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 #include "hb-ot.h"
31
32 #include <string.h>
33
34
35
36 /* hb_script_t */
37
38 static hb_tag_t
hb_ot_old_tag_from_script(hb_script_t script)39 hb_ot_old_tag_from_script (hb_script_t script)
40 {
41 /* This seems to be accurate as of end of 2012. */
42
43 switch ((hb_tag_t) script) {
44 case HB_SCRIPT_INVALID: return HB_OT_TAG_DEFAULT_SCRIPT;
45
46 /* KATAKANA and HIRAGANA both map to 'kana' */
47 case HB_SCRIPT_HIRAGANA: return HB_TAG('k','a','n','a');
48
49 /* Spaces at the end are preserved, unlike ISO 15924 */
50 case HB_SCRIPT_LAO: return HB_TAG('l','a','o',' ');
51 case HB_SCRIPT_YI: return HB_TAG('y','i',' ',' ');
52 /* Unicode-5.0 additions */
53 case HB_SCRIPT_NKO: return HB_TAG('n','k','o',' ');
54 /* Unicode-5.1 additions */
55 case HB_SCRIPT_VAI: return HB_TAG('v','a','i',' ');
56 /* Unicode-5.2 additions */
57 /* Unicode-6.0 additions */
58 }
59
60 /* Else, just change first char to lowercase and return */
61 return ((hb_tag_t) script) | 0x20000000;
62 }
63
64 static hb_script_t
hb_ot_old_tag_to_script(hb_tag_t tag)65 hb_ot_old_tag_to_script (hb_tag_t tag)
66 {
67 if (unlikely (tag == HB_OT_TAG_DEFAULT_SCRIPT))
68 return HB_SCRIPT_INVALID;
69
70 /* This side of the conversion is fully algorithmic. */
71
72 /* Any spaces at the end of the tag are replaced by repeating the last
73 * letter. Eg 'nko ' -> 'Nkoo' */
74 if (unlikely ((tag & 0x0000FF00) == 0x00002000))
75 tag |= (tag >> 8) & 0x0000FF00; /* Copy second letter to third */
76 if (unlikely ((tag & 0x000000FF) == 0x00000020))
77 tag |= (tag >> 8) & 0x000000FF; /* Copy third letter to fourth */
78
79 /* Change first char to uppercase and return */
80 return (hb_script_t) (tag & ~0x20000000);
81 }
82
83 static hb_tag_t
hb_ot_new_tag_from_script(hb_script_t script)84 hb_ot_new_tag_from_script (hb_script_t script)
85 {
86 switch ((hb_tag_t) script) {
87 case HB_SCRIPT_BENGALI: return HB_TAG('b','n','g','2');
88 case HB_SCRIPT_DEVANAGARI: return HB_TAG('d','e','v','2');
89 case HB_SCRIPT_GUJARATI: return HB_TAG('g','j','r','2');
90 case HB_SCRIPT_GURMUKHI: return HB_TAG('g','u','r','2');
91 case HB_SCRIPT_KANNADA: return HB_TAG('k','n','d','2');
92 case HB_SCRIPT_MALAYALAM: return HB_TAG('m','l','m','2');
93 case HB_SCRIPT_ORIYA: return HB_TAG('o','r','y','2');
94 case HB_SCRIPT_TAMIL: return HB_TAG('t','m','l','2');
95 case HB_SCRIPT_TELUGU: return HB_TAG('t','e','l','2');
96 case HB_SCRIPT_MYANMAR: return HB_TAG('m','y','m','2');
97 }
98
99 return HB_OT_TAG_DEFAULT_SCRIPT;
100 }
101
102 static hb_script_t
hb_ot_new_tag_to_script(hb_tag_t tag)103 hb_ot_new_tag_to_script (hb_tag_t tag)
104 {
105 switch (tag) {
106 case HB_TAG('b','n','g','2'): return HB_SCRIPT_BENGALI;
107 case HB_TAG('d','e','v','2'): return HB_SCRIPT_DEVANAGARI;
108 case HB_TAG('g','j','r','2'): return HB_SCRIPT_GUJARATI;
109 case HB_TAG('g','u','r','2'): return HB_SCRIPT_GURMUKHI;
110 case HB_TAG('k','n','d','2'): return HB_SCRIPT_KANNADA;
111 case HB_TAG('m','l','m','2'): return HB_SCRIPT_MALAYALAM;
112 case HB_TAG('o','r','y','2'): return HB_SCRIPT_ORIYA;
113 case HB_TAG('t','m','l','2'): return HB_SCRIPT_TAMIL;
114 case HB_TAG('t','e','l','2'): return HB_SCRIPT_TELUGU;
115 case HB_TAG('m','y','m','2'): return HB_SCRIPT_MYANMAR;
116 }
117
118 return HB_SCRIPT_UNKNOWN;
119 }
120
121 /*
122 * Complete list at:
123 * https://www.microsoft.com/typography/otspec/scripttags.htm
124 * https://www.microsoft.com/typography/otspec160/scripttagsProposed.htm
125 *
126 * Most of the script tags are the same as the ISO 15924 tag but lowercased.
127 * So we just do that, and handle the exceptional cases in a switch.
128 */
129
130 void
hb_ot_tags_from_script(hb_script_t script,hb_tag_t * script_tag_1,hb_tag_t * script_tag_2)131 hb_ot_tags_from_script (hb_script_t script,
132 hb_tag_t *script_tag_1,
133 hb_tag_t *script_tag_2)
134 {
135 hb_tag_t new_tag;
136
137 *script_tag_2 = HB_OT_TAG_DEFAULT_SCRIPT;
138 *script_tag_1 = hb_ot_old_tag_from_script (script);
139
140 new_tag = hb_ot_new_tag_from_script (script);
141 if (unlikely (new_tag != HB_OT_TAG_DEFAULT_SCRIPT)) {
142 *script_tag_2 = *script_tag_1;
143 *script_tag_1 = new_tag;
144 }
145 }
146
147 hb_script_t
hb_ot_tag_to_script(hb_tag_t tag)148 hb_ot_tag_to_script (hb_tag_t tag)
149 {
150 if (unlikely ((tag & 0x000000FF) == '2'))
151 return hb_ot_new_tag_to_script (tag);
152
153 return hb_ot_old_tag_to_script (tag);
154 }
155
156
157 /* hb_language_t */
158
159 typedef struct {
160 char language[6];
161 hb_tag_t tag;
162 } LangTag;
163
164 /*
165 * Complete list at:
166 * http://www.microsoft.com/typography/otspec/languagetags.htm
167 *
168 * Generated by intersecting the OpenType language tag list from
169 * Draft OpenType 1.5 spec, with with the ISO 639-3 codes from
170 * 2008/08/04, matching on name, and finally adjusted manually.
