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 {"cco", HB_TAG('C','C','H','N')}, /* Chinantec */
274 {"ce", HB_TAG('C','H','E',' ')}, /* Chechen */
275 {"ceb", HB_TAG('C','E','B',' ')}, /* Cebuano */
276 {"cfm", HB_TAG('H','A','L',' ')}, /* Halam/Falam Chin */
277 {"cgg", HB_TAG('C','G','G',' ')}, /* Chiga */
278 {"ch", HB_TAG('C','H','A',' ')}, /* Chamorro */
279 {"chj", HB_TAG('C','C','H','N')}, /* Chinantec */
280 {"chk", HB_TAG('C','H','K','0')}, /* Chuukese */
281 {"cho", HB_TAG('C','H','O',' ')}, /* Choctaw */
282 {"chp", HB_TAG('C','H','P',' ')}, /* Chipewyan */
283 {"chq", HB_TAG('C','C','H','N')}, /* Chinantec */
284 {"chr", HB_TAG('C','H','R',' ')}, /* Cherokee */
285 {"chy", HB_TAG('C','H','Y',' ')}, /* Cheyenne */
286 {"chz", HB_TAG('C','C','H','N')}, /* Chinantec */
287 {"cja", HB_TAG('C','J','A',' ')}, /* Western Cham */
288 {"cjm", HB_TAG('C','J','M',' ')}, /* Eastern Cham */
289 {"cka", HB_TAG('Q','I','N',' ')}, /* Khumi Awa Chin */
290 {"ckb", HB_TAG('K','U','R',' ')}, /* Central Kurdish (Sorani) */
291 {"ckt", HB_TAG('C','H','K',' ')}, /* Chukchi */
292 {"cld", HB_TAG('S','Y','R',' ')}, /* Chaldean Neo-Aramaic */
293 {"cle", HB_TAG('C','C','H','N')}, /* Chinantec */
294 {"cmr", HB_TAG('Q','I','N',' ')}, /* Mro-Khimi Chin */
295 {"cnb", HB_TAG('Q','I','N',' ')}, /* Chinbon Chin */
296 {"cnh", HB_TAG('Q','I','N',' ')}, /* Hakha Chin */
297 {"cnk", HB_TAG('Q','I','N',' ')}, /* Khumi Chin */
298 {"cnl", HB_TAG('C','C','H','N')}, /* Chinantec */
299 {"cnt", HB_TAG('C','C','H','N')}, /* Chinantec */
300 {"cnw", HB_TAG('Q','I','N',' ')}, /* Ngawn Chin */
301 {"cop", HB_TAG('C','O','P',' ')}, /* Coptic */
302 {"cpa", HB_TAG('C','C','H','N')}, /* Chinantec */
303 {"cpp", HB_TAG('C','P','P',' ')}, /* Creoles */
304 {"cr", HB_TAG('C','R','E',' ')}, /* Cree */
305 {"cre", HB_TAG('Y','C','R',' ')}, /* Y-Cree */
306 {"crh", HB_TAG('C','R','T',' ')}, /* Crimean Tatar */
307 {"crj", HB_TAG('E','C','R',' ')}, /* [Southern] East Cree */
308 {"crk", HB_TAG('W','C','R',' ')}, /* West-Cree */
309 {"crl", HB_TAG('E','C','R',' ')}, /* [Northern] East Cree */
310 {"crm", HB_TAG('M','C','R',' ')}, /* Moose Cree */
311 {"crx", HB_TAG('C','R','R',' ')}, /* Carrier */
312 {"cs", HB_TAG('C','S','Y',' ')}, /* Czech */
313 {"csa", HB_TAG('C','C','H','N')}, /* Chinantec */
314 {"csb", HB_TAG('C','S','B',' ')}, /* Kashubian */
315 {"csh", HB_TAG('Q','I','N',' ')}, /* Asho Chin */
316 {"cso", HB_TAG('C','C','H','N')}, /* Chinantec */
317 {"csy", HB_TAG('Q','I','N',' ')}, /* Siyin Chin */
318 {"ctd", HB_TAG('Q','I','N',' ')}, /* Tedim Chin */
319 {"cte", HB_TAG('C','C','H','N')}, /* Chinantec */
320 {"ctg", HB_TAG('C','T','G',' ')}, /* Chittagonian */
321 {"ctl", HB_TAG('C','C','H','N')}, /* Chinantec */
322 {"cts", HB_TAG('B','I','K',' ')}, /* Northern Catanduanes Bikol */
323 {"cu", HB_TAG('C','S','L',' ')}, /* Church Slavic */
324 {"cuc", HB_TAG('C','C','H','N')}, /* Chinantec */
325 {"cuk", HB_TAG('C','U','K',' ')}, /* San Blas Kuna */
326 {"cv", HB_TAG('C','H','U',' ')}, /* Chuvash */
327 {"cvn", HB_TAG('C','C','H','N')}, /* Chinantec */
328 {"cwd", HB_TAG('D','C','R',' ')}, /* Woods Cree */
329 {"cy", HB_TAG('W','E','L',' ')}, /* Welsh */
330 {"czt", HB_TAG('Q','I','N',' ')}, /* Zotung Chin */
331 {"da", HB_TAG('D','A','N',' ')}, /* Danish */
332 {"dao", HB_TAG('Q','I','N',' ')}, /* Daai Chin */
333 {"dap", HB_TAG('N','I','S',' ')}, /* Nisi (India) */
334 {"dar", HB_TAG('D','A','R',' ')}, /* Dargwa */
335 {"dax", HB_TAG('D','A','X',' ')}, /* Dayi */
336 {"de", HB_TAG('D','E','U',' ')}, /* German */
337 {"dgo", HB_TAG('D','G','O',' ')}, /* Dogri */
338 {"dhd", HB_TAG('M','A','W',' ')}, /* Dhundari */
339 {"dhg", HB_TAG('D','H','G',' ')}, /* Dhangu */
340 {"din", HB_TAG('D','N','K',' ')}, /* Dinka [macrolanguage] */
341 {"diq", HB_TAG('D','I','Q',' ')}, /* Dimli */
342 {"dje", HB_TAG('D','J','R',' ')}, /* Zarma */
343 {"djr", HB_TAG('D','J','R','0')}, /* Djambarrpuyngu */
344 {"dng", HB_TAG('D','U','N',' ')}, /* Dungan */
345 {"dnj", HB_TAG('D','N','J',' ')}, /* Dan */
346 {"doi", HB_TAG('D','G','R',' ')}, /* Dogri [macrolanguage] */
347 {"dsb", HB_TAG('L','S','B',' ')}, /* Lower Sorbian */
348 {"duj", HB_TAG('D','U','J',' ')}, /* Dhuwal */
349 {"dv", HB_TAG('D','I','V',' ')}, /* Dhivehi/Divehi/Maldivian */
350 {"dyu", HB_TAG('J','U','L',' ')}, /* Jula */
351 {"dz", HB_TAG('D','Z','N',' ')}, /* Dzongkha */
352 {"ee", HB_TAG('E','W','E',' ')}, /* Ewe */
353 {"efi", HB_TAG('E','F','I',' ')}, /* Efik */
354 {"ekk", HB_TAG('E','T','I',' ')}, /* Standard Estonian */
355 {"el", HB_TAG('E','L','L',' ')}, /* Modern Greek (1453-) */
356 {"emk", HB_TAG('M','N','K',' ')}, /* Eastern Maninkakan */
357 {"en", HB_TAG('E','N','G',' ')}, /* English */
358 {"enf", HB_TAG('F','N','E',' ')}, /* Forest Nenets */
359 {"enh", HB_TAG('T','N','E',' ')}, /* Tundra Nenets */
360 {"eo", HB_TAG('N','T','O',' ')}, /* Esperanto */
361 {"eot", HB_TAG('B','T','I',' ')}, /* Beti (Côte d'Ivoire) */
362 {"es", HB_TAG('E','S','P',' ')}, /* Spanish */
