1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/gfx/harfbuzz/src/hb-ot-tag.cc Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,903 @@
1.4 +/*
1.5 + * Copyright © 2009 Red Hat, Inc.
1.6 + * Copyright © 2011 Google, Inc.
1.7 + *
1.8 + * This is part of HarfBuzz, a text shaping library.
1.9 + *
1.10 + * Permission is hereby granted, without written agreement and without
1.11 + * license or royalty fees, to use, copy, modify, and distribute this
1.12 + * software and its documentation for any purpose, provided that the
1.13 + * above copyright notice and the following two paragraphs appear in
1.14 + * all copies of this software.
1.15 + *
1.16 + * IN NO EVENT SHALL THE COPYRIGHT HOLDER BE LIABLE TO ANY PARTY FOR
1.17 + * DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES
1.18 + * ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS DOCUMENTATION, EVEN
1.19 + * IF THE COPYRIGHT HOLDER HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH
1.20 + * DAMAGE.
1.21 + *
1.22 + * THE COPYRIGHT HOLDER SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING,
1.23 + * BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
1.24 + * FITNESS FOR A PARTICULAR PURPOSE. THE SOFTWARE PROVIDED HEREUNDER IS
1.25 + * ON AN "AS IS" BASIS, AND THE COPYRIGHT HOLDER HAS NO OBLIGATION TO
1.26 + * PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
1.27 + *
1.28 + * Red Hat Author(s): Behdad Esfahbod
1.29 + * Google Author(s): Behdad Esfahbod, Roozbeh Pournader
1.30 + */
1.31 +
1.32 +#include "hb-private.hh"
1.33 +
1.34 +#include <string.h>
1.35 +
1.36 +
1.37 +
1.38 +/* hb_script_t */
1.39 +
1.40 +static hb_tag_t
1.41 +hb_ot_old_tag_from_script (hb_script_t script)
1.42 +{
1.43 + /* This seems to be accurate as of end of 2012. */
1.44 +
1.45 + switch ((hb_tag_t) script) {
1.46 + case HB_SCRIPT_INVALID: return HB_OT_TAG_DEFAULT_SCRIPT;
1.47 +
1.48 + /* KATAKANA and HIRAGANA both map to 'kana' */
1.49 + case HB_SCRIPT_HIRAGANA: return HB_TAG('k','a','n','a');
1.50 +
1.51 + /* Spaces at the end are preserved, unlike ISO 15924 */
1.52 + case HB_SCRIPT_LAO: return HB_TAG('l','a','o',' ');
1.53 + case HB_SCRIPT_YI: return HB_TAG('y','i',' ',' ');
1.54 + /* Unicode-5.0 additions */
1.55 + case HB_SCRIPT_NKO: return HB_TAG('n','k','o',' ');
1.56 + /* Unicode-5.1 additions */
1.57 + case HB_SCRIPT_VAI: return HB_TAG('v','a','i',' ');
1.58 + /* Unicode-5.2 additions */
1.59 + /* Unicode-6.0 additions */
1.60 + }
1.61 +
1.62 + /* Else, just change first char to lowercase and return */
1.63 + return ((hb_tag_t) script) | 0x20000000;
1.64 +}
1.65 +
1.66 +static hb_script_t
1.67 +hb_ot_old_tag_to_script (hb_tag_t tag)
1.68 +{
1.69 + if (unlikely (tag == HB_OT_TAG_DEFAULT_SCRIPT))
1.70 + return HB_SCRIPT_INVALID;
1.71 +
1.72 + /* This side of the conversion is fully algorithmic. */
1.73 +
1.74 + /* Any spaces at the end of the tag are replaced by repeating the last
1.75 + * letter. Eg 'nko ' -> 'Nkoo' */
1.76 + if (unlikely ((tag & 0x0000FF00) == 0x00002000))
1.77 + tag |= (tag >> 8) & 0x0000FF00; /* Copy second letter to third */
1.78 + if (unlikely ((tag & 0x000000FF) == 0x00000020))
1.79 + tag |= (tag >> 8) & 0x000000FF; /* Copy third letter to fourth */
1.80 +
1.81 + /* Change first char to uppercase and return */
1.82 + return (hb_script_t) (tag & ~0x20000000);
1.83 +}
1.84 +
1.85 +static hb_tag_t
1.86 +hb_ot_new_tag_from_script (hb_script_t script)
1.87 +{
1.88 + switch ((hb_tag_t) script) {
1.89 + case HB_SCRIPT_BENGALI: return HB_TAG('b','n','g','2');
1.90 + case HB_SCRIPT_DEVANAGARI: return HB_TAG('d','e','v','2');
1.91 + case HB_SCRIPT_GUJARATI: return HB_TAG('g','j','r','2');
1.92 + case HB_SCRIPT_GURMUKHI: return HB_TAG('g','u','r','2');
1.93 + case HB_SCRIPT_KANNADA: return HB_TAG('k','n','d','2');
1.94 + case HB_SCRIPT_MALAYALAM: return HB_TAG('m','l','m','2');
1.95 + case HB_SCRIPT_ORIYA: return HB_TAG('o','r','y','2');
1.96 + case HB_SCRIPT_TAMIL: return HB_TAG('t','m','l','2');
1.97 + case HB_SCRIPT_TELUGU: return HB_TAG('t','e','l','2');
1.98 + case HB_SCRIPT_MYANMAR: return HB_TAG('m','y','m','2');
1.99 + }
1.100 +
1.101 + return HB_OT_TAG_DEFAULT_SCRIPT;
1.102 +}
1.103 +
1.104 +static hb_script_t
1.105 +hb_ot_new_tag_to_script (hb_tag_t tag)
1.106 +{
1.107 + switch (tag) {
1.108 + case HB_TAG('b','n','g','2'): return HB_SCRIPT_BENGALI;
1.109 + case HB_TAG('d','e','v','2'): return HB_SCRIPT_DEVANAGARI;
1.110 + case HB_TAG('g','j','r','2'): return HB_SCRIPT_GUJARATI;
1.111 + case HB_TAG('g','u','r','2'): return HB_SCRIPT_GURMUKHI;
1.112 + case HB_TAG('k','n','d','2'): return HB_SCRIPT_KANNADA;
1.113 + case HB_TAG('m','l','m','2'): return HB_SCRIPT_MALAYALAM;
1.114 + case HB_TAG('o','r','y','2'): return HB_SCRIPT_ORIYA;
1.115 + case HB_TAG('t','m','l','2'): return HB_SCRIPT_TAMIL;
1.116 + case HB_TAG('t','e','l','2'): return HB_SCRIPT_TELUGU;
1.117 + case HB_TAG('m','y','m','2'): return HB_SCRIPT_MYANMAR;
1.118 + }
1.119 +
1.120 + return HB_SCRIPT_UNKNOWN;
1.121 +}
1.122 +
1.123 +/*
1.124 + * Complete list at:
1.125 + * https://www.microsoft.com/typography/otspec/scripttags.htm
1.126 + * https://www.microsoft.com/typography/otspec160/scripttagsProposed.htm
1.127 + *
1.128 + * Most of the script tags are the same as the ISO 15924 tag but lowercased.
1.129 + * So we just do that, and handle the exceptional cases in a switch.
1.130 + */
1.131 +
1.132 +void
1.133 +hb_ot_tags_from_script (hb_script_t script,
1.134 + hb_tag_t *script_tag_1,
1.135 + hb_tag_t *script_tag_2)
1.136 +{
1.137 + hb_tag_t new_tag;
1.138 +
1.139 + *script_tag_2 = HB_OT_TAG_DEFAULT_SCRIPT;
1.140 + *script_tag_1 = hb_ot_old_tag_from_script (script);
1.141 +
1.142 + new_tag = hb_ot_new_tag_from_script (script);
1.143 + if (unlikely (new_tag != HB_OT_TAG_DEFAULT_SCRIPT)) {
1.144 + *script_tag_2 = *script_tag_1;
1.145 + *script_tag_1 = new_tag;
1.146 + }
1.147 +}
1.148 +
1.149 +hb_script_t
1.150 +hb_ot_tag_to_script (hb_tag_t tag)
1.151 +{
1.152 + if (unlikely ((tag & 0x000000FF) == '2'))
1.153 + return hb_ot_new_tag_to_script (tag);
1.154 +
1.155 + return hb_ot_old_tag_to_script (tag);
1.156 +}
1.157 +
1.158 +
1.159 +/* hb_language_t */
1.160 +
1.161 +typedef struct {
1.162 + char language[6];
1.163 + hb_tag_t tag;
1.164 +} LangTag;
1.165 +
1.166 +/*
1.167 + * Complete list at:
1.168 + * http://www.microsoft.com/typography/otspec/languagetags.htm
1.169 + *
1.170 + * Generated by intersecting the OpenType language tag list from
1.171 + * Draft OpenType 1.5 spec, with with the ISO 639-3 codes from
1.172 + * 2008-08-04, matching on name, and finally adjusted manually.
1.173 + *
1.174 + * Updated on 2012-12-07 with more research into remaining codes.
