_TAG('L','U','A',' ')},*/ /* Luba-Lulua */
817 /*{"mlq", HB_TAG('M','L','N',' ')},*/ /* Malinke */
818 /*{"nso", HB_TAG('N','S','O',' ')},*/ /* Sotho, Northern */
819 /*{"??", HB_TAG('M','A','L',' ')},*/ /* Malayalam Traditional */
820 /*{"csw", HB_TAG('N','C','R',' ')},*/ /* N-Cree */
821 /*{"csw", HB_TAG('N','H','C',' ')},*/ /* Norway House Cree */
822 /*{"el-polyton", HB_TAG('P','G','R',' ')},*/ /* Polytonic Greek */
823 /*{"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 */
824 /*{"??", HB_TAG('Y','I','C',' ')},*/ /* Yi Classic */
825 /*{"zh-Latn-pinyin", HB_TAG('Z','H','P',' ')},*/ /* Chinese Phonetic */
826 };
827
828 typedef struct {
829 char language[11];
830 hb_tag_t tag;
831 } LangTagLong;
832 static const LangTagLong ot_languages_zh[] = {
833 /* Store longest-first, if one is a prefix of another. */
834 {"zh-cn", HB_TAG('Z','H','S',' ')}, /* Chinese (China) */
835 {"zh-hk", HB_TAG('Z','H','H',' ')}, /* Chinese (Hong Kong) */
836 {"zh-mo", HB_TAG('Z','H','H',' ')}, /* Chinese (Macao) */
837 {"zh-sg", HB_TAG('Z','H','S',' ')}, /* Chinese (Singapore) */
838 {"zh-tw", HB_TAG('Z','H','T',' ')}, /* Chinese (Taiwan) */
839 {"zh-hans", HB_TAG('Z','H','S',' ')}, /* Chinese (Simplified) */
840 {"zh-hant-hk",HB_TAG('Z','H','H',' ')}, /* Chinese (Hong Kong) */
841 {"zh-hant-mo",HB_TAG('Z','H','H',' ')}, /* Chinese (Macao) */
842 {"zh-hant", HB_TAG('Z','H','T',' ')}, /* Chinese (Traditional) */
843 };
844
845 static int
846 lang_compare_first_component (const char *a,
847 const char *b)
848 {
849 unsigned int da, db;
850 const char *p;
851
852 p = strchr (a, '-');
853 da = p ? (unsigned int) (p - a) : strlen (a);
854
855 p = strchr (b, '-');
856 db = p ? (unsigned int) (p - b) : strlen (b);
857
858 return strncmp (a, b, MAX (da, db));
859 }
860
861 static hb_bool_t
862 lang_matches (const char *lang_str, const char *spec)
863 {
864 unsigned int len = strlen (spec);
865
866 return strncmp (lang_str, spec, len) == 0 &&
867 (lang_str[len] == '\0' || lang_str[len] == '-');
868 }
869
870 hb_tag_t
871 hb_ot_tag_from_language (hb_language_t language)
872 {
873 const char *lang_str, *s;
874
875 if (language == HB_LANGUAGE_INVALID)
876 return HB_OT_TAG_DEFAULT_LANGUAGE;
877
878 lang_str = hb_language_to_string (language);
879
880 s = strstr (lang_str, "x-hbot");
881 if (s) {
882 char tag[4];
883 int i;
884 s += 6;
885 for (i = 0; i < 4 && ISALPHA (s[i]); i++)
886 tag[i] = TOUPPER (s[i]);
887 if (i) {
888 for (; i < 4; i++)
889 tag[i] = ' ';
890 return HB_TAG_CHAR4 (tag);
891 }
892 }
893
894 /*
895 * "fonipa" is a variant tag in BCP-47, meaning the International Phonetic Alphabet.
896 * It can be applied to any language.
897 */
898 if (strstr (lang_str, "-fonipa")) {
899 return HB_TAG('I','P','P','H'); /* Phonetic transcription—IPA conventions */
900 }
901
902 /*
903 * "fonnapa" is a variant tag in BCP-47, meaning the North American Phonetic Alphabet
904 * also known as Americanist Phonetic Notation. It can be applied to any language.