363 {"esu", HB_TAG('E','S','U',' ')}, /* Central Yupik */
364 {"et", HB_TAG('E','T','I',' ')}, /* Estonian [macrolanguage] */
365 {"eu", HB_TAG('E','U','Q',' ')}, /* Basque */
366 {"eve", HB_TAG('E','V','N',' ')}, /* Even */
367 {"evn", HB_TAG('E','V','K',' ')}, /* Evenki */
368 {"fa", HB_TAG('F','A','R',' ')}, /* Persian [macrolanguage] */
369 {"fan", HB_TAG('F','A','N','0')}, /* Fang */
370 {"fat", HB_TAG('F','A','T',' ')}, /* Fanti */
371 {"ff", HB_TAG('F','U','L',' ')}, /* Fulah [macrolanguage] */
372 {"fi", HB_TAG('F','I','N',' ')}, /* Finnish */
373 {"fil", HB_TAG('P','I','L',' ')}, /* Filipino */
374 {"fj", HB_TAG('F','J','I',' ')}, /* Fijian */
375 {"flm", HB_TAG('H','A','L',' ')}, /* Halam/Falam Chin [retired ISO639 code] */
376 {"fo", HB_TAG('F','O','S',' ')}, /* Faroese */
377 {"fon", HB_TAG('F','O','N',' ')}, /* Fon */
378 {"fr", HB_TAG('F','R','A',' ')}, /* French */
379 {"frc", HB_TAG('F','R','C',' ')}, /* Cajun French */
380 {"frp", HB_TAG('F','R','P',' ')}, /* Arpitan/Francoprovençal */
381 {"fuf", HB_TAG('F','T','A',' ')}, /* Futa */
382 {"fur", HB_TAG('F','R','L',' ')}, /* Friulian */
383 {"fuv", HB_TAG('F','U','V',' ')}, /* Nigerian Fulfulde */
384 {"fy", HB_TAG('F','R','I',' ')}, /* Western Frisian */
385 {"ga", HB_TAG('I','R','I',' ')}, /* Irish */
386 {"gaa", HB_TAG('G','A','D',' ')}, /* Ga */
387 {"gag", HB_TAG('G','A','G',' ')}, /* Gagauz */
388 {"gbm", HB_TAG('G','A','W',' ')}, /* Garhwali */
389 {"gd", HB_TAG('G','A','E',' ')}, /* Scottish Gaelic */
390 {"gez", HB_TAG('G','E','Z',' ')}, /* Ge'ez */
391 {"ggo", HB_TAG('G','O','N',' ')}, /* Southern Gondi */
392 {"gih", HB_TAG('G','I','H',' ')}, /* Githabul */
393 {"gil", HB_TAG('G','I','L','0')}, /* Kiribati (Gilbertese) */
394 {"gkp", HB_TAG('G','K','P',' ')}, /* Kpelle (Guinea) */
395 {"gl", HB_TAG('G','A','L',' ')}, /* Galician */
396 {"gld", HB_TAG('N','A','N',' ')}, /* Nanai */
397 {"glk", HB_TAG('G','L','K',' ')}, /* Gilaki */
398 {"gn", HB_TAG('G','U','A',' ')}, /* Guarani [macrolanguage] */
399 {"gnn", HB_TAG('G','N','N',' ')}, /* Gumatj */
400 {"gno", HB_TAG('G','O','N',' ')}, /* Northern Gondi */
401 {"gog", HB_TAG('G','O','G',' ')}, /* Gogo */
402 {"gon", HB_TAG('G','O','N',' ')}, /* Gondi [macrolanguage] */
403 {"grt", HB_TAG('G','R','O',' ')}, /* Garo */
404 {"gru", HB_TAG('S','O','G',' ')}, /* Sodo Gurage */
405 {"gsw", HB_TAG('A','L','S',' ')}, /* Alsatian */
406 {"gu", HB_TAG('G','U','J',' ')}, /* Gujarati */
407 {"guc", HB_TAG('G','U','C',' ')}, /* Wayuu */
408 {"guf", HB_TAG('G','U','F',' ')}, /* Gupapuyngu */
409 {"guk", HB_TAG('G','M','Z',' ')}, /* Gumuz */
410 /*{"guk", HB_TAG('G','U','K',' ')},*/ /* Gumuz (in SIL fonts) */
411 {"guz", HB_TAG('G','U','Z',' ')}, /* Ekegusii/Gusii */
412 {"gv", HB_TAG('M','N','X',' ')}, /* Manx */
413 {"ha", HB_TAG('H','A','U',' ')}, /* Hausa */
414 {"har", HB_TAG('H','R','I',' ')}, /* Harari */
415 {"haw", HB_TAG('H','A','W',' ')}, /* Hawaiian */
416 {"hay", HB_TAG('H','A','Y',' ')}, /* Haya */
417 {"haz", HB_TAG('H','A','Z',' ')}, /* Hazaragi */
418 {"he", HB_TAG('I','W','R',' ')}, /* Hebrew */
419 {"hi", HB_TAG('H','I','N',' ')}, /* Hindi */
420 {"hil", HB_TAG('H','I','L',' ')}, /* Hiligaynon */
421 {"hlt", HB_TAG('Q','I','N',' ')}, /* Matu Chin */
422 {"hmn", HB_TAG('H','M','N',' ')}, /* Hmong */
423 {"hnd", HB_TAG('H','N','D',' ')}, /* [Southern] Hindko */
424 {"hne", HB_TAG('C','H','H',' ')}, /* Chattisgarhi */
425 {"hno", HB_TAG('H','N','D',' ')}, /* [Northern] Hindko */
426 {"ho", HB_TAG('H','M','O',' ')}, /* Hiri Motu */
427 {"hoc", HB_TAG('H','O',' ',' ')}, /* Ho */
428 {"hoj", HB_TAG('H','A','R',' ')}, /* Harauti */
429 {"hr", HB_TAG('H','R','V',' ')}, /* Croatian */
430 {"hsb", HB_TAG('U','S','B',' ')}, /* Upper Sorbian */
431 {"ht", HB_TAG('H','A','I',' ')}, /* Haitian/Haitian Creole */
432 {"hu", HB_TAG('H','U','N',' ')}, /* Hungarian */
433 {"hy", HB_TAG('H','Y','E',' ')}, /* Armenian */
434 {"hz", HB_TAG('H','E','R',' ')}, /* Herero */
435 {"ia", HB_TAG('I','N','A',' ')}, /* Interlingua (International Auxiliary Language Association) */
436 {"iba", HB_TAG('I','B','A',' ')}, /* Iban */
437 {"ibb", HB_TAG('I','B','B',' ')}, /* Ibibio */
438 {"id", HB_TAG('I','N','D',' ')}, /* Indonesian */
439 {"ie", HB_TAG('I','L','E',' ')}, /* Interlingue/Occidental */
440 {"ig", HB_TAG('I','B','O',' ')}, /* Igbo */
441 {"igb", HB_TAG('E','B','I',' ')}, /* Ebira */
442 {"ii", HB_TAG('Y','I','M',' ')}, /* Yi Modern */
443 {"ijc", HB_TAG('I','J','O',' ')}, /* Izon */
444 {"ijo", HB_TAG('I','J','O',' ')}, /* Ijo [family] */
445 {"ik", HB_TAG('I','P','K',' ')}, /* Inupiaq [macrolanguage] */
446 {"ilo", HB_TAG('I','L','O',' ')}, /* Ilokano */
447 {"inh", HB_TAG('I','N','G',' ')}, /* Ingush */
448 {"io", HB_TAG('I','D','O',' ')}, /* Ido */
449 {"is", HB_TAG('I','S','L',' ')}, /* Icelandic */
450 {"it", HB_TAG('I','T','A',' ')}, /* Italian */
451 {"iu", HB_TAG('I','N','U',' ')}, /* Inuktitut [macrolanguage] */
452 {"ja", HB_TAG('J','A','N',' ')}, /* Japanese */