1.175 + *
1.176 + * Updated on 2013-11-23 based on usage in SIL and Microsoft fonts,
1.177 + * the new proposal from Microsoft, and latest ISO 639-3 names.
1.178 + *
1.179 + * Some items still missing. Those are commented out at the end.
1.180 + * Keep sorted for bsearch.
1.181 + */
1.182 +
1.183 +static const LangTag ot_languages[] = {
1.184 + {"aa", HB_TAG('A','F','R',' ')}, /* Afar */
1.185 + {"ab", HB_TAG('A','B','K',' ')}, /* Abkhazian */
1.186 + {"abq", HB_TAG('A','B','A',' ')}, /* Abaza */
1.187 + {"ach", HB_TAG('A','C','H',' ')}, /* Acoli */
1.188 + {"ada", HB_TAG('D','N','G',' ')}, /* Dangme */
1.189 + {"ady", HB_TAG('A','D','Y',' ')}, /* Adyghe */
1.190 + {"af", HB_TAG('A','F','K',' ')}, /* Afrikaans */
1.191 + {"aii", HB_TAG('S','W','A',' ')}, /* Swadaya Aramaic */
1.192 + {"aio", HB_TAG('A','I','O',' ')}, /* Aiton */
1.193 + {"aiw", HB_TAG('A','R','I',' ')}, /* Aari */
1.194 + {"ak", HB_TAG('T','W','I',' ')}, /* Akan [macrolanguage] */
1.195 + {"alt", HB_TAG('A','L','T',' ')}, /* [Southern] Altai */
1.196 + {"am", HB_TAG('A','M','H',' ')}, /* Amharic */
1.197 + {"amf", HB_TAG('H','B','N',' ')}, /* Hammer-Banna */
1.198 + {"an", HB_TAG('A','R','G',' ')}, /* Aragonese */
1.199 + {"ang", HB_TAG('A','N','G',' ')}, /* Old English (ca. 450-1100) */
1.200 + {"ar", HB_TAG('A','R','A',' ')}, /* Arabic [macrolanguage] */
1.201 + {"arb", HB_TAG('A','R','A',' ')}, /* Standard Arabic */
1.202 + {"arn", HB_TAG('M','A','P',' ')}, /* Mapudungun */
1.203 + {"ary", HB_TAG('M','O','R',' ')}, /* Moroccan Arabic */
1.204 + {"as", HB_TAG('A','S','M',' ')}, /* Assamese */
1.205 + {"ast", HB_TAG('A','S','T',' ')}, /* Asturian/Asturleonese/Bable/Leonese */
1.206 + {"ath", HB_TAG('A','T','H',' ')}, /* Athapaskan [family] */
1.207 + {"atv", HB_TAG('A','L','T',' ')}, /* [Northern] Altai */
1.208 + {"av", HB_TAG('A','V','R',' ')}, /* Avaric */
1.209 + {"awa", HB_TAG('A','W','A',' ')}, /* Awadhi */
1.210 + {"ay", HB_TAG('A','Y','M',' ')}, /* Aymara [macrolanguage] */
1.211 + {"az", HB_TAG('A','Z','E',' ')}, /* Azerbaijani [macrolanguage] */
1.212 + {"azb", HB_TAG('A','Z','B',' ')}, /* South Azerbaijani */
1.213 + {"azj", HB_TAG('A','Z','E',' ')}, /* North Azerbaijani */
1.214 + {"ba", HB_TAG('B','S','H',' ')}, /* Bashkir */
1.215 + {"bai", HB_TAG('B','M','L',' ')}, /* Bamileke [family] */
1.216 + {"bal", HB_TAG('B','L','I',' ')}, /* Baluchi [macrolangauge] */
1.217 + {"ban", HB_TAG('B','A','N',' ')}, /* Balinese */
1.218 + {"bar", HB_TAG('B','A','R',' ')}, /* Bavarian */
1.219 + {"bbc", HB_TAG('B','B','C',' ')}, /* Batak Toba */
1.220 + {"bci", HB_TAG('B','A','U',' ')}, /* Baoulé */
1.221 + {"bcl", HB_TAG('B','I','K',' ')}, /* Central Bikol */
1.222 + {"bcq", HB_TAG('B','C','H',' ')}, /* Bench */
1.223 + {"be", HB_TAG('B','E','L',' ')}, /* Belarusian */
1.224 + {"bem", HB_TAG('B','E','M',' ')}, /* Bemba (Zambia) */
1.225 + {"ber", HB_TAG('B','E','R',' ')}, /* Berber [family] */
1.226 + {"bfq", HB_TAG('B','A','D',' ')}, /* Badaga */
1.227 + {"bft", HB_TAG('B','L','T',' ')}, /* Balti */
1.228 + {"bfy", HB_TAG('B','A','G',' ')}, /* Baghelkhandi */
1.229 + {"bg", HB_TAG('B','G','R',' ')}, /* Bulgarian */
1.230 + {"bgc", HB_TAG('B','G','C',' ')}, /* Haryanvi */
1.231 + {"bgq", HB_TAG('B','G','Q',' ')}, /* Bagri */
1.232 + {"bhb", HB_TAG('B','H','I',' ')}, /* Bhili */
1.233 + {"bhk", HB_TAG('B','I','K',' ')}, /* Albay Bicolano (retired code) */
1.234 + {"bho", HB_TAG('B','H','O',' ')}, /* Bhojpuri */
1.235 + {"bi", HB_TAG('B','I','S',' ')}, /* Bislama */
1.236 + {"bik", HB_TAG('B','I','K',' ')}, /* Bikol [macrolanguage] */
1.237 + {"bin", HB_TAG('E','D','O',' ')}, /* Bini */
1.238 + {"bjj", HB_TAG('B','J','J',' ')}, /* Kanauji */
1.239 + {"bjt", HB_TAG('B','L','N',' ')}, /* Balanta-Ganja */
1.240 + {"bla", HB_TAG('B','K','F',' ')}, /* Blackfoot */
1.241 + {"ble", HB_TAG('B','L','N',' ')}, /* Balanta-Kentohe */
1.242 + {"blk", HB_TAG('B','L','K',' ')}, /* Pa'O/Pa'o Karen */
1.243 + {"bln", HB_TAG('B','I','K',' ')}, /* Southern Catanduanes Bikol */
1.244 + {"bm", HB_TAG('B','M','B',' ')}, /* Bambara */
1.245 + {"bn", HB_TAG('B','E','N',' ')}, /* Bengali */
1.246 + {"bo", HB_TAG('T','I','B',' ')}, /* Tibetan */
1.247 + {"bpy", HB_TAG('B','P','Y',' ')}, /* Bishnupriya */
1.248 + {"bqi", HB_TAG('L','R','C',' ')}, /* Bakhtiari */
1.249 + {"br", HB_TAG('B','R','E',' ')}, /* Breton */
1.250 + {"bra", HB_TAG('B','R','I',' ')}, /* Braj Bhasha */
1.251 + {"brh", HB_TAG('B','R','H',' ')}, /* Brahui */
1.252 + {"brx", HB_TAG('B','R','X',' ')}, /* Bodo (India) */
1.253 + {"bs", HB_TAG('B','O','S',' ')}, /* Bosnian */
1.254 + {"btb", HB_TAG('B','T','I',' ')}, /* Beti (Cameroon) */
1.255 + {"bto", HB_TAG('B','I','K',' ')}, /* Rinconada Bikol */
1.256 + {"bts", HB_TAG('B','T','S',' ')}, /* Batak Simalungun */
1.257 + {"bug", HB_TAG('B','U','G',' ')}, /* Buginese */
1.258 + {"bxr", HB_TAG('R','B','U',' ')}, /* Russian Buriat */
1.259 + {"byn", HB_TAG('B','I','L',' ')}, /* Bilen */
1.260 + {"ca", HB_TAG('C','A','T',' ')}, /* Catalan */
1.261 + {"cbk", HB_TAG('C','B','K',' ')}, /* Chavacano */
1.262 + {"ce", HB_TAG('C','H','E',' ')}, /* Chechen */
1.263 + {"ceb", HB_TAG('C','E','B',' ')}, /* Cebuano */
1.264 + {"cgg", HB_TAG('C','G','G',' ')}, /* Chiga */
1.265 + {"ch", HB_TAG('C','H','A',' ')}, /* Chamorro */
1.266 + {"cho", HB_TAG('C','H','O',' ')}, /* Choctaw */
1.267 + {"chp", HB_TAG('C','H','P',' ')}, /* Chipewyan */
1.268 + {"chr", HB_TAG('C','H','R',' ')}, /* Cherokee */
1.269 + {"chy", HB_TAG('C','H','Y',' ')}, /* Cheyenne */
1.270 + {"ckb", HB_TAG('K','U','R',' ')}, /* Central Kurdish (Sorani) */
1.271 + {"ckt", HB_TAG('C','H','K',' ')}, /* Chukchi */
