1 /*
2 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
3 *
4 * This code is free software; you can redistribute it and/or modify it
5 * under the terms of the GNU General Public License version 2 only, as
6 * published by the Free Software Foundation. Oracle designates this
7 * particular file as subject to the "Classpath" exception as provided
8 * by Oracle in the LICENSE file that accompanied this code.
9 *
10 * This code is distributed in the hope that it will be useful, but WITHOUT
11 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
12 * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
13 * version 2 for more details (a copy is included in the LICENSE file that
14 * accompanied this code).
15 *
16 * You should have received a copy of the GNU General Public License version
17 * 2 along with this work; if not, write to the Free Software Foundation,
18 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
19 *
20 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
21 * or visit www.oracle.com if you need additional information or have any
22 * questions.
23 */
24
25 // This file is available under and governed by the GNU General Public
26 // License version 2 only, as published by the Free Software Foundation.
27 // However, the following notice accompanied the original version of this
28 // file:
29 //
30 /*
31 * Copyright © 2009 Red Hat, Inc.
32 * Copyright © 2011 Google, Inc.
33 *
34 * This is part of HarfBuzz, a text shaping library.
35 *
36 * Permission is hereby granted, without written agreement and without
37 * license or royalty fees, to use, copy, modify, and distribute this
38 * software and its documentation for any purpose, provided that the
39 * above copyright notice and the following two paragraphs appear in
40 * all copies of this software.
41 *
42 * IN NO EVENT SHALL THE COPYRIGHT HOLDER BE LIABLE TO ANY PARTY FOR
43 * DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES
44 * ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS DOCUMENTATION, EVEN
45 * IF THE COPYRIGHT HOLDER HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH
46 * DAMAGE.
47 *
48 * THE COPYRIGHT HOLDER SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING,
49 * BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
50 * FITNESS FOR A PARTICULAR PURPOSE. THE SOFTWARE PROVIDED HEREUNDER IS
51 * ON AN "AS IS" BASIS, AND THE COPYRIGHT HOLDER HAS NO OBLIGATION TO
52 * PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
53 *
54 * Red Hat Author(s): Behdad Esfahbod
55 * Google Author(s): Behdad Esfahbod, Roozbeh Pournader
56 */
57
58 #include "hb-private.hh"
59
60 #include <string.h>
61
62
63
64 /* hb_script_t */
65
66 static hb_tag_t
67 hb_ot_old_tag_from_script (hb_script_t script)
68 {
69 /* This seems to be accurate as of end of 2012. */
70
71 switch ((hb_tag_t) script) {
72 case HB_SCRIPT_INVALID: return HB_OT_TAG_DEFAULT_SCRIPT;
73
74 /* KATAKANA and HIRAGANA both map to 'kana' */
75 case HB_SCRIPT_HIRAGANA: return HB_TAG('k','a','n','a');
76
77 /* Spaces at the end are preserved, unlike ISO 15924 */
78 case HB_SCRIPT_LAO: return HB_TAG('l','a','o',' ');
79 case HB_SCRIPT_YI: return HB_TAG('y','i',' ',' ');
80 /* Unicode-5.0 additions */
81 case HB_SCRIPT_NKO: return HB_TAG('n','k','o',' ');
82 /* Unicode-5.1 additions */
83 case HB_SCRIPT_VAI: return HB_TAG('v','a','i',' ');
84 /* Unicode-5.2 additions */
85 /* Unicode-6.0 additions */
86 }
87
88 /* Else, just change first char to lowercase and return */
89 return ((hb_tag_t) script) | 0x20000000u;
90 }
91
92 static hb_script_t
93 hb_ot_old_tag_to_script (hb_tag_t tag)
94 {
95 if (unlikely (tag == HB_OT_TAG_DEFAULT_SCRIPT))
96 return HB_SCRIPT_INVALID;
97
98 /* This side of the conversion is fully algorithmic. */
99
100 /* Any spaces at the end of the tag are replaced by repeating the last
101 * letter. Eg 'nko ' -> 'Nkoo' */
102 if (unlikely ((tag & 0x0000FF00u) == 0x00002000u))
103 tag |= (tag >> 8) & 0x0000FF00u; /* Copy second letter to third */
104 if (unlikely ((tag & 0x000000FFu) == 0x00000020u))
105 tag |= (tag >> 8) & 0x000000FFu; /* Copy third letter to fourth */
106
107 /* Change first char to uppercase and return */
108 return (hb_script_t) (tag & ~0x20000000u);
109 }
110
111 static hb_tag_t
112 hb_ot_new_tag_from_script (hb_script_t script)
113 {
114 switch ((hb_tag_t) script) {
115 case HB_SCRIPT_BENGALI: return HB_TAG('b','n','g','2');
116 case HB_SCRIPT_DEVANAGARI: return HB_TAG('d','e','v','2');
117 case HB_SCRIPT_GUJARATI: return HB_TAG('g','j','r','2');
118 case HB_SCRIPT_GURMUKHI: return HB_TAG('g','u','r','2');
119 case HB_SCRIPT_KANNADA: return HB_TAG('k','n','d','2');
120 case HB_SCRIPT_MALAYALAM: return HB_TAG('m','l','m','2');
121 case HB_SCRIPT_ORIYA: return HB_TAG('o','r','y','2');
122 case HB_SCRIPT_TAMIL: return HB_TAG('t','m','l','2');
123 case HB_SCRIPT_TELUGU: return HB_TAG('t','e','l','2');
124 case HB_SCRIPT_MYANMAR: return HB_TAG('m','y','m','2');
125 }
126
127 return HB_OT_TAG_DEFAULT_SCRIPT;
128 }
129
130 static hb_script_t
131 hb_ot_new_tag_to_script (hb_tag_t tag)
132 {
133 switch (tag) {
134 case HB_TAG('b','n','g','2'): return HB_SCRIPT_BENGALI;
135 case HB_TAG('d','e','v','2'): return HB_SCRIPT_DEVANAGARI;
136 case HB_TAG('g','j','r','2'): return HB_SCRIPT_GUJARATI;
137 case HB_TAG('g','u','r','2'): return HB_SCRIPT_GURMUKHI;
138 case HB_TAG('k','n','d','2'): return HB_SCRIPT_KANNADA;
139 case HB_TAG('m','l','m','2'): return HB_SCRIPT_MALAYALAM;
140 case HB_TAG('o','r','y','2'): return HB_SCRIPT_ORIYA;
141 case HB_TAG('t','m','l','2'): return HB_SCRIPT_TAMIL;
142 case HB_TAG('t','e','l','2'): return HB_SCRIPT_TELUGU;
143 case HB_TAG('m','y','m','2'): return HB_SCRIPT_MYANMAR;
144 }
145
146 return HB_SCRIPT_UNKNOWN;
147 }
148
149 /*
150 * Complete list at:
151 * https://www.microsoft.com/typography/otspec/scripttags.htm
152 * https://www.microsoft.com/typography/otspec160/scripttagsProposed.htm
153 *
154 * Most of the script tags are the same as the ISO 15924 tag but lowercased.
155 * So we just do that, and handle the exceptional cases in a switch.
156 */
157
158 void
159 hb_ot_tags_from_script (hb_script_t script,
160 hb_tag_t *script_tag_1,
161 hb_tag_t *script_tag_2)
162 {
163 hb_tag_t new_tag;
164
165 *script_tag_2 = HB_OT_TAG_DEFAULT_SCRIPT;
166 *script_tag_1 = hb_ot_old_tag_from_script (script);
167
168 new_tag = hb_ot_new_tag_from_script (script);
169 if (unlikely (new_tag != HB_OT_TAG_DEFAULT_SCRIPT)) {
170 *script_tag_2 = *script_tag_1;
171 *script_tag_1 = new_tag;
172 }
173 }
174
175 hb_script_t
176 hb_ot_tag_to_script (hb_tag_t tag)
177 {
178 if (unlikely ((tag & 0x000000FFu) == '2'))
179 return hb_ot_new_tag_to_script (tag);
180
181 return hb_ot_old_tag_to_script (tag);
182 }
183
184
185 /* hb_language_t */
186
187 typedef struct {
188 char language[4];
189 hb_tag_t tag;
190 } LangTag;
191
192 /*
193 * Complete list at:
194 * http://www.microsoft.com/typography/otspec/languagetags.htm
195 *
196 * Generated by intersecting the OpenType language tag list from
197 * Draft OpenType 1.5 spec, with with the ISO 639-3 codes from
198 * 2008-08-04, matching on name, and finally adjusted manually.