453 {"jam", HB_TAG('J','A','M',' ')}, /* Jamaican Creole English */
454 {"jbo", HB_TAG('J','B','O',' ')}, /* Lojban */
455 {"jv", HB_TAG('J','A','V',' ')}, /* Javanese */
456 {"ka", HB_TAG('K','A','T',' ')}, /* Georgian */
457 {"kaa", HB_TAG('K','R','K',' ')}, /* Karakalpak */
458 {"kab", HB_TAG('K','A','B','0')}, /* Kabyle */
459 {"kam", HB_TAG('K','M','B',' ')}, /* Kamba (Kenya) */
460 {"kar", HB_TAG('K','R','N',' ')}, /* Karen [family] */
461 {"kat", HB_TAG('K','G','E',' ')}, /* Khutsuri Georgian */
462 {"kbd", HB_TAG('K','A','B',' ')}, /* Kabardian */
463 {"kde", HB_TAG('K','D','E',' ')}, /* Makonde */
464 {"kdr", HB_TAG('K','R','M',' ')}, /* Karaim */
465 {"kdt", HB_TAG('K','U','Y',' ')}, /* Kuy */
466 {"kea", HB_TAG('K','E','A',' ')}, /* Kabuverdianu (Crioulo) */
467 {"kek", HB_TAG('K','E','K',' ')}, /* Kekchi */
468 {"kex", HB_TAG('K','K','N',' ')}, /* Kokni */
469 {"kfa", HB_TAG('K','O','D',' ')}, /* Kodagu */
470 {"kfr", HB_TAG('K','A','C',' ')}, /* Kachchi */
471 {"kfx", HB_TAG('K','U','L',' ')}, /* Kulvi */
472 {"kfy", HB_TAG('K','M','N',' ')}, /* Kumaoni */
473 {"kg", HB_TAG('K','O','N',' ')}, /* Kongo [macrolanguage] */
474 {"kha", HB_TAG('K','S','I',' ')}, /* Khasi */
475 {"khb", HB_TAG('X','B','D',' ')}, /* Lü */
476 {"kht", HB_TAG('K','H','N',' ')}, /* Khamti (Microsoft fonts) */
477 /*{"kht", HB_TAG('K','H','T',' ')},*/ /* Khamti (OpenType spec and SIL fonts) */
478 {"khw", HB_TAG('K','H','W',' ')}, /* Khowar */
479 {"ki", HB_TAG('K','I','K',' ')}, /* Gikuyu/Kikuyu */
480 {"kiu", HB_TAG('K','I','U',' ')}, /* Kirmanjki */
481 {"kj", HB_TAG('K','U','A',' ')}, /* Kuanyama/Kwanyama */
482 {"kjd", HB_TAG('K','J','D',' ')}, /* Southern Kiwai */
483 {"kjh", HB_TAG('K','H','A',' ')}, /* Khakass */
484 {"kjp", HB_TAG('K','J','P',' ')}, /* Pwo Eastern Karen */
485 {"kk", HB_TAG('K','A','Z',' ')}, /* Kazakh */
486 {"kl", HB_TAG('G','R','N',' ')}, /* Kalaallisut */
487 {"kln", HB_TAG('K','A','L',' ')}, /* Kalenjin */
488 {"km", HB_TAG('K','H','M',' ')}, /* Central Khmer */
489 {"kmb", HB_TAG('M','B','N',' ')}, /* Kimbundu */
490 {"kmw", HB_TAG('K','M','O',' ')}, /* Komo (Democratic Republic of Congo) */
491 {"kn", HB_TAG('K','A','N',' ')}, /* Kannada */
492 {"knn", HB_TAG('K','O','K',' ')}, /* Konkani */
493 {"ko", HB_TAG('K','O','R',' ')}, /* Korean */
494 {"koi", HB_TAG('K','O','P',' ')}, /* Komi-Permyak */
495 {"kok", HB_TAG('K','O','K',' ')}, /* Konkani [macrolanguage] */
496 {"kon", HB_TAG('K','O','N','0')}, /* Kongo */
497 {"kos", HB_TAG('K','O','S',' ')}, /* Kosraean */
498 {"kpe", HB_TAG('K','P','L',' ')}, /* Kpelle [macrolanguage] */
499 {"kpv", HB_TAG('K','O','Z',' ')}, /* Komi-Zyrian */
500 {"kpy", HB_TAG('K','Y','K',' ')}, /* Koryak */
501 {"kqy", HB_TAG('K','R','T',' ')}, /* Koorete */
502 {"kr", HB_TAG('K','N','R',' ')}, /* Kanuri [macrolanguage] */
503 {"kri", HB_TAG('K','R','I',' ')}, /* Krio */
504 {"krl", HB_TAG('K','R','L',' ')}, /* Karelian */
505 {"kru", HB_TAG('K','U','U',' ')}, /* Kurukh */
506 {"ks", HB_TAG('K','S','H',' ')}, /* Kashmiri */
507 {"ksh", HB_TAG('K','S','H','0')}, /* Ripuarian, Kölsch */
508 /*{"ksw", HB_TAG('K','R','N',' ')},*/ /* S'gaw Karen (Microsoft fonts?) */
509 {"ksw", HB_TAG('K','S','W',' ')}, /* S'gaw Karen (OpenType spec and SIL fonts) */
510 {"ktb", HB_TAG('K','E','B',' ')}, /* Kebena */
511 {"ktu", HB_TAG('K','O','N',' ')}, /* Kikongo */
512 {"ku", HB_TAG('K','U','R',' ')}, /* Kurdish [macrolanguage] */
513 {"kum", HB_TAG('K','U','M',' ')}, /* Kumyk */
514 {"kv", HB_TAG('K','O','M',' ')}, /* Komi [macrolanguage] */
515 {"kvd", HB_TAG('K','U','I',' ')}, /* Kui (Indonesia) */
516 {"kw", HB_TAG('C','O','R',' ')}, /* Cornish */
517 {"kxc", HB_TAG('K','M','S',' ')}, /* Komso */
518 {"kxu", HB_TAG('K','U','I',' ')}, /* Kui (India) */
519 {"ky", HB_TAG('K','I','R',' ')}, /* Kirghiz/Kyrgyz */
520 {"kyu", HB_TAG('K','Y','U',' ')}, /* Western Kayah */
521 {"la", HB_TAG('L','A','T',' ')}, /* Latin */
522 {"lad", HB_TAG('J','U','D',' ')}, /* Ladino */
523 {"lb", HB_TAG('L','T','Z',' ')}, /* Luxembourgish */
524 {"lbe", HB_TAG('L','A','K',' ')}, /* Lak */
525 {"lbj", HB_TAG('L','D','K',' ')}, /* Ladakhi */
526 {"lez", HB_TAG('L','E','Z',' ')}, /* Lezgi */
527 {"lg", HB_TAG('L','U','G',' ')}, /* Ganda */
528 {"li", HB_TAG('L','I','M',' ')}, /* Limburgan/Limburger/Limburgish */
529 {"lif", HB_TAG('L','M','B',' ')}, /* Limbu */
530 {"lij", HB_TAG('L','I','J',' ')}, /* Ligurian */
531 {"lis", HB_TAG('L','I','S',' ')}, /* Lisu */
532 {"ljp", HB_TAG('L','J','P',' ')}, /* Lampung Api */
533 {"lki", HB_TAG('L','K','I',' ')}, /* Laki */
534 {"lld", HB_TAG('L','A','D',' ')}, /* Ladin */
535 {"lmn", HB_TAG('L','A','M',' ')}, /* Lambani */
536 {"lmo", HB_TAG('L','M','O',' ')}, /* Lombard */
537 {"ln", HB_TAG('L','I','N',' ')}, /* Lingala */
538 {"lo", HB_TAG('L','A','O',' ')}, /* Lao */
539 {"lom", HB_TAG('L','O','M',' ')}, /* Loma */
540 {"lrc", HB_TAG('L','R','C',' ')}, /* Northern Luri */
541 {"lt", HB_TAG('L','T','H',' ')}, /* Lithuanian */
542 {"lu", HB_TAG('L','U','B',' ')}, /* Luba-Katanga */
