Merge pull request #55 from amire80/langdb_aliases_cleanup

Add basic support for redirects

data/langdb.yaml

@@ -2,11 +2,10 @@ languages:
   aa: [Latn, [AF], Qafár af]
   ab: [Cyrl, [EU], Аҧсшәа]
   ace: [Latn, [AS, PA], Acèh]
-  ady-cyrl: [Cyrl, [EU], Адыгэбзэ]
+ # FIXME: alias
-  ady-latn: [Latn, [EU], Adygabze]
+  ady: [Cyrl, [EU, ME], Адыгэбзэ]
- # XXX multiple script
+  ady-cyrl: [Cyrl, [EU, ME], Адыгэбзэ]
- # ady: [[Cyrl, Latn], [EU], [Адыгэбзэ, Adygabze]]
+  ady-latn: [Latn, [EU, ME], Adygabze]
-  ady: [Cyrl, [EU], Адыгэбзэ]
   aeb: [Arab, [AF], زَوُن]
   af: [Latn, [AF], Afrikaans]
   ahr: [Deva, [AS], अहिराणी]
@@ -30,8 +29,12 @@ languages:
   av: [Cyrl, [EU], авар]
   avk: [Latn, [WW], Kotava]
   ay: [Latn, [AM], Aymar aru]
- # also Arab, and in the past - Cyrl
+ # Latin is the default in Azerbaijani Wikipedia
+ # FIXME: alias
   az: [Latn, [EU, ME], azərbaycanca]
+  az-latn: [Latn, [EU, ME], azərbaycanca]
+  az-arab: [Arab, [EU, ME], آذربايجانجا]
+  az-cyrl: [Latn, [EU, ME], азәрбајҹанҹа]
   ba: [Cyrl, [EU], башҡортса]
   bar: [Latn, [EU], Boarisch]
   bbc-latn: [Latn, [AS], Batak Toba]
@@ -39,13 +42,14 @@ languages:
   bbc: [Batk, [AS], Batak Toba/Batak autonym unknown]
   bcc: [Arab, [AS, ME], بلوچی مکرانی]
   bcl: [Latn, [AS], Bikol Central]
-# FIXME: which is the correct code?
+ # FIXME: alias
   be-tarask: [Cyrl, [EU], беларуская (тарашкевіца)]
   be-x-old: [Cyrl, [EU], беларуская (тарашкевіца)]
   be: [Cyrl, [EU], беларуская]
   bew: [Latn, [AS], Bahasa Betawi]
   bfq: [Taml, [AS], படகா]
   bg: [Cyrl, [EU], български]
+ # FIXME - currently says Bhojpuri, but it's a macrolanguage.
   bh: [Deva, [AS], भोजपुरी]
   bho: [Deva, [AS], भोजपुरी]
   bi: [Latn, [PA], Bislama]
@@ -73,17 +77,14 @@ languages:
   ckb: [Arab, [ME], کوردی]
   co: [Latn, [EU], corsu]
   cps: [Latn, [AS], Capiceño]
+ # FIXME: alias
+  cr: [Cans, [AM], ᓀᐦᐃᔭᐍᐏᐣ]
   cr-cans: [Cans, [AM], ᓀᐦᐃᔭᐍᐏᐣ]
   cr-latn: [Latn, [AM], Nēhiyawēwin]
- # XXX multiple script
+ # FIXME: alias
- # cr: [[Latn, Cans], [AM], [Nēhiyawēwin, ᓀᐦᐃᔭᐍᐏᐣ]]
+  crh: [Latn, [EU], qırımtatarca]
-  cr: [Cans, [AM], Nēhiyawēwin / ᓀᐦᐃᔭᐍᐏᐣ]
+  crh-cyrl: [Cyrl, [EU], къырымтатарджа]
-  crh-cyrl: [Cyrl, [EU], къырымтатарджа (Кирилл)]
+  crh-latn: [Latn, [EU], qırımtatarca]
-  crh-latn: [Latn, [EU], qırımtatarca (Latin)]
- # XXX multiple script
- # Latn is default in Wikipedia, Cyrl is common IRL
- # crh: [[Cyrl, Latn], [EU], [къырымтатарджа, qırımtatarca]]
-  crh: [Latn, [EU], къырымтатарджа / qırımtatarca]
   cs: [Latn, [EU], česky]
   csb: [Latn, [EU], kaszëbsczi]
  # FIXME: what script?
@@ -106,7 +107,6 @@ languages:
   eml: [Latn, [EU], emiliàn e rumagnòl]
   en-ca: [Latn, [AM], Canadian English]
   en-gb: [Latn, [EU, AS, PA], British English]
- # world?
   en: [Latn, [EU, AM, AF, ME, AS, PA, WW], English]
   eo: [Latn, [WW], Esperanto]
   es-419: [Latn, [AM], español de America Latina]
@@ -133,6 +133,7 @@ languages:
   ga: [Latn, [EU], Gaeilge]
   gag: [Latn, [EU], Gagauz]
   gah: [Latn, [AS], Alekano]
+ # FIXME: alias
   gan-hans: [Hans, [AS], 赣语（简体）]
   gan-hant: [Hant, [AS], 贛語（繁體）]
   gan: [Hant, [AS], 贛語]
@@ -142,12 +143,10 @@ languages:
   gl: [Latn, [EU], galego]
   glk: [Arab, [ME], گیلکی]
   gn: [Latn, [AM], "Avañe'ẽ"]
+ # FIXME: alias
+  gom: [Deva, [AS], कोंकणी]
   gom-deva: [Deva, [AS], कोंकणी]
   gom-latn: [Latn, [AS], Konknni]
- # XXX multiple script