199 *
200 * Updated on 2012-12-07 with more research into remaining codes.
201 *
202 * Updated on 2013-11-23 based on usage in SIL and Microsoft fonts,
203 * the new proposal from Microsoft, and latest ISO 639-3 names.
204 *
205 * Some items still missing. Those are commented out at the end.
206 * Keep sorted for bsearch.
207 *
208 * Updated as of 2015-05-06: OT1.7 on MS website has some newer
209 * items that we don't have here, eg. Zazaki. This is the new
210 * items in OpenType 1.7 (red items), most of which we have:
211 * http://www.microsoft.com/typography/otspec170/languagetags.htm
212 */
213
214 static const LangTag ot_languages[] = {
215 {"aa", HB_TAG('A','F','R',' ')}, /* Afar */
216 {"ab", HB_TAG('A','B','K',' ')}, /* Abkhazian */
217 {"abq", HB_TAG('A','B','A',' ')}, /* Abaza */
218 {"ach", HB_TAG('A','C','H',' ')}, /* Acoli */
219 {"ada", HB_TAG('D','N','G',' ')}, /* Dangme */
220 {"ady", HB_TAG('A','D','Y',' ')}, /* Adyghe */
221 {"af", HB_TAG('A','F','K',' ')}, /* Afrikaans */
222 {"aii", HB_TAG('S','W','A',' ')}, /* Swadaya Aramaic */
223 {"aio", HB_TAG('A','I','O',' ')}, /* Aiton */
224 {"aiw", HB_TAG('A','R','I',' ')}, /* Aari */
225 {"ak", HB_TAG('T','W','I',' ')}, /* Akan [macrolanguage] */
226 {"alt", HB_TAG('A','L','T',' ')}, /* [Southern] Altai */
227 {"am", HB_TAG('A','M','H',' ')}, /* Amharic */
228 {"amf", HB_TAG('H','B','N',' ')}, /* Hammer-Banna */
229 {"an", HB_TAG('A','R','G',' ')}, /* Aragonese */
230 {"ang", HB_TAG('A','N','G',' ')}, /* Old English (ca. 450-1100) */
231 {"ar", HB_TAG('A','R','A',' ')}, /* Arabic [macrolanguage] */
232 {"arb", HB_TAG('A','R','A',' ')}, /* Standard Arabic */
233 {"arn", HB_TAG('M','A','P',' ')}, /* Mapudungun */
234 {"ary", HB_TAG('M','O','R',' ')}, /* Moroccan Arabic */
235 {"as", HB_TAG('A','S','M',' ')}, /* Assamese */
236 {"ast", HB_TAG('A','S','T',' ')}, /* Asturian/Asturleonese/Bable/Leonese */
237 {"ath", HB_TAG('A','T','H',' ')}, /* Athapaskan [family] */
238 {"atv", HB_TAG('A','L','T',' ')}, /* [Northern] Altai */
239 {"av", HB_TAG('A','V','R',' ')}, /* Avaric */
240 {"awa", HB_TAG('A','W','A',' ')}, /* Awadhi */
241 {"ay", HB_TAG('A','Y','M',' ')}, /* Aymara [macrolanguage] */
242 {"az", HB_TAG('A','Z','E',' ')}, /* Azerbaijani [macrolanguage] */
243 {"azb", HB_TAG('A','Z','B',' ')}, /* South Azerbaijani */
244 {"azj", HB_TAG('A','Z','E',' ')}, /* North Azerbaijani */
245 {"ba", HB_TAG('B','S','H',' ')}, /* Bashkir */
246 {"bai", HB_TAG('B','M','L',' ')}, /* Bamileke [family] */
247 {"bal", HB_TAG('B','L','I',' ')}, /* Baluchi [macrolangauge] */
248 {"ban", HB_TAG('B','A','N',' ')}, /* Balinese */
249 {"bar", HB_TAG('B','A','R',' ')}, /* Bavarian */
250 {"bbc", HB_TAG('B','B','C',' ')}, /* Batak Toba */
251 {"bci", HB_TAG('B','A','U',' ')}, /* Baoulé */
252 {"bcl", HB_TAG('B','I','K',' ')}, /* Central Bikol */
253 {"bcq", HB_TAG('B','C','H',' ')}, /* Bench */
254 {"be", HB_TAG('B','E','L',' ')}, /* Belarusian */
255 {"bem", HB_TAG('B','E','M',' ')}, /* Bemba (Zambia) */
256 {"ber", HB_TAG('B','E','R',' ')}, /* Berber [family] */
257 {"bfq", HB_TAG('B','A','D',' ')}, /* Badaga */
258 {"bft", HB_TAG('B','L','T',' ')}, /* Balti */
259 {"bfy", HB_TAG('B','A','G',' ')}, /* Baghelkhandi */
260 {"bg", HB_TAG('B','G','R',' ')}, /* Bulgarian */
261 {"bgc", HB_TAG('B','G','C',' ')}, /* Haryanvi */
262 {"bgq", HB_TAG('B','G','Q',' ')}, /* Bagri */
263 {"bhb", HB_TAG('B','H','I',' ')}, /* Bhili */
264 {"bhk", HB_TAG('B','I','K',' ')}, /* Albay Bicolano (retired code) */
265 {"bho", HB_TAG('B','H','O',' ')}, /* Bhojpuri */
266 {"bi", HB_TAG('B','I','S',' ')}, /* Bislama */
267 {"bik", HB_TAG('B','I','K',' ')}, /* Bikol [macrolanguage] */
268 {"bin", HB_TAG('E','D','O',' ')}, /* Bini */
269 {"bjj", HB_TAG('B','J','J',' ')}, /* Kanauji */
270 {"bjt", HB_TAG('B','L','N',' ')}, /* Balanta-Ganja */
271 {"bla", HB_TAG('B','K','F',' ')}, /* Blackfoot */
272 {"ble", HB_TAG('B','L','N',' ')}, /* Balanta-Kentohe */
273 {"blk", HB_TAG('B','L','K',' ')}, /* Pa'O/Pa'o Karen */
274 {"bln", HB_TAG('B','I','K',' ')}, /* Southern Catanduanes Bikol */
275 {"bm", HB_TAG('B','M','B',' ')}, /* Bambara */
276 {"bn", HB_TAG('B','E','N',' ')}, /* Bengali */
277 {"bo", HB_TAG('T','I','B',' ')}, /* Tibetan */
278 {"bpy", HB_TAG('B','P','Y',' ')}, /* Bishnupriya */
279 {"bqi", HB_TAG('L','R','C',' ')}, /* Bakhtiari */
280 {"br", HB_TAG('B','R','E',' ')}, /* Breton */
281 {"bra", HB_TAG('B','R','I',' ')}, /* Braj Bhasha */
282 {"brh", HB_TAG('B','R','H',' ')}, /* Brahui */
283 {"brx", HB_TAG('B','R','X',' ')}, /* Bodo (India) */
284 {"bs", HB_TAG('B','O','S',' ')}, /* Bosnian */
285 {"btb", HB_TAG('B','T','I',' ')}, /* Beti (Cameroon) */
286 {"bto", HB_TAG('B','I','K',' ')}, /* Rinconada Bikol */
287 {"bts", HB_TAG('B','T','S',' ')}, /* Batak Simalungun */
288 {"bug", HB_TAG('B','U','G',' ')}, /* Buginese */
289 {"bxr", HB_TAG('R','B','U',' ')}, /* Russian Buriat */
290 {"byn", HB_TAG('B','I','L',' ')}, /* Bilen */
291 {"ca", HB_TAG('C','A','T',' ')}, /* Catalan */
292 {"cbk", HB_TAG('C','B','K',' ')}, /* Chavacano */
293 {"ce", HB_TAG('C','H','E',' ')}, /* Chechen */
294 {"ceb", HB_TAG('C','E','B',' ')}, /* Cebuano */
295 {"cgg", HB_TAG('C','G','G',' ')}, /* Chiga */
296 {"ch", HB_TAG('C','H','A',' ')}, /* Chamorro */
297 {"cho", HB_TAG('C','H','O',' ')}, /* Choctaw */
298 {"chp", HB_TAG('C','H','P',' ')}, /* Chipewyan */
299 {"chr", HB_TAG('C','H','R',' ')}, /* Cherokee */
