Bug 470895 – Improve OpenType language tag list

2008-08-18  Behdad Esfahbod  <behdad@gnome.org>

        Bug 470895 – Improve OpenType language tag list

        * pango/pango-ot-tag.c: Update to OpenType 1.5 draft, intersected with
        ISO639-3 of 2008/08/04.

svn path=/trunk/; revision=2691

2 files changed, 416 insertions, 201 deletions

diff --git a/ChangeLog b/ChangeLog
index ef762f05..0f0193a9 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,10 @@
+2008-08-18  Behdad Esfahbod  <behdad@gnome.org>
+
+	Bug 470895 – Improve OpenType language tag list
+
+	* pango/pango-ot-tag.c: Update to OpenType 1.5 draft, intersected with
+	ISO639-3 of 2008/08/04.
+
 2008-08-12  Sven Herzberg  <sven@imendio.com>
 
 	Bug 547432 – Deprecation of pango_(backend)_font_map_create_context()
diff --git a/pango/pango-ot-tag.c b/pango/pango-ot-tag.c
index c4787d42..c150bd7b 100644
--- a/pango/pango-ot-tag.c
+++ b/pango/pango-ot-tag.c
@@ -181,209 +181,417 @@ typedef struct {
  * complete list at:
  * http://www.microsoft.com/OpenType/OTSpec/languagetags.htm
  *
- * Generated by intersecting the above list with the ISO 639-2 codes
- * and then adjusting manually.  A lot of items missing still, feel
- * free to add.  Keep sorted for bsearch purpose.
+ * Generated by intersecting the OpenType language tag list from
+ * Draft OpenType 1.5 spec, with with the ISO 639-3 codes from
+ * 2008/08/04, matching on name, and finally adjusted manually.
+ *
+ * Many items still missing.  Those are commented out at the end.
+ * Keep sorted for bsearch.
  */
 static const LangTag ot_languages[] = {
-  {"aa",	{"AFR "}},
-  {"ab",	{"ABK "}},
-  {"ady",	{"ADY "}},
-  {"af",	{"AFK "}},
-  {"am",	{"AMH "}},
-  {"ar",	{"ARA "}},
-  {"as",	{"ASM "}},
-  {"awa",	{"AWA "}},
-  {"ay",	{"AYM "}},
-  {"az",	{"AZE "}},
-  {"ba",	{"BSH "}},
-  {"bal",	{"BLI "}},
-  {"bem",	{"BEM "}},
-  {"ber",	{"BBR "}},
-  {"bg",	{"BGR "}},
-  {"bho",	{"BHO "}},
-  {"bik",	{"BIK "}},
-  {"bin",	{"EDO "}},
-  {"bm",	{"BMB "}},
-  {"bn",	{"BEN "}},
-  {"bo",	{"TIB "}},
-  {"br",	{"BRE "}},
-  {"brh",	{"BRH "}},
-  {"ca",	{"CAT "}},
-  {"ce",	{"CHE "}},
-  {"ceb",	{"CEB "}},
-  {"chp",	{"CHP "}},
-  {"chr",	{"CHR "}},
-  {"cop",	{"COP "}},
-  {"cr",	{"CRE "}},
-  {"crh",	{"CRT "}},
-  {"cs",	{"CSY "}},
-  {"cu",	{"CSL "}},
-  {"cv",	{"CHU "}},
-  {"cy",	{"WEL "}},
-  {"da",	{"DAN "}},
-  {"dar",	{"DAR "}},
-  {"de",	{"DEU "}},
-  {"din",	{"DNK "}},
-  {"doi",	{"DGR "}},
-  {"dsb",	{"LSB "}},
-  {"dv",	{"DHV "}},
-  {"dz",	{"DZN "}},
-  {"ee",	{"EWE "}},
-  {"efi",	{"EFI "}},
-  {"el",	{"ELL "}},