- # gom: [[Deva, Latn], [AS], [कोंकणी, Konknni]]
-  gom: [Deva, [AS], कोंकणी / Konknni]
- # hmph?..
   got: [Goth, [EU], 𐌲𐌿𐍄𐌹𐍃𐌺]
   grc: [Grek, [EU], Ἀρχαία ἑλληνικὴ]
   gsw: [Latn, [EU], Alemannisch]
@@ -155,19 +154,18 @@ languages:
   guc: [Latn, [AM], Wayúu]
   gur: [Latn, [AF], Gurenɛ]
   gv: [Latn, [EU], Gaelg]
+ # FIXME: alias
   ha-arab: [Arab, [AF], هَوُسَ]
   ha-latn: [Latn, [AF], Hausa]
   ha: [Latn, [AF], Hausa]
   hak: [Latn, [AS], Hak-kâ-fa]
   haw: [Latn, [AM, PA], Hawai`i]
   he: [Hebr, [ME], עברית]
- # Or maybe world?
   hi: [Deva, [AS], हिन्दी]
+ # FIXME: alias
+  hif: [Latn, [PA, AS], Fiji Hindi]
   hif-deva: [Deva, [AS], फ़ीजी हिन्दी]
   hif-latn: [Latn, [PA, AS], Fiji Hindi]
- # XXX multiple script
- # hif: [[Deva, Latn], [PA, AS], [फ़ीजी हिन्दी, Fiji Hindi]]
-  hif: [Latn, [PA, AS], फ़ीजी हिन्दी / Fiji Hindi]
   hil: [Latn, [AS], Ilonggo]
   hne: [Deva, [AS], छत्तीसगढ़ी]
   ho: [Latn, [PA], Hiri Motu]
@@ -194,13 +192,14 @@ languages:
   is: [Latn, [EU], íslenska]
   it: [Latn, [EU], italiano]
  # For variants ike-* is used
-  iu: [Cans, [AM], ᐃᓄᒃᑎᑐᑦ/inuktitut]
+  iu: [Cans, [AM], ᐃᓄᒃᑎᑐᑦ]
   ja: [Jpan, [AS], 日本語]
   jam: [Latn, [AM], Patois]
   jbo: [Latn, [WW], Lojban]
   jut: [Latn, [EU], jysk]
- # also in the Javanese script (Java), but the Wikipedia is in Latn
   jv: [Latn, [AS, PA], Basa Jawa]
+ # For support in webfonts.
+  jv-java: [Java, [AS, PA], ꦧꦱꦗꦮ]
   ka: [Geor, [EU], ქართული]
   kaa: [Latn, [AS], Qaraqalpaqsha]
  # Can also be Tfng, but the Wikipedia is mostly Latn
@@ -215,15 +214,14 @@ languages:
   ki: [Latn, [AF], Gĩkũyũ]
   kiu: [Latn, [EU, ME], Kırmancki]
   kj: [Latn, [AF], Kwanyama]
+ # FIXME: alias
+  kk: [Cyrl, [EU, AS], қазақша]
   kk-arab: [Arab, [EU, AS], قازاقشا (تٴوتە)]
   kk-cn: [Arab, [EU, AS, ME], قازاقشا (جۇنگو)]
-  kk-cyrl: [Cyrl, [EU, AS], қазақша (кирил)]
+  kk-cyrl: [Cyrl, [EU, AS], қазақша]
   kk-kz: [Cyrl, [EU, AS], қазақша (Қазақстан)]
-  kk-latn: [Latn, [EU, AS, ME], qazaqşa (latın)]
+  kk-latn: [Latn, [EU, AS, ME], qazaqşa]
   kk-tr: [Latn, [EU, AS, ME], qazaqşa (Türkïya)]
- # XXX multiple script
- # kk: [[Arab, Cyrl, Latn] [EU, AS], [قازاقشا, қазақша, qazaqşa]]
-  kk: [Cyrl, [EU, AS], қазақша / قازاقشا / qazaqşa]
   kl: [Latn, [AM, EU], kalaallisut]
   km: [Khmr, [AS], ភាសាខ្មែរ]
   kn: [Knda, [AS], ಕನ್ನಡ]
@@ -239,25 +237,22 @@ languages:
   krl: [Latn, [EU], Karjala]
   ks-arab: [Arab, [AS], کٲشُر]
   ks-deva: [Deva, [AS], कॉशुर]
- # XXX multiple script
+ # FIXME: alias
- # Arab is first here just because it's the current default in the Wikipedia. Deva may be needed, too.
+  ks: [Arab, [AS], کٲشُر]
- # ks: [[Deva, Arab], [AS], [कॉशुर, کٲشُر]]
-  ks: [Arab, [AS], कॉशुर / کٲشُر]
   ksf: [Latn, [AF], Bafia]
   ksh: [Latn, [EU], Ripoarisch]
-  ku-arab: [Arab, [EU, ME], كوردي (عەرەبی)]
+ # FIXME: alias
-  ku-latn: [Latn, [EU, ME], Kurdî (latînî)]
+  ku: [Latn, [EU, ME], Kurdî]
- # XXX multiple script
+  ku-arab: [Arab, [EU, ME], كوردي]
- # ku: [[Arab, Latn], [EU, ME], [كوردي , Kurdî]]
+  ku-latn: [Latn, [EU, ME], Kurdî]
-  ku: [Latn, [EU, ME], كوردي / Kurdî]
   kv: [Cyrl, [EU], коми]
   kw: [Latn, [EU], kernowek]
   ky: [Cyrl, [AS], Кыргызча]
   la: [Latn, [EU], Latina]
- # Most identified with Turkey, Bulgaria, Greece, Spain and Israel,
+ # FIXME: alias
- # but also spoken in Latin America and elsewhere.
- # Wikipedia is mostly in Latn, but also in Hebr. (Comparable to az.)
   lad: [Latn, [ME, EU, AM], Ladino]