171 *
172 * Updated on 2012/12/07 with more research into remaining codes.
173 *
174 * Some items still missing. Those are commented out at the end.
175 * Keep sorted for bsearch.
176 */
177
178 static const LangTag ot_languages[] = {
179 {"aa", HB_TAG('A','F','R',' ')}, /* Afar */
180 {"ab", HB_TAG('A','B','K',' ')}, /* Abkhazian */
181 {"abq", HB_TAG('A','B','A',' ')}, /* Abaza */
182 {"ada", HB_TAG('D','N','G',' ')}, /* Dangme */
183 {"ady", HB_TAG('A','D','Y',' ')}, /* Adyghe */
184 {"af", HB_TAG('A','F','K',' ')}, /* Afrikaans */
185 {"aii", HB_TAG('S','W','A',' ')}, /* Swadaya Aramaic */
186 {"aiw", HB_TAG('A','R','I',' ')}, /* Aari */
187 {"alt", HB_TAG('A','L','T',' ')}, /* [Southern] Altai */
188 {"am", HB_TAG('A','M','H',' ')}, /* Amharic */
189 {"amf", HB_TAG('H','B','N',' ')}, /* Hammer-Banna */
190 {"ar", HB_TAG('A','R','A',' ')}, /* Arabic */
191 {"arn", HB_TAG('M','A','P',' ')}, /* Mapudungun */
192 {"as", HB_TAG('A','S','M',' ')}, /* Assamese */
193 {"ath", HB_TAG('A','T','H',' ')}, /* Athapaskan [family] */
194 {"atv", HB_TAG('A','L','T',' ')}, /* [Northern] Altai */
195 {"av", HB_TAG('A','V','R',' ')}, /* Avaric */
196 {"awa", HB_TAG('A','W','A',' ')}, /* Awadhi */
197 {"ay", HB_TAG('A','Y','M',' ')}, /* Aymara */
198 {"az", HB_TAG('A','Z','E',' ')}, /* Azerbaijani */
199 {"ba", HB_TAG('B','S','H',' ')}, /* Bashkir */
200 {"bai", HB_TAG('B','M','L',' ')}, /* Bamileke [family] */
201 {"bal", HB_TAG('B','L','I',' ')}, /* Baluchi */
202 {"bci", HB_TAG('B','A','U',' ')}, /* Baule */
203 {"bcq", HB_TAG('B','C','H',' ')}, /* Bench */
204 {"be", HB_TAG('B','E','L',' ')}, /* Belarussian */
205 {"bem", HB_TAG('B','E','M',' ')}, /* Bemba (Zambia) */
206 {"ber", HB_TAG('B','E','R',' ')}, /* Berber [family] */
207 {"bfq", HB_TAG('B','A','D',' ')}, /* Badaga */
208 {"bft", HB_TAG('B','L','T',' ')}, /* Balti */
209 {"bfy", HB_TAG('B','A','G',' ')}, /* Baghelkhandi */
210 {"bg", HB_TAG('B','G','R',' ')}, /* Bulgarian */
211 {"bhb", HB_TAG('B','H','I',' ')}, /* Bhili */
212 {"bho", HB_TAG('B','H','O',' ')}, /* Bhojpuri */
213 {"bik", HB_TAG('B','I','K',' ')}, /* Bikol */
214 {"bin", HB_TAG('E','D','O',' ')}, /* Bini */
215 {"bjt", HB_TAG('B','L','N',' ')}, /* Balanta-Ganja */
216 {"bla", HB_TAG('B','K','F',' ')}, /* Blackfoot */
217 {"ble", HB_TAG('B','L','N',' ')}, /* Balanta-Kentohe */
218 {"bm", HB_TAG('B','M','B',' ')}, /* Bambara */
219 {"bn", HB_TAG('B','E','N',' ')}, /* Bengali */
220 {"bo", HB_TAG('T','I','B',' ')}, /* Tibetan */
221 {"br", HB_TAG('B','R','E',' ')}, /* Breton */
222 {"bra", HB_TAG('B','R','I',' ')}, /* Braj Bhasha */
223 {"brh", HB_TAG('B','R','H',' ')}, /* Brahui */
224 {"bs", HB_TAG('B','O','S',' ')}, /* Bosnian */
225 {"btb", HB_TAG('B','T','I',' ')}, /* Beti (Cameroon) */
226 {"bxr", HB_TAG('R','B','U',' ')}, /* Russian Buriat */
227 {"byn", HB_TAG('B','I','L',' ')}, /* Bilen */
228 {"ca", HB_TAG('C','A','T',' ')}, /* Catalan */
229 {"ce", HB_TAG('C','H','E',' ')}, /* Chechen */
230 {"ceb", HB_TAG('C','E','B',' ')}, /* Cebuano */
231 {"chp", HB_TAG('C','H','P',' ')}, /* Chipewyan */
232 {"chr", HB_TAG('C','H','R',' ')}, /* Cherokee */
233 {"ckt", HB_TAG('C','H','K',' ')}, /* Chukchi */
234 {"cop", HB_TAG('C','O','P',' ')}, /* Coptic */
235 {"cr", HB_TAG('C','R','E',' ')}, /* Cree */
236 {"crh", HB_TAG('C','R','T',' ')}, /* Crimean Tatar */
237 {"crj", HB_TAG('E','C','R',' ')}, /* [Southern] East Cree */
238 {"crl", HB_TAG('E','C','R',' ')}, /* [Northern] East Cree */
239 {"crm", HB_TAG('M','C','R',' ')}, /* Moose Cree */
240 {"crx", HB_TAG('C','R','R',' ')}, /* Carrier */
241 {"cs", HB_TAG('C','S','Y',' ')}, /* Czech */
242 {"cu", HB_TAG('C','S','L',' ')}, /* Church Slavic */