300 {"chy", HB_TAG('C','H','Y',' ')}, /* Cheyenne */
301 {"ckb", HB_TAG('K','U','R',' ')}, /* Central Kurdish (Sorani) */
302 {"ckt", HB_TAG('C','H','K',' ')}, /* Chukchi */
303 {"cop", HB_TAG('C','O','P',' ')}, /* Coptic */
304 {"cr", HB_TAG('C','R','E',' ')}, /* Cree */
305 {"crh", HB_TAG('C','R','T',' ')}, /* Crimean Tatar */
306 {"crj", HB_TAG('E','C','R',' ')}, /* [Southern] East Cree */
307 {"crl", HB_TAG('E','C','R',' ')}, /* [Northern] East Cree */
308 {"crm", HB_TAG('M','C','R',' ')}, /* Moose Cree */
309 {"crx", HB_TAG('C','R','R',' ')}, /* Carrier */
310 {"cs", HB_TAG('C','S','Y',' ')}, /* Czech */
311 {"csb", HB_TAG('C','S','B',' ')}, /* Kashubian */
312 {"ctg", HB_TAG('C','T','G',' ')}, /* Chittagonian */
313 {"cts", HB_TAG('B','I','K',' ')}, /* Northern Catanduanes Bikol */
314 {"cu", HB_TAG('C','S','L',' ')}, /* Church Slavic */
315 {"cv", HB_TAG('C','H','U',' ')}, /* Chuvash */
316 {"cwd", HB_TAG('D','C','R',' ')}, /* Woods Cree */
317 {"cy", HB_TAG('W','E','L',' ')}, /* Welsh */
318 {"da", HB_TAG('D','A','N',' ')}, /* Danish */
319 {"dap", HB_TAG('N','I','S',' ')}, /* Nisi (India) */
320 {"dar", HB_TAG('D','A','R',' ')}, /* Dargwa */
321 {"de", HB_TAG('D','E','U',' ')}, /* German */
322 {"dgo", HB_TAG('D','G','O',' ')}, /* Dogri */
323 {"dhd", HB_TAG('M','A','W',' ')}, /* Dhundari */
324 {"din", HB_TAG('D','N','K',' ')}, /* Dinka [macrolanguage] */
325 {"diq", HB_TAG('D','I','Q',' ')}, /* Dimli */
326 {"dje", HB_TAG('D','J','R',' ')}, /* Zarma */
327 {"dng", HB_TAG('D','U','N',' ')}, /* Dungan */
328 {"doi", HB_TAG('D','G','R',' ')}, /* Dogri [macrolanguage] */
329 {"dsb", HB_TAG('L','S','B',' ')}, /* Lower Sorbian */
330 {"dv", HB_TAG('D','I','V',' ')}, /* Dhivehi/Divehi/Maldivian */
331 {"dyu", HB_TAG('J','U','L',' ')}, /* Jula */
332 {"dz", HB_TAG('D','Z','N',' ')}, /* Dzongkha */
333 {"ee", HB_TAG('E','W','E',' ')}, /* Ewe */
334 {"efi", HB_TAG('E','F','I',' ')}, /* Efik */
335 {"ekk", HB_TAG('E','T','I',' ')}, /* Standard Estonian */
336 {"el", HB_TAG('E','L','L',' ')}, /* Modern Greek (1453-) */
337 {"emk", HB_TAG('M','N','K',' ')}, /* Eastern Maninkakan */
338 {"en", HB_TAG('E','N','G',' ')}, /* English */
339 {"eo", HB_TAG('N','T','O',' ')}, /* Esperanto */
340 {"eot", HB_TAG('B','T','I',' ')}, /* Beti (Côte d'Ivoire) */
341 {"es", HB_TAG('E','S','P',' ')}, /* Spanish */
342 {"et", HB_TAG('E','T','I',' ')}, /* Estonian [macrolanguage] */
343 {"eu", HB_TAG('E','U','Q',' ')}, /* Basque */
344 {"eve", HB_TAG('E','V','N',' ')}, /* Even */
345 {"evn", HB_TAG('E','V','K',' ')}, /* Evenki */
346 {"fa", HB_TAG('F','A','R',' ')}, /* Persian [macrolanguage] */
347 {"ff", HB_TAG('F','U','L',' ')}, /* Fulah [macrolanguage] */
348 {"fi", HB_TAG('F','I','N',' ')}, /* Finnish */
349 {"fil", HB_TAG('P','I','L',' ')}, /* Filipino */
350 {"fj", HB_TAG('F','J','I',' ')}, /* Fijian */
351 {"fo", HB_TAG('F','O','S',' ')}, /* Faroese */
352 {"fon", HB_TAG('F','O','N',' ')}, /* Fon */
353 {"fr", HB_TAG('F','R','A',' ')}, /* French */
354 {"frc", HB_TAG('F','R','C',' ')}, /* Cajun French */
355 {"frp", HB_TAG('F','R','P',' ')}, /* Arpitan/Francoprovençal */
356 {"fur", HB_TAG('F','R','L',' ')}, /* Friulian */
357 {"fuv", HB_TAG('F','U','V',' ')}, /* Nigerian Fulfulde */
358 {"fy", HB_TAG('F','R','I',' ')}, /* Western Frisian */
359 {"ga", HB_TAG('I','R','I',' ')}, /* Irish */
360 {"gaa", HB_TAG('G','A','D',' ')}, /* Ga */
361 {"gag", HB_TAG('G','A','G',' ')}, /* Gagauz */
362 {"gbm", HB_TAG('G','A','W',' ')}, /* Garhwali */
363 {"gd", HB_TAG('G','A','E',' ')}, /* Scottish Gaelic */
364 {"gez", HB_TAG('G','E','Z',' ')}, /* Ge'ez */
365 {"ggo", HB_TAG('G','O','N',' ')}, /* Southern Gondi */
366 {"gl", HB_TAG('G','A','L',' ')}, /* Galician */
367 {"gld", HB_TAG('N','A','N',' ')}, /* Nanai */
368 {"glk", HB_TAG('G','L','K',' ')}, /* Gilaki */
369 {"gn", HB_TAG('G','U','A',' ')}, /* Guarani [macrolanguage] */
370 {"gno", HB_TAG('G','O','N',' ')}, /* Northern Gondi */
371 {"gog", HB_TAG('G','O','G',' ')}, /* Gogo */
372 {"gon", HB_TAG('G','O','N',' ')}, /* Gondi [macrolanguage] */
373 {"grt", HB_TAG('G','R','O',' ')}, /* Garo */
374 {"gru", HB_TAG('S','O','G',' ')}, /* Sodo Gurage */
375 {"gu", HB_TAG('G','U','J',' ')}, /* Gujarati */
376 {"guc", HB_TAG('G','U','C',' ')}, /* Wayuu */
377 {"guk", HB_TAG('G','M','Z',' ')}, /* Gumuz */
378 /*{"guk", HB_TAG('G','U','K',' ')},*/ /* Gumuz (in SIL fonts) */
379 {"guz", HB_TAG('G','U','Z',' ')}, /* Ekegusii/Gusii */
380 {"gv", HB_TAG('M','N','X',' ')}, /* Manx */
381 {"ha", HB_TAG('H','A','U',' ')}, /* Hausa */
382 {"har", HB_TAG('H','R','I',' ')}, /* Harari */
383 {"haw", HB_TAG('H','A','W',' ')}, /* Hawaiian */
384 {"hay", HB_TAG('H','A','Y',' ')}, /* Haya */
385 {"haz", HB_TAG('H','A','Z',' ')}, /* Hazaragi */
386 {"he", HB_TAG('I','W','R',' ')}, /* Hebrew */
387 {"hz", HB_TAG('H','E','R',' ')}, /* Herero */
388 {"hi", HB_TAG('H','I','N',' ')}, /* Hindi */
389 {"hil", HB_TAG('H','I','L',' ')}, /* Hiligaynon */
390 {"hnd", HB_TAG('H','N','D',' ')}, /* [Southern] Hindko */
391 {"hne", HB_TAG('C','H','H',' ')}, /* Chattisgarhi */
392 {"hno", HB_TAG('H','N','D',' ')}, /* [Northern] Hindko */
393 {"ho", HB_TAG('H','M','O',' ')}, /* Hiri Motu */
394 {"hoc", HB_TAG('H','O',' ',' ')}, /* Ho */
395 {"hoj", HB_TAG('H','A','R',' ')}, /* Harauti */
396 {"hr", HB_TAG('H','R','V',' ')}, /* Croatian */
397 {"hsb", HB_TAG('U','S','B',' ')}, /* Upper Sorbian */