+  lad-latn: [Latn, [ME, EU, AM], Ladino]
+  lad-hebr: [Hebr, [ME, EU, AM], לאדינו]
   lb: [Latn, [EU], Lëtzebuergesch]
   lbe: [Cyrl, [EU], лакку]
   lez: [Cyrl, [EU], лезги]
@@ -333,7 +328,8 @@ languages:
   om: [Latn, [AF], Oromoo]
   or: [Orya, [AS], ଓଡ଼ିଆ]
   os: [Cyrl, [EU], Ирон]
-  pa: [Guru, [AS], ਪੰਜਾਬੀ]
+  pa: [pa-guru]
+  pa-guru: [Guru, [AS], ਪੰਜਾਬੀ]
   pag: [Latn, [AS], Pangasinan]
   pam: [Latn, [AS], Kapampangan]
   pap: [Latn, [AM], Papiamentu]
@@ -354,7 +350,6 @@ languages:
   pru: [Latn, [EU], Prūsiskan]
   ps: [Arab, [AS, ME], پښتو]
   pt-br: [Latn, [AM], português do Brasil]
- # world?
   pt: [Latn, [EU, AM, AS, PA, AF, WW], português]
   qu: [Latn, [AM], Runa Simi]
   qug: [Latn, [AM], Runa shimi]
@@ -397,13 +392,11 @@ languages:
   sgs: [Latn, [EU], žemaitėška]
   sh-cyrl: [Cyrl, [EU], српскохрватски]
   sh-latn: [Latn, [EU], srpskohrvatski]
- # XXX multiple script
+ # FIXME: alias
- # sh: [[Latn, Cyrl], [EU], [srpskohrvatski, српскохрватски]]
+  sh: [Latn, [EU], srpskohrvatski]
-  sh: [Latn, [EU], srpskohrvatski / српскохрватски]
   shi-latn: [Latn, [AF], Tašlḥiyt]
   shi-tfng: [Tfng, [AF], ⵜⴰⵛⵍⵃⵉⵜ]
- # XXX multiple script
+ # FIXME: which one to pick and alias?
- # shi: [[Latn, Tfng], [AF], [Tašlḥiyt, ⵜⴰⵛⵍⵃⵉⵜ]]
   shi: [Latn, [AF], Tašlḥiyt / ⵜⴰⵛⵍⵃⵉⵜ]
   shn: [Mymr, [AS], လိၵ်ႈတႆး]
   si: [Sinh, [AS], සිංහල]
@@ -421,11 +414,12 @@ languages:
   sn: [Latn, [AF], chiShona]
   so: [Latn, [AF], Soomaaliga]
   sq: [Latn, [EU], shqip]
-  sr-ec: [Cyrl, [EU], српски (ћирилица)]
+ # FIXME: do we need the double name here?
-  sr-el: [Latn, [EU], srpski (latinica)]
+  sr: [Cyrl, [EU], српски]
- # XXX multiple script
+  sr-ec: [sr-cyrl]
- # sr: [[Cyrl, Latn], [EU], [српски, srpski]]
+  sr-cyrl: [Cyrl, [EU], српски]
-  sr: [Cyrl, [EU], српски / srpski]
+  sr-el: [sr-latn]
+  sr-latn: [Latn, [EU], srpski]
   srn: [Latn, [AM, EU], Sranantongo]
   ss: [Latn, [AF], SiSwati]
   st: [Latn, [AF], Sesotho]
@@ -461,9 +455,10 @@ languages:
   tsd: [Grek, [EU], Τσακωνικά]
   tt-cyrl: [Cyrl, [EU], татарча]
   tt-latn: [Latn, [EU], tatarça]
- # XXX multiple script
+ # FIXME: alias
- # tt: [[Cyrl, Latn] [EU], [татарча, tatarça]]
+  tt: [Cyrl, [EU], татарча]
-  tt: [Cyrl, [EU], татарча / tatarça]
+  tt-cyrl: [Cyrl, [EU], татарча]
+  tt-latn: [Latn, [EU], tatarça]
   ttt: [Cyrl, [AS], Tati]
   tum: [Latn, [AF], chiTumbuka]
   tw: [Latn, [AF], Twi]
@@ -472,11 +467,10 @@ languages:
   tyv: [Cyrl, [AS], тыва дыл]
   tzm: [Tfng, [AF], ⵜⴰⵎⴰⵣⵉⵖⵜ]
   udm: [Cyrl, [EU], удмурт]
-  ug-arab: [Arab, [AS], ئۇيغۇرچە]
+ # FIXME: alias
-  ug-latn: [Latn, [AS], Uyghurche]
+  ug: [Arab, [AS], ئۇيغۇرچە ]
- # XXX multiple script
+  ug-arab: [Arab, [AS], ئۇيغۇرچە ]
- # ug: [[Arab, Latn], [AS], [ئۇيغۇرچە, Uyghurche]]
+  ug-latn: [Latn, [AS], Uyghurche ]
-  ug: [Arab, [AS], ئۇيغۇرچە / Uyghurche ]
   uk: [Cyrl, [EU], українська]
   ur: [Arab, [AS, ME], اردو]
   uz: [Latn, [AS], oʻzbekcha]
@@ -507,6 +501,8 @@ languages:
   yue: [Hant, [AS], 粵語]
   za: [Latn, [AS], Vahcuengh]
   zea: [Latn, [EU], Zeêuws]
+ # world? (may apply to many varieties of Chinese)
+  zh: [Hans, [AS], 中文]
   zh-classical: [Hant, [AS], 文言]
   zh-cn: [Hans, [AS], 中文（中国大陆）]
   zh-hans: [Hans, [AS], 中文（简体）]
@@ -518,17 +514,21 @@ languages:
   zh-sg: [Hans, [AS], 中文（新加坡）]
   zh-tw: [Hant, [AS], 中文（台灣）]
   zh-yue: [Hans, [AS], 粵語]
-  zh: [Hans, [AS], 中文]
   zu: [Latn, [AF], isiZulu]
 