1.272 + {"cop", HB_TAG('C','O','P',' ')}, /* Coptic */
1.273 + {"cr", HB_TAG('C','R','E',' ')}, /* Cree */
1.274 + {"crh", HB_TAG('C','R','T',' ')}, /* Crimean Tatar */
1.275 + {"crj", HB_TAG('E','C','R',' ')}, /* [Southern] East Cree */
1.276 + {"crl", HB_TAG('E','C','R',' ')}, /* [Northern] East Cree */
1.277 + {"crm", HB_TAG('M','C','R',' ')}, /* Moose Cree */
1.278 + {"crx", HB_TAG('C','R','R',' ')}, /* Carrier */
1.279 + {"cs", HB_TAG('C','S','Y',' ')}, /* Czech */
1.280 + {"csb", HB_TAG('C','S','B',' ')}, /* Kashubian */
1.281 + {"ctg", HB_TAG('C','T','G',' ')}, /* Chittagonian */
1.282 + {"cts", HB_TAG('B','I','K',' ')}, /* Northern Catanduanes Bikol */
1.283 + {"cu", HB_TAG('C','S','L',' ')}, /* Church Slavic */
1.284 + {"cv", HB_TAG('C','H','U',' ')}, /* Chuvash */
1.285 + {"cwd", HB_TAG('D','C','R',' ')}, /* Woods Cree */
1.286 + {"cy", HB_TAG('W','E','L',' ')}, /* Welsh */
1.287 + {"da", HB_TAG('D','A','N',' ')}, /* Danish */
1.288 + {"dap", HB_TAG('N','I','S',' ')}, /* Nisi (India) */
1.289 + {"dar", HB_TAG('D','A','R',' ')}, /* Dargwa */
1.290 + {"de", HB_TAG('D','E','U',' ')}, /* German */
1.291 + {"dgo", HB_TAG('D','G','O',' ')}, /* Dogri */
1.292 + {"dhd", HB_TAG('M','A','W',' ')}, /* Dhundari */
1.293 + {"din", HB_TAG('D','N','K',' ')}, /* Dinka [macrolanguage] */
1.294 + {"diq", HB_TAG('D','I','Q',' ')}, /* Dimli */
1.295 + {"dje", HB_TAG('D','J','R',' ')}, /* Zarma */
1.296 + {"dng", HB_TAG('D','U','N',' ')}, /* Dungan */
1.297 + {"doi", HB_TAG('D','G','R',' ')}, /* Dogri [macrolanguage] */
1.298 + {"dsb", HB_TAG('L','S','B',' ')}, /* Lower Sorbian */
1.299 + {"dv", HB_TAG('D','I','V',' ')}, /* Dhivehi/Divehi/Maldivian */
1.300 + {"dyu", HB_TAG('J','U','L',' ')}, /* Jula */
1.301 + {"dz", HB_TAG('D','Z','N',' ')}, /* Dzongkha */
1.302 + {"ee", HB_TAG('E','W','E',' ')}, /* Ewe */
1.303 + {"efi", HB_TAG('E','F','I',' ')}, /* Efik */
1.304 + {"ekk", HB_TAG('E','T','I',' ')}, /* Standard Estonian */
1.305 + {"el", HB_TAG('E','L','L',' ')}, /* Modern Greek (1453-) */
1.306 + {"emk", HB_TAG('M','N','K',' ')}, /* Eastern Maninkakan */
1.307 + {"en", HB_TAG('E','N','G',' ')}, /* English */
1.308 + {"eo", HB_TAG('N','T','O',' ')}, /* Esperanto */
1.309 + {"eot", HB_TAG('B','T','I',' ')}, /* Beti (Côte d'Ivoire) */
1.310 + {"es", HB_TAG('E','S','P',' ')}, /* Spanish */
1.311 + {"et", HB_TAG('E','T','I',' ')}, /* Estonian [macrolanguage] */
1.312 + {"eu", HB_TAG('E','U','Q',' ')}, /* Basque */
1.313 + {"eve", HB_TAG('E','V','N',' ')}, /* Even */
1.314 + {"evn", HB_TAG('E','V','K',' ')}, /* Evenki */
1.315 + {"fa", HB_TAG('F','A','R',' ')}, /* Persian [macrolanguage] */
1.316 + {"ff", HB_TAG('F','U','L',' ')}, /* Fulah [macrolanguage] */
1.317 + {"fi", HB_TAG('F','I','N',' ')}, /* Finnish */
1.318 + {"fil", HB_TAG('P','I','L',' ')}, /* Filipino */
1.319 + {"fj", HB_TAG('F','J','I',' ')}, /* Fijian */
1.320 + {"fo", HB_TAG('F','O','S',' ')}, /* Faroese */
1.321 + {"fon", HB_TAG('F','O','N',' ')}, /* Fon */
1.322 + {"fr", HB_TAG('F','R','A',' ')}, /* French */
1.323 + {"frc", HB_TAG('F','R','C',' ')}, /* Cajun French */
1.324 + {"frp", HB_TAG('F','R','P',' ')}, /* Arpitan/Francoprovençal */
1.325 + {"fur", HB_TAG('F','R','L',' ')}, /* Friulian */
1.326 + {"fuv", HB_TAG('F','U','V',' ')}, /* Nigerian Fulfulde */
1.327 + {"fy", HB_TAG('F','R','I',' ')}, /* Western Frisian */
1.328 + {"ga", HB_TAG('I','R','I',' ')}, /* Irish */
1.329 + {"gaa", HB_TAG('G','A','D',' ')}, /* Ga */
1.330 + {"gag", HB_TAG('G','A','G',' ')}, /* Gagauz */
1.331 + {"gbm", HB_TAG('G','A','W',' ')}, /* Garhwali */
1.332 + {"gd", HB_TAG('G','A','E',' ')}, /* Scottish Gaelic */
1.333 + {"gez", HB_TAG('G','E','Z',' ')}, /* Ge'ez */
1.334 + {"ggo", HB_TAG('G','O','N',' ')}, /* Southern Gondi */
1.335 + {"gl", HB_TAG('G','A','L',' ')}, /* Galician */
1.336 + {"gld", HB_TAG('N','A','N',' ')}, /* Nanai */
1.337 + {"glk", HB_TAG('G','L','K',' ')}, /* Gilaki */
1.338 + {"gn", HB_TAG('G','U','A',' ')}, /* Guarani [macrolanguage] */
1.339 + {"gno", HB_TAG('G','O','N',' ')}, /* Northern Gondi */
1.340 + {"gog", HB_TAG('G','O','G',' ')}, /* Gogo */
1.341 + {"gon", HB_TAG('G','O','N',' ')}, /* Gondi [macrolanguage] */
1.342 + {"grt", HB_TAG('G','R','O',' ')}, /* Garo */
1.343 + {"gru", HB_TAG('S','O','G',' ')}, /* Sodo Gurage */
1.344 + {"gu", HB_TAG('G','U','J',' ')}, /* Gujarati */
1.345 + {"guc", HB_TAG('G','U','C',' ')}, /* Wayuu */
1.346 + {"guk", HB_TAG('G','M','Z',' ')}, /* Gumuz */
1.347 +/*{"guk", HB_TAG('G','U','K',' ')},*/ /* Gumuz (in SIL fonts) */
1.348 + {"guz", HB_TAG('G','U','Z',' ')}, /* Ekegusii/Gusii */
1.349 + {"gv", HB_TAG('M','N','X',' ')}, /* Manx */
1.350 + {"ha", HB_TAG('H','A','U',' ')}, /* Hausa */
1.351 + {"har", HB_TAG('H','R','I',' ')}, /* Harari */
1.352 + {"haw", HB_TAG('H','A','W',' ')}, /* Hawaiian */
1.353 + {"hay", HB_TAG('H','A','Y',' ')}, /* Haya */
1.354 + {"haz", HB_TAG('H','A','Z',' ')}, /* Hazaragi */
1.355 + {"he", HB_TAG('I','W','R',' ')}, /* Hebrew */
1.356 + {"hz", HB_TAG('H','E','R',' ')}, /* Herero */
1.357 + {"hi", HB_TAG('H','I','N',' ')}, /* Hindi */
1.358 + {"hil", HB_TAG('H','I','L',' ')}, /* Hiligaynon */
1.359 + {"hnd", HB_TAG('H','N','D',' ')}, /* [Southern] Hindko */
1.360 + {"hne", HB_TAG('C','H','H',' ')}, /* Chattisgarhi */
1.361 + {"hno", HB_TAG('H','N','D',' ')}, /* [Northern] Hindko */
1.362 + {"ho", HB_TAG('H','M','O',' ')}, /* Hiri Motu */
1.363 + {"hoc", HB_TAG('H','O',' ',' ')}, /* Ho */
1.364 + {"hoj", HB_TAG('H','A','R',' ')}, /* Harauti */
1.365 + {"hr", HB_TAG('H','R','V',' ')}, /* Croatian */
1.366 + {"hsb", HB_TAG('U','S','B',' ')}, /* Upper Sorbian */
1.367 + {"ht", HB_TAG('H','A','I',' ')}, /* Haitian/Haitian Creole */