398 {"ht", HB_TAG('H','A','I',' ')}, /* Haitian/Haitian Creole */
399 {"hu", HB_TAG('H','U','N',' ')}, /* Hungarian */
400 {"hy", HB_TAG('H','Y','E',' ')}, /* Armenian */
401 {"hz", HB_TAG('H','E','R',' ')}, /* Herero */
402 {"ia", HB_TAG('I','N','A',' ')}, /* Interlingua (International Auxiliary Language Association) */
403 {"ibb", HB_TAG('I','B','B',' ')}, /* Ibibio */
404 {"id", HB_TAG('I','N','D',' ')}, /* Indonesian */
405 {"ie", HB_TAG('I','L','E',' ')}, /* Interlingue/Occidental */
406 {"ig", HB_TAG('I','B','O',' ')}, /* Igbo */
407 {"igb", HB_TAG('E','B','I',' ')}, /* Ebira */
408 {"ijc", HB_TAG('I','J','O',' ')}, /* Izon */
409 {"ijo", HB_TAG('I','J','O',' ')}, /* Ijo [family] */
410 {"ik", HB_TAG('I','P','K',' ')}, /* Inupiaq [macrolanguage] */
411 {"ilo", HB_TAG('I','L','O',' ')}, /* Ilokano */
412 {"inh", HB_TAG('I','N','G',' ')}, /* Ingush */
413 {"io", HB_TAG('I','D','O',' ')}, /* Ido */
414 {"is", HB_TAG('I','S','L',' ')}, /* Icelandic */
415 {"it", HB_TAG('I','T','A',' ')}, /* Italian */
416 {"iu", HB_TAG('I','N','U',' ')}, /* Inuktitut [macrolanguage] */
417 {"ja", HB_TAG('J','A','N',' ')}, /* Japanese */
418 {"jam", HB_TAG('J','A','M',' ')}, /* Jamaican Creole English */
419 {"jbo", HB_TAG('J','B','O',' ')}, /* Lojban */
420 {"jv", HB_TAG('J','A','V',' ')}, /* Javanese */
421 {"ka", HB_TAG('K','A','T',' ')}, /* Georgian */
422 {"kaa", HB_TAG('K','R','K',' ')}, /* Karakalpak */
423 {"kab", HB_TAG('K','A','B',' ')}, /* Kabyle */
424 {"kam", HB_TAG('K','M','B',' ')}, /* Kamba (Kenya) */
425 {"kar", HB_TAG('K','R','N',' ')}, /* Karen [family] */
426 {"kbd", HB_TAG('K','A','B',' ')}, /* Kabardian */
427 {"kde", HB_TAG('K','D','E',' ')}, /* Makonde */
428 {"kdr", HB_TAG('K','R','M',' ')}, /* Karaim */
429 {"kdt", HB_TAG('K','U','Y',' ')}, /* Kuy */
430 {"kex", HB_TAG('K','K','N',' ')}, /* Kokni */
431 {"kfr", HB_TAG('K','A','C',' ')}, /* Kachchi */
432 {"kfy", HB_TAG('K','M','N',' ')}, /* Kumaoni */
433 {"kg", HB_TAG('K','O','N',' ')}, /* Kongo [macrolanguage] */
434 {"kha", HB_TAG('K','S','I',' ')}, /* Khasi */
435 {"khb", HB_TAG('X','B','D',' ')}, /* Lü */
436 {"kht", HB_TAG('K','H','N',' ')}, /* Khamti (Microsoft fonts) */
437 /*{"kht", HB_TAG('K','H','T',' ')},*/ /* Khamti (OpenType spec and SIL fonts) */
438 {"khw", HB_TAG('K','H','W',' ')}, /* Khowar */
439 {"ki", HB_TAG('K','I','K',' ')}, /* Gikuyu/Kikuyu */
440 {"kj", HB_TAG('K','U','A',' ')}, /* Kuanyama/Kwanyama */
441 {"kjh", HB_TAG('K','H','A',' ')}, /* Khakass */
442 {"kjp", HB_TAG('K','J','P',' ')}, /* Pwo Eastern Karen */
443 {"kk", HB_TAG('K','A','Z',' ')}, /* Kazakh */
444 {"kl", HB_TAG('G','R','N',' ')}, /* Kalaallisut */
445 {"kln", HB_TAG('K','A','L',' ')}, /* Kalenjin */
446 {"km", HB_TAG('K','H','M',' ')}, /* Central Khmer */
447 {"kmb", HB_TAG('M','B','N',' ')}, /* Kimbundu */
448 {"kmw", HB_TAG('K','M','O',' ')}, /* Komo (Democratic Republic of Congo) */
449 {"kn", HB_TAG('K','A','N',' ')}, /* Kannada */
450 {"knn", HB_TAG('K','O','K',' ')}, /* Konkani */
451 {"ko", HB_TAG('K','O','R',' ')}, /* Korean */
452 {"koi", HB_TAG('K','O','P',' ')}, /* Komi-Permyak */
453 {"kok", HB_TAG('K','O','K',' ')}, /* Konkani [macrolanguage] */
454 {"kpe", HB_TAG('K','P','L',' ')}, /* Kpelle [macrolanguage] */
455 {"kpv", HB_TAG('K','O','Z',' ')}, /* Komi-Zyrian */
456 {"kpy", HB_TAG('K','Y','K',' ')}, /* Koryak */
457 {"kqy", HB_TAG('K','R','T',' ')}, /* Koorete */
458 {"kr", HB_TAG('K','N','R',' ')}, /* Kanuri [macrolanguage] */
459 {"kri", HB_TAG('K','R','I',' ')}, /* Krio */
460 {"krl", HB_TAG('K','R','L',' ')}, /* Karelian */
461 {"kru", HB_TAG('K','U','U',' ')}, /* Kurukh */
462 {"ks", HB_TAG('K','S','H',' ')}, /* Kashmiri */
463 {"ksh", HB_TAG('K','S','H',' ')}, /* Kölsch */
464 /*{"ksw", HB_TAG('K','R','N',' ')},*/ /* S'gaw Karen (Microsoft fonts?) */
465 {"ksw", HB_TAG('K','S','W',' ')}, /* S'gaw Karen (OpenType spec and SIL fonts) */
466 {"ku", HB_TAG('K','U','R',' ')}, /* Kurdish [macrolanguage] */
467 {"kum", HB_TAG('K','U','M',' ')}, /* Kumyk */
468 {"kv", HB_TAG('K','O','M',' ')}, /* Komi [macrolanguage] */
469 {"kvd", HB_TAG('K','U','I',' ')}, /* Kui (Indonesia) */
470 {"kw", HB_TAG('C','O','R',' ')}, /* Cornish */
471 {"kxc", HB_TAG('K','M','S',' ')}, /* Komso */
472 {"kxu", HB_TAG('K','U','I',' ')}, /* Kui (India) */
473 {"ky", HB_TAG('K','I','R',' ')}, /* Kirghiz/Kyrgyz */
474 {"kyu", HB_TAG('K','Y','U',' ')}, /* Western Kayah */
475 {"la", HB_TAG('L','A','T',' ')}, /* Latin */
476 {"lad", HB_TAG('J','U','D',' ')}, /* Ladino */
477 {"lb", HB_TAG('L','T','Z',' ')}, /* Luxembourgish */
478 {"lbe", HB_TAG('L','A','K',' ')}, /* Lak */
479 {"lbj", HB_TAG('L','D','K',' ')}, /* Ladakhi */
480 {"lez", HB_TAG('L','E','Z',' ')}, /* Lezgi */
481 {"lg", HB_TAG('L','U','G',' ')}, /* Ganda */
482 {"li", HB_TAG('L','I','M',' ')}, /* Limburgan/Limburger/Limburgish */
483 {"lif", HB_TAG('L','M','B',' ')}, /* Limbu */
484 {"lij", HB_TAG('L','I','J',' ')}, /* Ligurian */
485 {"lis", HB_TAG('L','I','S',' ')}, /* Lisu */
486 {"ljp", HB_TAG('L','J','P',' ')}, /* Lampung Api */
487 {"lki", HB_TAG('L','K','I',' ')}, /* Laki */
488 {"lld", HB_TAG('L','A','D',' ')}, /* Ladin */
489 {"lmn", HB_TAG('L','A','M',' ')}, /* Lambani */
490 {"lmo", HB_TAG('L','M','O',' ')}, /* Lombard */
491 {"ln", HB_TAG('L','I','N',' ')}, /* Lingala */
492 {"lo", HB_TAG('L','A','O',' ')}, /* Lao */