+ # All the supported scripts, grouped logically.
+ #
  # The codes are taken from http://unicode.org/iso15924/iso15924-codes.html .
  #
  # The classification is roughly based on http://www.unicode.org/charts/
  # with some practical corrections.
- # The order of the groups affects display. It was suggested by Pau to distance
+ #
- # the largest groups from one another to improve discoverability.
+ # The order of the groups affects display. Pau Giner suggested the order;
-scriptgroups:
+ # the rationale of the order is to distance the largest groups from
+ # one another to improve discoverability.
+ #
  # The group name "Other" is reserved.
+scriptgroups:
  # It's hard to find a better place for Goth except the Latin group.
   Latin: [Latn, Goth]
  # Greek is probalby different enough from Latin and Cyrillic, but user testing

src/jquery.uls.data.utils.js

@@ -18,28 +18,44 @@
  */
 
 ( function ( $ ) {
-	"use strict";
+	'use strict';
 
 	/**
 	 * Log deprecated functions
 	 */
 	function deprecated( oldFunc, newFunc ) {
 		if ( window.console && window.console.log ) {
-			window.console.log( oldFunc + " is deprecated. Please use " + newFunc );
+			window.console.log( oldFunc + ' is deprecated. Please use ' + newFunc );
 		}
 	}
 
+	/**
+	 * Is this language a redirect to another language?
+	 * @param string language code
+	 * @return Target language code if it's a redirect or false if it's not
+	 */
+	$.uls.data.isRedirect = function( language ) {
+		return ( $.uls.data.languages[language] !== undefined &&
+			$.uls.data.languages[language].length === 1 ) ? $.uls.data.languages[language][0] : false;
+	};
+
 	/**
 	 * Returns the script of the language.
 	 * @param string language code
 	 * @return string
 	 */
 	$.uls.data.getScript = function( language ) {
+		var target = $.uls.data.isRedirect( language );
+
+		if ( target ) {
+			return $.uls.data.getScript( target );
+		}
+
 		return $.uls.data.languages[language][0];
 	};
 