-  {"en",	{"ENG "}},
-  {"eo",	{"NTO "}},
-  {"es",	{"ESP "}},
-  {"et",	{"ETI "}},
-  {"eu",	{"EUQ "}},
-  {"fa",	{"FAR "}},
-  {"ff",	{"FUL "}},
-  {"fi",	{"FIN "}},
-  {"fil",	{"PIL "}},
-  {"fj",	{"FJI "}},
-  {"fo",	{"FOS "}},
-  {"fon",	{"FON "}},
-  {"fr",	{"FRA "}},
-  {"fur",	{"FRL "}},
-  {"fy",	{"FRI "}},
-  {"ga",	{"IRI "}},
-  {"gaa",	{"GAD "}},
-  {"gd",	{"GAE "}},
-  {"gl",	{"GAL "}},
-  {"gn",	{"GUA "}},
-  {"gon",	{"GON "}},
-  {"gu",	{"GUJ "}},
-  {"ha",	{"HAU "}},
-  {"he",	{"IWR "}},
-  {"hi",	{"HIN "}},
-  {"hil",	{"HIL "}},
-  {"hr",	{"HRV "}},
-  {"hsb",	{"USB "}},
-  {"ht",	{"HAI "}},
-  {"hu",	{"HUN "}},
-  {"hy",	{"HYE "}},
-  {"id",	{"IND "}},
-  {"ig",	{"IBO "}},
-  {"inc",	{"SRK "}},
-  {"ine",	{"KHW "}},
-  {"inh",	{"ING "}},
-  {"is",	{"ISL "}},
-  {"it",	{"ITA "}},
-  {"iu",	{"INU "}},
-  {"ja",	{"JAN "}},
-  {"jv",	{"JAV "}},
-  {"ka",	{"KAT "}},
-  {"kam",	{"KMB "}},
-  {"kbd",	{"KAB "}},
-  {"kha",	{"KSI "}},
-  {"ki",	{"KIK "}},
-  {"kk",	{"KAZ "}},
-  {"kl",	{"GRN "}},
-  {"km",	{"KHM "}},
-  {"kn",	{"KAN "}},
-  {"ko",	{"KOR "}},
-  {"kok",	{"KOK "}},
-  {"kpe",	{"KPL "}},
-  {"kr",	{"KNR "}},
-  {"krl",	{"KRL "}},
-  {"kru",	{"KUU "}},
-  {"ks",	{"KSH "}},
-  {"ku",	{"KUR "}},
-  {"kum",	{"KUM "}},
-  {"ky",	{"KIR "}},
-  {"la",	{"LAT "}},
-  {"lad",	{"JUD "}},
-  {"lbj",	{"LDK "}},
-  {"ln",	{"LIN "}},
-  {"lo",	{"LAO "}},
-  {"lt",	{"LTH "}},
-  {"lv",	{"LVI "}},
-  {"mai",	{"MTH "}},
-  {"mdf",	{"MOK "}},
-  {"men",	{"MDE "}},
-  {"mg",	{"MLG "}},
-  {"mi",	{"MRI "}},
-  {"mk",	{"MKD "}},
-  {"mkh",	{"KUY "}},
-  {"ml",	{"MLR "}},
-  {"mnc",	{"MCH "}},
-  {"mni",	{"MNI "}},
-  {"mnk",	{"MND "}},
-  {"mo",	{"MOL "}},
-  {"mr",	{"MAR "}},
-  {"ms",	{"MLY "}},
-  {"mt",	{"MTS "}},
-  {"mwr",	{"MAW "}},
-  {"my",	{"BRM "}},
-  {"myv",	{"ERZ "}},
-  {"ne",	{"NEP "}},
-  {"nl",	{"NLD "}},
-  {"no",	{"NOR "}},
-  {"ny",	{"CHI "}},
-  {"oc",	{"PRO "}},
-  {"om",	{"ORO "}},
-  {"or",	{"ORI "}},
-  {"os",	{"OSS "}},
-  {"pa",	{"PAN "}},
-  {"pi",	{"PAL "}},
-  {"pl",	{"PLK "}},
-  {"ps",	{"PAS "}},
-  {"pt",	{"PTG "}},
-  {"ro",	{"ROM "}},
-  {"rom",	{"ROY "}},
-  {"ru",	{"RUS "}},
-  {"sa",	{"SAN "}},
-  {"sat",	{"SAT "}},
-  {"sd",	{"SND "}},
-  {"sel",	{"SEL "}},
-  {"sg",	{"SGO "}},
-  {"shn",	{"SHN "}},
-  {"si",	{"SNH "}},
-  {"sid",	{"SID "}},
-  {"sk",	{"SKY "}},
-  {"sl",	{"SLV "}},
-  {"sm",	{"SMO "}},
-  {"smj",	{"LSM "}},
-  {"smn",	{"ISM "}},
-  {"sms",	{"SKS "}},
-  {"snk",	{"SNK "}},
-  {"so",	{"SML "}},
-  {"sq",	{"SQI "}},
-  {"sr",	{"SRB "}},
-  {"srr",	{"SRR "}},
-  {"sv",	{"SVE "}},
-  {"sw",	{"SWK "}},