543 {"lua", HB_TAG('L','U','B',' ')}, /* Luba-Kasai */
544 {"luo", HB_TAG('L','U','O',' ')}, /* Luo (Kenya and Tanzania) */
545 {"lus", HB_TAG('M','I','Z',' ')}, /* Mizo */
546 {"luy", HB_TAG('L','U','H',' ')}, /* Luyia/Oluluyia [macrolanguage] */
547 {"luz", HB_TAG('L','R','C',' ')}, /* Southern Luri */
548 {"lv", HB_TAG('L','V','I',' ')}, /* Latvian */
549 {"lzz", HB_TAG('L','A','Z',' ')}, /* Laz */
550 {"mad", HB_TAG('M','A','D',' ')}, /* Madurese */
551 {"mag", HB_TAG('M','A','G',' ')}, /* Magahi */
552 {"mai", HB_TAG('M','T','H',' ')}, /* Maithili */
553 {"mak", HB_TAG('M','K','R',' ')}, /* Makasar */
554 {"mam", HB_TAG('M','A','M',' ')}, /* Mam */
555 {"man", HB_TAG('M','N','K',' ')}, /* Manding/Mandingo [macrolanguage] */
556 {"mdc", HB_TAG('M','L','E',' ')}, /* Male (Papua New Guinea) */
557 {"mdf", HB_TAG('M','O','K',' ')}, /* Moksha */
558 {"mdr", HB_TAG('M','D','R',' ')}, /* Mandar */
559 {"mdy", HB_TAG('M','L','E',' ')}, /* Male (Ethiopia) */
560 {"men", HB_TAG('M','D','E',' ')}, /* Mende (Sierra Leone) */
561 {"mer", HB_TAG('M','E','R',' ')}, /* Meru */
562 {"mfe", HB_TAG('M','F','E',' ')}, /* Morisyen */
563 {"mg", HB_TAG('M','L','G',' ')}, /* Malagasy [macrolanguage] */
564 {"mh", HB_TAG('M','A','H',' ')}, /* Marshallese */
565 {"mhr", HB_TAG('L','M','A',' ')}, /* Low Mari */
566 {"mi", HB_TAG('M','R','I',' ')}, /* Maori */
567 {"min", HB_TAG('M','I','N',' ')}, /* Minangkabau */
568 {"mk", HB_TAG('M','K','D',' ')}, /* Macedonian */
569 {"mku", HB_TAG('M','N','K',' ')}, /* Konyanka Maninka */
570 {"mkw", HB_TAG('M','K','W',' ')}, /* Kituba (Congo) */
571 {"ml", HB_TAG('M','L','R',' ')}, /* Malayalam */
572 {"mlq", HB_TAG('M','N','K',' ')}, /* Western Maninkakan */
573 {"mn", HB_TAG('M','N','G',' ')}, /* Mongolian [macrolanguage] */
574 {"mnc", HB_TAG('M','C','H',' ')}, /* Manchu */
575 {"mni", HB_TAG('M','N','I',' ')}, /* Manipuri */
576 {"mnk", HB_TAG('M','N','D',' ')}, /* Mandinka */
577 {"mns", HB_TAG('M','A','N',' ')}, /* Mansi */
578 {"mnw", HB_TAG('M','O','N',' ')}, /* Mon */
579 {"mo", HB_TAG('M','O','L',' ')}, /* Moldavian */
580 {"moh", HB_TAG('M','O','H',' ')}, /* Mohawk */
581 {"mos", HB_TAG('M','O','S',' ')}, /* Mossi */
582 {"mpe", HB_TAG('M','A','J',' ')}, /* Majang */
583 {"mr", HB_TAG('M','A','R',' ')}, /* Marathi */
584 {"mrh", HB_TAG('Q','I','N',' ')}, /* Mara Chin */
585 {"mrj", HB_TAG('H','M','A',' ')}, /* High Mari */
586 {"ms", HB_TAG('M','L','Y',' ')}, /* Malay [macrolanguage] */
587 {"msc", HB_TAG('M','N','K',' ')}, /* Sankaran Maninka */
588 {"mt", HB_TAG('M','T','S',' ')}, /* Maltese */
589 {"mtr", HB_TAG('M','A','W',' ')}, /* Mewari */
590 {"mus", HB_TAG('M','U','S',' ')}, /* Creek */
591 {"mve", HB_TAG('M','A','W',' ')}, /* Marwari (Pakistan) */
592 {"mwk", HB_TAG('M','N','K',' ')}, /* Kita Maninkakan */
593 {"mwl", HB_TAG('M','W','L',' ')}, /* Mirandese */
594 {"mwr", HB_TAG('M','A','W',' ')}, /* Marwari [macrolanguage] */
595 {"mww", HB_TAG('M','W','W',' ')}, /* Hmong Daw */
596 {"my", HB_TAG('B','R','M',' ')}, /* Burmese */
597 {"mym", HB_TAG('M','E','N',' ')}, /* Me'en */
598 {"myn", HB_TAG('M','Y','N',' ')}, /* Mayan */
599 {"myq", HB_TAG('M','N','K',' ')}, /* Forest Maninka (retired code) */
600 {"myv", HB_TAG('E','R','Z',' ')}, /* Erzya */
601 {"mzn", HB_TAG('M','Z','N',' ')}, /* Mazanderani */
602 {"na", HB_TAG('N','A','U',' ')}, /* Nauru */
603 {"nag", HB_TAG('N','A','G',' ')}, /* Naga-Assamese */
604 {"nah", HB_TAG('N','A','H',' ')}, /* Nahuatl [family] */
605 {"nap", HB_TAG('N','A','P',' ')}, /* Neapolitan */
606 {"nb", HB_TAG('N','O','R',' ')}, /* Norwegian Bokmål */
607 {"nco", HB_TAG('S','I','B',' ')}, /* Sibe */
608 {"nd", HB_TAG('N','D','B',' ')}, /* [North] Ndebele */
609 {"ndc", HB_TAG('N','D','C',' ')}, /* Ndau */
610 {"nds", HB_TAG('N','D','S',' ')}, /* Low German/Low Saxon */
611 {"ne", HB_TAG('N','E','P',' ')}, /* Nepali */
612 {"new", HB_TAG('N','E','W',' ')}, /* Newari */
613 {"ng", HB_TAG('N','D','G',' ')}, /* Ndonga */
614 {"nga", HB_TAG('N','G','A',' ')}, /* Ngabaka */
615 {"ngl", HB_TAG('L','M','W',' ')}, /* Lomwe */
616 {"ngo", HB_TAG('S','X','T',' ')}, /* Sutu */
617 {"niu", HB_TAG('N','I','U',' ')}, /* Niuean */
618 {"niv", HB_TAG('G','I','L',' ')}, /* Gilyak */
619 {"nl", HB_TAG('N','L','D',' ')}, /* Dutch */
620 {"nn", HB_TAG('N','Y','N',' ')}, /* Norwegian Nynorsk */
621 {"no", HB_TAG('N','O','R',' ')}, /* Norwegian [macrolanguage] */
622 {"nod", HB_TAG('N','T','A',' ')}, /* Northern Thai */
623 {"noe", HB_TAG('N','O','E',' ')}, /* Nimadi */
624 {"nog", HB_TAG('N','O','G',' ')}, /* Nogai */
625 {"nov", HB_TAG('N','O','V',' ')}, /* Novial */
626 {"nqo", HB_TAG('N','K','O',' ')}, /* N'Ko */
627 {"nr", HB_TAG('N','D','B',' ')}, /* [South] Ndebele */
628 {"nsk", HB_TAG('N','A','S',' ')}, /* Naskapi */
629 {"nso", HB_TAG('S','O','T',' ')}, /* [Northern] Sotho */
630 {"nv", HB_TAG('N','A','V',' ')}, /* Navajo */
631 {"ny", HB_TAG('C','H','I',' ')}, /* Chewa/Chichwa/Nyanja */
632 {"nym", HB_TAG('N','Y','M',' ')}, /* Nyamwezi */