 	$.uls.data.script = function( language ) { // deprecated
-		deprecated( "script", "getScript" );
+		deprecated( 'script', 'getScript' );
 		return $.uls.data.getScript( language );
 	};
 
@@ -49,11 +65,17 @@
 	 * @return array|string 'UNKNOWN'
 	 */
 	$.uls.data.getRegions = function( language ) {
+		var target = $.uls.data.isRedirect( language );
+
+		if ( target ) {
+			return $.uls.data.getRegions( target );
+		}
+
 		return ( $.uls.data.languages[language] && $.uls.data.languages[language][1] ) || 'UNKNOWN';
 	};
 
 	$.uls.data.regions = function( language ) { // deprecated
-		deprecated( "regions", "getRegions" );
+		deprecated( 'regions', 'getRegions' );
 		return $.uls.data.getRegions( language );
 	};
 
@@ -63,11 +85,17 @@
 	 * @return string
 	 */
 	$.uls.data.getAutonym = function( language ) {
+		var target = $.uls.data.isRedirect( language );
+
+		if ( target ) {
+			return $.uls.data.getAutonym( target );
+		}
+
 		return ( $.uls.data.languages[language] && $.uls.data.languages[language][2] ) || language;
 	};
 
 	$.uls.data.autonym = function( language ) { // deprecated
-		deprecated( "autonym", "getAutonym" );
+		deprecated( 'autonym', 'getAutonym' );
 		return $.uls.data.getAutonym( language );
 	};
 
@@ -79,6 +107,10 @@
 		var autonymsByCode = {};
 
 		for ( var language in $.uls.data.languages ) {
+			if ( $.uls.data.isRedirect( language ) ) {
+				continue;
+			}
+
 			autonymsByCode[language] = $.uls.data.getAutonym( language );
 		}
 
@@ -86,7 +118,7 @@
 	};
 
 	$.uls.data.autonyms = function() { // deprecated
-		deprecated( "autonyms", "getAutonyms" );
+		deprecated( 'autonyms', 'getAutonyms' );
 		return $.uls.data.getAutonyms();
 	};
 
@@ -105,7 +137,7 @@
 	};
 
 	$.uls.data.allRegions = function() { // deprecated
-		deprecated( "allRegions", "getAllRegions" );
+		deprecated( 'allRegions', 'getAllRegions' );
 		return $.uls.data.getAllRegions();
 	};
 
@@ -119,7 +151,7 @@
 	};
 
 	$.uls.data.languagesInScript = function( script ) { // deprecated
-		deprecated( "languagesInScript", "getLanguagesInScript" );
+		deprecated( 'languagesInScript', 'getLanguagesInScript' );
 		return $.uls.data.getLanguagesInScript( script );
 	};
 
@@ -132,6 +164,10 @@
 		var languagesInScripts = [];
 
 		for ( var language in $.uls.data.languages ) {
+			if ( $.uls.data.isRedirect( language ) ) {
+				continue;
+			}
+
 			for ( var i = 0; i < scripts.length; i++ ) {
 				if ( scripts[i] === $.uls.data.getScript( language ) ) {
 					languagesInScripts.push( language );
@@ -144,7 +180,7 @@
 	};
 
 	$.uls.data.languagesInScripts = function( scripts ) { // deprecated
-		deprecated( "languagesInScripts", "getLanguagesInScripts" );
+		deprecated( 'languagesInScripts', 'getLanguagesInScripts' );
 		return $.uls.data.getLanguagesInScripts( scripts );
 	};
 
@@ -158,7 +194,7 @@
 	};
 
 	$.uls.data.languagesInRegion = function( region ) { // deprecated
-		deprecated( "languagesInRegion", "getLanguagesInRegion" );
+		deprecated( 'languagesInRegion', 'getLanguagesInRegion' );
 		return $.uls.data.getLanguagesInRegion( region );
 	};
 
@@ -171,6 +207,10 @@
 		var languagesInRegions = [];
 
 		for ( var language in $.uls.data.languages ) {
+			if ( $.uls.data.isRedirect( language ) ) {
+				continue;
+			}
+
 			for ( var i = 0; i < regions.length; i++ ) {
 				if ( $.inArray( regions[i], $.uls.data.getRegions( language ) ) !== -1 ) {
 					languagesInRegions.push( language );
@@ -183,7 +223,7 @@
 	};
 
 	$.uls.data.languagesInRegions = function( regions ) { // deprecated
-		deprecated( "languagesInRegions", "getLanguagesInRegions" );
+		deprecated( 'languagesInRegions', 'getLanguagesInRegions' );
 		return $.uls.data.getLanguagesInRegions( regions );
 	};
 
@@ -197,7 +237,7 @@
 	};
 