243 {"cv", HB_TAG('C','H','U',' ')}, /* Chuvash */
244 {"cwd", HB_TAG('D','C','R',' ')}, /* Woods Cree */
245 {"cy", HB_TAG('W','E','L',' ')}, /* Welsh */
246 {"da", HB_TAG('D','A','N',' ')}, /* Danish */
247 {"dap", HB_TAG('N','I','S',' ')}, /* Nisi (India) */
248 {"dar", HB_TAG('D','A','R',' ')}, /* Dargwa */
249 {"de", HB_TAG('D','E','U',' ')}, /* German */
250 {"din", HB_TAG('D','N','K',' ')}, /* Dinka */
251 {"dje", HB_TAG('D','J','R',' ')}, /* Djerma */
252 {"dng", HB_TAG('D','U','N',' ')}, /* Dungan */
253 {"doi", HB_TAG('D','G','R',' ')}, /* Dogri */
254 {"dsb", HB_TAG('L','S','B',' ')}, /* Lower Sorbian */
255 {"dv", HB_TAG('D','I','V',' ')}, /* Dhivehi */
256 {"dyu", HB_TAG('J','U','L',' ')}, /* Jula */
257 {"dz", HB_TAG('D','Z','N',' ')}, /* Dzongkha */
258 {"ee", HB_TAG('E','W','E',' ')}, /* Ewe */
259 {"efi", HB_TAG('E','F','I',' ')}, /* Efik */
260 {"el", HB_TAG('E','L','L',' ')}, /* Modern Greek (1453-) */
261 {"en", HB_TAG('E','N','G',' ')}, /* English */
262 {"eo", HB_TAG('N','T','O',' ')}, /* Esperanto */
263 {"eot", HB_TAG('B','T','I',' ')}, /* Beti (Côte d'Ivoire) */
264 {"es", HB_TAG('E','S','P',' ')}, /* Spanish */
265 {"et", HB_TAG('E','T','I',' ')}, /* Estonian */
266 {"eu", HB_TAG('E','U','Q',' ')}, /* Basque */
267 {"eve", HB_TAG('E','V','N',' ')}, /* Even */
268 {"evn", HB_TAG('E','V','K',' ')}, /* Evenki */
269 {"fa", HB_TAG('F','A','R',' ')}, /* Persian */
270 {"ff", HB_TAG('F','U','L',' ')}, /* Fulah */
271 {"fi", HB_TAG('F','I','N',' ')}, /* Finnish */
272 {"fil", HB_TAG('P','I','L',' ')}, /* Filipino */
273 {"fj", HB_TAG('F','J','I',' ')}, /* Fijian */
274 {"fo", HB_TAG('F','O','S',' ')}, /* Faroese */
275 {"fon", HB_TAG('F','O','N',' ')}, /* Fon */
276 {"fr", HB_TAG('F','R','A',' ')}, /* French */
277 {"fur", HB_TAG('F','R','L',' ')}, /* Friulian */
278 {"fy", HB_TAG('F','R','I',' ')}, /* Western Frisian */
279 {"ga", HB_TAG('I','R','I',' ')}, /* Irish */
280 {"gaa", HB_TAG('G','A','D',' ')}, /* Ga */
281 {"gag", HB_TAG('G','A','G',' ')}, /* Gagauz */
282 {"gbm", HB_TAG('G','A','W',' ')}, /* Garhwali */
283 {"gd", HB_TAG('G','A','E',' ')}, /* Scottish Gaelic */
284 {"gez", HB_TAG('G','E','Z',' ')}, /* Ge'ez */
285 {"gl", HB_TAG('G','A','L',' ')}, /* Galician */
286 {"gld", HB_TAG('N','A','N',' ')}, /* Nanai */
287 {"gn", HB_TAG('G','U','A',' ')}, /* Guarani */
288 {"gon", HB_TAG('G','O','N',' ')}, /* Gondi */
289 {"grt", HB_TAG('G','R','O',' ')}, /* Garo */
290 {"gru", HB_TAG('S','O','G',' ')}, /* Sodo Gurage */
291 {"gu", HB_TAG('G','U','J',' ')}, /* Gujarati */
292 {"guk", HB_TAG('G','M','Z',' ')}, /* Gumuz */
293 {"gv", HB_TAG('M','N','X',' ')}, /* Manx Gaelic */
294 {"ha", HB_TAG('H','A','U',' ')}, /* Hausa */
295 {"har", HB_TAG('H','R','I',' ')}, /* Harari */
296 {"haw", HB_TAG('H','A','W',' ')}, /* Hawaiin */
297 {"he", HB_TAG('I','W','R',' ')}, /* Hebrew */
298 {"hi", HB_TAG('H','I','N',' ')}, /* Hindi */
299 {"hil", HB_TAG('H','I','L',' ')}, /* Hiligaynon */
300 {"hnd", HB_TAG('H','N','D',' ')}, /* [Southern] Hindko */
301 {"hne", HB_TAG('C','H','H',' ')}, /* Chattisgarhi */
302 {"hno", HB_TAG('H','N','D',' ')}, /* [Northern] Hindko */
303 {"hoc", HB_TAG('H','O',' ',' ')}, /* Ho */
304 {"hoj", HB_TAG('H','A','R',' ')}, /* Harauti */
305 {"hr", HB_TAG('H','R','V',' ')}, /* Croatian */
306 {"hsb", HB_TAG('U','S','B',' ')}, /* Upper Sorbian */
307 {"ht", HB_TAG('H','A','I',' ')}, /* Haitian */
308 {"hu", HB_TAG('H','U','N',' ')}, /* Hungarian */
309 {"hy", HB_TAG('H','Y','E',' ')}, /* Armenian */
310 {"id", HB_TAG('I','N','D',' ')}, /* Indonesian */
311 {"ig", HB_TAG('I','B','O',' ')}, /* Igbo */
312 {"igb", HB_TAG('E','B','I',' ')}, /* Ebira */
313 {"ijo", HB_TAG('I','J','O',' ')}, /* Ijo [family] */