-  {"syr",	{"SYR "}},
-  {"ta",	{"TAM "}},
-  {"te",	{"TEL "}},
-  {"tg",	{"TAJ "}},
-  {"th",	{"THA "}},
-  {"ti",	{"TGY "}},
-  {"tig",	{"TGR "}},
-  {"tk",	{"TKM "}},
-  {"tn",	{"TNA "}},
-  {"tr",	{"TRK "}},
-  {"ts",	{"TSG "}},
-  {"tw",	{"TWI "}},
-  {"udm",	{"UDM "}},
-  {"ug",	{"UYG "}},
-  {"uk",	{"UKR "}},
-  {"ur",	{"URD "}},
-  {"uz",	{"UZB "}},
-  {"ve",	{"VEN "}},
-  {"vi",	{"VIT "}},
-  {"wo",	{"WLF "}},
-  {"xal",	{"KLM "}},
-  {"xh",	{"XHS "}},
-  {"yi",	{"JII "}},
-  {"yo",	{"YBA "}},
-  {"zh-cn",	{"ZHS "}},
-  {"zh-hk",	{"ZHH "}},
-  {"zh-mo",	{"ZHT "}},
-  {"zh-sg",	{"ZHS "}},
-  {"zh-tw",	{"ZHT "}},
-  {"zu",	{"ZUL "}}
+  {"aa",	{"AFR "}},	/* Afar */
+  {"ab",	{"ABK "}},	/* Abkhazian */
+  {"abq",	{"ABA "}},	/* Abaza */
+  {"ady",	{"ADY "}},	/* Adyghe */
+  {"af",	{"AFK "}},	/* Afrikaans */
+  {"aiw",	{"ARI "}},	/* Aari */
+  {"am",	{"AMH "}},	/* Amharic */
+  {"ar",	{"ARA "}},	/* Arabic */
+  {"arn",	{"MAP "}},	/* Mapudungun */
+  {"as",	{"ASM "}},	/* Assamese */
+  {"av",	{"AVR "}},	/* Avaric */
+  {"awa",	{"AWA "}},	/* Awadhi */
+  {"ay",	{"AYM "}},	/* Aymara */
+  {"az",	{"AZE "}},	/* Azerbaijani */
+  {"ba",	{"BSH "}},	/* Bashkir */
+  {"bal",	{"BLI "}},	/* Baluchi */
+  {"bcq",	{"BCH "}},	/* Bench */
+  {"bem",	{"BEM "}},	/* Bemba (Zambia) */
+  {"bfq",	{"BAD "}},	/* Badaga */
+  {"bft",	{"BLT "}},	/* Balti */
+  {"bg",	{"BGR "}},	/* Bulgarian */
+  {"bhb",	{"BHI "}},	/* Bhili */
+  {"bho",	{"BHO "}},	/* Bhojpuri */
+  {"bik",	{"BIK "}},	/* Bikol */
+  {"bin",	{"EDO "}},	/* Bini */
+  {"bm",	{"BMB "}},	/* Bambara */
+  {"bn",	{"BEN "}},	/* Bengali */
+  {"bo",	{"TIB "}},	/* Tibetan */
+  {"br",	{"BRE "}},	/* Breton */
+  {"brh",	{"BRH "}},	/* Brahui */
+  {"bs",	{"BOS "}},	/* Bosnian */
+  {"btb",	{"BTI "}},	/* Beti (Cameroon) */
+  {"ca",	{"CAT "}},	/* Catalan */
+  {"ce",	{"CHE "}},	/* Chechen */
+  {"ceb",	{"CEB "}},	/* Cebuano */
+  {"chp",	{"CHP "}},	/* Chipewyan */
+  {"chr",	{"CHR "}},	/* Cherokee */
+  {"cop",	{"COP "}},	/* Coptic */
+  {"cr",	{"CRE "}},	/* Cree */
+  {"crh",	{"CRT "}},	/* Crimean Tatar */
+  {"crm",	{"MCR "}},	/* Moose Cree */
+  {"crx",	{"CRR "}},	/* Carrier */
+  {"cs",	{"CSY "}},	/* Czech */
+  {"cu",	{"CSL "}},	/* Church Slavic */
+  {"cv",	{"CHU "}},	/* Chuvash */
+  {"cwd",	{"DCR "}},	/* Woods Cree */
+  {"cy",	{"WEL "}},	/* Welsh */
+  {"da",	{"DAN "}},	/* Danish */
+  {"dap",	{"NIS "}},	/* Nisi (India) */
+  {"dar",	{"DAR "}},	/* Dargwa */
+  {"de",	{"DEU "}},	/* German */
+  {"din",	{"DNK "}},	/* Dinka */
+  {"dng",	{"DUN "}},	/* Dungan */
+  {"doi",	{"DGR "}},	/* Dogri (macrolanguage) */