 	$.uls.data.languagesInRegionGroup = function( groupNum ) { // deprecated
-		deprecated( "languagesInRegionGroup", "getLanguagesInRegionGroup" );
+		deprecated( 'languagesInRegionGroup', 'getLanguagesInRegionGroup' );
 		return $.uls.data.getLanguagesInRegionGroup( groupNum );
 	};
 
@@ -211,6 +251,10 @@
 		var languagesByScriptInRegion = {};
 
 		for ( var language in $.uls.data.languages ) {
+			if ( $.uls.data.isRedirect( language ) ) {
+				continue;
+			}
+
 			if ( $.inArray( region, $.uls.data.getRegions( language ) ) !== -1 ) {
 				var script = $.uls.data.getScript( language );
 				if ( languagesByScriptInRegion[script] === undefined ) {
@@ -224,7 +268,7 @@
 	};
 
 	$.uls.data.languagesByScriptInRegion = function( region ) { // deprecated
-		deprecated( "languagesByScriptInRegion", "getLanguagesByScriptInRegion" );
+		deprecated( 'languagesByScriptInRegion', 'getLanguagesByScriptInRegion' );
 		return $.uls.data.getLanguagesByScriptInRegion( region );
 	};
 
@@ -239,7 +283,7 @@
 	};
 
 	$.uls.data.languagesByScriptGroupInRegion = function( region ) { // deprecated
-		deprecated( "languagesByScriptGroupInRegion", "getLanguagesByScriptGroupInRegion" );
+		deprecated( 'languagesByScriptGroupInRegion', 'getLanguagesByScriptGroupInRegion' );
 		return $.uls.data.getLanguagesByScriptGroupInRegion( region );
 	};
 
@@ -253,7 +297,7 @@
 	};
 
 	$.uls.data.allLanguagesByScriptGroup = function() { // deprecated
-		deprecated( "allLanguagesByScriptGroup", "getAllLanguagesByScriptGroup" );
+		deprecated( 'allLanguagesByScriptGroup', 'getAllLanguagesByScriptGroup' );
 		return $.uls.data.getAllLanguagesByScriptGroup();
 	};
 
@@ -285,7 +329,7 @@
 	};
 
 	$.uls.data.languagesByScriptGroup = function( languages ) { // deprecated
-		deprecated( "languagesByScriptGroup", "getLanguagesByScriptGroup" );
+		deprecated( 'languagesByScriptGroup', 'getLanguagesByScriptGroup' );
 		return $.uls.data.getLanguagesByScriptGroup( languages );
 	};
 
@@ -299,12 +343,18 @@
 		var languagesByScriptGroupInRegions = {};
 
 		for ( var language in $.uls.data.languages ) {
+			if ( $.uls.data.isRedirect( language ) ) {
+				continue;
+			}
+
 			for ( var i = 0; i < regions.length; i++ ) {
 				if ( $.inArray( regions[i], $.uls.data.getRegions( language ) ) !== -1 ) {
 					var scriptGroup = $.uls.data.getScriptGroupOfLanguage( language );
+
 					if ( languagesByScriptGroupInRegions[scriptGroup] === undefined ) {
 						languagesByScriptGroupInRegions[scriptGroup] = [];
 					}
+
 					languagesByScriptGroupInRegions[scriptGroup].push( language );
 					break;
 				}
@@ -315,7 +365,7 @@
 	};
 
 	$.uls.data.languagesByScriptGroupInRegions = function( regions ) { // deprecated
-		deprecated( "languagesByScriptGroupInRegions", "getLanguagesByScriptGroupInRegions" );
+		deprecated( 'languagesByScriptGroupInRegions', 'getLanguagesByScriptGroupInRegions' );
 		return $.uls.data.getLanguagesByScriptGroupInRegions( regions );
 	};
 
@@ -331,13 +381,19 @@
 
 		for ( region in $.uls.data.regiongroups ) {
 			regionGroup = $.uls.data.regiongroups[region];
+
 			if ( allLanguagesByRegionAndScript[regionGroup] === undefined ) {
 				allLanguagesByRegionAndScript[regionGroup] = {};
 			}
+
 			allLanguagesByRegionAndScript[regionGroup][region] = {};
 		}
 
 		for ( var language in $.uls.data.languages ) {
+			if ( $.uls.data.isRedirect( language ) ) {
+				continue;
+			}
+
 			var script = $.uls.data.getScript( language );
 			var scriptGroup = $.uls.data.getGroupOfScript( script );
 			var regions = $.uls.data.getRegions( language );
@@ -362,7 +418,7 @@
 	};
 
 	$.uls.data.allLanguagesByRegionAndScript = function() { // deprecated
-		deprecated( "allLanguagesByRegionAndScript", "getAllLanguagesByRegionAndScript" );
+		deprecated( 'allLanguagesByRegionAndScript', 'getAllLanguagesByRegionAndScript' );
 		return $.uls.data.getAllLanguagesByRegionAndScript();
 	};
 