314 {"ilo", HB_TAG('I','L','O',' ')}, /* Ilokano */
315 {"inh", HB_TAG('I','N','G',' ')}, /* Ingush */
316 {"is", HB_TAG('I','S','L',' ')}, /* Icelandic */
317 {"it", HB_TAG('I','T','A',' ')}, /* Italian */
318 {"iu", HB_TAG('I','N','U',' ')}, /* Inuktitut */
319 {"ja", HB_TAG('J','A','N',' ')}, /* Japanese */
320 {"jv", HB_TAG('J','A','V',' ')}, /* Javanese */
321 {"ka", HB_TAG('K','A','T',' ')}, /* Georgian */
322 {"kaa", HB_TAG('K','R','K',' ')}, /* Karakalpak */
323 {"kam", HB_TAG('K','M','B',' ')}, /* Kamba (Kenya) */
324 {"kar", HB_TAG('K','R','N',' ')}, /* Karen [family] */
325 {"kbd", HB_TAG('K','A','B',' ')}, /* Kabardian */
326 {"kdr", HB_TAG('K','R','M',' ')}, /* Karaim */
327 {"kdt", HB_TAG('K','U','Y',' ')}, /* Kuy */
328 {"kex", HB_TAG('K','K','N',' ')}, /* Kokni */
329 {"kfr", HB_TAG('K','A','C',' ')}, /* Kachchi */
330 {"kfy", HB_TAG('K','M','N',' ')}, /* Kumaoni */
331 {"kha", HB_TAG('K','S','I',' ')}, /* Khasi */
332 {"khb", HB_TAG('X','B','D',' ')}, /* Tai Lue */
333 {"khw", HB_TAG('K','H','W',' ')}, /* Khowar */
334 {"ki", HB_TAG('K','I','K',' ')}, /* Kikuyu */
335 {"kjh", HB_TAG('K','H','A',' ')}, /* Khakass */
336 {"kk", HB_TAG('K','A','Z',' ')}, /* Kazakh */
337 {"kl", HB_TAG('G','R','N',' ')}, /* Kalaallisut */
338 {"kln", HB_TAG('K','A','L',' ')}, /* Kalenjin */
339 {"km", HB_TAG('K','H','M',' ')}, /* Central Khmer */
340 {"kmb", HB_TAG('M','B','N',' ')}, /* [North] Mbundu */
341 {"kmw", HB_TAG('K','M','O',' ')}, /* Komo (Democratic Republic of Congo) */
342 {"kn", HB_TAG('K','A','N',' ')}, /* Kannada */
343 {"ko", HB_TAG('K','O','R',' ')}, /* Korean */
344 {"koi", HB_TAG('K','O','P',' ')}, /* Komi-Permyak */
345 {"kok", HB_TAG('K','O','K',' ')}, /* Konkani */
346 {"kpe", HB_TAG('K','P','L',' ')}, /* Kpelle */
347 {"kpv", HB_TAG('K','O','Z',' ')}, /* Komi-Zyrian */
348 {"kpy", HB_TAG('K','Y','K',' ')}, /* Koryak */
349 {"kqy", HB_TAG('K','R','T',' ')}, /* Koorete */
350 {"kr", HB_TAG('K','N','R',' ')}, /* Kanuri */
351 {"kri", HB_TAG('K','R','I',' ')}, /* Krio */
352 {"krl", HB_TAG('K','R','L',' ')}, /* Karelian */
353 {"kru", HB_TAG('K','U','U',' ')}, /* Kurukh */
354 {"ks", HB_TAG('K','S','H',' ')}, /* Kashmiri */
355 {"ku", HB_TAG('K','U','R',' ')}, /* Kurdish */
356 {"kum", HB_TAG('K','U','M',' ')}, /* Kumyk */
357 {"kvd", HB_TAG('K','U','I',' ')}, /* Kui (Indonesia) */
358 {"kxc", HB_TAG('K','M','S',' ')}, /* Komso */
359 {"kxu", HB_TAG('K','U','I',' ')}, /* Kui (India) */
360 {"ky", HB_TAG('K','I','R',' ')}, /* Kirghiz */
361 {"la", HB_TAG('L','A','T',' ')}, /* Latin */
362 {"lad", HB_TAG('J','U','D',' ')}, /* Ladino */
363 {"lb", HB_TAG('L','T','Z',' ')}, /* Luxembourgish */
364 {"lbe", HB_TAG('L','A','K',' ')}, /* Lak */
365 {"lbj", HB_TAG('L','D','K',' ')}, /* Ladakhi */
366 {"lez", HB_TAG('L','E','Z',' ')}, /* Lezgi */
367 {"lg", HB_TAG('L','U','G',' ')}, /* Luganda */
368 {"lif", HB_TAG('L','M','B',' ')}, /* Limbu */
369 {"lld", HB_TAG('L','A','D',' ')}, /* Ladin */
370 {"lmn", HB_TAG('L','A','M',' ')}, /* Lambani */
371 {"ln", HB_TAG('L','I','N',' ')}, /* Lingala */
372 {"lo", HB_TAG('L','A','O',' ')}, /* Lao */
373 {"lt", HB_TAG('L','T','H',' ')}, /* Lithuanian */
374 {"lu", HB_TAG('L','U','B',' ')}, /* Luba-Katanga */
375 {"lua", HB_TAG('L','U','B',' ')}, /* Luba-Kasai */
376 {"luo", HB_TAG('L','U','O',' ')}, /* Luo (Kenya and Tanzania) */
377 {"lus", HB_TAG('M','I','Z',' ')}, /* Mizo */
378 {"luy", HB_TAG('L','U','H',' ')}, /* Luhya [macrolanguage] */
379 {"lv", HB_TAG('L','V','I',' ')}, /* Latvian */
380 {"lzz", HB_TAG('L','A','Z',' ')}, /* Laz */
381 {"mai", HB_TAG('M','T','H',' ')}, /* Maithili */
382 {"mdc", HB_TAG('M','L','E',' ')}, /* Male (Papua New Guinea) */