+  {"dsb",	{"LSB "}},	/* Lower Sorbian */
+  {"dv",	{"DIV "}},	/* Dhivehi */
+  {"dz",	{"DZN "}},	/* Dzongkha */
+  {"ee",	{"EWE "}},	/* Ewe */
+  {"efi",	{"EFI "}},	/* Efik */
+  {"el",	{"ELL "}},	/* Modern Greek (1453-) */
+  {"en",	{"ENG "}},	/* English */
+  {"eo",	{"NTO "}},	/* Esperanto */
+  {"eot",	{"BTI "}},	/* Beti (Côte d'Ivoire) */
+  {"es",	{"ESP "}},	/* Spanish */
+  {"et",	{"ETI "}},	/* Estonian */
+  {"eu",	{"EUQ "}},	/* Basque */
+  {"eve",	{"EVN "}},	/* Even */
+  {"evn",	{"EVK "}},	/* Evenki */
+  {"fa",	{"FAR "}},	/* Persian */
+  {"ff",	{"FUL "}},	/* Fulah */
+  {"fi",	{"FIN "}},	/* Finnish */
+  {"fil",	{"PIL "}},	/* Filipino */
+  {"fj",	{"FJI "}},	/* Fijian */
+  {"fo",	{"FOS "}},	/* Faroese */
+  {"fon",	{"FON "}},	/* Fon */
+  {"fr",	{"FRA "}},	/* French */
+  {"fur",	{"FRL "}},	/* Friulian */
+  {"fy",	{"FRI "}},	/* Western Frisian */
+  {"ga",	{"IRI "}},	/* Irish */
+  {"gaa",	{"GAD "}},	/* Ga */
+  {"gag",	{"GAG "}},	/* Gagauz */
+  {"gbm",	{"GAW "}},	/* Garhwali */
+  {"gd",	{"GAE "}},	/* Scottish Gaelic */
+  {"gl",	{"GAL "}},	/* Galician */
+  {"gld",	{"NAN "}},	/* Nanai */
+  {"gn",	{"GUA "}},	/* Guarani */
+  {"gon",	{"GON "}},	/* Gondi */
+  {"grt",	{"GRO "}},	/* Garo */
+  {"gu",	{"GUJ "}},	/* Gujarati */
+  {"guk",	{"GMZ "}},	/* Gumuz */
+  {"ha",	{"HAU "}},	/* Hausa */
+  {"har",	{"HRI "}},	/* Harari */
+  {"he",	{"IWR "}},	/* Hebrew */
+  {"hi",	{"HIN "}},	/* Hindi */
+  {"hil",	{"HIL "}},	/* Hiligaynon */
+  {"hoc",	{"HO  "}},	/* Ho */
+  {"hr",	{"HRV "}},	/* Croatian */
+  {"hsb",	{"USB "}},	/* Upper Sorbian */
+  {"ht",	{"HAI "}},	/* Haitian */
+  {"hu",	{"HUN "}},	/* Hungarian */
+  {"hy",	{"HYE "}},	/* Armenian */
+  {"id",	{"IND "}},	/* Indonesian */
+  {"ig",	{"IBO "}},	/* Igbo */
+  {"igb",	{"EBI "}},	/* Ebira */
+  {"inh",	{"ING "}},	/* Ingush */
+  {"is",	{"ISL "}},	/* Icelandic */
+  {"it",	{"ITA "}},	/* Italian */
+  {"iu",	{"INU "}},	/* Inuktitut */
+  {"ja",	{"JAN "}},	/* Japanese */
+  {"jv",	{"JAV "}},	/* Javanese */
+  {"ka",	{"KAT "}},	/* Georgian */
+  {"kam",	{"KMB "}},	/* Kamba (Kenya) */
+  {"kbd",	{"KAB "}},	/* Kabardian */
+  {"kdr",	{"KRM "}},	/* Karaim */
+  {"kdt",	{"KUY "}},	/* Kuy */
+  {"kfr",	{"KAC "}},	/* Kachchi */
+  {"kfy",	{"KMN "}},	/* Kumaoni */
+  {"kha",	{"KSI "}},	/* Khasi */
+  {"khw",	{"KHW "}},	/* Khowar */
+  {"ki",	{"KIK "}},	/* Kikuyu */
+  {"kk",	{"KAZ "}},	/* Kazakh */
+  {"kl",	{"GRN "}},	/* Kalaallisut */
+  {"kln",	{"KAL "}},	/* Kalenjin */
+  {"km",	{"KHM "}},	/* Central Khmer */
+  {"kmw",	{"KMO "}},	/* Komo (Democratic Republic of Congo) */