633 {"nyn", HB_TAG('N','K','L',' ')}, /* Nyankole */
634 {"oc", HB_TAG('O','C','I',' ')}, /* Occitan (post 1500) */
635 {"oj", HB_TAG('O','J','B',' ')}, /* Ojibwa [macrolanguage] */
636 {"ojs", HB_TAG('O','C','R',' ')}, /* Oji-Cree */
637 {"okm", HB_TAG('K','O','H',' ')}, /* Korean Old Hangul */
638 {"om", HB_TAG('O','R','O',' ')}, /* Oromo [macrolanguage] */
639 {"or", HB_TAG('O','R','I',' ')}, /* Oriya */
640 {"os", HB_TAG('O','S','S',' ')}, /* Ossetian */
641 {"pa", HB_TAG('P','A','N',' ')}, /* Panjabi */
642 {"pag", HB_TAG('P','A','G',' ')}, /* Pangasinan */
643 {"pam", HB_TAG('P','A','M',' ')}, /* Kapampangan/Pampanga */
644 {"pap", HB_TAG('P','A','P','0')}, /* Papiamento */
645 {"pau", HB_TAG('P','A','U',' ')}, /* Palauan */
646 {"pcc", HB_TAG('P','C','C',' ')}, /* Bouyei */
647 {"pcd", HB_TAG('P','C','D',' ')}, /* Picard */
648 {"pce", HB_TAG('P','L','G',' ')}, /* [Ruching] Palaung */
649 {"pck", HB_TAG('Q','I','N',' ')}, /* Paite Chin */
650 {"pdc", HB_TAG('P','D','C',' ')}, /* Pennsylvania German */
651 {"pes", HB_TAG('F','A','R',' ')}, /* Iranian Persian */
652 {"phk", HB_TAG('P','H','K',' ')}, /* Phake */
653 {"pi", HB_TAG('P','A','L',' ')}, /* Pali */
654 {"pih", HB_TAG('P','I','H',' ')}, /* Pitcairn-Norfolk */
655 {"pl", HB_TAG('P','L','K',' ')}, /* Polish */
656 {"pll", HB_TAG('P','L','G',' ')}, /* [Shwe] Palaung */
657 {"plp", HB_TAG('P','A','P',' ')}, /* Palpa */
658 {"pms", HB_TAG('P','M','S',' ')}, /* Piemontese */
659 {"pnb", HB_TAG('P','N','B',' ')}, /* Western Panjabi */
660 {"poh", HB_TAG('P','O','H',' ')}, /* Pocomchi */
661 {"pon", HB_TAG('P','O','N',' ')}, /* Pohnpeian */
662 {"prs", HB_TAG('D','R','I',' ')}, /* Afghan Persian/Dari */
663 {"ps", HB_TAG('P','A','S',' ')}, /* Pashto/Pushto [macrolanguage] */
664 {"pt", HB_TAG('P','T','G',' ')}, /* Portuguese */
665 {"pwo", HB_TAG('P','W','O',' ')}, /* Pwo Western Karen */
666 {"qu", HB_TAG('Q','U','Z',' ')}, /* Quechua [macrolanguage] */
667 {"quc", HB_TAG('Q','U','C',' ')}, /* K'iche'/Quiché */
668 {"quh", HB_TAG('Q','U','H',' ')}, /* Quechua (Bolivia) */
669 {"quz", HB_TAG('Q','U','Z',' ')}, /* Cusco Quechua */
670 {"qvi", HB_TAG('Q','V','I',' ')}, /* Quechua (Ecuador) */
671 {"qwh", HB_TAG('Q','W','H',' ')}, /* Quechua (Peru) */
672 {"raj", HB_TAG('R','A','J',' ')}, /* Rajasthani [macrolanguage] */
673 {"rar", HB_TAG('R','A','R',' ')}, /* Rarotongan */
674 {"rbb", HB_TAG('P','L','G',' ')}, /* Rumai Palaung */
675 {"rej", HB_TAG('R','E','J',' ')}, /* Rejang */
676 {"ria", HB_TAG('R','I','A',' ')}, /* Riang (India) */
677 {"rif", HB_TAG('R','I','F',' ')}, /* Tarifit */
678 {"ril", HB_TAG('R','I','A',' ')}, /* Riang (Myanmar) */
679 {"rit", HB_TAG('R','I','T',' ')}, /* Ritarungo */
680 {"rki", HB_TAG('A','R','K',' ')}, /* Rakhine */
681 {"rkw", HB_TAG('R','K','W',' ')}, /* Arakwal */
682 {"rm", HB_TAG('R','M','S',' ')}, /* Romansh */
683 {"rmy", HB_TAG('R','M','Y',' ')}, /* Vlax Romani */
684 {"rn", HB_TAG('R','U','N',' ')}, /* Rundi */
685 {"ro", HB_TAG('R','O','M',' ')}, /* Romanian */
686 {"rom", HB_TAG('R','O','Y',' ')}, /* Romany [macrolanguage] */
687 {"rtm", HB_TAG('R','T','M',' ')}, /* Rotuman */
688 {"ru", HB_TAG('R','U','S',' ')}, /* Russian */
689 {"rue", HB_TAG('R','S','Y',' ')}, /* Rusyn */
690 {"rup", HB_TAG('R','U','P',' ')}, /* Aromanian/Arumanian/Macedo-Romanian */
691 {"rw", HB_TAG('R','U','A',' ')}, /* Kinyarwanda */
692 {"rwr", HB_TAG('M','A','W',' ')}, /* Marwari (India) */
693 {"sa", HB_TAG('S','A','N',' ')}, /* Sanskrit */
694 {"sah", HB_TAG('Y','A','K',' ')}, /* Yakut */
695 {"sam", HB_TAG('P','A','A',' ')}, /* Palestinian Aramaic */
696 {"sas", HB_TAG('S','A','S',' ')}, /* Sasak */
697 {"sat", HB_TAG('S','A','T',' ')}, /* Santali */
698 {"sc", HB_TAG('S','R','D',' ')}, /* Sardinian [macrolanguage] */
699 {"sck", HB_TAG('S','A','D',' ')}, /* Sadri */
700 {"scn", HB_TAG('S','C','N',' ')}, /* Sicilian */
701 {"sco", HB_TAG('S','C','O',' ')}, /* Scots */
702 {"scs", HB_TAG('S','L','A',' ')}, /* [North] Slavey */
703 {"sd", HB_TAG('S','N','D',' ')}, /* Sindhi */
704 {"se", HB_TAG('N','S','M',' ')}, /* Northern Sami */
705 {"seh", HB_TAG('S','N','A',' ')}, /* Sena */
706 {"sel", HB_TAG('S','E','L',' ')}, /* Selkup */
707 {"sez", HB_TAG('Q','I','N',' ')}, /* Senthang Chin */
708 {"sg", HB_TAG('S','G','O',' ')}, /* Sango */
709 {"sga", HB_TAG('S','G','A',' ')}, /* Old Irish (to 900) */
710 {"sgs", HB_TAG('S','G','S',' ')}, /* Samogitian */
711 {"sgw", HB_TAG('C','H','G',' ')}, /* Sebat Bet Gurage */
712 /*{"sgw", HB_TAG('S','G','W',' ')},*/ /* Sebat Bet Gurage (in SIL fonts) */
713 {"shi", HB_TAG('S','H','I',' ')}, /* Tachelhit */
714 {"shn", HB_TAG('S','H','N',' ')}, /* Shan */
715 {"si", HB_TAG('S','N','H',' ')}, /* Sinhala */
716 {"sid", HB_TAG('S','I','D',' ')}, /* Sidamo */
717 {"sjd", HB_TAG('K','S','M',' ')}, /* Kildin Sami */
718 {"sk", HB_TAG('S','K','Y',' ')}, /* Slovak */
719 {"skr", HB_TAG('S','R','K',' ')}, /* Seraiki */
720 {"sl", HB_TAG('S','L','V',' ')}, /* Slovenian */
721 {"sm", HB_TAG('S','M','O',' ')}, /* Samoan */