383 {"mdf", HB_TAG('M','O','K',' ')}, /* Moksha */
384 {"mdy", HB_TAG('M','L','E',' ')}, /* Male (Ethiopia) */
385 {"men", HB_TAG('M','D','E',' ')}, /* Mende (Sierra Leone) */
386 {"mg", HB_TAG('M','L','G',' ')}, /* Malagasy */
387 {"mhr", HB_TAG('L','M','A',' ')}, /* Low Mari */
388 {"mi", HB_TAG('M','R','I',' ')}, /* Maori */
389 {"mk", HB_TAG('M','K','D',' ')}, /* Macedonian */
390 {"ml", HB_TAG('M','L','R',' ')}, /* Malayalam */
391 {"mn", HB_TAG('M','N','G',' ')}, /* Mongolian */
392 {"mnc", HB_TAG('M','C','H',' ')}, /* Manchu */
393 {"mni", HB_TAG('M','N','I',' ')}, /* Manipuri */
394 {"mnk", HB_TAG('M','N','D',' ')}, /* Mandinka */
395 {"mns", HB_TAG('M','A','N',' ')}, /* Mansi */
396 {"mnw", HB_TAG('M','O','N',' ')}, /* Mon */
397 {"mo", HB_TAG('M','O','L',' ')}, /* Moldavian */
398 {"moh", HB_TAG('M','O','H',' ')}, /* Mohawk */
399 {"mpe", HB_TAG('M','A','J',' ')}, /* Majang */
400 {"mr", HB_TAG('M','A','R',' ')}, /* Marathi */
401 {"mrj", HB_TAG('H','M','A',' ')}, /* High Mari */
402 {"ms", HB_TAG('M','L','Y',' ')}, /* Malay */
403 {"mt", HB_TAG('M','T','S',' ')}, /* Maltese */
404 {"mwr", HB_TAG('M','A','W',' ')}, /* Marwari */
405 {"my", HB_TAG('B','R','M',' ')}, /* Burmese */
406 {"mym", HB_TAG('M','E','N',' ')}, /* Me'en */
407 {"myv", HB_TAG('E','R','Z',' ')}, /* Erzya */
408 {"nag", HB_TAG('N','A','G',' ')}, /* Naga-Assamese */
409 {"nb", HB_TAG('N','O','R',' ')}, /* Norwegian Bokmål */
410 {"nco", HB_TAG('S','I','B',' ')}, /* Sibe */
411 {"nd", HB_TAG('N','D','B',' ')}, /* [North] Ndebele */
412 {"ne", HB_TAG('N','E','P',' ')}, /* Nepali */
413 {"new", HB_TAG('N','E','W',' ')}, /* Newari */
414 {"ng", HB_TAG('N','D','G',' ')}, /* Ndonga */
415 {"ngl", HB_TAG('L','M','W',' ')}, /* Lomwe */
416 {"niu", HB_TAG('N','I','U',' ')}, /* Niuean */
417 {"niv", HB_TAG('G','I','L',' ')}, /* Gilyak */
418 {"nl", HB_TAG('N','L','D',' ')}, /* Dutch */
419 {"nn", HB_TAG('N','Y','N',' ')}, /* Norwegian Nynorsk */
420 {"no", HB_TAG('N','O','R',' ')}, /* Norwegian (deprecated) */
421 {"nod", HB_TAG('N','T','A',' ')}, /* Northern Tai */
422 {"nog", HB_TAG('N','O','G',' ')}, /* Nogai */
423 {"nqo", HB_TAG('N','K','O',' ')}, /* N'Ko */
424 {"nr", HB_TAG('N','D','B',' ')}, /* [South] Ndebele */
425 {"nsk", HB_TAG('N','A','S',' ')}, /* Naskapi */
426 {"nso", HB_TAG('S','O','T',' ')}, /* [Northern] Sotho */
427 {"ny", HB_TAG('C','H','I',' ')}, /* Nyanja */
428 {"nyn", HB_TAG('N','K','L',' ')}, /* Nkole */
429 {"oc", HB_TAG('O','C','I',' ')}, /* Occitan (post 1500) */
430 {"oj", HB_TAG('O','J','B',' ')}, /* Ojibwa */
431 {"ojs", HB_TAG('O','C','R',' ')}, /* Oji-Cree */
432 {"om", HB_TAG('O','R','O',' ')}, /* Oromo */
433 {"or", HB_TAG('O','R','I',' ')}, /* Oriya */
434 {"os", HB_TAG('O','S','S',' ')}, /* Ossetian */
435 {"pa", HB_TAG('P','A','N',' ')}, /* Panjabi */
436 {"pce", HB_TAG('P','L','G',' ')}, /* [Ruching] Palaung */
437 {"pi", HB_TAG('P','A','L',' ')}, /* Pali */
438 {"pl", HB_TAG('P','L','K',' ')}, /* Polish */
439 {"pll", HB_TAG('P','L','G',' ')}, /* [Shwe] Palaung */
440 {"plp", HB_TAG('P','A','P',' ')}, /* Palpa */
441 {"prs", HB_TAG('D','R','I',' ')}, /* Dari */
442 {"ps", HB_TAG('P','A','S',' ')}, /* Pushto */
443 {"pt", HB_TAG('P','T','G',' ')}, /* Portuguese */
444 {"raj", HB_TAG('R','A','J',' ')}, /* Rajasthani */
445 {"rbb", HB_TAG('P','L','G',' ')}, /* [Rumai] Palaung */
446 {"ria", HB_TAG('R','I','A',' ')}, /* Riang (India) */
447 {"ril", HB_TAG('R','I','A',' ')}, /* Riang (Myanmar) */
448 {"rki", HB_TAG('A','R','K',' ')}, /* Arakanese */
449 {"rm", HB_TAG('R','M','S',' ')}, /* Rhaeto-Romanic */
450 {"ro", HB_TAG('R','O','M',' ')}, /* Romanian */
451 {"rom", HB_TAG('R','O','Y',' ')}, /* Romany */
452 {"ru", HB_TAG('R','U','S',' ')}, /* Russian */