+  {"kn",	{"KAN "}},	/* Kannada */
+  {"ko",	{"KOR "}},	/* Korean */
+  {"koi",	{"KOP "}},	/* Komi-Permyak */
+  {"kok",	{"KOK "}},	/* Konkani (macrolanguage) */
+  {"kpe",	{"KPL "}},	/* Kpelle */
+  {"kpv",	{"KOZ "}},	/* Komi-Zyrian */
+  {"kpy",	{"KYK "}},	/* Koryak */
+  {"kqy",	{"KRT "}},	/* Koorete */
+  {"kr",	{"KNR "}},	/* Kanuri */
+  {"kri",	{"KRI "}},	/* Krio */
+  {"krl",	{"KRL "}},	/* Karelian */
+  {"kru",	{"KUU "}},	/* Kurukh */
+  {"ks",	{"KSH "}},	/* Kashmiri */
+  {"ku",	{"KUR "}},	/* Kurdish */
+  {"kum",	{"KUM "}},	/* Kumyk */
+  {"kvd",	{"KUI "}},	/* Kui (Indonesia) */
+  {"kxu",	{"KUI "}},	/* Kui (India) */
+  {"ky",	{"KIR "}},	/* Kirghiz */
+  {"la",	{"LAT "}},	/* Latin */
+  {"lad",	{"JUD "}},	/* Ladino */
+  {"lb",	{"LTZ "}},	/* Luxembourgish */
+  {"lbe",	{"LAK "}},	/* Lak */
+  {"lbj",	{"LDK "}},	/* Ladakhi */
+  {"lif",	{"LMB "}},	/* Limbu */
+  {"lld",	{"LAD "}},	/* Ladin */
+  {"ln",	{"LIN "}},	/* Lingala */
+  {"lo",	{"LAO "}},	/* Lao */
+  {"lt",	{"LTH "}},	/* Lithuanian */
+  {"luo",	{"LUO "}},	/* Luo (Kenya and Tanzania) */
+  {"luw",	{"LUO "}},	/* Luo (Cameroon) */
+  {"lv",	{"LVI "}},	/* Latvian */
+  {"lzz",	{"LAZ "}},	/* Laz */
+  {"mai",	{"MTH "}},	/* Maithili */
+  {"mdc",	{"MLE "}},	/* Male (Papua New Guinea) */
+  {"mdf",	{"MOK "}},	/* Moksha */
+  {"mdy",	{"MLE "}},	/* Male (Ethiopia) */
+  {"men",	{"MDE "}},	/* Mende (Sierra Leone) */
+  {"mg",	{"MLG "}},	/* Malagasy */
+  {"mi",	{"MRI "}},	/* Maori */
+  {"mk",	{"MKD "}},	/* Macedonian */
+  {"ml",	{"MLR "}},	/* Malayalam */
+  {"mn",	{"MNG "}},	/* Mongolian */
+  {"mnc",	{"MCH "}},	/* Manchu */
+  {"mni",	{"MNI "}},	/* Manipuri */
+  {"mnk",	{"MND "}},	/* Mandinka */
+  {"mns",	{"MAN "}},	/* Mansi */
+  {"mnw",	{"MON "}},	/* Mon */
+  {"mo",	{"MOL "}},	/* Moldavian */
+  {"moh",	{"MOH "}},	/* Mohawk */
+  {"mpe",	{"MAJ "}},	/* Majang */
+  {"mr",	{"MAR "}},	/* Marathi */
+  {"ms",	{"MLY "}},	/* Malay (macrolanguage) */
+  {"mt",	{"MTS "}},	/* Maltese */
+  {"mwr",	{"MAW "}},	/* Marwari */
+  {"my",	{"BRM "}},	/* Burmese */
+  {"mym",	{"MEN "}},	/* Me'en */
+  {"myv",	{"ERZ "}},	/* Erzya */
+  {"nco",	{"SIB "}},	/* Sibe */
+  {"ne",	{"NEP "}},	/* Nepali */
+  {"new",	{"NEW "}},	/* Newari */
+  {"ng",	{"NDG "}},	/* Ndonga */
+  {"ngl",	{"LMW "}},	/* Lomwe */
+  {"niu",	{"NIU "}},	/* Niuean */
+  {"niv",	{"GIL "}},	/* Gilyak */
+  {"nl",	{"NLD "}},	/* Dutch */
+  {"nn",	{"NYN "}},	/* Norwegian Nynorsk */
+  {"no",	{"NOR "}},	/* Norwegian */
+  {"nog",	{"NOG "}},	/* Nogai */
+  {"nqo",	{"NKO "}},	/* N'Ko */
+  {"nsk",	{"NAS "}},	/* Naskapi */
