Merge pull request #265 from amire80/unique-autonyms

Make all autonyms unique and add a test for this

data/generated-langdb.json

@@ -316,14 +316,10 @@
             [
                 "AS"
             ],
-            "Batak Toba"
+            "ᯅᯖᯂ᯲ ᯖᯬᯅ"
         ],
         "bbc": [
-            "Latn",
+            "bbc-latn"
-            [
-                "AS"
-            ],
-            "Batak Toba"
         ],
         "bcc": [
             "Arab",
@@ -1445,11 +1441,7 @@
             "italiano"
         ],
         "iu": [
-            "Cans",
+            "ike-cans"
-            [
-                "AM"
-            ],
-            "ᐃᓄᒃᑎᑐᑦ"
         ],
         "ja": [
             "Jpan",
@@ -1756,11 +1748,7 @@
             "कॉशुर"
         ],
         "ks": [
-            "Arab",
+            "ks-arab"
-            [
-                "AS"
-            ],
-            "کٲشُر"
         ],
         "ksf": [
             "Latn",
@@ -3239,11 +3227,7 @@
             "tojikī"
         ],
         "tg": [
-            "Cyrl",
+            "tg-cyrl"
-            [
-                "AS"
-            ],
-            "тоҷикӣ"
         ],
         "th": [
             "Thai",
@@ -3700,7 +3684,7 @@
             [
                 "AF"
             ],
-            "ⵜⴰⵎⴰⵣⵉⵖⵜ"
+            "ⵜⴰⵎⴰⵣⵉⵖⵜ ⵜⴰⵏⴰⵡⴰⵢⵜ"
         ],
         "zh": [
             "Hans",
@@ -3710,11 +3694,7 @@
             "中文"
         ],
         "zh-classical": [
-            "Hant",
+            "lzh"
-            [
-                "AS"
-            ],
-            "文言"
         ],
         "zh-cn": [
             "Hans",
@@ -4076,7 +4056,7 @@
             "pdt",
             "cr",
             "yi",
-            "iu",
+            "ike-cans",
             "atj"
         ],
         "CC": [
@@ -4356,7 +4336,7 @@
             "zh",
             "el",
             "it",
-            "ks",
+            "ks-arab",
             "gd",
             "yi",
             "ml",
@@ -4483,7 +4463,7 @@
             "bug",
             "bjn",
             "ace",
-            "bbc",
+            "bbc-latn",
             "zh-hant",
             "zh",
             "sly",
@@ -4531,7 +4511,7 @@
             "hne",
             "ne",
             "sat",
-            "ks",
+            "ks-arab",
             "gom-deva",
             "sd",
             "doi",
@@ -4984,10 +4964,10 @@
             "brh",
             "fa",
             "bgn",
-            "tg",
+            "tg-cyrl",
             "bft",
             "khw",
-            "ks"
+            "ks-arab"
         ],
         "PL": [
             "pl",
@@ -5242,7 +5222,7 @@
             "shn"
         ],
         "TJ": [
-            "tg",
+            "tg-cyrl",
             "ru",
             "fa",
             "ar"