722 {"sma", HB_TAG('S','S','M',' ')}, /* Southern Sami */
723 {"smj", HB_TAG('L','S','M',' ')}, /* Lule Sami */
724 {"smn", HB_TAG('I','S','M',' ')}, /* Inari Sami */
725 {"sms", HB_TAG('S','K','S',' ')}, /* Skolt Sami */
726 {"sn", HB_TAG('S','N','A','0')}, /* Shona */
727 {"snk", HB_TAG('S','N','K',' ')}, /* Soninke */
728 {"so", HB_TAG('S','M','L',' ')}, /* Somali */
729 {"sop", HB_TAG('S','O','P',' ')}, /* Songe */
730 {"sq", HB_TAG('S','Q','I',' ')}, /* Albanian [macrolanguage] */
731 {"sr", HB_TAG('S','R','B',' ')}, /* Serbian */
732 {"srr", HB_TAG('S','R','R',' ')}, /* Serer */
733 {"ss", HB_TAG('S','W','Z',' ')}, /* Swati */
734 {"st", HB_TAG('S','O','T',' ')}, /* [Southern] Sotho */
735 {"stq", HB_TAG('S','T','Q',' ')}, /* Saterfriesisch */
736 {"stv", HB_TAG('S','I','G',' ')}, /* Silt'e */
737 {"su", HB_TAG('S','U','N',' ')}, /* Sundanese */
738 {"suk", HB_TAG('S','U','K',' ')}, /* Sukama */
739 {"suq", HB_TAG('S','U','R',' ')}, /* Suri */
740 {"sv", HB_TAG('S','V','E',' ')}, /* Swedish */
741 {"sva", HB_TAG('S','V','A',' ')}, /* Svan */
742 {"sw", HB_TAG('S','W','K',' ')}, /* Swahili [macrolanguage] */
743 {"swb", HB_TAG('C','M','R',' ')}, /* Comorian */
744 {"swh", HB_TAG('S','W','K',' ')}, /* Kiswahili/Swahili */
745 {"swv", HB_TAG('M','A','W',' ')}, /* Shekhawati */
746 {"sxu", HB_TAG('S','X','U',' ')}, /* Upper Saxon */
747 {"syc", HB_TAG('S','Y','R',' ')}, /* Classical Syriac */
748 {"syl", HB_TAG('S','Y','L',' ')}, /* Sylheti */
749 {"syr", HB_TAG('S','Y','R',' ')}, /* Syriac [macrolanguage] */
750 {"szl", HB_TAG('S','Z','L',' ')}, /* Silesian */
751 {"ta", HB_TAG('T','A','M',' ')}, /* Tamil */
752 {"tab", HB_TAG('T','A','B',' ')}, /* Tabasaran */
753 {"tcp", HB_TAG('Q','I','N',' ')}, /* Tawr Chin */
754 {"tcy", HB_TAG('T','U','L',' ')}, /* Tulu */
755 {"tcz", HB_TAG('Q','I','N',' ')}, /* Thado Chin */
756 {"tdd", HB_TAG('T','D','D',' ')}, /* Tai Nüa */
757 {"te", HB_TAG('T','E','L',' ')}, /* Telugu */
758 {"tem", HB_TAG('T','M','N',' ')}, /* Temne */
759 {"tet", HB_TAG('T','E','T',' ')}, /* Tetum */
760 {"tg", HB_TAG('T','A','J',' ')}, /* Tajik */
761 {"th", HB_TAG('T','H','A',' ')}, /* Thai */
762 {"ti", HB_TAG('T','G','Y',' ')}, /* Tigrinya */
763 {"tig", HB_TAG('T','G','R',' ')}, /* Tigre */
764 {"tiv", HB_TAG('T','I','V',' ')}, /* Tiv */
765 {"tk", HB_TAG('T','K','M',' ')}, /* Turkmen */
766 {"tl", HB_TAG('T','G','L',' ')}, /* Tagalog */
767 {"tmh", HB_TAG('T','M','H',' ')}, /* Tamashek */
768 {"tn", HB_TAG('T','N','A',' ')}, /* Tswana */
769 {"to", HB_TAG('T','G','N',' ')}, /* Tonga (Tonga Islands) */
770 {"tod", HB_TAG('T','O','D','0')}, /* Toma */
771 {"toi", HB_TAG('T','N','G',' ')}, /* Tonga */
772 {"tpi", HB_TAG('T','P','I',' ')}, /* Tok Pisin */
773 {"tr", HB_TAG('T','R','K',' ')}, /* Turkish */
774 {"tru", HB_TAG('T','U','A',' ')}, /* Turoyo Aramaic */
775 {"ts", HB_TAG('T','S','G',' ')}, /* Tsonga */
776 {"tt", HB_TAG('T','A','T',' ')}, /* Tatar */
777 {"tum", HB_TAG('T','U','M',' ')}, /* Tumbuka */
778 {"tvl", HB_TAG('T','V','L',' ')}, /* Tuvalu */
779 {"tw", HB_TAG('T','W','I',' ')}, /* Twi */
780 {"ty", HB_TAG('T','H','T',' ')}, /* Tahitian */
781 {"tyv", HB_TAG('T','U','V',' ')}, /* Tuvin */
782 {"tyz", HB_TAG('T','Y','Z',' ')}, /* Tày */
783 {"tzm", HB_TAG('T','Z','M',' ')}, /* Central Atlas Tamazight */
784 {"tzo", HB_TAG('T','Z','O',' ')}, /* Tzotzil */
785 {"udm", HB_TAG('U','D','M',' ')}, /* Udmurt */
786 {"ug", HB_TAG('U','Y','G',' ')}, /* Uighur */
787 {"uk", HB_TAG('U','K','R',' ')}, /* Ukrainian */
788 {"umb", HB_TAG('U','M','B',' ')}, /* Umbundu */
789 {"unr", HB_TAG('M','U','N',' ')}, /* Mundari */
790 {"ur", HB_TAG('U','R','D',' ')}, /* Urdu */
791 {"uz", HB_TAG('U','Z','B',' ')}, /* Uzbek [macrolanguage] */
792 {"uzn", HB_TAG('U','Z','B',' ')}, /* Northern Uzbek */
793 {"uzs", HB_TAG('U','Z','B',' ')}, /* Southern Uzbek */
794 {"ve", HB_TAG('V','E','N',' ')}, /* Venda */
795 {"vec", HB_TAG('V','E','C',' ')}, /* Venetian */
796 {"vi", HB_TAG('V','I','T',' ')}, /* Vietnamese */
797 {"vls", HB_TAG('F','L','E',' ')}, /* Vlaams */
798 {"vmw", HB_TAG('M','A','K',' ')}, /* Makhuwa */
799 {"vo", HB_TAG('V','O','L',' ')}, /* Volapük */
800 {"vro", HB_TAG('V','R','O',' ')}, /* Võro */
801 {"wa", HB_TAG('W','L','N',' ')}, /* Walloon */
802 {"war", HB_TAG('W','A','R',' ')}, /* Waray (Philippines) */
803 {"wbm", HB_TAG('W','A',' ',' ')}, /* Wa */
804 {"wbr", HB_TAG('W','A','G',' ')}, /* Wagdi */
805 {"wle", HB_TAG('S','I','G',' ')}, /* Wolane */
806 {"wo", HB_TAG('W','L','F',' ')}, /* Wolof */
807 {"wry", HB_TAG('M','A','W',' ')}, /* Merwari */
808 {"wtm", HB_TAG('W','T','M',' ')}, /* Mewati */
809 {"xal", HB_TAG('K','L','M',' ')}, /* Kalmyk */
810 {"xan", HB_TAG('S','E','K',' ')}, /* Sekota */
811 {"xh", HB_TAG('X','H','S',' ')}, /* Xhosa */
812 {"xjb", HB_TAG('X','J','B',' ')}, /* Minjangbal */
813 {"xog", HB_TAG('X','O','G',' ')}, /* Soga */
814 {"xom", HB_TAG('K','M','O',' ')}, /* Komo (Sudan) */
815 {"xpe", HB_TAG('X','P','E',' ')}, /* Kpelle (Liberia) */