1.368 + {"hu", HB_TAG('H','U','N',' ')}, /* Hungarian */
1.369 + {"hy", HB_TAG('H','Y','E',' ')}, /* Armenian */
1.370 + {"hz", HB_TAG('H','E','R',' ')}, /* Herero */
1.371 + {"ia", HB_TAG('I','N','A',' ')}, /* Interlingua (International Auxiliary Language Association) */
1.372 + {"ibb", HB_TAG('I','B','B',' ')}, /* Ibibio */
1.373 + {"id", HB_TAG('I','N','D',' ')}, /* Indonesian */
1.374 + {"ie", HB_TAG('I','L','E',' ')}, /* Interlingue/Occidental */
1.375 + {"ig", HB_TAG('I','B','O',' ')}, /* Igbo */
1.376 + {"igb", HB_TAG('E','B','I',' ')}, /* Ebira */
1.377 + {"ijc", HB_TAG('I','J','O',' ')}, /* Izon */
1.378 + {"ijo", HB_TAG('I','J','O',' ')}, /* Ijo [family] */
1.379 + {"ik", HB_TAG('I','P','K',' ')}, /* Inupiaq [macrolanguage] */
1.380 + {"ilo", HB_TAG('I','L','O',' ')}, /* Ilokano */
1.381 + {"inh", HB_TAG('I','N','G',' ')}, /* Ingush */
1.382 + {"io", HB_TAG('I','D','O',' ')}, /* Ido */
1.383 + {"is", HB_TAG('I','S','L',' ')}, /* Icelandic */
1.384 + {"it", HB_TAG('I','T','A',' ')}, /* Italian */
1.385 + {"iu", HB_TAG('I','N','U',' ')}, /* Inuktitut [macrolanguage] */
1.386 + {"ja", HB_TAG('J','A','N',' ')}, /* Japanese */
1.387 + {"jam", HB_TAG('J','A','M',' ')}, /* Jamaican Creole English */
1.388 + {"jbo", HB_TAG('J','B','O',' ')}, /* Lojban */
1.389 + {"jv", HB_TAG('J','A','V',' ')}, /* Javanese */
1.390 + {"ka", HB_TAG('K','A','T',' ')}, /* Georgian */
1.391 + {"kaa", HB_TAG('K','R','K',' ')}, /* Karakalpak */
1.392 + {"kab", HB_TAG('K','A','B',' ')}, /* Kabyle */
1.393 + {"kam", HB_TAG('K','M','B',' ')}, /* Kamba (Kenya) */
1.394 + {"kar", HB_TAG('K','R','N',' ')}, /* Karen [family] */
1.395 + {"kbd", HB_TAG('K','A','B',' ')}, /* Kabardian */
1.396 + {"kde", HB_TAG('K','D','E',' ')}, /* Makonde */
1.397 + {"kdr", HB_TAG('K','R','M',' ')}, /* Karaim */
1.398 + {"kdt", HB_TAG('K','U','Y',' ')}, /* Kuy */
1.399 + {"kex", HB_TAG('K','K','N',' ')}, /* Kokni */
1.400 + {"kfr", HB_TAG('K','A','C',' ')}, /* Kachchi */
1.401 + {"kfy", HB_TAG('K','M','N',' ')}, /* Kumaoni */
1.402 + {"kg", HB_TAG('K','O','N',' ')}, /* Kongo [macrolanguage] */
1.403 + {"kha", HB_TAG('K','S','I',' ')}, /* Khasi */
1.404 + {"khb", HB_TAG('X','B','D',' ')}, /* Lü */
1.405 + {"kht", HB_TAG('K','H','N',' ')}, /* Khamti (Microsoft fonts) */
1.406 +/*{"kht", HB_TAG('K','H','T',' ')},*/ /* Khamti (OpenType spec and SIL fonts) */
1.407 + {"khw", HB_TAG('K','H','W',' ')}, /* Khowar */
1.408 + {"ki", HB_TAG('K','I','K',' ')}, /* Gikuyu/Kikuyu */
1.409 + {"kj", HB_TAG('K','U','A',' ')}, /* Kuanyama/Kwanyama */
1.410 + {"kjh", HB_TAG('K','H','A',' ')}, /* Khakass */
1.411 + {"kjp", HB_TAG('K','J','P',' ')}, /* Pwo Eastern Karen */
1.412 + {"kk", HB_TAG('K','A','Z',' ')}, /* Kazakh */
1.413 + {"kl", HB_TAG('G','R','N',' ')}, /* Kalaallisut */
1.414 + {"kln", HB_TAG('K','A','L',' ')}, /* Kalenjin */
1.415 + {"km", HB_TAG('K','H','M',' ')}, /* Central Khmer */
1.416 + {"kmb", HB_TAG('M','B','N',' ')}, /* Kimbundu */
1.417 + {"kmw", HB_TAG('K','M','O',' ')}, /* Komo (Democratic Republic of Congo) */
1.418 + {"kn", HB_TAG('K','A','N',' ')}, /* Kannada */
1.419 + {"knn", HB_TAG('K','O','K',' ')}, /* Konkani */
1.420 + {"ko", HB_TAG('K','O','R',' ')}, /* Korean */
1.421 + {"koi", HB_TAG('K','O','P',' ')}, /* Komi-Permyak */
1.422 + {"kok", HB_TAG('K','O','K',' ')}, /* Konkani [macrolanguage] */
1.423 + {"kpe", HB_TAG('K','P','L',' ')}, /* Kpelle [macrolanguage] */
1.424 + {"kpv", HB_TAG('K','O','Z',' ')}, /* Komi-Zyrian */
1.425 + {"kpy", HB_TAG('K','Y','K',' ')}, /* Koryak */
1.426 + {"kqy", HB_TAG('K','R','T',' ')}, /* Koorete */
1.427 + {"kr", HB_TAG('K','N','R',' ')}, /* Kanuri [macrolanguage] */
1.428 + {"kri", HB_TAG('K','R','I',' ')}, /* Krio */
1.429 + {"krl", HB_TAG('K','R','L',' ')}, /* Karelian */
1.430 + {"kru", HB_TAG('K','U','U',' ')}, /* Kurukh */
1.431 + {"ks", HB_TAG('K','S','H',' ')}, /* Kashmiri */
1.432 + {"ksh", HB_TAG('K','S','H',' ')}, /* Kölsch */
1.433 +/*{"ksw", HB_TAG('K','R','N',' ')},*/ /* S'gaw Karen (Microsoft fonts?) */
1.434 + {"ksw", HB_TAG('K','S','W',' ')}, /* S'gaw Karen (OpenType spec and SIL fonts) */
1.435 + {"ku", HB_TAG('K','U','R',' ')}, /* Kurdish [macrolanguage] */
1.436 + {"kum", HB_TAG('K','U','M',' ')}, /* Kumyk */
1.437 + {"kv", HB_TAG('K','O','M',' ')}, /* Komi [macrolanguage] */
1.438 + {"kvd", HB_TAG('K','U','I',' ')}, /* Kui (Indonesia) */
1.439 + {"kw", HB_TAG('C','O','R',' ')}, /* Cornish */
1.440 + {"kxc", HB_TAG('K','M','S',' ')}, /* Komso */
1.441 + {"kxu", HB_TAG('K','U','I',' ')}, /* Kui (India) */
1.442 + {"ky", HB_TAG('K','I','R',' ')}, /* Kirghiz/Kyrgyz */
1.443 + {"kyu", HB_TAG('K','Y','U',' ')}, /* Western Kayah */
1.444 + {"la", HB_TAG('L','A','T',' ')}, /* Latin */
1.445 + {"lad", HB_TAG('J','U','D',' ')}, /* Ladino */
1.446 + {"lb", HB_TAG('L','T','Z',' ')}, /* Luxembourgish */
1.447 + {"lbe", HB_TAG('L','A','K',' ')}, /* Lak */
1.448 + {"lbj", HB_TAG('L','D','K',' ')}, /* Ladakhi */
1.449 + {"lez", HB_TAG('L','E','Z',' ')}, /* Lezgi */
1.450 + {"lg", HB_TAG('L','U','G',' ')}, /* Ganda */
1.451 + {"li", HB_TAG('L','I','M',' ')}, /* Limburgan/Limburger/Limburgish */
1.452 + {"lif", HB_TAG('L','M','B',' ')}, /* Limbu */
1.453 + {"lij", HB_TAG('L','I','J',' ')}, /* Ligurian */
1.454 + {"lis", HB_TAG('L','I','S',' ')}, /* Lisu */
1.455 + {"ljp", HB_TAG('L','J','P',' ')}, /* Lampung Api */
1.456 + {"lki", HB_TAG('L','K','I',' ')}, /* Laki */
1.457 + {"lld", HB_TAG('L','A','D',' ')}, /* Ladin */
1.458 + {"lmn", HB_TAG('L','A','M',' ')}, /* Lambani */
1.459 + {"lmo", HB_TAG('L','M','O',' ')}, /* Lombard */
1.460 + {"ln", HB_TAG('L','I','N',' ')}, /* Lingala */
1.461 + {"lo", HB_TAG('L','A','O',' ')}, /* Lao */