905 */
906 if (strstr (lang_str, "-fonnapa")) {
907 return HB_TAG('A','P','P','H'); /* Phonetic transcription—Americanist conventions */
908 }
909
910 /* Find a language matching in the first component */
911 {
912 const LangTag *lang_tag;
913 lang_tag = (LangTag *) bsearch (lang_str, ot_languages,
914 ARRAY_LENGTH (ot_languages), sizeof (LangTag),
915 (hb_compare_func_t) lang_compare_first_component);
916 if (lang_tag)
917 return lang_tag->tag;
918 }
919
920 /* Otherwise, check the Chinese ones */
921 if (0 == lang_compare_first_component (lang_str, "zh"))
922 {
923 unsigned int i;
924
925 for (i = 0; i < ARRAY_LENGTH (ot_languages_zh); i++)
926 {
927 const LangTagLong *lang_tag;
928 lang_tag = &ot_languages_zh[i];
929 if (lang_matches (lang_str, lang_tag->language))
930 return lang_tag->tag;
931 }
932
933 /* Otherwise just return 'ZHS ' */
934 return HB_TAG('Z','H','S',' ');
935 }
936
937 s = strchr (lang_str, '-');
938 if (!s)
939 s = lang_str + strlen (lang_str);
940 if (s - lang_str == 3) {
941 /* Assume it's ISO-639-3 and upper-case and use it. */
942 return hb_tag_from_string (lang_str, s - lang_str) & ~0x20202000u;
943 }
944
945 return HB_OT_TAG_DEFAULT_LANGUAGE;
946 }
947
948 /**
949 * hb_ot_tag_to_language:
950 *
951 *
952 *
953 * Return value: (transfer none):
954 *
955 * Since: 0.9.2
956 **/
957 hb_language_t
958 hb_ot_tag_to_language (hb_tag_t tag)
959 {
960 unsigned int i;
961
962 if (tag == HB_OT_TAG_DEFAULT_LANGUAGE)
963 return NULL;
964
965 for (i = 0; i < ARRAY_LENGTH (ot_languages); i++)
966 if (ot_languages[i].tag == tag)
967 return hb_language_from_string (ot_languages[i].language, -1);
968
969 /* If tag starts with ZH, it's Chinese */
970 if ((tag & 0xFFFF0000u) == 0x5A480000u) {
971 switch (tag) {
972 case HB_TAG('Z','H','H',' '): return hb_language_from_string ("zh-hk", -1); /* Hong Kong */
973 case HB_TAG('Z','H','S',' '): return hb_language_from_string ("zh-Hans", -1); /* Simplified */
974 case HB_TAG('Z','H','T',' '): return hb_language_from_string ("zh-Hant", -1); /* Traditional */
975 default: break; /* Fall through */
976 }
977 }
978
979 /* struct LangTag has only room for 3-letter language tags. */
980 switch (tag) {
981 case HB_TAG('A','P','P','H'): /* Phonetic transcription—Americanist conventions */
982 return hb_language_from_string ("und-fonnapa", -1);
983 case HB_TAG('I','P','P','H'): /* Phonetic transcription—IPA conventions */
984 return hb_language_from_string ("und-fonipa", -1);
985 }
986
987 /* Else return a custom language in the form of "x-hbotABCD" */
988 {
989 unsigned char buf[11] = "x-hbot";
990 buf[6] = tag >> 24;
991 buf[7] = (tag >> 16) & 0xFF;
992 buf[8] = (tag >> 8) & 0xFF;
993 buf[9] = tag & 0xFF;
994 if (buf[9] == 0x20)
995 buf[9] = '\0';
996 buf[10] = '\0';
997 return hb_language_from_string ((char *) buf, -1);
998 }
999 }
1000
1001 #ifdef MAIN
1002 static inline void
1003 test_langs_sorted (void)
1004 {
1005 for (unsigned int i = 1; i < ARRAY_LENGTH (ot_languages); i++)
1006 {
1007 int c = lang_compare_first_component (ot_languages[i-1].language, ot_languages[i].language);
1008 if (c >= 0)
1009 {
1010 fprintf (stderr, "ot_languages not sorted at index %d: %s %d %s\n",
1011 i, ot_languages[i-1].language, c, ot_languages[i].language);
1012 abort();
1013 }
1014 }
1015 }
1016
1017 int
1018 main (void)
1019 {
1020 test_langs_sorted ();
1021 return 0;
1022 }
1023
1024 #endif