816 {"xsl", HB_TAG('S','S','L',' ')}, /* South Slavey */
817 {"xst", HB_TAG('S','I','G',' ')}, /* Silt'e (retired code) */
818 {"xwo", HB_TAG('T','O','D',' ')}, /* Written Oirat (Todo) */
819 {"yao", HB_TAG('Y','A','O',' ')}, /* Yao */
820 {"yap", HB_TAG('Y','A','P',' ')}, /* Yapese */
821 {"yi", HB_TAG('J','I','I',' ')}, /* Yiddish [macrolanguage] */
822 {"yo", HB_TAG('Y','B','A',' ')}, /* Yoruba */
823 {"yos", HB_TAG('Q','I','N',' ')}, /* Yos, deprecated by IANA in favor of Zou [zom] */
824 {"yso", HB_TAG('N','I','S',' ')}, /* Nisi (China) */
825 {"za", HB_TAG('Z','H','A',' ')}, /* Chuang/Zhuang [macrolanguage] */
826 {"zea", HB_TAG('Z','E','A',' ')}, /* Zeeuws */
827 {"zgh", HB_TAG('Z','G','H',' ')}, /* Standard Morrocan Tamazigh */
828 {"zne", HB_TAG('Z','N','D',' ')}, /* Zande */
829 {"zom", HB_TAG('Q','I','N',' ')}, /* Zou */
830 {"zu", HB_TAG('Z','U','L',' ')}, /* Zulu */
831 {"zum", HB_TAG('L','R','C',' ')}, /* Kumzari */
832 {"zza", HB_TAG('Z','Z','A',' ')}, /* Zazaki */
833
834 /* The corresponding languages IDs for the following IDs are unclear,
835 * overlap, or are architecturally weird. Needs more research. */
836
837 /*{"chp", HB_TAG('S','A','Y',' ')},*/ /* Sayisi */
838 /*{"cwd", HB_TAG('T','C','R',' ')},*/ /* TH-Cree */
839 /*{"emk", HB_TAG('E','M','K',' ')},*/ /* Eastern Maninkakan */
840 /*{"krc", HB_TAG('B','A','L',' ')},*/ /* Balkar */
841 /*{"??", HB_TAG('B','C','R',' ')},*/ /* Bible Cree */
842 /*{"zh?", HB_TAG('C','H','N',' ')},*/ /* Chinese (seen in Microsoft fonts) */
843 /*{"ar-Syrc?", HB_TAG('G','A','R',' ')},*/ /* Garshuni */
844 /*{"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) */
845 /*{"ga-Latg?/" HB_TAG('I','R','T',' ')},*/ /* Irish Traditional */
846 /*{"krc", HB_TAG('K','A','R',' ')},*/ /* Karachay */
847 /*{"ka-Geok?", HB_TAG('K','G','E',' ')},*/ /* Khutsuri Georgian */
848 /*{"kca", HB_TAG('K','H','K',' ')},*/ /* Khanty-Kazim */
849 /*{"kca", HB_TAG('K','H','S',' ')},*/ /* Khanty-Shurishkar */
850 /*{"kca", HB_TAG('K','H','V',' ')},*/ /* Khanty-Vakhi */
851 /*{"kqs, kss", HB_TAG('K','I','S',' ')},*/ /* Kisii */
852 /*{"lua", HB_TAG('L','U','A',' ')},*/ /* Luba-Lulua */
853 /*{"mlq", HB_TAG('M','L','N',' ')},*/ /* Malinke */
854 /*{"nso", HB_TAG('N','S','O',' ')},*/ /* Sotho, Northern */
855 /*{"??", HB_TAG('M','A','L',' ')},*/ /* Malayalam Traditional */
856 /*{"csw", HB_TAG('N','C','R',' ')},*/ /* N-Cree */
857 /*{"csw", HB_TAG('N','H','C',' ')},*/ /* Norway House Cree */
858 /*{"el-polyton", HB_TAG('P','G','R',' ')},*/ /* Polytonic Greek */
859 /*{"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 */
860 /*{"??", HB_TAG('Y','I','C',' ')},*/ /* Yi Classic */
861 /*{"zh-Latn-pinyin", HB_TAG('Z','H','P',' ')},*/ /* Chinese Phonetic */
862 };
863
864 typedef struct {
865 char language[11];
866 hb_tag_t tag;
867 } LangTagLong;
868 static const LangTagLong ot_languages_zh[] = {
869 /* Store longest-first, if one is a prefix of another. */
870 {"zh-cn", HB_TAG('Z','H','S',' ')}, /* Chinese (China) */
871 {"zh-hk", HB_TAG('Z','H','H',' ')}, /* Chinese (Hong Kong) */
872 {"zh-mo", HB_TAG('Z','H','H',' ')}, /* Chinese (Macao) */
873 {"zh-sg", HB_TAG('Z','H','S',' ')}, /* Chinese (Singapore) */
874 {"zh-tw", HB_TAG('Z','H','T',' ')}, /* Chinese (Taiwan) */
875 {"zh-hans", HB_TAG('Z','H','S',' ')}, /* Chinese (Simplified) */
876 {"zh-hant-hk",HB_TAG('Z','H','H',' ')}, /* Chinese (Hong Kong) */
877 {"zh-hant-mo",HB_TAG('Z','H','H',' ')}, /* Chinese (Macao) */
878 {"zh-hant", HB_TAG('Z','H','T',' ')}, /* Chinese (Traditional) */
879 };
880
881 static int
lang_compare_first_component(const void * pa,const void * pb)882 lang_compare_first_component (const void *pa,
883 const void *pb)
884 {
885 const char *a = (const char *) pa;
886 const char *b = (const char *) pb;
887 unsigned int da, db;
888 const char *p;
889
890 p = strchr (a, '-');
891 da = p ? (unsigned int) (p - a) : strlen (a);
892
893 p = strchr (b, '-');
894 db = p ? (unsigned int) (p - b) : strlen (b);
895
896 return strncmp (a, b, MAX (da, db));
897 }
898
899 static hb_bool_t
lang_matches(const char * lang_str,const char * spec)900 lang_matches (const char *lang_str, const char *spec)
901 {
902 unsigned int len = strlen (spec);
903
904 return strncmp (lang_str, spec, len) == 0 &&
905 (lang_str[len] == '\0' || lang_str[len] == '-');
906 }
907
908 hb_tag_t
hb_ot_tag_from_language(hb_language_t language)909 hb_ot_tag_from_language (hb_language_t language)
910 {
911 const char *lang_str, *s;
912
913 if (language == HB_LANGUAGE_INVALID)
914 return HB_OT_TAG_DEFAULT_LANGUAGE;
915
916 lang_str = hb_language_to_string (language);
917
918 s = strstr (lang_str, "x-hbot");
919 if (s) {
920 char tag[4];
921 int i;
922 s += 6;
923 for (i = 0; i < 4 && ISALNUM (s[i]); i++)
924 tag[i] = TOUPPER (s[i]);
925 if (i) {
926 for (; i < 4; i++)
927 tag[i] = ' ';
928 return HB_TAG (tag[0], tag[1], tag[2], tag[3]);
929 }
930 }
931
932 /*
933 * "fonipa" is a variant tag in BCP-47, meaning the International Phonetic Alphabet.