+  {"ny",	{"CHI "}},	/* Nyanja */
+  {"oc",	{"OCI "}},	/* Occitan (post 1500) */
+  {"oc",	{"PRO "}},	/* Occitan (post 1500) */
+  {"oj",	{"OJB "}},	/* Ojibwa */
+  {"om",	{"ORO "}},	/* Oromo */
+  {"or",	{"ORI "}},	/* Oriya */
+  {"os",	{"OSS "}},	/* Ossetian */
+  {"pa",	{"PAN "}},	/* Panjabi */
+  {"pi",	{"PAL "}},	/* Pali */
+  {"pl",	{"PLK "}},	/* Polish */
+  {"plp",	{"PAP "}},	/* Palpa */
+  {"prs",	{"DRI "}},	/* Dari */
+  {"ps",	{"PAS "}},	/* Pushto */
+  {"pt",	{"PTG "}},	/* Portuguese */
+  {"raj",	{"RAJ "}},	/* Rajasthani */
+  {"ria",	{"RIA "}},	/* Riang (India) */
+  {"ril",	{"RIA "}},	/* Riang (Myanmar) */
+  {"ro",	{"ROM "}},	/* Romanian */
+  {"rom",	{"ROY "}},	/* Romany */
+  {"ru",	{"RUS "}},	/* Russian */
+  {"rue",	{"RSY "}},	/* Rusyn */
+  {"sa",	{"SAN "}},	/* Sanskrit */
+  {"sah",	{"YAK "}},	/* Yakut */
+  {"sat",	{"SAT "}},	/* Santali */
+  {"sck",	{"SAD "}},	/* Sadri */
+  {"sd",	{"SND "}},	/* Sindhi */
+  {"se",	{"NSM "}},	/* Northern Sami */
+  {"seh",	{"SNA "}},	/* Sena */
+  {"sel",	{"SEL "}},	/* Selkup */
+  {"sg",	{"SGO "}},	/* Sango */
+  {"shn",	{"SHN "}},	/* Shan */
+  {"si",	{"SNH "}},	/* Sinhala */
+  {"sid",	{"SID "}},	/* Sidamo */
+  {"sjd",	{"KSM "}},	/* Kildin Sami */
+  {"sk",	{"SKY "}},	/* Slovak */
+  {"skr",	{"SRK "}},	/* Seraiki */
+  {"sl",	{"SLV "}},	/* Slovenian */
+  {"sm",	{"SMO "}},	/* Samoan */
+  {"sma",	{"SSM "}},	/* Southern Sami */
+  {"smj",	{"LSM "}},	/* Lule Sami */
+  {"smn",	{"ISM "}},	/* Inari Sami */
+  {"sms",	{"SKS "}},	/* Skolt Sami */
+  {"snk",	{"SNK "}},	/* Soninke */
+  {"so",	{"SML "}},	/* Somali */
+  {"sq",	{"SQI "}},	/* Albanian */
+  {"sr",	{"SRB "}},	/* Serbian */
+  {"srr",	{"SRR "}},	/* Serer */
+  {"suq",	{"SUR "}},	/* Suri */
+  {"sv",	{"SVE "}},	/* Swedish */
+  {"sva",	{"SVA "}},	/* Svan */
+  {"sw",	{"SWK "}},	/* Swahili (macrolanguage) */
+  {"swb",	{"CMR "}},	/* Comorian */
+  {"syr",	{"SYR "}},	/* Syriac */
+  {"ta",	{"TAM "}},	/* Tamil */
+  {"tcy",	{"TUL "}},	/* Tulu */
+  {"te",	{"TEL "}},	/* Telugu */
+  {"tg",	{"TAJ "}},	/* Tajik */
+  {"th",	{"THA "}},	/* Thai */
+  {"ti",	{"TGY "}},	/* Tigrinya */
+  {"tig",	{"TGR "}},	/* Tigre */
+  {"tk",	{"TKM "}},	/* Turkmen */
+  {"tn",	{"TNA "}},	/* Tswana */
+  {"tnz",	{"TNG "}},	/* Tonga (Thailand) */
+  {"to",	{"TNG "}},	/* Tonga (Tonga Islands) */
+  {"tog",	{"TNG "}},	/* Tonga (Nyasa) */
+  {"toi",	{"TNG "}},	/* Tonga (Zambia) */
+  {"tr",	{"TRK "}},	/* Turkish */
+  {"ts",	{"TSG "}},	/* Tsonga */
+  {"tt",	{"TAT "}},	/* Tatar */
+  {"tw",	{"TWI "}},	/* Twi */
+  {"ty",	{"THT "}},	/* Tahitian */
+  {"udm",	{"UDM "}},	/* Udmurt */