493 {"lrc", HB_TAG('L','R','C',' ')}, /* Northern Luri */
494 {"lt", HB_TAG('L','T','H',' ')}, /* Lithuanian */
495 {"lu", HB_TAG('L','U','B',' ')}, /* Luba-Katanga */
496 {"lua", HB_TAG('L','U','B',' ')}, /* Luba-Kasai */
497 {"luo", HB_TAG('L','U','O',' ')}, /* Luo (Kenya and Tanzania) */
498 {"lus", HB_TAG('M','I','Z',' ')}, /* Mizo */
499 {"luy", HB_TAG('L','U','H',' ')}, /* Luyia/Oluluyia [macrolanguage] */
500 {"luz", HB_TAG('L','R','C',' ')}, /* Southern Luri */
501 {"lv", HB_TAG('L','V','I',' ')}, /* Latvian */
502 {"lzz", HB_TAG('L','A','Z',' ')}, /* Laz */
503 {"mad", HB_TAG('M','A','D',' ')}, /* Madurese */
504 {"mag", HB_TAG('M','A','G',' ')}, /* Magahi */
505 {"mai", HB_TAG('M','T','H',' ')}, /* Maithili */
506 {"mak", HB_TAG('M','K','R',' ')}, /* Makasar */
507 {"man", HB_TAG('M','N','K',' ')}, /* Manding/Mandingo [macrolanguage] */
508 {"mdc", HB_TAG('M','L','E',' ')}, /* Male (Papua New Guinea) */
509 {"mdf", HB_TAG('M','O','K',' ')}, /* Moksha */
510 {"mdr", HB_TAG('M','D','R',' ')}, /* Mandar */
511 {"mdy", HB_TAG('M','L','E',' ')}, /* Male (Ethiopia) */
512 {"men", HB_TAG('M','D','E',' ')}, /* Mende (Sierra Leone) */
513 {"mer", HB_TAG('M','E','R',' ')}, /* Meru */
514 {"mfe", HB_TAG('M','F','E',' ')}, /* Morisyen */
515 {"mg", HB_TAG('M','L','G',' ')}, /* Malagasy [macrolanguage] */
516 {"mh", HB_TAG('M','A','H',' ')}, /* Marshallese */
517 {"mhr", HB_TAG('L','M','A',' ')}, /* Low Mari */
518 {"mi", HB_TAG('M','R','I',' ')}, /* Maori */
519 {"min", HB_TAG('M','I','N',' ')}, /* Minangkabau */
520 {"mk", HB_TAG('M','K','D',' ')}, /* Macedonian */
521 {"mku", HB_TAG('M','N','K',' ')}, /* Konyanka Maninka */
522 {"mkw", HB_TAG('M','K','W',' ')}, /* Kituba (Congo) */
523 {"ml", HB_TAG('M','L','R',' ')}, /* Malayalam */
524 {"mlq", HB_TAG('M','N','K',' ')}, /* Western Maninkakan */
525 {"mn", HB_TAG('M','N','G',' ')}, /* Mongolian [macrolanguage] */
526 {"mnc", HB_TAG('M','C','H',' ')}, /* Manchu */
527 {"mni", HB_TAG('M','N','I',' ')}, /* Manipuri */
528 {"mnk", HB_TAG('M','N','D',' ')}, /* Mandinka */
529 {"mns", HB_TAG('M','A','N',' ')}, /* Mansi */
530 {"mnw", HB_TAG('M','O','N',' ')}, /* Mon */
531 {"mo", HB_TAG('M','O','L',' ')}, /* Moldavian */
532 {"moh", HB_TAG('M','O','H',' ')}, /* Mohawk */
533 {"mos", HB_TAG('M','O','S',' ')}, /* Mossi */
534 {"mpe", HB_TAG('M','A','J',' ')}, /* Majang */
535 {"mr", HB_TAG('M','A','R',' ')}, /* Marathi */
536 {"mrj", HB_TAG('H','M','A',' ')}, /* High Mari */
537 {"ms", HB_TAG('M','L','Y',' ')}, /* Malay [macrolanguage] */
538 {"msc", HB_TAG('M','N','K',' ')}, /* Sankaran Maninka */
539 {"mt", HB_TAG('M','T','S',' ')}, /* Maltese */
540 {"mtr", HB_TAG('M','A','W',' ')}, /* Mewari */
541 {"mus", HB_TAG('M','U','S',' ')}, /* Creek */
542 {"mve", HB_TAG('M','A','W',' ')}, /* Marwari (Pakistan) */
543 {"mwk", HB_TAG('M','N','K',' ')}, /* Kita Maninkakan */
544 {"mwl", HB_TAG('M','W','L',' ')}, /* Mirandese */
545 {"mwr", HB_TAG('M','A','W',' ')}, /* Marwari [macrolanguage] */
546 {"mww", HB_TAG('M','W','W',' ')}, /* Hmong Daw */
547 {"my", HB_TAG('B','R','M',' ')}, /* Burmese */
548 {"mym", HB_TAG('M','E','N',' ')}, /* Me'en */
549 {"myq", HB_TAG('M','N','K',' ')}, /* Forest Maninka (retired code) */
550 {"myv", HB_TAG('E','R','Z',' ')}, /* Erzya */
551 {"mzn", HB_TAG('M','Z','N',' ')}, /* Mazanderani */
552 {"na", HB_TAG('N','A','U',' ')}, /* Nauru */
553 {"nag", HB_TAG('N','A','G',' ')}, /* Naga-Assamese */
554 {"nah", HB_TAG('N','A','H',' ')}, /* Nahuatl [family] */
555 {"nap", HB_TAG('N','A','P',' ')}, /* Neapolitan */
556 {"nb", HB_TAG('N','O','R',' ')}, /* Norwegian Bokmål */
557 {"nco", HB_TAG('S','I','B',' ')}, /* Sibe */
558 {"nd", HB_TAG('N','D','B',' ')}, /* [North] Ndebele */
559 {"ndc", HB_TAG('N','D','C',' ')}, /* Ndau */
560 {"nds", HB_TAG('N','D','S',' ')}, /* Low German/Low Saxon */
561 {"ne", HB_TAG('N','E','P',' ')}, /* Nepali */
562 {"new", HB_TAG('N','E','W',' ')}, /* Newari */
563 {"ng", HB_TAG('N','D','G',' ')}, /* Ndonga */
564 {"nga", HB_TAG('N','G','A',' ')}, /* Ngabaka */
565 {"ngl", HB_TAG('L','M','W',' ')}, /* Lomwe */
566 {"niu", HB_TAG('N','I','U',' ')}, /* Niuean */
567 {"niv", HB_TAG('G','I','L',' ')}, /* Gilyak */
568 {"nl", HB_TAG('N','L','D',' ')}, /* Dutch */
569 {"nn", HB_TAG('N','Y','N',' ')}, /* Norwegian Nynorsk */
570 {"no", HB_TAG('N','O','R',' ')}, /* Norwegian [macrolanguage] */
571 {"nod", HB_TAG('N','T','A',' ')}, /* Northern Thai */
572 {"noe", HB_TAG('N','O','E',' ')}, /* Nimadi */
573 {"nog", HB_TAG('N','O','G',' ')}, /* Nogai */
574 {"nov", HB_TAG('N','O','V',' ')}, /* Novial */
575 {"nqo", HB_TAG('N','K','O',' ')}, /* N'Ko */
576 {"nr", HB_TAG('N','D','B',' ')}, /* [South] Ndebele */
577 {"nsk", HB_TAG('N','A','S',' ')}, /* Naskapi */
578 {"nso", HB_TAG('S','O','T',' ')}, /* [Northern] Sotho */
579 {"nv", HB_TAG('N','A','V',' ')}, /* Navajo */
580 {"ny", HB_TAG('C','H','I',' ')}, /* Chewa/Chichwa/Nyanja */
581 {"nym", HB_TAG('N','Y','M',' ')}, /* Nyamwezi */
582 {"nyn", HB_TAG('N','K','L',' ')}, /* Nyankole */
583 {"oc", HB_TAG('O','C','I',' ')}, /* Occitan (post 1500) */
584 {"oj", HB_TAG('O','J','B',' ')}, /* Ojibwa [macrolanguage] */
585 {"ojs", HB_TAG('O','C','R',' ')}, /* Oji-Cree */
586 {"om", HB_TAG('O','R','O',' ')}, /* Oromo [macrolanguage] */
587 {"or", HB_TAG('O','R','I',' ')}, /* Oriya */
588 {"os", HB_TAG('O','S','S',' ')}, /* Ossetian */
589 {"pa", HB_TAG('P','A','N',' ')}, /* Panjabi */