1.462 + {"lrc", HB_TAG('L','R','C',' ')}, /* Northern Luri */
1.463 + {"lt", HB_TAG('L','T','H',' ')}, /* Lithuanian */
1.464 + {"lu", HB_TAG('L','U','B',' ')}, /* Luba-Katanga */
1.465 + {"lua", HB_TAG('L','U','B',' ')}, /* Luba-Kasai */
1.466 + {"luo", HB_TAG('L','U','O',' ')}, /* Luo (Kenya and Tanzania) */
1.467 + {"lus", HB_TAG('M','I','Z',' ')}, /* Mizo */
1.468 + {"luy", HB_TAG('L','U','H',' ')}, /* Luyia/Oluluyia [macrolanguage] */
1.469 + {"luz", HB_TAG('L','R','C',' ')}, /* Southern Luri */
1.470 + {"lv", HB_TAG('L','V','I',' ')}, /* Latvian */
1.471 + {"lzz", HB_TAG('L','A','Z',' ')}, /* Laz */
1.472 + {"mad", HB_TAG('M','A','D',' ')}, /* Madurese */
1.473 + {"mag", HB_TAG('M','A','G',' ')}, /* Magahi */
1.474 + {"mai", HB_TAG('M','T','H',' ')}, /* Maithili */
1.475 + {"mak", HB_TAG('M','K','R',' ')}, /* Makasar */
1.476 + {"man", HB_TAG('M','N','K',' ')}, /* Manding/Mandingo [macrolanguage] */
1.477 + {"mdc", HB_TAG('M','L','E',' ')}, /* Male (Papua New Guinea) */
1.478 + {"mdf", HB_TAG('M','O','K',' ')}, /* Moksha */
1.479 + {"mdr", HB_TAG('M','D','R',' ')}, /* Mandar */
1.480 + {"mdy", HB_TAG('M','L','E',' ')}, /* Male (Ethiopia) */
1.481 + {"men", HB_TAG('M','D','E',' ')}, /* Mende (Sierra Leone) */
1.482 + {"mer", HB_TAG('M','E','R',' ')}, /* Meru */
1.483 + {"mfe", HB_TAG('M','F','E',' ')}, /* Morisyen */
1.484 + {"mg", HB_TAG('M','L','G',' ')}, /* Malagasy [macrolanguage] */
1.485 + {"mh", HB_TAG('M','A','H',' ')}, /* Marshallese */
1.486 + {"mhr", HB_TAG('L','M','A',' ')}, /* Low Mari */
1.487 + {"mi", HB_TAG('M','R','I',' ')}, /* Maori */
1.488 + {"min", HB_TAG('M','I','N',' ')}, /* Minangkabau */
1.489 + {"mk", HB_TAG('M','K','D',' ')}, /* Macedonian */
1.490 + {"mku", HB_TAG('M','N','K',' ')}, /* Konyanka Maninka */
1.491 + {"mkw", HB_TAG('M','K','W',' ')}, /* Kituba (Congo) */
1.492 + {"ml", HB_TAG('M','L','R',' ')}, /* Malayalam */
1.493 + {"mlq", HB_TAG('M','N','K',' ')}, /* Western Maninkakan */
1.494 + {"mn", HB_TAG('M','N','G',' ')}, /* Mongolian [macrolanguage] */
1.495 + {"mnc", HB_TAG('M','C','H',' ')}, /* Manchu */
1.496 + {"mni", HB_TAG('M','N','I',' ')}, /* Manipuri */
1.497 + {"mnk", HB_TAG('M','N','D',' ')}, /* Mandinka */
1.498 + {"mns", HB_TAG('M','A','N',' ')}, /* Mansi */
1.499 + {"mnw", HB_TAG('M','O','N',' ')}, /* Mon */
1.500 + {"mo", HB_TAG('M','O','L',' ')}, /* Moldavian */
1.501 + {"moh", HB_TAG('M','O','H',' ')}, /* Mohawk */
1.502 + {"mos", HB_TAG('M','O','S',' ')}, /* Mossi */
1.503 + {"mpe", HB_TAG('M','A','J',' ')}, /* Majang */
1.504 + {"mr", HB_TAG('M','A','R',' ')}, /* Marathi */
1.505 + {"mrj", HB_TAG('H','M','A',' ')}, /* High Mari */
1.506 + {"ms", HB_TAG('M','L','Y',' ')}, /* Malay [macrolanguage] */
1.507 + {"msc", HB_TAG('M','N','K',' ')}, /* Sankaran Maninka */
1.508 + {"mt", HB_TAG('M','T','S',' ')}, /* Maltese */
1.509 + {"mtr", HB_TAG('M','A','W',' ')}, /* Mewari */
1.510 + {"mus", HB_TAG('M','U','S',' ')}, /* Creek */
1.511 + {"mve", HB_TAG('M','A','W',' ')}, /* Marwari (Pakistan) */
1.512 + {"mwk", HB_TAG('M','N','K',' ')}, /* Kita Maninkakan */
1.513 + {"mwl", HB_TAG('M','W','L',' ')}, /* Mirandese */
1.514 + {"mwr", HB_TAG('M','A','W',' ')}, /* Marwari [macrolanguage] */
1.515 + {"mww", HB_TAG('M','W','W',' ')}, /* Hmong Daw */
1.516 + {"my", HB_TAG('B','R','M',' ')}, /* Burmese */
1.517 + {"mym", HB_TAG('M','E','N',' ')}, /* Me'en */
1.518 + {"myq", HB_TAG('M','N','K',' ')}, /* Forest Maninka (retired code) */
1.519 + {"myv", HB_TAG('E','R','Z',' ')}, /* Erzya */
1.520 + {"mzn", HB_TAG('M','Z','N',' ')}, /* Mazanderani */
1.521 + {"na", HB_TAG('N','A','U',' ')}, /* Nauru */
1.522 + {"nag", HB_TAG('N','A','G',' ')}, /* Naga-Assamese */
1.523 + {"nah", HB_TAG('N','A','H',' ')}, /* Nahuatl [family] */
1.524 + {"nap", HB_TAG('N','A','P',' ')}, /* Neapolitan */
1.525 + {"nb", HB_TAG('N','O','R',' ')}, /* Norwegian Bokmål */
1.526 + {"nco", HB_TAG('S','I','B',' ')}, /* Sibe */
1.527 + {"nd", HB_TAG('N','D','B',' ')}, /* [North] Ndebele */
1.528 + {"ndc", HB_TAG('N','D','C',' ')}, /* Ndau */
1.529 + {"nds", HB_TAG('N','D','S',' ')}, /* Low German/Low Saxon */
1.530 + {"ne", HB_TAG('N','E','P',' ')}, /* Nepali */
1.531 + {"new", HB_TAG('N','E','W',' ')}, /* Newari */
1.532 + {"ng", HB_TAG('N','D','G',' ')}, /* Ndonga */
1.533 + {"nga", HB_TAG('N','G','A',' ')}, /* Ngabaka */
1.534 + {"ngl", HB_TAG('L','M','W',' ')}, /* Lomwe */
1.535 + {"niu", HB_TAG('N','I','U',' ')}, /* Niuean */
1.536 + {"niv", HB_TAG('G','I','L',' ')}, /* Gilyak */
1.537 + {"nl", HB_TAG('N','L','D',' ')}, /* Dutch */
1.538 + {"nn", HB_TAG('N','Y','N',' ')}, /* Norwegian Nynorsk */
1.539 + {"no", HB_TAG('N','O','R',' ')}, /* Norwegian [macrolanguage] */
1.540 + {"nod", HB_TAG('N','T','A',' ')}, /* Northern Thai */
1.541 + {"noe", HB_TAG('N','O','E',' ')}, /* Nimadi */
1.542 + {"nog", HB_TAG('N','O','G',' ')}, /* Nogai */
1.543 + {"nov", HB_TAG('N','O','V',' ')}, /* Novial */
1.544 + {"nqo", HB_TAG('N','K','O',' ')}, /* N'Ko */
1.545 + {"nr", HB_TAG('N','D','B',' ')}, /* [South] Ndebele */
1.546 + {"nsk", HB_TAG('N','A','S',' ')}, /* Naskapi */
1.547 + {"nso", HB_TAG('S','O','T',' ')}, /* [Northern] Sotho */
1.548 + {"ny", HB_TAG('C','H','I',' ')}, /* Chewa/Chichwa/Nyanja */
1.549 + {"nym", HB_TAG('N','Y','M',' ')}, /* Nyamwezi */
1.550 + {"nyn", HB_TAG('N','K','L',' ')}, /* Nyankole */
1.551 + {"oc", HB_TAG('O','C','I',' ')}, /* Occitan (post 1500) */
1.552 + {"oj", HB_TAG('O','J','B',' ')}, /* Ojibwa [macrolanguage] */
1.553 + {"ojs", HB_TAG('O','C','R',' ')}, /* Oji-Cree */
1.554 + {"om", HB_TAG('O','R','O',' ')}, /* Oromo [macrolanguage] */
1.555 + {"or", HB_TAG('O','R','I',' ')}, /* Oriya */