453 {"rue", HB_TAG('R','S','Y',' ')}, /* Rusyn */
454 {"rw", HB_TAG('R','U','A',' ')}, /* Ruanda */
455 {"sa", HB_TAG('S','A','N',' ')}, /* Sanskrit */
456 {"sah", HB_TAG('Y','A','K',' ')}, /* Yakut */
457 {"sat", HB_TAG('S','A','T',' ')}, /* Santali */
458 {"sck", HB_TAG('S','A','D',' ')}, /* Sadri */
459 {"scs", HB_TAG('S','L','A',' ')}, /* [North] Slavey */
460 {"sd", HB_TAG('S','N','D',' ')}, /* Sindhi */
461 {"se", HB_TAG('N','S','M',' ')}, /* Northern Sami */
462 {"seh", HB_TAG('S','N','A',' ')}, /* Sena */
463 {"sel", HB_TAG('S','E','L',' ')}, /* Selkup */
464 {"sg", HB_TAG('S','G','O',' ')}, /* Sango */
465 {"shn", HB_TAG('S','H','N',' ')}, /* Shan */
466 {"si", HB_TAG('S','N','H',' ')}, /* Sinhala */
467 {"sid", HB_TAG('S','I','D',' ')}, /* Sidamo */
468 {"sjd", HB_TAG('K','S','M',' ')}, /* Kildin Sami */
469 {"sk", HB_TAG('S','K','Y',' ')}, /* Slovak */
470 {"skr", HB_TAG('S','R','K',' ')}, /* Seraiki */
471 {"sl", HB_TAG('S','L','V',' ')}, /* Slovenian */
472 {"sm", HB_TAG('S','M','O',' ')}, /* Samoan */
473 {"sma", HB_TAG('S','S','M',' ')}, /* Southern Sami */
474 {"smj", HB_TAG('L','S','M',' ')}, /* Lule Sami */
475 {"smn", HB_TAG('I','S','M',' ')}, /* Inari Sami */
476 {"sms", HB_TAG('S','K','S',' ')}, /* Skolt Sami */
477 {"snk", HB_TAG('S','N','K',' ')}, /* Soninke */
478 {"so", HB_TAG('S','M','L',' ')}, /* Somali */
479 {"sq", HB_TAG('S','Q','I',' ')}, /* Albanian */
480 {"sr", HB_TAG('S','R','B',' ')}, /* Serbian */
481 {"srr", HB_TAG('S','R','R',' ')}, /* Serer */
482 {"ss", HB_TAG('S','W','Z',' ')}, /* Swazi */
483 {"st", HB_TAG('S','O','T',' ')}, /* [Southern] Sotho */
484 {"suq", HB_TAG('S','U','R',' ')}, /* Suri */
485 {"sv", HB_TAG('S','V','E',' ')}, /* Swedish */
486 {"sva", HB_TAG('S','V','A',' ')}, /* Svan */
487 {"sw", HB_TAG('S','W','K',' ')}, /* Swahili */
488 {"swb", HB_TAG('C','M','R',' ')}, /* Comorian */
489 {"syr", HB_TAG('S','Y','R',' ')}, /* Syriac */
490 {"ta", HB_TAG('T','A','M',' ')}, /* Tamil */
491 {"tab", HB_TAG('T','A','B',' ')}, /* Tabasaran */
492 {"tcy", HB_TAG('T','U','L',' ')}, /* Tulu */
493 {"te", HB_TAG('T','E','L',' ')}, /* Telugu */
494 {"tem", HB_TAG('T','M','N',' ')}, /* Temne */
495 {"tg", HB_TAG('T','A','J',' ')}, /* Tajik */
496 {"th", HB_TAG('T','H','A',' ')}, /* Thai */
497 {"ti", HB_TAG('T','G','Y',' ')}, /* Tigrinya */
498 {"tig", HB_TAG('T','G','R',' ')}, /* Tigre */
499 {"tk", HB_TAG('T','K','M',' ')}, /* Turkmen */
500 {"tn", HB_TAG('T','N','A',' ')}, /* Tswana */
501 {"to", HB_TAG('T','G','N',' ')}, /* Tonga (Tonga Islands) */
502 {"tr", HB_TAG('T','R','K',' ')}, /* Turkish */
503 {"tru", HB_TAG('T','U','A',' ')}, /* Turoyo Aramaic */
504 {"ts", HB_TAG('T','S','G',' ')}, /* Tsonga */
505 {"tt", HB_TAG('T','A','T',' ')}, /* Tatar */
506 {"tw", HB_TAG('T','W','I',' ')}, /* Twi */
507 {"ty", HB_TAG('T','H','T',' ')}, /* Tahitian */
508 {"tyv", HB_TAG('T','U','V',' ')}, /* Tuvin */
509 {"udm", HB_TAG('U','D','M',' ')}, /* Udmurt */
510 {"ug", HB_TAG('U','Y','G',' ')}, /* Uighur */
511 {"uk", HB_TAG('U','K','R',' ')}, /* Ukrainian */
512 {"umb", HB_TAG('M','B','N',' ')}, /* [South] Mbundu */
513 {"unr", HB_TAG('M','U','N',' ')}, /* Mundari */
514 {"ur", HB_TAG('U','R','D',' ')}, /* Urdu */
515 {"uz", HB_TAG('U','Z','B',' ')}, /* Uzbek */
516 {"ve", HB_TAG('V','E','N',' ')}, /* Venda */
517 {"vi", HB_TAG('V','I','T',' ')}, /* Vietnamese */
518 {"vmw", HB_TAG('M','A','K',' ')}, /* Makua */
519 {"wbm", HB_TAG('W','A',' ',' ')}, /* Wa */
520 {"wbr", HB_TAG('W','A','G',' ')}, /* Wagdi */
521 {"wo", HB_TAG('W','L','F',' ')}, /* Wolof */
522 {"xal", HB_TAG('K','L','M',' ')}, /* Kalmyk */
523 {"xh", HB_TAG('X','H','S',' ')}, /* Xhosa */