590 {"pag", HB_TAG('P','A','G',' ')}, /* Pangasinan */
591 {"pam", HB_TAG('P','A','M',' ')}, /* Kapampangan/Pampanga */
592 {"pap", HB_TAG('P','A','P',' ')}, /* Papiamento */
593 {"pcc", HB_TAG('P','C','C',' ')}, /* Bouyei */
594 {"pcd", HB_TAG('P','C','D',' ')}, /* Picard */
595 {"pce", HB_TAG('P','L','G',' ')}, /* [Ruching] Palaung */
596 {"pdc", HB_TAG('P','D','C',' ')}, /* Pennsylvania German */
597 {"pes", HB_TAG('F','A','R',' ')}, /* Iranian Persian */
598 {"phk", HB_TAG('P','H','K',' ')}, /* Phake */
599 {"pi", HB_TAG('P','A','L',' ')}, /* Pali */
600 {"pih", HB_TAG('P','I','H',' ')}, /* Pitcairn-Norfolk */
601 {"pl", HB_TAG('P','L','K',' ')}, /* Polish */
602 {"pll", HB_TAG('P','L','G',' ')}, /* [Shwe] Palaung */
603 {"plp", HB_TAG('P','A','P',' ')}, /* Palpa */
604 {"pms", HB_TAG('P','M','S',' ')}, /* Piemontese */
605 {"pnb", HB_TAG('P','N','B',' ')}, /* Western Panjabi */
606 {"prs", HB_TAG('D','R','I',' ')}, /* Afghan Persian/Dari */
607 {"ps", HB_TAG('P','A','S',' ')}, /* Pashto/Pushto [macrolanguage] */
608 {"pt", HB_TAG('P','T','G',' ')}, /* Portuguese */
609 {"pwo", HB_TAG('P','W','O',' ')}, /* Pwo Western Karen */
610 {"qu", HB_TAG('Q','U','Z',' ')}, /* Quechua [macrolanguage] */
611 {"quc", HB_TAG('Q','U','C',' ')}, /* K'iche'/Quiché */
612 {"quz", HB_TAG('Q','U','Z',' ')}, /* Cusco Quechua */
613 {"raj", HB_TAG('R','A','J',' ')}, /* Rajasthani [macrolanguage] */
614 {"rbb", HB_TAG('P','L','G',' ')}, /* Rumai Palaung */
615 {"rej", HB_TAG('R','E','J',' ')}, /* Rejang */
616 {"ria", HB_TAG('R','I','A',' ')}, /* Riang (India) */
617 {"ril", HB_TAG('R','I','A',' ')}, /* Riang (Myanmar) */
618 {"rki", HB_TAG('A','R','K',' ')}, /* Rakhine */
619 {"rm", HB_TAG('R','M','S',' ')}, /* Romansh */
620 {"rmy", HB_TAG('R','M','Y',' ')}, /* Vlax Romani */
621 {"rn", HB_TAG('R','U','N',' ')}, /* Rundi */
622 {"ro", HB_TAG('R','O','M',' ')}, /* Romanian */
623 {"rom", HB_TAG('R','O','Y',' ')}, /* Romany [macrolanguage] */
624 {"ru", HB_TAG('R','U','S',' ')}, /* Russian */
625 {"rue", HB_TAG('R','S','Y',' ')}, /* Rusyn */
626 {"rup", HB_TAG('R','U','P',' ')}, /* Aromanian/Arumanian/Macedo-Romanian */
627 {"rw", HB_TAG('R','U','A',' ')}, /* Kinyarwanda */
628 {"rwr", HB_TAG('M','A','W',' ')}, /* Marwari (India) */
629 {"sa", HB_TAG('S','A','N',' ')}, /* Sanskrit */
630 {"sah", HB_TAG('Y','A','K',' ')}, /* Yakut */
631 {"sas", HB_TAG('S','A','S',' ')}, /* Sasak */
632 {"sat", HB_TAG('S','A','T',' ')}, /* Santali */
633 {"sck", HB_TAG('S','A','D',' ')}, /* Sadri */
634 {"sc", HB_TAG('S','R','D',' ')}, /* Sardinian [macrolanguage] */
635 {"scn", HB_TAG('S','C','N',' ')}, /* Sicilian */
636 {"sco", HB_TAG('S','C','O',' ')}, /* Scots */
637 {"scs", HB_TAG('S','L','A',' ')}, /* [North] Slavey */
638 {"sd", HB_TAG('S','N','D',' ')}, /* Sindhi */
639 {"se", HB_TAG('N','S','M',' ')}, /* Northern Sami */
640 {"seh", HB_TAG('S','N','A',' ')}, /* Sena */
641 {"sel", HB_TAG('S','E','L',' ')}, /* Selkup */
642 {"sg", HB_TAG('S','G','O',' ')}, /* Sango */
643 {"sga", HB_TAG('S','G','A',' ')}, /* Old Irish (to 900) */
644 {"sgs", HB_TAG('S','G','S',' ')}, /* Samogitian */
645 {"sgw", HB_TAG('C','H','G',' ')}, /* Sebat Bet Gurage */
646 /*{"sgw", HB_TAG('S','G','W',' ')},*/ /* Sebat Bet Gurage (in SIL fonts) */
647 {"shn", HB_TAG('S','H','N',' ')}, /* Shan */
648 {"si", HB_TAG('S','N','H',' ')}, /* Sinhala */
649 {"sid", HB_TAG('S','I','D',' ')}, /* Sidamo */
650 {"sjd", HB_TAG('K','S','M',' ')}, /* Kildin Sami */
651 {"sk", HB_TAG('S','K','Y',' ')}, /* Slovak */
652 {"skr", HB_TAG('S','R','K',' ')}, /* Seraiki */
653 {"sl", HB_TAG('S','L','V',' ')}, /* Slovenian */
654 {"sm", HB_TAG('S','M','O',' ')}, /* Samoan */
655 {"sma", HB_TAG('S','S','M',' ')}, /* Southern Sami */
656 {"smj", HB_TAG('L','S','M',' ')}, /* Lule Sami */
657 {"smn", HB_TAG('I','S','M',' ')}, /* Inari Sami */
658 {"sms", HB_TAG('S','K','S',' ')}, /* Skolt Sami */
659 {"sn", HB_TAG('S','N','A',' ')}, /* Shona */
660 {"snk", HB_TAG('S','N','K',' ')}, /* Soninke */
661 {"so", HB_TAG('S','M','L',' ')}, /* Somali */
662 {"sop", HB_TAG('S','O','P',' ')}, /* Songe */
663 {"sq", HB_TAG('S','Q','I',' ')}, /* Albanian [macrolanguage] */
664 {"sr", HB_TAG('S','R','B',' ')}, /* Serbian */
665 {"srr", HB_TAG('S','R','R',' ')}, /* Serer */
666 {"ss", HB_TAG('S','W','Z',' ')}, /* Swati */
667 {"st", HB_TAG('S','O','T',' ')}, /* [Southern] Sotho */
668 {"stq", HB_TAG('S','T','Q',' ')}, /* Saterfriesisch */
669 {"stv", HB_TAG('S','I','G',' ')}, /* Silt'e */
670 {"su", HB_TAG('S','U','N',' ')}, /* Sundanese */
671 {"suk", HB_TAG('S','U','K',' ')}, /* Sukama */
672 {"suq", HB_TAG('S','U','R',' ')}, /* Suri */
673 {"sv", HB_TAG('S','V','E',' ')}, /* Swedish */
674 {"sva", HB_TAG('S','V','A',' ')}, /* Svan */
675 {"sw", HB_TAG('S','W','K',' ')}, /* Swahili [macrolanguage] */
676 {"swb", HB_TAG('C','M','R',' ')}, /* Comorian */
677 {"swh", HB_TAG('S','W','K',' ')}, /* Kiswahili/Swahili */
678 {"swv", HB_TAG('M','A','W',' ')}, /* Shekhawati */
679 {"sxu", HB_TAG('S','X','U',' ')}, /* Upper Saxon */
680 {"syl", HB_TAG('S','Y','L',' ')}, /* Sylheti */
681 {"syr", HB_TAG('S','Y','R',' ')}, /* Syriac [macrolanguage] */
682 {"szl", HB_TAG('S','Z','L',' ')}, /* Silesian */
683 {"ta", HB_TAG('T','A','M',' ')}, /* Tamil */
684 {"tab", HB_TAG('T','A','B',' ')}, /* Tabasaran */
685 {"tcy", HB_TAG('T','U','L',' ')}, /* Tulu */
686 {"tdd", HB_TAG('T','D','D',' ')}, /* Tai Nüa */