1.556 + {"os", HB_TAG('O','S','S',' ')}, /* Ossetian */
1.557 + {"pa", HB_TAG('P','A','N',' ')}, /* Panjabi */
1.558 + {"pag", HB_TAG('P','A','G',' ')}, /* Pangasinan */
1.559 + {"pam", HB_TAG('P','A','M',' ')}, /* Kapampangan/Pampanga */
1.560 + {"pap", HB_TAG('P','A','P',' ')}, /* Papiamento */
1.561 + {"pcc", HB_TAG('P','C','C',' ')}, /* Bouyei */
1.562 + {"pcd", HB_TAG('P','C','D',' ')}, /* Picard */
1.563 + {"pce", HB_TAG('P','L','G',' ')}, /* [Ruching] Palaung */
1.564 + {"pdc", HB_TAG('P','D','C',' ')}, /* Pennsylvania German */
1.565 + {"pes", HB_TAG('F','A','R',' ')}, /* Iranian Persian */
1.566 + {"phk", HB_TAG('P','H','K',' ')}, /* Phake */
1.567 + {"pi", HB_TAG('P','A','L',' ')}, /* Pali */
1.568 + {"pih", HB_TAG('P','I','H',' ')}, /* Pitcairn-Norfolk */
1.569 + {"pl", HB_TAG('P','L','K',' ')}, /* Polish */
1.570 + {"pll", HB_TAG('P','L','G',' ')}, /* [Shwe] Palaung */
1.571 + {"plp", HB_TAG('P','A','P',' ')}, /* Palpa */
1.572 + {"pms", HB_TAG('P','M','S',' ')}, /* Piemontese */
1.573 + {"pnb", HB_TAG('P','N','B',' ')}, /* Western Panjabi */
1.574 + {"prs", HB_TAG('D','R','I',' ')}, /* Afghan Persian/Dari */
1.575 + {"ps", HB_TAG('P','A','S',' ')}, /* Pashto/Pushto [macrolanguage] */
1.576 + {"pt", HB_TAG('P','T','G',' ')}, /* Portuguese */
1.577 + {"pwo", HB_TAG('P','W','O',' ')}, /* Pwo Western Karen */
1.578 + {"qu", HB_TAG('Q','U','Z',' ')}, /* Quechua [macrolanguage] */
1.579 + {"quc", HB_TAG('Q','U','C',' ')}, /* K'iche'/Quiché */
1.580 + {"quz", HB_TAG('Q','U','Z',' ')}, /* Cusco Quechua */
1.581 + {"raj", HB_TAG('R','A','J',' ')}, /* Rajasthani [macrolanguage] */
1.582 + {"rbb", HB_TAG('P','L','G',' ')}, /* Rumai Palaung */
1.583 + {"rej", HB_TAG('R','E','J',' ')}, /* Rejang */
1.584 + {"ria", HB_TAG('R','I','A',' ')}, /* Riang (India) */
1.585 + {"ril", HB_TAG('R','I','A',' ')}, /* Riang (Myanmar) */
1.586 + {"rki", HB_TAG('A','R','K',' ')}, /* Rakhine */
1.587 + {"rm", HB_TAG('R','M','S',' ')}, /* Romansh */
1.588 + {"rmy", HB_TAG('R','M','Y',' ')}, /* Vlax Romani */
1.589 + {"rn", HB_TAG('R','U','N',' ')}, /* Rundi */
1.590 + {"ro", HB_TAG('R','O','M',' ')}, /* Romanian */
1.591 + {"rom", HB_TAG('R','O','Y',' ')}, /* Romany [macrolanguage] */
1.592 + {"ru", HB_TAG('R','U','S',' ')}, /* Russian */
1.593 + {"rue", HB_TAG('R','S','Y',' ')}, /* Rusyn */
1.594 + {"rup", HB_TAG('R','U','P',' ')}, /* Aromanian/Arumanian/Macedo-Romanian */
1.595 + {"rw", HB_TAG('R','U','A',' ')}, /* Kinyarwanda */
1.596 + {"rwr", HB_TAG('M','A','W',' ')}, /* Marwari (India) */
1.597 + {"sa", HB_TAG('S','A','N',' ')}, /* Sanskrit */
1.598 + {"sah", HB_TAG('Y','A','K',' ')}, /* Yakut */
1.599 + {"sas", HB_TAG('S','A','S',' ')}, /* Sasak */
1.600 + {"sat", HB_TAG('S','A','T',' ')}, /* Santali */
1.601 + {"sck", HB_TAG('S','A','D',' ')}, /* Sadri */
1.602 + {"sc", HB_TAG('S','R','D',' ')}, /* Sardinian [macrolanguage] */
1.603 + {"scn", HB_TAG('S','C','N',' ')}, /* Sicilian */
1.604 + {"sco", HB_TAG('S','C','O',' ')}, /* Scots */
1.605 + {"scs", HB_TAG('S','L','A',' ')}, /* [North] Slavey */
1.606 + {"sd", HB_TAG('S','N','D',' ')}, /* Sindhi */
1.607 + {"se", HB_TAG('N','S','M',' ')}, /* Northern Sami */
1.608 + {"seh", HB_TAG('S','N','A',' ')}, /* Sena */
1.609 + {"sel", HB_TAG('S','E','L',' ')}, /* Selkup */
1.610 + {"sg", HB_TAG('S','G','O',' ')}, /* Sango */
1.611 + {"sga", HB_TAG('S','G','A',' ')}, /* Old Irish (to 900) */
1.612 + {"sgs", HB_TAG('S','G','S',' ')}, /* Samogitian */
1.613 + {"sgw", HB_TAG('C','H','G',' ')}, /* Sebat Bet Gurage */
1.614 +/*{"sgw", HB_TAG('S','G','W',' ')},*/ /* Sebat Bet Gurage (in SIL fonts) */
1.615 + {"shn", HB_TAG('S','H','N',' ')}, /* Shan */
1.616 + {"si", HB_TAG('S','N','H',' ')}, /* Sinhala */
1.617 + {"sid", HB_TAG('S','I','D',' ')}, /* Sidamo */
1.618 + {"sjd", HB_TAG('K','S','M',' ')}, /* Kildin Sami */
1.619 + {"sk", HB_TAG('S','K','Y',' ')}, /* Slovak */
1.620 + {"skr", HB_TAG('S','R','K',' ')}, /* Seraiki */
1.621 + {"sl", HB_TAG('S','L','V',' ')}, /* Slovenian */
1.622 + {"sm", HB_TAG('S','M','O',' ')}, /* Samoan */
1.623 + {"sma", HB_TAG('S','S','M',' ')}, /* Southern Sami */
1.624 + {"smj", HB_TAG('L','S','M',' ')}, /* Lule Sami */
1.625 + {"smn", HB_TAG('I','S','M',' ')}, /* Inari Sami */
1.626 + {"sms", HB_TAG('S','K','S',' ')}, /* Skolt Sami */
1.627 + {"sn", HB_TAG('S','N','A',' ')}, /* Shona */
1.628 + {"snk", HB_TAG('S','N','K',' ')}, /* Soninke */
1.629 + {"so", HB_TAG('S','M','L',' ')}, /* Somali */
1.630 + {"sop", HB_TAG('S','O','P',' ')}, /* Songe */
1.631 + {"sq", HB_TAG('S','Q','I',' ')}, /* Albanian [macrolanguage] */
1.632 + {"sr", HB_TAG('S','R','B',' ')}, /* Serbian */
1.633 + {"srr", HB_TAG('S','R','R',' ')}, /* Serer */
1.634 + {"ss", HB_TAG('S','W','Z',' ')}, /* Swati */
1.635 + {"st", HB_TAG('S','O','T',' ')}, /* [Southern] Sotho */
1.636 + {"stq", HB_TAG('S','T','Q',' ')}, /* Saterfriesisch */
1.637 + {"stv", HB_TAG('S','I','G',' ')}, /* Silt'e */
1.638 + {"su", HB_TAG('S','U','N',' ')}, /* Sundanese */
1.639 + {"suk", HB_TAG('S','U','K',' ')}, /* Sukama */
1.640 + {"suq", HB_TAG('S','U','R',' ')}, /* Suri */
1.641 + {"sv", HB_TAG('S','V','E',' ')}, /* Swedish */
1.642 + {"sva", HB_TAG('S','V','A',' ')}, /* Svan */
1.643 + {"sw", HB_TAG('S','W','K',' ')}, /* Swahili [macrolanguage] */
1.644 + {"swb", HB_TAG('C','M','R',' ')}, /* Comorian */
1.645 + {"swh", HB_TAG('S','W','K',' ')}, /* Kiswahili/Swahili */
1.646 + {"swv", HB_TAG('M','A','W',' ')}, /* Shekhawati */
1.647 + {"sxu", HB_TAG('S','X','U',' ')}, /* Upper Saxon */
1.648 + {"syl", HB_TAG('S','Y','L',' ')}, /* Sylheti */
1.649 + {"syr", HB_TAG('S','Y','R',' ')}, /* Syriac [macrolanguage] */