@@ -384,7 +440,7 @@
 	};
 
 	$.uls.data.regionsInGroup = function( groupNum ) { // deprecated
-		deprecated( "regionsInGroup", "getRegionsInGroup" );
+		deprecated( 'regionsInGroup', 'getRegionsInGroup' );
 		return $.uls.data.getRegionsInGroup( groupNum );
 	};
 
@@ -405,7 +461,7 @@
 	};
 
 	$.uls.data.groupOfScript = function( script ) { // deprecated
-		deprecated( "groupOfScript", "getGroupOfScript" );
+		deprecated( 'groupOfScript', 'getGroupOfScript' );
 		return $.uls.data.getGroupOfScript( script );
 	};
 
@@ -419,7 +475,7 @@
 	};
 
 	$.uls.data.scriptGroupOfLanguage = function( language ) { // deprecated
-		deprecated( "scriptGroupOfLanguage", "getScriptGroupOfLanguage" );
+		deprecated( 'scriptGroupOfLanguage', 'getScriptGroupOfLanguage' );
 		return $.uls.data.getScriptGroupOfLanguage( language );
 	};
 
@@ -462,7 +518,7 @@
 	};
 
 	$.uls.data.languagesInTerritory = function( territory ) { // deprecated
-		deprecated( "languagesInTerritory", "getLanguagesInTerritory" );
+		deprecated( 'languagesInTerritory', 'getLanguagesInTerritory' );
 		return $.uls.data.getLanguagesInTerritory( territory );
 	};
 } ( jQuery ) );

test/jquery.uls.test.js

@@ -34,6 +34,23 @@
 		return result;
 	};
 
+	/*
+	 * Runs over all languages and checks that all redirects have a valid target.
+	 */
+	var badRedirects = function () {
+		var result = [];
+
+		for ( var language in $.uls.data.languages ) {
+			var target = $.uls.data.isRedirect( language );
+
+			if ( target && !$.uls.data.languages[target] ) {
+				result.push( language );
+			}
+		}
+
+		return result;
+	};
+
 	/*
 	 * Runs over all script codes mentioned in langdb and checks whether
 	 * they have something that looks like an autonym.
@@ -54,13 +71,18 @@
 		assert.ok( $.fn.uls, "$.fn.uls is defined" );
 	} );
 
-	test( "-- $.uls.data testing", 27, function ( assert ) {
+	test( "-- $.uls.data testing", 40, function ( assert ) {
 
-		assert.strictEqual( $.uls.data.getAutonyms()['he'], 'עברית', 'Correct autonym is returned for Hebrew using getAutonyms().' );
+		assert.strictEqual( $.uls.data.isRedirect( 'sr-ec' ), 'sr-cyrl', "'sr-ec' is a redirect to 'sr-cyrl'" );
+		var autonyms = $.uls.data.getAutonyms();
+		assert.strictEqual( autonyms['he'], 'עברית', 'Correct autonym is returned for Hebrew using getAutonyms().' );
+		assert.strictEqual( autonyms['pa'], undefined, "Language 'pa' is not listed in autonyms, because it's a redirect" );
+		assert.strictEqual( autonyms['pa-guru'], 'ਪੰਜਾਬੀ', "Language 'pa-guru' has the correct autonym" );
 
 		// This test assumes that we don't want any scripts to be in the 'Other'
 		// group. Actually, this may become wrong some day.
 		assert.deepEqual( orphanScripts(), [], 'All scripts belong to script groups.' );
+		assert.deepEqual( badRedirects(), [], 'All redirects have valid targets.' );
 		assert.deepEqual( languagesWithoutAutonym(), [], 'All languages have autonyms.' );
 
 		assert.strictEqual(
@@ -84,20 +106,27 @@
 		assert.deepEqual( allLanguagesByRegionAndScript['4']['AS']['SouthEastAsian']['Bugi'], [
 			'bug'
 		], 'All languages in the Buginese script in Asia were selected' );
+		assert.deepEqual( allLanguagesByRegionAndScript['4']['AS']['SouthAsian']['Guru'], [
+			'pa-guru'
+		], 'Only language pa-guru appears as a language in script Guru in SouthAsian languages in Asia' );
+
+		var languagesInEU = $.uls.data.getLanguagesInRegion( 'EU' );
+		assert.strictEqual( $.inArray( 'sr-ec', languagesInEU ), -1, "Language 'sr-ec' doesn't appear in languages of region EU, because it's a redirect" );
+		assert.ok( $.inArray( 'sr-cyrl', languagesInEU ) > -1, "Language 'sr-cyrl' appears in languages of region EU, because it's not a redirect" );
 