data/langdb.yaml

@@ -51,9 +51,8 @@ languages:
   bar: [Latn, [EU], Boarisch]
   bat-smg: [sgs]
   bbc-latn: [Latn, [AS], Batak Toba]
- # FIXME - autonym in the Batak script is unknown
+  bbc-batk: [Batk, [AS], ᯅᯖᯂ᯲ ᯖᯬᯅ]
-  bbc-batk: [Batk, [AS], Batak Toba]
+  bbc: [bbc-latn]
-  bbc: [Latn, [AS], Batak Toba]
   bcc: [Arab, [AS, ME], جهلسری بلوچی]
   bcl: [Latn, [AS], Bikol Central]
   be-tarask: [Cyrl, [EU], беларуская (тарашкевіца)]
@@ -228,8 +227,7 @@ languages:
   io: [Latn, [WW], Ido]
   is: [Latn, [EU], íslenska]
   it: [Latn, [EU], italiano]
- # For variants ike-* is used
+  iu: [ike-cans]
-  iu: [Cans, [AM], ᐃᓄᒃᑎᑐᑦ]
   ja: [Jpan, [AS], 日本語]
   jam: [Latn, [AM], Patois]
   jbo: [Latn, [WW], lojban]
@@ -279,7 +277,7 @@ languages:
   krl: [Latn, [EU], Karjala]
   ks-arab: [Arab, [AS], کٲشُر]
   ks-deva: [Deva, [AS], कॉशुर]
-  ks: [Arab, [AS], کٲشُر]
+  ks: [ks-arab]
   ksf: [Latn, [AF], Bafia]
   ksh: [Latn, [EU], Ripoarisch]
  # CLDR uses ku-latn and ku-arab. Latin is more common and is used in Wikipedia.
@@ -502,7 +500,7 @@ languages:
   tet: [Latn, [AS, PA], tetun]
   tg-cyrl: [Cyrl, [AS], тоҷикӣ]
   tg-latn: [Latn, [AS], tojikī]
-  tg: [Cyrl, [AS], тоҷикӣ]
+  tg: [tg-cyrl]
   th: [Thai, [AS], ไทย]
   ti: [Ethi, [AF], ትግርኛ]
   tig: [Ethi, [AF], ትግረ]
@@ -570,10 +568,10 @@ languages:
   yue: [Hant, [AS], 粵語]
   za: [Latn, [AS], Vahcuengh]
   zea: [Latn, [EU], Zeêuws]
-  zgh: [Tfng, [AF], ⵜⴰⵎⴰⵣⵉⵖⵜ]
+  zgh: [Tfng, [AF], ⵜⴰⵎⴰⵣⵉⵖⵜ ⵜⴰⵏⴰⵡⴰⵢⵜ]
  # world? (may apply to many varieties of Chinese)
   zh: [Hans, [AS], 中文]
-  zh-classical: [Hant, [AS], 文言]
+  zh-classical: [lzh]
   zh-cn: [Hans, [AS], 中文（中国大陆）]
   zh-hans: [Hans, [AS], 中文（简体）]
   zh-hant: [Hant, [AS], 中文（繁體）]

test/jquery.uls.test.js

@@ -20,7 +20,7 @@
 
 	module( 'jquery.uls' );
 
-	var orphanScripts, badRedirects, doubleRedirects, languagesWithoutAutonym;
+	var orphanScripts, badRedirects, doubleRedirects, doubleAutonyms, languagesWithoutAutonym;
 
 	/*
 	 * Runs over all script codes mentioned in langdb and checks whether
@@ -59,6 +59,31 @@
 		return result;
 	};
 
+	/*
+	 * Runs over all languages and checks that all autonyms are unique.
+	 */
+	doubleAutonyms = function () {
+		var language, autonym,
+			autonyms = [],
+			duplicateAutonyms = [];
+
+		for ( language in $.uls.data.languages ) {
+			if ( $.uls.data.isRedirect( language ) ) {
+				continue;
+			}
+
+			autonym = $.uls.data.getAutonym( language );
+
+			if ( $.inArray( autonym, autonyms ) > -1 ) {
+				duplicateAutonyms.push( language );
+			}
+
+			autonyms.push( autonym );
+		}
+
+		return duplicateAutonyms;
+	};
+
 	/*
 	 * Runs over all languages and checks that all redirects point to a language.
 	 * There's no reason to have double redirects.
@@ -99,7 +124,7 @@
 		assert.ok( $.fn.uls, '$.fn.uls is defined' );
 	} );
 
-	test( '-- $.uls.data testing', 30, function ( assert ) {
+	test( '-- $.uls.data testing', 31, function ( assert ) {
 		var autonyms,
 			languagesToGroup, groupedLanguages;
 
@@ -126,6 +151,7 @@
 		assert.deepEqual( badRedirects(), [], 'All redirects have valid targets.' );
 		assert.deepEqual( doubleRedirects(), [], 'There are no double redirects.' );
 		assert.deepEqual( languagesWithoutAutonym(), [], 'All languages have autonyms.' );
+		assert.deepEqual( doubleAutonyms(), [], 'All languages have distinct autonyms.' );
 
 		assert.strictEqual(
 			$.uls.data.getGroupOfScript( 'Beng' ),