1.650 + {"szl", HB_TAG('S','Z','L',' ')}, /* Silesian */
1.651 + {"ta", HB_TAG('T','A','M',' ')}, /* Tamil */
1.652 + {"tab", HB_TAG('T','A','B',' ')}, /* Tabasaran */
1.653 + {"tcy", HB_TAG('T','U','L',' ')}, /* Tulu */
1.654 + {"tdd", HB_TAG('T','D','D',' ')}, /* Tai Nüa */
1.655 + {"te", HB_TAG('T','E','L',' ')}, /* Telugu */
1.656 + {"tem", HB_TAG('T','M','N',' ')}, /* Temne */
1.657 + {"tet", HB_TAG('T','E','T',' ')}, /* Tetum */
1.658 + {"tg", HB_TAG('T','A','J',' ')}, /* Tajik */
1.659 + {"th", HB_TAG('T','H','A',' ')}, /* Thai */
1.660 + {"ti", HB_TAG('T','G','Y',' ')}, /* Tigrinya */
1.661 + {"tig", HB_TAG('T','G','R',' ')}, /* Tigre */
1.662 + {"tiv", HB_TAG('T','I','V',' ')}, /* Tiv */
1.663 + {"tk", HB_TAG('T','K','M',' ')}, /* Turkmen */
1.664 + {"tl", HB_TAG('T','G','L',' ')}, /* Tagalog */
1.665 + {"tmh", HB_TAG('t','m','h',' ')}, /* Tamashek [macrolanguage] */
1.666 + {"tn", HB_TAG('T','N','A',' ')}, /* Tswana */
1.667 + {"to", HB_TAG('T','G','N',' ')}, /* Tonga (Tonga Islands) */
1.668 + {"tpi", HB_TAG('T','P','I',' ')}, /* Tok Pisin */
1.669 + {"tr", HB_TAG('T','R','K',' ')}, /* Turkish */
1.670 + {"tru", HB_TAG('T','U','A',' ')}, /* Turoyo Aramaic */
1.671 + {"ts", HB_TAG('T','S','G',' ')}, /* Tsonga */
1.672 + {"tt", HB_TAG('T','A','T',' ')}, /* Tatar */
1.673 + {"tum", HB_TAG('T','U','M',' ')}, /* Tumbuka */
1.674 + {"tw", HB_TAG('T','W','I',' ')}, /* Twi */
1.675 + {"ty", HB_TAG('T','H','T',' ')}, /* Tahitian */
1.676 + {"tyv", HB_TAG('T','U','V',' ')}, /* Tuvin */
1.677 + {"tyz", HB_TAG('T','Y','Z',' ')}, /* Tày */
1.678 + {"tzm", HB_TAG('T','Z','M',' ')}, /* Central Atlas Tamazight */
1.679 + {"udm", HB_TAG('U','D','M',' ')}, /* Udmurt */
1.680 + {"ug", HB_TAG('U','Y','G',' ')}, /* Uighur */
1.681 + {"uk", HB_TAG('U','K','R',' ')}, /* Ukrainian */
1.682 + {"umb", HB_TAG('U','M','B',' ')}, /* Umbundu */
1.683 + {"unr", HB_TAG('M','U','N',' ')}, /* Mundari */
1.684 + {"ur", HB_TAG('U','R','D',' ')}, /* Urdu */
1.685 + {"uz", HB_TAG('U','Z','B',' ')}, /* Uzbek [macrolanguage] */
1.686 + {"uzn", HB_TAG('U','Z','B',' ')}, /* Northern Uzbek */
1.687 + {"uzs", HB_TAG('U','Z','B',' ')}, /* Southern Uzbek */
1.688 + {"ve", HB_TAG('V','E','N',' ')}, /* Venda */
1.689 + {"vec", HB_TAG('V','E','C',' ')}, /* Venetian */
1.690 + {"vls", HB_TAG('F','L','E',' ')}, /* Vlaams */
1.691 + {"vi", HB_TAG('V','I','T',' ')}, /* Vietnamese */
1.692 + {"vmw", HB_TAG('M','A','K',' ')}, /* Makhuwa */
1.693 + {"vo", HB_TAG('V','O','L',' ')}, /* Volapük */
1.694 + {"vro", HB_TAG('V','R','O',' ')}, /* Võro */
1.695 + {"wa", HB_TAG('W','L','N',' ')}, /* Walloon */
1.696 + {"war", HB_TAG('W','A','R',' ')}, /* Waray (Philippines) */
1.697 + {"wbm", HB_TAG('W','A',' ',' ')}, /* Wa */
1.698 + {"wbr", HB_TAG('W','A','G',' ')}, /* Wagdi */
1.699 + {"wle", HB_TAG('S','I','G',' ')}, /* Wolane */
1.700 + {"wry", HB_TAG('M','A','W',' ')}, /* Merwari */
1.701 + {"wtm", HB_TAG('W','T','M',' ')}, /* Mewati */
1.702 + {"wo", HB_TAG('W','L','F',' ')}, /* Wolof */
1.703 + {"xal", HB_TAG('K','L','M',' ')}, /* Kalmyk */
1.704 + {"xh", HB_TAG('X','H','S',' ')}, /* Xhosa */
1.705 + {"xog", HB_TAG('X','O','G',' ')}, /* Soga */
1.706 + {"xom", HB_TAG('K','M','O',' ')}, /* Komo (Sudan) */
1.707 + {"xsl", HB_TAG('S','S','L',' ')}, /* South Slavey */
1.708 + {"xst", HB_TAG('S','I','G',' ')}, /* Silt'e (retired code) */
1.709 + {"xwo", HB_TAG('T','O','D',' ')}, /* Written Oirat (Todo) */
1.710 + {"yao", HB_TAG('Y','A','O',' ')}, /* Yao */
1.711 + {"yi", HB_TAG('J','I','I',' ')}, /* Yiddish [macrolanguage] */
1.712 + {"yo", HB_TAG('Y','B','A',' ')}, /* Yoruba */
1.713 + {"yso", HB_TAG('N','I','S',' ')}, /* Nisi (China) */
1.714 + {"za", HB_TAG('Z','H','A',' ')}, /* Chuang/Zhuang [macrolanguage] */
1.715 + {"zea", HB_TAG('Z','E','A',' ')}, /* Zeeuws */
1.716 + {"zne", HB_TAG('Z','N','D',' ')}, /* Zande */
1.717 + {"zu", HB_TAG('Z','U','L',' ')}, /* Zulu */
1.718 + {"zum", HB_TAG('L','R','C',' ')} /* Kumzari */
1.719 +
1.720 + /* The corresponding languages IDs for the following IDs are unclear,
1.721 + * overlap, or are architecturally weird. Needs more research. */
1.722 +
1.723 +/*{"ahg/awn/xan?", HB_TAG('A','G','W',' ')},*/ /* Agaw */
1.724 +/*{"gsw?/gsw-FR?", HB_TAG('A','L','S',' ')},*/ /* Alsatian */
1.725 +/*{"krc", HB_TAG('B','A','L',' ')},*/ /* Balkar */
1.726 +/*{"??", HB_TAG('B','C','R',' ')},*/ /* Bible Cree */
1.727 +/*{"zh?", HB_TAG('C','H','N',' ')},*/ /* Chinese (seen in Microsoft fonts) */
1.728 +/*{"acf/gcf?", HB_TAG('F','A','N',' ')},*/ /* French Antillean */
1.729 +/*{"enf?/yrk?", HB_TAG('F','N','E',' ')},*/ /* Forest Nenets */
1.730 +/*{"fuf?", HB_TAG('F','T','A',' ')},*/ /* Futa */
1.731 +/*{"ar-Syrc?", HB_TAG('G','A','R',' ')},*/ /* Garshuni */
1.732 +/*{"cfm/rnl?", HB_TAG('H','A','L',' ')},*/ /* Halam */
1.733 +/*{"fonipa", HB_TAG('I','P','P','H')},*/ /* Phonetic transcription—IPA conventions */
1.734 +/*{"ga-Latg?/Latg?", HB_TAG('I','R','T',' ')},*/ /* Irish Traditional */
1.735 +/*{"krc", HB_TAG('K','A','R',' ')},*/ /* Karachay */
1.736 +/*{"alw?/ktb?", HB_TAG('K','E','B',' ')},*/ /* Kebena */
1.737 +/*{"Geok", HB_TAG('K','G','E',' ')},*/ /* Khutsuri Georgian */
1.738 +/*{"kca", HB_TAG('K','H','K',' ')},*/ /* Khanty-Kazim */
1.739 +/*{"kca", HB_TAG('K','H','S',' ')},*/ /* Khanty-Shurishkar */
1.740 +/*{"kca", HB_TAG('K','H','V',' ')},*/ /* Khanty-Vakhi */
1.741 +/*{"guz?/kqs?/kss?", HB_TAG('K','I','S',' ')},*/ /* Kisii */
1.742 +/*{"kfa/kfi?/kpb?/xua?/xuj?", HB_TAG('K','O','D',' ')},*/ /* Kodagu */
1.743 +/*{"okm?/oko?", HB_TAG('K','O','H',' ')},*/ /* Korean Old Hangul */
1.744 +/*{"kon?/ktu?/...", HB_TAG('K','O','N',' ')},*/ /* Kikongo */