524 {"xom", HB_TAG('K','M','O',' ')}, /* Komo (Sudan) */
525 {"xsl", HB_TAG('S','S','L',' ')}, /* South Slavey */
526 {"yi", HB_TAG('J','I','I',' ')}, /* Yiddish */
527 {"yo", HB_TAG('Y','B','A',' ')}, /* Yoruba */
528 {"yso", HB_TAG('N','I','S',' ')}, /* Nisi (China) */
529 {"zne", HB_TAG('Z','N','D',' ')}, /* Zande */
530 {"zu", HB_TAG('Z','U','L',' ')} /* Zulu */
531
532 /* The corresponding languages IDs for the following IDs are unclear,
533 * overlap, or are architecturally weird. Needs more research. */
534
535 /*{"ahg/awn/xan?", HB_TAG('A','G','W',' ')},*/ /* Agaw */
536 /*{"gsw?/gsw-FR?", HB_TAG('A','L','S',' ')},*/ /* Alsatian */
537 /*{"krc", HB_TAG('B','A','L',' ')},*/ /* Balkar */
538 /*{"??", HB_TAG('B','C','R',' ')},*/ /* Bible Cree */
539 /*{"sgw?", HB_TAG('C','H','G',' ')},*/ /* Chaha Gurage */
540 /*{"acf/gcf?", HB_TAG('F','A','N',' ')},*/ /* French Antillean */
541 /*{"vls/nl-be", HB_TAG('F','L','E',' ')},*/ /* Flemish */
542 /*{"enf?/yrk?", HB_TAG('F','N','E',' ')},*/ /* Forest Nenets */
543 /*{"fuf?", HB_TAG('F','T','A',' ')},*/ /* Futa */
544 /*{"ar-Syrc?", HB_TAG('G','A','R',' ')},*/ /* Garshuni */
545 /*{"cfm/rnl?", HB_TAG('H','A','L',' ')},*/ /* Halam */
546 /*{"ga-Latg?/Latg?", HB_TAG('I','R','T',' ')},*/ /* Irish Traditional */
547 /*{"krc", HB_TAG('K','A','R',' ')},*/ /* Karachay */
548 /*{"alw?/ktb?", HB_TAG('K','E','B',' ')},*/ /* Kebena */
549 /*{"Geok", HB_TAG('K','G','E',' ')},*/ /* Khutsuri Georgian */
550 /*{"kca", HB_TAG('K','H','K',' ')},*/ /* Khanty-Kazim */
551 /*{"kca", HB_TAG('K','H','S',' ')},*/ /* Khanty-Shurishkar */
552 /*{"kca", HB_TAG('K','H','V',' ')},*/ /* Khanty-Vakhi */
553 /*{"guz?/kqs?/kss?", HB_TAG('K','I','S',' ')},*/ /* Kisii */
554 /*{"kfa/kfi?/kpb?/xua?/xuj?", HB_TAG('K','O','D',' ')},*/ /* Kodagu */
555 /*{"okm?/oko?", HB_TAG('K','O','H',' ')},*/ /* Korean Old Hangul */
556 /*{"kon?/ktu?/...", HB_TAG('K','O','N',' ')},*/ /* Kikongo */
557 /*{"kfx?", HB_TAG('K','U','L',' ')},*/ /* Kulvi */
558 /*{"??", HB_TAG('L','A','H',' ')},*/ /* Lahuli */
559 /*{"??", HB_TAG('L','C','R',' ')},*/ /* L-Cree */
560 /*{"??", HB_TAG('M','A','L',' ')},*/ /* Malayalam Traditional */
561 /*{"mnk?/mlq?/...", HB_TAG('M','L','N',' ')},*/ /* Malinke */
562 /*{"man?/myq?/mku?/msc?/...", HB_TAG('M','N','K',' ')},*/ /* Maninka */
563 /*{"??", HB_TAG('M','O','R',' ')},*/ /* Moroccan */
564 /*{"??", HB_TAG('N','C','R',' ')},*/ /* N-Cree */
565 /*{"??", HB_TAG('N','H','C',' ')},*/ /* Norway House Cree */
566 /*{"jpa?/sam?", HB_TAG('P','A','A',' ')},*/ /* Palestinian Aramaic */
567 /*{"polyton", HB_TAG('P','G','R',' ')},*/ /* Polytonic Greek */
568 /*{"??", HB_TAG('Q','I','N',' ')},*/ /* Asho Chin */
569 /*{"??", HB_TAG('R','C','R',' ')},*/ /* R-Cree */
570 /*{"chp?", HB_TAG('S','A','Y',' ')},*/ /* Sayisi */
571 /*{"xan?", HB_TAG('S','E','K',' ')},*/ /* Sekota */
572 /*{"stv/wle?/xst?", HB_TAG('S','I','G',' ')},*/ /* Silte Gurage */
573 /*{"ngo?", HB_TAG('S','X','T',' ')},*/ /* Sutu */
574 /*{"??", HB_TAG('T','C','R',' ')},*/ /* TH-Cree */
575 /*{"tnz?/tog?/toi?", HB_TAG('T','N','G',' ')},*/ /* Tonga */
576 /*{"enh?/yrk?", HB_TAG('T','N','E',' ')},*/ /* Tundra Nenets */
577 /*{"??", HB_TAG('T','O','D',' ')},*/ /* Todo */
578 /*{"??", HB_TAG('W','C','R',' ')},*/ /* West-Cree */
579 /*{"??", HB_TAG('Y','C','R',' ')},*/ /* Y-Cree */
580 /*{"??", HB_TAG('Y','I','C',' ')},*/ /* Yi Classic */
581 /*{"ii?/Yiii?", HB_TAG('Y','I','M',' ')},*/ /* Yi Modern */
582 /*{"??", HB_TAG('Z','H','P',' ')},*/ /* Chinese Phonetic */
583 };
584
585 static const LangTag ot_languages_zh[] = {
586 {"zh-cn", HB_TAG('Z','H','S',' ')}, /* Chinese (China) */
587 {"zh-hk", HB_TAG('Z','H','H',' ')}, /* Chinese (Hong Kong) */
588 {"zh-mo", HB_TAG('Z','H','T',' ')}, /* Chinese (Macao) */