934 * It can be applied to any language.
935 */
936 if (strstr (lang_str, "-fonipa")) {
937 return HB_TAG('I','P','P','H'); /* Phonetic transcription—IPA conventions */
938 }
939
940 /*
941 * "fonnapa" is a variant tag in BCP-47, meaning the North American Phonetic Alphabet
942 * also known as Americanist Phonetic Notation. It can be applied to any language.
943 */
944 if (strstr (lang_str, "-fonnapa")) {
945 return HB_TAG('A','P','P','H'); /* Phonetic transcription—Americanist conventions */
946 }
947
948 /*
949 * "Syre" is a BCP-47 script tag, meaning the Estrangela variant of the Syriac script.
950 * It can be applied to any language.
951 */
952 if (strstr (lang_str, "-syre")) {
953 return HB_TAG('S','Y','R','E'); /* Estrangela Syriac */
954 }
955
956 /*
957 * "Syrj" is a BCP-47 script tag, meaning the Western variant of the Syriac script.
958 * It can be applied to any language.
959 */
960 if (strstr (lang_str, "-syrj")) {
961 return HB_TAG('S','Y','R','J'); /* Western Syriac */
962 }
963
964 /*
965 * "Syrn" is a BCP-47 script tag, meaning the Eastern variant of the Syriac script.
966 * It can be applied to any language.
967 */
968 if (strstr (lang_str, "-syrn")) {
969 return HB_TAG('S','Y','R','N'); /* Eastern Syriac */
970 }
971
972 /* Find a language matching in the first component */
973 {
974 const LangTag *lang_tag;
975 lang_tag = (LangTag *) bsearch (lang_str, ot_languages,
976 ARRAY_LENGTH (ot_languages), sizeof (LangTag),
977 lang_compare_first_component);
978 if (lang_tag)
979 return lang_tag->tag;
980 }
981
982 /* Otherwise, check the Chinese ones */
983 if (0 == lang_compare_first_component (lang_str, "zh"))
984 {
985 unsigned int i;
986
987 for (i = 0; i < ARRAY_LENGTH (ot_languages_zh); i++)
988 {
989 const LangTagLong *lang_tag;
990 lang_tag = &ot_languages_zh[i];
991 if (lang_matches (lang_str, lang_tag->language))
992 return lang_tag->tag;
993 }
994
995 /* Otherwise just return 'ZHS ' */
996 return HB_TAG('Z','H','S',' ');
997 }
998
999 s = strchr (lang_str, '-');
1000 if (!s)
1001 s = lang_str + strlen (lang_str);
1002 if (s - lang_str == 3) {
1003 /* Assume it's ISO-639-3 and upper-case and use it. */
1004 return hb_tag_from_string (lang_str, s - lang_str) & ~0x20202000u;
1005 }
1006
1007 return HB_OT_TAG_DEFAULT_LANGUAGE;
1008 }
1009
1010 /**
1011 * hb_ot_tag_to_language:
1012 *
1013 *
1014 *
1015 * Return value: (transfer none):
1016 *
1017 * Since: 0.9.2
1018 **/
1019 hb_language_t
hb_ot_tag_to_language(hb_tag_t tag)1020 hb_ot_tag_to_language (hb_tag_t tag)
1021 {
1022 unsigned int i;
1023
1024 if (tag == HB_OT_TAG_DEFAULT_LANGUAGE)
1025 return nullptr;
1026
1027 /* struct LangTag has only room for 3-letter language tags. */
1028 switch (tag) {
1029 case HB_TAG('A','P','P','H'): /* Phonetic transcription—Americanist conventions */
1030 return hb_language_from_string ("und-fonnapa", -1);
1031 case HB_TAG('I','P','P','H'): /* Phonetic transcription—IPA conventions */
1032 return hb_language_from_string ("und-fonipa", -1);
1033 case HB_TAG('S','Y','R',' '): /* Syriac [macrolanguage] */
1034 return hb_language_from_string ("syr", -1);
1035 case HB_TAG('S','Y','R','E'): /* Estrangela Syriac */
1036 return hb_language_from_string ("und-Syre", -1);
1037 case HB_TAG('S','Y','R','J'): /* Western Syriac */
1038 return hb_language_from_string ("und-Syrj", -1);
1039 case HB_TAG('S','Y','R','N'): /* Eastern Syriac */
1040 return hb_language_from_string ("und-Syrn", -1);
1041 }
1042
1043 for (i = 0; i < ARRAY_LENGTH (ot_languages); i++)
1044 if (ot_languages[i].tag == tag)
1045 return hb_language_from_string (ot_languages[i].language, -1);
1046
1047 /* If tag starts with ZH, it's Chinese */
1048 if ((tag & 0xFFFF0000u) == 0x5A480000u) {
1049 switch (tag) {
1050 case HB_TAG('Z','H','H',' '): return hb_language_from_string ("zh-hk", -1); /* Hong Kong */
1051 case HB_TAG('Z','H','S',' '): return hb_language_from_string ("zh-Hans", -1); /* Simplified */
1052 case HB_TAG('Z','H','T',' '): return hb_language_from_string ("zh-Hant", -1); /* Traditional */
1053 default: break; /* Fall through */
1054 }
1055 }
1056
1057 /* Else return a custom language in the form of "x-hbotABCD" */
1058 {
1059 unsigned char buf[11] = "x-hbot";
1060 buf[6] = tag >> 24;
1061 buf[7] = (tag >> 16) & 0xFF;
1062 buf[8] = (tag >> 8) & 0xFF;
1063 buf[9] = tag & 0xFF;
1064 if (buf[9] == 0x20)
1065 buf[9] = '\0';
1066 buf[10] = '\0';
1067 return hb_language_from_string ((char *) buf, -1);
1068 }
1069 }
1070
1071 #ifdef MAIN
1072 static inline void
test_langs_sorted(void)1073 test_langs_sorted (void)
1074 {
1075 for (unsigned int i = 1; i < ARRAY_LENGTH (ot_languages); i++)
1076 {
1077 int c = lang_compare_first_component (ot_languages[i-1].language, ot_languages[i].language);
1078 if (c >= 0)
1079 {
1080 fprintf (stderr, "ot_languages not sorted at index %d: %s %d %s\n",
1081 i, ot_languages[i-1].language, c, ot_languages[i].language);
1082 abort();
1083 }
1084 }
1085 }
1086
1087 int
main(void)1088 main (void)
1089 {
1090 test_langs_sorted ();
1091 return 0;
1092 }
1093
1094 #endif
1095