687 {"te", HB_TAG('T','E','L',' ')}, /* Telugu */
688 {"tem", HB_TAG('T','M','N',' ')}, /* Temne */
689 {"tet", HB_TAG('T','E','T',' ')}, /* Tetum */
690 {"tg", HB_TAG('T','A','J',' ')}, /* Tajik */
691 {"th", HB_TAG('T','H','A',' ')}, /* Thai */
692 {"ti", HB_TAG('T','G','Y',' ')}, /* Tigrinya */
693 {"tig", HB_TAG('T','G','R',' ')}, /* Tigre */
694 {"tiv", HB_TAG('T','I','V',' ')}, /* Tiv */
695 {"tk", HB_TAG('T','K','M',' ')}, /* Turkmen */
696 {"tl", HB_TAG('T','G','L',' ')}, /* Tagalog */
697 {"tmh", HB_TAG('t','m','h',' ')}, /* Tamashek [macrolanguage] */
698 {"tn", HB_TAG('T','N','A',' ')}, /* Tswana */
699 {"to", HB_TAG('T','G','N',' ')}, /* Tonga (Tonga Islands) */
700 {"tpi", HB_TAG('T','P','I',' ')}, /* Tok Pisin */
701 {"tr", HB_TAG('T','R','K',' ')}, /* Turkish */
702 {"tru", HB_TAG('T','U','A',' ')}, /* Turoyo Aramaic */
703 {"ts", HB_TAG('T','S','G',' ')}, /* Tsonga */
704 {"tt", HB_TAG('T','A','T',' ')}, /* Tatar */
705 {"tum", HB_TAG('T','U','M',' ')}, /* Tumbuka */
706 {"tw", HB_TAG('T','W','I',' ')}, /* Twi */
707 {"ty", HB_TAG('T','H','T',' ')}, /* Tahitian */
708 {"tyv", HB_TAG('T','U','V',' ')}, /* Tuvin */
709 {"tyz", HB_TAG('T','Y','Z',' ')}, /* Tày */
710 {"tzm", HB_TAG('T','Z','M',' ')}, /* Central Atlas Tamazight */
711 {"udm", HB_TAG('U','D','M',' ')}, /* Udmurt */
712 {"ug", HB_TAG('U','Y','G',' ')}, /* Uighur */
713 {"uk", HB_TAG('U','K','R',' ')}, /* Ukrainian */
714 {"umb", HB_TAG('U','M','B',' ')}, /* Umbundu */
715 {"unr", HB_TAG('M','U','N',' ')}, /* Mundari */
716 {"ur", HB_TAG('U','R','D',' ')}, /* Urdu */
717 {"uz", HB_TAG('U','Z','B',' ')}, /* Uzbek [macrolanguage] */
718 {"uzn", HB_TAG('U','Z','B',' ')}, /* Northern Uzbek */
719 {"uzs", HB_TAG('U','Z','B',' ')}, /* Southern Uzbek */
720 {"ve", HB_TAG('V','E','N',' ')}, /* Venda */
721 {"vec", HB_TAG('V','E','C',' ')}, /* Venetian */
722 {"vls", HB_TAG('F','L','E',' ')}, /* Vlaams */
723 {"vi", HB_TAG('V','I','T',' ')}, /* Vietnamese */
724 {"vmw", HB_TAG('M','A','K',' ')}, /* Makhuwa */
725 {"vo", HB_TAG('V','O','L',' ')}, /* Volapük */
726 {"vro", HB_TAG('V','R','O',' ')}, /* Võro */
727 {"wa", HB_TAG('W','L','N',' ')}, /* Walloon */
728 {"war", HB_TAG('W','A','R',' ')}, /* Waray (Philippines) */
729 {"wbm", HB_TAG('W','A',' ',' ')}, /* Wa */
730 {"wbr", HB_TAG('W','A','G',' ')}, /* Wagdi */
731 {"wle", HB_TAG('S','I','G',' ')}, /* Wolane */
732 {"wry", HB_TAG('M','A','W',' ')}, /* Merwari */
733 {"wtm", HB_TAG('W','T','M',' ')}, /* Mewati */
734 {"wo", HB_TAG('W','L','F',' ')}, /* Wolof */
735 {"xal", HB_TAG('K','L','M',' ')}, /* Kalmyk */
736 {"xh", HB_TAG('X','H','S',' ')}, /* Xhosa */
737 {"xog", HB_TAG('X','O','G',' ')}, /* Soga */
738 {"xom", HB_TAG('K','M','O',' ')}, /* Komo (Sudan) */
739 {"xsl", HB_TAG('S','S','L',' ')}, /* South Slavey */
740 {"xst", HB_TAG('S','I','G',' ')}, /* Silt'e (retired code) */
741 {"xwo", HB_TAG('T','O','D',' ')}, /* Written Oirat (Todo) */
742 {"yao", HB_TAG('Y','A','O',' ')}, /* Yao */
743 {"yi", HB_TAG('J','I','I',' ')}, /* Yiddish [macrolanguage] */
744 {"yo", HB_TAG('Y','B','A',' ')}, /* Yoruba */
745 {"yso", HB_TAG('N','I','S',' ')}, /* Nisi (China) */
746 {"za", HB_TAG('Z','H','A',' ')}, /* Chuang/Zhuang [macrolanguage] */
747 {"zea", HB_TAG('Z','E','A',' ')}, /* Zeeuws */
748 {"zne", HB_TAG('Z','N','D',' ')}, /* Zande */
749 {"zu", HB_TAG('Z','U','L',' ')}, /* Zulu */
750 {"zum", HB_TAG('L','R','C',' ')} /* Kumzari */
751
752 /* The corresponding languages IDs for the following IDs are unclear,
753 * overlap, or are architecturally weird. Needs more research. */
754
755 /*{"ahg/awn/xan?", HB_TAG('A','G','W',' ')},*/ /* Agaw */
756 /*{"gsw?/gsw-FR?", HB_TAG('A','L','S',' ')},*/ /* Alsatian */
757 /*{"krc", HB_TAG('B','A','L',' ')},*/ /* Balkar */
758 /*{"??", HB_TAG('B','C','R',' ')},*/ /* Bible Cree */
759 /*{"zh?", HB_TAG('C','H','N',' ')},*/ /* Chinese (seen in Microsoft fonts) */
760 /*{"acf/gcf?", HB_TAG('F','A','N',' ')},*/ /* French Antillean */
761 /*{"enf?/yrk?", HB_TAG('F','N','E',' ')},*/ /* Forest Nenets */
762 /*{"fuf?", HB_TAG('F','T','A',' ')},*/ /* Futa */
763 /*{"ar-Syrc?", HB_TAG('G','A','R',' ')},*/ /* Garshuni */
764 /*{"cfm/rnl?", HB_TAG('H','A','L',' ')},*/ /* Halam */
765 /*{"ga-Latg?/Latg?", HB_TAG('I','R','T',' ')},*/ /* Irish Traditional */
766 /*{"krc", HB_TAG('K','A','R',' ')},*/ /* Karachay */
767 /*{"alw?/ktb?", HB_TAG('K','E','B',' ')},*/ /* Kebena */
768 /*{"Geok", HB_TAG('K','G','E',' ')},*/ /* Khutsuri Georgian */
769 /*{"kca", HB_TAG('K','H','K',' ')},*/ /* Khanty-Kazim */
770 /*{"kca", HB_TAG('K','H','S',' ')},*/ /* Khanty-Shurishkar */
771 /*{"kca", HB_TAG('K','H','V',' ')},*/ /* Khanty-Vakhi */
772 /*{"guz?/kqs?/kss?", HB_TAG('K','I','S',' ')},*/ /* Kisii */
773 /*{"kfa/kfi?/kpb?/xua?/xuj?", HB_TAG('K','O','D',' ')},*/ /* Kodagu */
774 /*{"okm?/oko?", HB_TAG('K','O','H',' ')},*/ /* Korean Old Hangul */
775 /*{"kon?/ktu?/...", HB_TAG('K','O','N',' ')},*/ /* Kikongo */
776 /*{"kfx?", HB_TAG('K','U','L',' ')},*/ /* Kulvi */
777 /*{"??", HB_TAG('L','A','H',' ')},*/ /* Lahuli */
778 /*{"??", HB_TAG('L','C','R',' ')},*/ /* L-Cree */
779 /*{"??", HB_TAG('M','A','L',' ')},*/ /* Malayalam Traditional */
780 /*{"mnk?/mlq?/...", HB_TAG('M','L','N',' ')},*/ /* Malinke */
781 /*{"??", HB_TAG('N','C','R',' ')},*/ /* N-Cree */
782 /*{"??", HB_TAG('N','H','C',' ')},*/ /* Norway House Cree */