+  {"ug",	{"UYG "}},	/* Uighur */
+  {"uk",	{"UKR "}},	/* Ukrainian */
+  {"unr",	{"MUN "}},	/* Mundari */
+  {"ur",	{"URD "}},	/* Urdu */
+  {"uz",	{"UZB "}},	/* Uzbek */
+  {"ve",	{"VEN "}},	/* Venda */
+  {"vi",	{"VIT "}},	/* Vietnamese */
+  {"wbm",	{"WA  "}},	/* Wa */
+  {"wbr",	{"WAG "}},	/* Wagdi */
+  {"wo",	{"WLF "}},	/* Wolof */
+  {"xal",	{"KLM "}},	/* Kalmyk */
+  {"xh",	{"XHS "}},	/* Xhosa */
+  {"xom",	{"KMO "}},	/* Komo (Sudan) */
+  {"xsl",	{"SSL "}},	/* South Slavey */
+  {"yi",	{"JII "}},	/* Yiddish */
+  {"yo",	{"YBA "}},	/* Yoruba */
+  {"yso",	{"NIS "}},	/* Nisi (China) */
+  {"zh-cn",	{"ZHS "}},	/* Chinese (China) */
+  {"zh-hk",	{"ZHH "}},	/* Chinese (Hong Kong) */
+  {"zh-mo",	{"ZHT "}},	/* Chinese (Macao) */
+  {"zh-sg",	{"ZHS "}},	/* Chinese (Singapore) */
+  {"zh-tw",	{"ZHT "}},	/* Chinese (Taiwan) */
+  {"zne",	{"ZND "}},	/* Zande */
+  {"zu",	{"ZUL "}} 	/* Zulu */
+
+  /* I couldn't find the language id for these */
+
+/*{"??",	{"AGW "}},*/	/* Agaw */
+/*{"??",	{"ALS "}},*/	/* Alsatian */
+/*{"??",	{"ALT "}},*/	/* Altai */
+/*{"??",	{"ARK "}},*/	/* Arakanese */
+/*{"??",	{"ATH "}},*/	/* Athapaskan */
+/*{"??",	{"BAG "}},*/	/* Baghelkhandi */
+/*{"??",	{"BAL "}},*/	/* Balkar */
+/*{"??",	{"BAU "}},*/	/* Baule */
+/*{"??",	{"BBR "}},*/	/* Berber */
+/*{"??",	{"BCR "}},*/	/* Bible Cree */
+/*{"??",	{"BEL "}},*/	/* Belarussian */
+/*{"??",	{"BIL "}},*/	/* Bilen */
+/*{"??",	{"BKF "}},*/	/* Blackfoot */
+/*{"??",	{"BLN "}},*/	/* Balante */
+/*{"??",	{"BML "}},*/	/* Bamileke */
+/*{"??",	{"BRI "}},*/	/* Braj Bhasha */
+/*{"??",	{"CHG "}},*/	/* Chaha Gurage */
+/*{"??",	{"CHH "}},*/	/* Chattisgarhi */
+/*{"??",	{"CHK "}},*/	/* Chukchi */
+/*{"??",	{"DJR "}},*/	/* Djerma */
+/*{"??",	{"DNG "}},*/	/* Dangme */
+/*{"??",	{"ECR "}},*/	/* Eastern Cree */
+/*{"??",	{"FAN "}},*/	/* French Antillean */
+/*{"??",	{"FLE "}},*/	/* Flemish */
+/*{"??",	{"FNE "}},*/	/* Forest Nenets */
+/*{"??",	{"FTA "}},*/	/* Futa */
+/*{"??",	{"GAR "}},*/	/* Garshuni */
+/*{"??",	{"GEZ "}},*/	/* Ge'ez */
+/*{"??",	{"HAL "}},*/	/* Halam */
+/*{"??",	{"HAR "}},*/	/* Harauti */
+/*{"??",	{"HAW "}},*/	/* Hawaiin */
+/*{"??",	{"HBN "}},*/	/* Hammer-Banna */
+/*{"??",	{"HMA "}},*/	/* High Mari */
+/*{"??",	{"HND "}},*/	/* Hindko */
+/*{"??",	{"IJO "}},*/	/* Ijo */
+/*{"??",	{"ILO "}},*/	/* Ilokano */
+/*{"??",	{"IRT "}},*/	/* Irish Traditional */
+/*{"??",	{"JUL "}},*/	/* Jula */
+/*{"??",	{"KAR "}},*/	/* Karachay */
+/*{"??",	{"KEB "}},*/	/* Kebena */
+/*{"??",	{"KGE "}},*/	/* Khutsuri Georgian */
+/*{"??",	{"KHA "}},*/	/* Khakass */