1.745 +/*{"kfx?", HB_TAG('K','U','L',' ')},*/ /* Kulvi */
1.746 +/*{"??", HB_TAG('L','A','H',' ')},*/ /* Lahuli */
1.747 +/*{"??", HB_TAG('L','C','R',' ')},*/ /* L-Cree */
1.748 +/*{"??", HB_TAG('M','A','L',' ')},*/ /* Malayalam Traditional */
1.749 +/*{"mnk?/mlq?/...", HB_TAG('M','L','N',' ')},*/ /* Malinke */
1.750 +/*{"??", HB_TAG('N','C','R',' ')},*/ /* N-Cree */
1.751 +/*{"??", HB_TAG('N','H','C',' ')},*/ /* Norway House Cree */
1.752 +/*{"jpa?/sam?", HB_TAG('P','A','A',' ')},*/ /* Palestinian Aramaic */
1.753 +/*{"polyton", HB_TAG('P','G','R',' ')},*/ /* Polytonic Greek */
1.754 +/*{"??", HB_TAG('Q','I','N',' ')},*/ /* Asho Chin */
1.755 +/*{"??", HB_TAG('R','C','R',' ')},*/ /* R-Cree */
1.756 +/*{"chp?", HB_TAG('S','A','Y',' ')},*/ /* Sayisi */
1.757 +/*{"xan?", HB_TAG('S','E','K',' ')},*/ /* Sekota */
1.758 +/*{"ngo?", HB_TAG('S','X','T',' ')},*/ /* Sutu */
1.759 +/*{"??", HB_TAG('T','C','R',' ')},*/ /* TH-Cree */
1.760 +/*{"tnz?/tog?/toi?", HB_TAG('T','N','G',' ')},*/ /* Tonga */
1.761 +/*{"enh?/yrk?", HB_TAG('T','N','E',' ')},*/ /* Tundra Nenets */
1.762 +/*{"??", HB_TAG('W','C','R',' ')},*/ /* West-Cree */
1.763 +/*{"cre?", HB_TAG('Y','C','R',' ')},*/ /* Y-Cree */
1.764 +/*{"??", HB_TAG('Y','I','C',' ')},*/ /* Yi Classic */
1.765 +/*{"ii?/Yiii?", HB_TAG('Y','I','M',' ')},*/ /* Yi Modern */
1.766 +/*{"??", HB_TAG('Z','H','P',' ')},*/ /* Chinese Phonetic */
1.767 +};
1.768 +
1.769 +static const LangTag ot_languages_zh[] = {
1.770 + {"zh-cn", HB_TAG('Z','H','S',' ')}, /* Chinese (China) */
1.771 + {"zh-hk", HB_TAG('Z','H','H',' ')}, /* Chinese (Hong Kong) */
1.772 + {"zh-mo", HB_TAG('Z','H','T',' ')}, /* Chinese (Macao) */
1.773 + {"zh-sg", HB_TAG('Z','H','S',' ')}, /* Chinese (Singapore) */
1.774 + {"zh-tw", HB_TAG('Z','H','T',' ')} /* Chinese (Taiwan) */
1.775 +};
1.776 +
1.777 +static int
1.778 +lang_compare_first_component (const char *a,
1.779 + const char *b)
1.780 +{
1.781 + unsigned int da, db;
1.782 + const char *p;
1.783 +
1.784 + p = strchr (a, '-');
1.785 + da = p ? (unsigned int) (p - a) : strlen (a);
1.786 +
1.787 + p = strchr (b, '-');
1.788 + db = p ? (unsigned int) (p - b) : strlen (b);
1.789 +
1.790 + return strncmp (a, b, MAX (da, db));
1.791 +}
1.792 +
1.793 +static hb_bool_t
1.794 +lang_matches (const char *lang_str, const char *spec)
1.795 +{
1.796 + unsigned int len = strlen (spec);
1.797 +
1.798 + return strncmp (lang_str, spec, len) == 0 &&
1.799 + (lang_str[len] == '\0' || lang_str[len] == '-');
1.800 +}
1.801 +
1.802 +hb_tag_t
1.803 +hb_ot_tag_from_language (hb_language_t language)
1.804 +{
1.805 + const char *lang_str, *s;
1.806 + const LangTag *lang_tag;
1.807 +
1.808 + if (language == HB_LANGUAGE_INVALID)
1.809 + return HB_OT_TAG_DEFAULT_LANGUAGE;
1.810 +
1.811 + lang_str = hb_language_to_string (language);
1.812 +
1.813 + s = strstr (lang_str, "x-hbot");
1.814 + if (s) {
1.815 + char tag[4];
1.816 + int i;
1.817 + s += 6;
1.818 + for (i = 0; i < 4 && ISALPHA (s[i]); i++)
1.819 + tag[i] = TOUPPER (s[i]);
1.820 + if (i) {
1.821 + for (; i < 4; i++)
1.822 + tag[i] = ' ';
1.823 + return HB_TAG_CHAR4 (tag);
1.824 + }
1.825 + }
1.826 +
1.827 + /* Find a language matching in the first component */
1.828 + lang_tag = (LangTag *) bsearch (lang_str, ot_languages,
1.829 + ARRAY_LENGTH (ot_languages), sizeof (LangTag),
1.830 + (hb_compare_func_t) lang_compare_first_component);
1.831 + if (lang_tag)
1.832 + return lang_tag->tag;
1.833 +
1.834 + /* Otherwise, check the Chinese ones */
1.835 + if (0 == lang_compare_first_component (lang_str, "zh"))
1.836 + {
1.837 + unsigned int i;
1.838 +
1.839 + for (i = 0; i < ARRAY_LENGTH (ot_languages_zh); i++)
1.840 + {
1.841 + lang_tag = &ot_languages_zh[i];
1.842 + if (lang_matches (lang_tag->language, lang_str))
1.843 + return lang_tag->tag;
1.844 + }
1.845 +
1.846 + /* Otherwise just return 'ZHS ' */
1.847 + return HB_TAG('Z','H','S',' ');
1.848 + }
1.849 +
1.850 + s = strchr (lang_str, '-');
1.851 + if (!s)
1.852 + s = lang_str + strlen (lang_str);
1.853 + if (s - lang_str == 3) {
1.854 + /* Assume it's ISO-639-3 and upper-case and use it. */
1.855 + return hb_tag_from_string (lang_str, s - lang_str) & ~0x20202000;
1.856 + }
1.857 +
1.858 + return HB_OT_TAG_DEFAULT_LANGUAGE;
1.859 +}
1.860 +
1.861 +hb_language_t
1.862 +hb_ot_tag_to_language (hb_tag_t tag)
1.863 +{
1.864 + unsigned int i;
1.865 +
1.866 + if (tag == HB_OT_TAG_DEFAULT_LANGUAGE)
1.867 + return NULL;
1.868 +
1.869 + for (i = 0; i < ARRAY_LENGTH (ot_languages); i++)
1.870 + if (ot_languages[i].tag == tag)
1.871 + return hb_language_from_string (ot_languages[i].language, -1);
1.872 +
1.873 + /* If tag starts with ZH, it's Chinese */
1.874 + if ((tag & 0xFFFF0000) == 0x5A480000) {
1.875 + switch (tag) {
1.876 + case HB_TAG('Z','H','H',' '): return hb_language_from_string ("zh-hk", -1); /* Hong Kong */
1.877 + default: {
1.878 + /* Encode the tag... */
1.879 + unsigned char buf[14] = "zh-x-hbot";
1.880 + buf[9] = tag >> 24;
1.881 + buf[10] = (tag >> 16) & 0xFF;
1.882 + buf[11] = (tag >> 8) & 0xFF;
1.883 + buf[12] = tag & 0xFF;
1.884 + if (buf[12] == 0x20)
1.885 + buf[12] = '\0';
1.886 + buf[13] = '\0';
1.887 + return hb_language_from_string ((char *) buf, -1);
1.888 + }
1.889 + }
1.890 + }
1.891 +
1.892 + /* Else return a custom language in the form of "x-hbotABCD" */
1.893 + {
1.894 + unsigned char buf[11] = "x-hbot";
1.895 + buf[6] = tag >> 24;
1.896 + buf[7] = (tag >> 16) & 0xFF;
1.897 + buf[8] = (tag >> 8) & 0xFF;
1.898 + buf[9] = tag & 0xFF;
1.899 + if (buf[9] == 0x20)
1.900 + buf[9] = '\0';
1.901 + buf[10] = '\0';
1.902 + return hb_language_from_string ((char *) buf, -1);
1.903 + }
1.904 +}
1.905 +
1.906 +