783 /*{"jpa?/sam?", HB_TAG('P','A','A',' ')},*/ /* Palestinian Aramaic */
784 /*{"polyton", HB_TAG('P','G','R',' ')},*/ /* Polytonic Greek */
785 /*{"??", HB_TAG('Q','I','N',' ')},*/ /* Asho Chin */
786 /*{"??", HB_TAG('R','C','R',' ')},*/ /* R-Cree */
787 /*{"chp?", HB_TAG('S','A','Y',' ')},*/ /* Sayisi */
788 /*{"xan?", HB_TAG('S','E','K',' ')},*/ /* Sekota */
789 /*{"ngo?", HB_TAG('S','X','T',' ')},*/ /* Sutu */
790 /*{"??", HB_TAG('T','C','R',' ')},*/ /* TH-Cree */
791 /*{"tnz?/tog?/toi?", HB_TAG('T','N','G',' ')},*/ /* Tonga */
792 /*{"enh?/yrk?", HB_TAG('T','N','E',' ')},*/ /* Tundra Nenets */
793 /*{"??", HB_TAG('W','C','R',' ')},*/ /* West-Cree */
794 /*{"cre?", HB_TAG('Y','C','R',' ')},*/ /* Y-Cree */
795 /*{"??", HB_TAG('Y','I','C',' ')},*/ /* Yi Classic */
796 /*{"ii?/Yiii?", HB_TAG('Y','I','M',' ')},*/ /* Yi Modern */
797 /*{"??", HB_TAG('Z','H','P',' ')},*/ /* Chinese Phonetic */
798 };
799
800 typedef struct {
801 char language[8];
802 hb_tag_t tag;
803 } LangTagLong;
804 static const LangTagLong ot_languages_zh[] = {
805 {"zh-cn", HB_TAG('Z','H','S',' ')}, /* Chinese (China) */
806 {"zh-hk", HB_TAG('Z','H','H',' ')}, /* Chinese (Hong Kong) */
807 {"zh-mo", HB_TAG('Z','H','T',' ')}, /* Chinese (Macao) */
808 {"zh-sg", HB_TAG('Z','H','S',' ')}, /* Chinese (Singapore) */
809 {"zh-tw", HB_TAG('Z','H','T',' ')}, /* Chinese (Taiwan) */
810 {"zh-hans", HB_TAG('Z','H','S',' ')}, /* Chinese (Simplified) */
811 {"zh-hant", HB_TAG('Z','H','T',' ')}, /* Chinese (Traditional) */
812 };
813
814 static int
815 lang_compare_first_component (const char *a,
816 const char *b)
817 {
818 unsigned int da, db;
819 const char *p;
820
821 p = strchr (a, '-');
822 da = p ? (unsigned int) (p - a) : strlen (a);
823
824 p = strchr (b, '-');
825 db = p ? (unsigned int) (p - b) : strlen (b);
826
827 return strncmp (a, b, MAX (da, db));
828 }
829
830 static hb_bool_t
831 lang_matches (const char *lang_str, const char *spec)
832 {
833 unsigned int len = strlen (spec);
834
835 return strncmp (lang_str, spec, len) == 0 &&
836 (lang_str[len] == '\0' || lang_str[len] == '-');
837 }
838
839 hb_tag_t
840 hb_ot_tag_from_language (hb_language_t language)
841 {
842 const char *lang_str, *s;
843
844 if (language == HB_LANGUAGE_INVALID)
845 return HB_OT_TAG_DEFAULT_LANGUAGE;
846
847 lang_str = hb_language_to_string (language);
848
849 s = strstr (lang_str, "x-hbot");
850 if (s) {
851 char tag[4];
852 int i;
853 s += 6;
854 for (i = 0; i < 4 && ISALPHA (s[i]); i++)
855 tag[i] = TOUPPER (s[i]);
856 if (i) {
857 for (; i < 4; i++)
858 tag[i] = ' ';
859 return HB_TAG_CHAR4 (tag);
860 }
861 }
862
863 /*
864 * The International Phonetic Alphabet is a variant tag in BCP-47,
865 * which can be applied to any language.
866 */
867 if (strstr (lang_str, "-fonipa")) {
868 return HB_TAG('I','P','P','H'); /* Phonetic transcription—IPA conventions */
869 }
870
871 /* Find a language matching in the first component */
872 {
873 const LangTag *lang_tag;
874 lang_tag = (LangTag *) bsearch (lang_str, ot_languages,
875 ARRAY_LENGTH (ot_languages), sizeof (LangTag),
876 (hb_compare_func_t) lang_compare_first_component);
877 if (lang_tag)
878 return lang_tag->tag;
879 }
880
881 /* Otherwise, check the Chinese ones */
882 if (0 == lang_compare_first_component (lang_str, "zh"))
883 {
884 unsigned int i;
885
886 for (i = 0; i < ARRAY_LENGTH (ot_languages_zh); i++)
887 {
888 const LangTagLong *lang_tag;
889 lang_tag = &ot_languages_zh[i];
890 if (lang_matches (lang_str, lang_tag->language))
891 return lang_tag->tag;
892 }
893
894 /* Otherwise just return 'ZHS ' */
895 return HB_TAG('Z','H','S',' ');
896 }
897
898 s = strchr (lang_str, '-');
899 if (!s)
900 s = lang_str + strlen (lang_str);
901 if (s - lang_str == 3) {
902 /* Assume it's ISO-639-3 and upper-case and use it. */
903 return hb_tag_from_string (lang_str, s - lang_str) & ~0x20202000u;
904 }
905
906 return HB_OT_TAG_DEFAULT_LANGUAGE;
907 }
908
909 /**
910 * hb_ot_tag_to_language:
911 *
912 *
913 *
914 * Return value: (transfer none):
915 *
916 * Since: 0.9.2
917 **/
918 hb_language_t
919 hb_ot_tag_to_language (hb_tag_t tag)
920 {
921 unsigned int i;
922
923 if (tag == HB_OT_TAG_DEFAULT_LANGUAGE)
924 return NULL;
925
926 for (i = 0; i < ARRAY_LENGTH (ot_languages); i++)
927 if (ot_languages[i].tag == tag)
928 return hb_language_from_string (ot_languages[i].language, -1);
929
930 /* If tag starts with ZH, it's Chinese */
931 if ((tag & 0xFFFF0000u) == 0x5A480000u) {
932 switch (tag) {
933 case HB_TAG('Z','H','H',' '): return hb_language_from_string ("zh-hk", -1); /* Hong Kong */
934 case HB_TAG('Z','H','S',' '): return hb_language_from_string ("zh-Hans", -1); /* Simplified */
935 case HB_TAG('Z','H','T',' '): return hb_language_from_string ("zh-Hant", -1); /* Traditional */
936 default: break; /* Fall through */
937 }
938 }
939
940 /* struct LangTag has only room for 3-letter language tags. */
941 switch (tag) {
942 case HB_TAG('I','P','P','H'): /* Phonetic transcription—IPA conventions */
943 return hb_language_from_string ("und-fonipa", -1);
944 }
945
946 /* Else return a custom language in the form of "x-hbotABCD" */
947 {
948 unsigned char buf[11] = "x-hbot";
949 buf[6] = tag >> 24;
950 buf[7] = (tag >> 16) & 0xFF;
951 buf[8] = (tag >> 8) & 0xFF;
952 buf[9] = tag & 0xFF;
953 if (buf[9] == 0x20)
954 buf[9] = '\0';
955 buf[10] = '\0';
956 return hb_language_from_string ((char *) buf, -1);
957 }
958 }
959
960