+/*{"??",	{"KHK "}},*/	/* Khanty-Kazim */
+/*{"??",	{"KHS "}},*/	/* Khanty-Shurishkar */
+/*{"??",	{"KHV "}},*/	/* Khanty-Vakhi */
+/*{"??",	{"KIS "}},*/	/* Kisii */
+/*{"??",	{"KKN "}},*/	/* Kokni */
+/*{"??",	{"KMS "}},*/	/* Komso */
+/*{"??",	{"KOD "}},*/	/* Kodagu */
+/*{"??",	{"KOH "}},*/	/* Korean Old Hangul */
+/*{"??",	{"KON "}},*/	/* Kikongo */
+/*{"??",	{"KRK "}},*/	/* Karakalpak */
+/*{"??",	{"KRN "}},*/	/* Karen */
+/*{"??",	{"KUL "}},*/	/* Kulvi */
+/*{"??",	{"LAH "}},*/	/* Lahuli */
+/*{"??",	{"LAM "}},*/	/* Lambani */
+/*{"??",	{"LCR "}},*/	/* L-Cree */
+/*{"??",	{"LEZ "}},*/	/* Lezgi */
+/*{"??",	{"LMA "}},*/	/* Low Mari */
+/*{"??",	{"LUB "}},*/	/* Luba */
+/*{"??",	{"LUG "}},*/	/* Luganda */
+/*{"??",	{"LUH "}},*/	/* Luhya */
+/*{"??",	{"MAK "}},*/	/* Makua */
+/*{"??",	{"MAL "}},*/	/* Malayalam Traditional */
+/*{"??",	{"MBN "}},*/	/* Mbundu */
+/*{"??",	{"MIZ "}},*/	/* Mizo */
+/*{"??",	{"MLN "}},*/	/* Malinke */
+/*{"??",	{"MNK "}},*/	/* Maninka */
+/*{"??",	{"MNX "}},*/	/* Manx Gaelic */
+/*{"??",	{"MOR "}},*/	/* Moroccan */
+/*{"??",	{"NAG "}},*/	/* Naga-Assamese */
+/*{"??",	{"NCR "}},*/	/* N-Cree */
+/*{"??",	{"NDB "}},*/	/* Ndebele */
+/*{"??",	{"NGR "}},*/	/* Nagari */
+/*{"??",	{"NHC "}},*/	/* Norway House Cree */
+/*{"??",	{"NKL "}},*/	/* Nkole */
+/*{"??",	{"NTA "}},*/	/* Northern Tai */
+/*{"??",	{"OCR "}},*/	/* Oji-Cree */
+/*{"??",	{"PAA "}},*/	/* Palestinian Aramaic */
+/*{"??",	{"PGR "}},*/	/* Polytonic Greek */
+/*{"??",	{"PLG "}},*/	/* Palaung */
+/*{"??",	{"QIN "}},*/	/* Chin */
+/*{"??",	{"RBU "}},*/	/* Russian Buriat */
+/*{"??",	{"RCR "}},*/	/* R-Cree */
+/*{"??",	{"RMS "}},*/	/* Rhaeto-Romanic */
+/*{"??",	{"RUA "}},*/	/* Ruanda */
+/*{"??",	{"SAY "}},*/	/* Sayisi */
+/*{"??",	{"SEK "}},*/	/* Sekota */
+/*{"??",	{"SIG "}},*/	/* Silte Gurage */
+/*{"??",	{"SLA "}},*/	/* Slavey */
+/*{"??",	{"SOG "}},*/	/* Sodo Gurage */
+/*{"??",	{"SOT "}},*/	/* Sotho */
+/*{"??",	{"SWA "}},*/	/* Swadaya Aramaic */
+/*{"??",	{"SWZ "}},*/	/* Swazi */
+/*{"??",	{"SXT "}},*/	/* Sutu */
+/*{"??",	{"TAB "}},*/	/* Tabasaran */
+/*{"??",	{"TCR "}},*/	/* TH-Cree */
+/*{"??",	{"TGN "}},*/	/* Tongan */
+/*{"??",	{"TMN "}},*/	/* Temne */
+/*{"??",	{"TNE "}},*/	/* Tundra Nenets */
+/*{"??",	{"TOD "}},*/	/* Todo */
+/*{"??",	{"TUA "}},*/	/* Turoyo Aramaic */
+/*{"??",	{"TUV "}},*/	/* Tuvin */
+/*{"??",	{"WCR "}},*/	/* West-Cree */
+/*{"??",	{"XBD "}},*/	/* Tai Lue */
+/*{"??",	{"YCR "}},*/	/* Y-Cree */
+/*{"??",	{"YIC "}},*/	/* Yi Classic */
+/*{"??",	{"YIM "}},*/	/* Yi Modern */
+/*{"??",	{"ZHP "}},*/	/* Chinese Phonetic */
 };
 
 static int