 		assert.deepEqual( $.uls.data.getLanguagesInRegion( "PA" ),
 			[
-				"ace", "bi", "ch", "en-gb", "en", "fj", "haw", "hif-latn", "hif", "ho", "jv",
+				"ace", "bi", "ch", "en-gb", "en", "fj", "haw", "hif", "hif-latn", "ho", "jv", "jv-java",
 				"mh", "mi", "na", "niu", "pih", "pis", "pt", "rtm", "sm", "tet",
 				"to", "tpi", "ty", "wls"
 			],
 			"languages of region PA are selected correctly" );
 		assert.deepEqual( $.uls.data.getLanguagesInRegions( ["AM", "WW"] ),
 			[
-				"akz", "arn", "aro", "ase", "avk", "ay", "cho", "chr", "chy", "cr-cans", "cr-latn",
+				"akz", "arn", "aro", "ase", "avk", "ay", "cho", "chr", "chy", "cr", "cr-cans", "cr-latn",
-				"cr", "en-ca", "en", "eo", "es-419", "es-formal", "es", "esu", "fr", "gcf", "gn",
+				"en-ca", "en", "eo", "es-419", "es-formal", "es", "esu", "fr", "gcf", "gn",
 				"guc", "haw", "ht", "ia", "ie", "ik", "ike-cans", "ike-latn", "io", "iu", "jam",
-				"jbo", "kgp", "kl", "lad", "lfn", "mfe", "mic", "mus", "nah", "nl-informal", "nl",
+				"jbo", "kgp", "kl", "lad", "lad-latn", "lad-hebr", "lfn", "mfe", "mic", "mus", "nah", "nl-informal", "nl",
 				"nov", "nv", "pap", "pdc", "pdt", "ppl", "pt-br", "pt", "qu", "qug", "rap", "sei",
 				"simple", "srn", "tokipona", "vo", "yi", "yrl", "yua"
 			],
@@ -107,6 +136,10 @@
 		assert.deepEqual( $.uls.data.getLanguagesInScript( 'Knda' ), [
 			"kn", "tcy"
 		], "languages in script Knda are selected correctly" );
+		assert.deepEqual( $.uls.data.getLanguagesInScript( 'Guru' ),
+			["pa-guru"],
+			"'pa-guru' is written in script Guru, and 'pa' is skipped as a redirect"
+		);
 		assert.deepEqual( $.uls.data.getLanguagesInScripts( ['Geor', 'Armn'] ),
 			["hy", "ka", "xmf"],
 			"languages in scripts Geor and Armn are selected correctly"
@@ -124,15 +157,26 @@
 
 		var languagesByScriptInAM = $.uls.data.getLanguagesByScriptInRegion( "AM" );
 		assert.deepEqual( languagesByScriptInAM['Cans'], [
-			"cr-cans", "cr", "ike-cans", "iu"
+			"cr", "cr-cans", "ike-cans", "iu"
 		], "correct languages in Cans in AM selected" );
 
-		assert.strictEqual( $.uls.data.getAutonym( 'pa' ), 'ਪੰਜਾਬੀ', 'Correct autonym of the Punjabi language was selected' );
+		var languagesByScriptInEU = $.uls.data.getLanguagesByScriptInRegion( "EU" );
+		assert.strictEqual( $.inArray( 'sr-el', languagesByScriptInEU['Latn'] ), -1,
+			"Language 'sr-el' doesn't appear as a Latin-script language in EU, because it's a redirect" );
+		assert.ok( $.inArray( 'sr-latn', languagesByScriptInEU['Latn'] ) > -1,
+			"Language 'sr-latn' appears as a Latin-script language in EU, because it's not a redirect" );
+
+		assert.strictEqual( $.uls.data.getAutonym( 'pa' ), 'ਪੰਜਾਬੀ', 'Correct autonym of the Punjabi language was selected using code pa.' );
+		assert.strictEqual( $.uls.data.getAutonym( 'pa-guru' ), 'ਪੰਜਾਬੀ', 'Correct autonym of the Punjabi language was selected using code pa.' );
 
 		var languagesByScriptGroupInEMEA = $.uls.data.getLanguagesByScriptGroupInRegions( $.uls.data.getRegionsInGroup( 3 ) );
 		assert.deepEqual( languagesByScriptGroupInEMEA['WestCaucasian'], [
 			'hy', 'ka', 'xmf'
 		], 'Correct languages in WestCaucasian script group in EMEA selected' );
+		assert.strictEqual( $.inArray( 'sr-ec', languagesByScriptGroupInEMEA['Cyrillic'] ), -1,
+			"Language 'sr-ec' doesn't appear as a Cyrillic-scriptgroup language in EMEA, because it's a redirect" );
+		assert.ok( $.inArray( 'sr-cyrl', languagesByScriptGroupInEMEA['Cyrillic'] ) > -1,
+			"Language 'sr-cyrl' appears as a Cyrillic-scriptgroup language in EMEA, because it's not a redirect" );
 
 		var allLanguagesByScriptGroup = $.uls.data.getAllLanguagesByScriptGroup();
 		assert.deepEqual( allLanguagesByScriptGroup['Greek'], [