589 {"zh-sg", HB_TAG('Z','H','S',' ')}, /* Chinese (Singapore) */
590 {"zh-tw", HB_TAG('Z','H','T',' ')} /* Chinese (Taiwan) */
591 };
592
593 static int
lang_compare_first_component(const char * a,const char * b)594 lang_compare_first_component (const char *a,
595 const char *b)
596 {
597 unsigned int da, db;
598 const char *p;
599
600 p = strchr (a, '-');
601 da = p ? (unsigned int) (p - a) : strlen (a);
602
603 p = strchr (b, '-');
604 db = p ? (unsigned int) (p - b) : strlen (b);
605
606 return strncmp (a, b, MAX (da, db));
607 }
608
609 static hb_bool_t
lang_matches(const char * lang_str,const char * spec)610 lang_matches (const char *lang_str, const char *spec)
611 {
612 unsigned int len = strlen (spec);
613
614 return strncmp (lang_str, spec, len) == 0 &&
615 (lang_str[len] == '\0' || lang_str[len] == '-');
616 }
617
618 hb_tag_t
hb_ot_tag_from_language(hb_language_t language)619 hb_ot_tag_from_language (hb_language_t language)
620 {
621 const char *lang_str, *s;
622 const LangTag *lang_tag;
623
624 if (language == HB_LANGUAGE_INVALID)
625 return HB_OT_TAG_DEFAULT_LANGUAGE;
626
627 lang_str = hb_language_to_string (language);
628
629 s = strstr (lang_str, "x-hbot");
630 if (s) {
631 char tag[4];
632 int i;
633 s += 6;
634 for (i = 0; i < 4 && ISALPHA (s[i]); i++)
635 tag[i] = TOUPPER (s[i]);
636 if (i) {
637 for (; i < 4; i++)
638 tag[i] = ' ';
639 return HB_TAG_CHAR4 (tag);
640 }
641 }
642
643 /* Find a language matching in the first component */
644 lang_tag = (LangTag *) bsearch (lang_str, ot_languages,
645 ARRAY_LENGTH (ot_languages), sizeof (LangTag),
646 (hb_compare_func_t) lang_compare_first_component);
647 if (lang_tag)
648 return lang_tag->tag;
649
650 /* Otherwise, check the Chinese ones */
651 if (0 == lang_compare_first_component (lang_str, "zh"))
652 {
653 unsigned int i;
654
655 for (i = 0; i < ARRAY_LENGTH (ot_languages_zh); i++)
656 {
657 lang_tag = &ot_languages_zh[i];
658 if (lang_matches (lang_tag->language, lang_str))
659 return lang_tag->tag;
660 }
661
662 /* Otherwise just return 'ZHS ' */
663 return HB_TAG('Z','H','S',' ');
664 }
665
666 s = strchr (lang_str, '-');
667 if (!s)
668 s = lang_str + strlen (lang_str);
669 if (s - lang_str == 3) {
670 /* Assume it's ISO-639-3 and upper-case and use it. */
671 return hb_tag_from_string (lang_str, s - lang_str) & ~0x20202000;
672 }
673
674 return HB_OT_TAG_DEFAULT_LANGUAGE;
675 }
676
677 hb_language_t
hb_ot_tag_to_language(hb_tag_t tag)678 hb_ot_tag_to_language (hb_tag_t tag)
679 {
680 unsigned int i;
681
682 if (tag == HB_OT_TAG_DEFAULT_LANGUAGE)
683 return NULL;
684
685 for (i = 0; i < ARRAY_LENGTH (ot_languages); i++)
686 if (ot_languages[i].tag == tag)
687 return hb_language_from_string (ot_languages[i].language, -1);
688
689 /* If tag starts with ZH, it's Chinese */
690 if ((tag & 0xFFFF0000) == 0x5A480000) {
691 switch (tag) {
692 case HB_TAG('Z','H','H',' '): return hb_language_from_string ("zh-hk", -1); /* Hong Kong */
693 default: {
694 /* Encode the tag... */
695 unsigned char buf[14] = "zh-x-hbot";
696 buf[9] = tag >> 24;
697 buf[10] = (tag >> 16) & 0xFF;
698 buf[11] = (tag >> 8) & 0xFF;
699 buf[12] = tag & 0xFF;
700 if (buf[12] == 0x20)
701 buf[12] = '\0';
702 buf[13] = '\0';
703 return hb_language_from_string ((char *) buf, -1);
704 }
705 }
706 }
707
708 /* Else return a custom language in the form of "x-hbotABCD" */
709 {
710 unsigned char buf[11] = "x-hbot";
711 buf[6] = tag >> 24;
712 buf[7] = (tag >> 16) & 0xFF;
713 buf[8] = (tag >> 8) & 0xFF;
714 buf[9] = tag & 0xFF;
715 if (buf[9] == 0x20)
716 buf[9] = '\0';
717 buf[10] = '\0';
718 return hb_language_from_string ((char *) buf, -1);
719 }
720 }
721
722
723