Restore "tonga" search test and update language search index

This test was removed in I62f6c9da90074258f13b7fb281e777b3bcf5d17d
to allow the easy updating of local English language names in
the CLDR extension in I2c09d409567efef2e0498cde15ecead4c438c657

Now this test is restored with the new name of the Tonga (toi)
language.

This also updates the search index with the languages
whose English names were updated in
I2c09d409567efef2e0498cde15ecead4c438c657

Bug: T374785
Depends-On: I2c09d409567efef2e0498cde15ecead4c438c657
Change-Id: I922570d7bf0fb00740c1411c36e2d2e889df4c9b

data/LanguageNameSearchData.php

@@ -174,6 +174,7 @@ class LanguageNameSearchData {
 				'adigueo' => 'ady',
 				'adigué' => 'ady',
 				'adigé' => 'ady',
+				'adilabad gondi' => 'wsg',
 				'adja' => 'ajg',
 				'adygabze' => 'ady-latn',
 				'adygai' => 'ady',
@@ -2186,6 +2187,7 @@ class LanguageNameSearchData {
 				'amērikas — lātiniskas amērikas špāniskan' => 'es-419',
 				'amẹ́ríka — látín amẹ́ríka spánish' => 'es-419',
 				'anaithnid — teanga anaithnid' => 'und',
+				'ancash — huaylas ancash quechua' => 'qwh',
 				'ancian — grec ancian' => 'grc',
 				'ancian — occitan ancian' => 'pro',
 				'ancien — grec ancien' => 'grc',
@@ -5623,6 +5625,7 @@ class LanguageNameSearchData {
 				'creole — louisiana creole' => 'lou',
 				'creole — mauritian creole' => 'mfe',
 				'creole — perancis seselwa creole' => 'crs',
+				'creole — réunion creole french' => 'rcf',
 				'creole — saint lucian creole' => 'acf',
 				'creole — seselwa creole franskt' => 'crs',
 				'creole — seselwa creole french' => 'crs',
@@ -5649,6 +5652,7 @@ class LanguageNameSearchData {
 				'criollu — francés criollu seselwa' => 'crs',
 				'criollu — inglés criollu xamaicanu' => 'jam',
 				'crioulo — francês crioulo seselwa' => 'crs',
+				'crioulo — upper guinea crioulo' => 'pov',
 				'croatia — tiếng croatia' => 'hr',
 				'croatian — afaan croatian' => 'hr',
 				'crìtheolach — beurla crìtheolach diameuga' => 'jam',
@@ -6173,6 +6177,8 @@ class LanguageNameSearchData {
 				'dasaga — èdè dasaga' => 'dzg',
 				'dataran — cree dataran' => 'crk',
 				'dataran — kri dataran' => 'crk',
+				'daw — hmong daw' => 'mww',
+				'daw — hmong daw (latin script)' => 'mww-latn',
 				'daw — hmong daw simi' => 'hmn',
 				'dawb — hmoob dawb' => 'mww-latn',
 				'dazaga — tiếng dazaga' => 'dzg',
@@ -7081,6 +7087,7 @@ class LanguageNameSearchData {
 				'elamščina' => 'elx',
 				'elbanî' => 'sq',
 				'eleme' => 'elm',
+				'elfdalian' => 'ovd',
 				'elizako eslaviera' => 'cu',
 				'elliceaansch' => 'tvl',
 				'elliceaansk' => 'tvl',
@@ -9288,6 +9295,7 @@ class LanguageNameSearchData {
 				'french — isi-swiss french' => 'fr-ch',
 				'french — middle french' => 'frm',
 				'french — old french' => 'fro',
+				'french — réunion creole french' => 'rcf',
 				'french — seselwa creole french' => 'crs',
 				'french — swiss french' => 'fr-ch',
 				'frens — kenedien frens' => 'fr-ca',
@@ -10299,6 +10307,7 @@ class LanguageNameSearchData {
 				'gogleddol — sami gogleddol' => 'se',
 				'gogleddol — sotho gogleddol' => 'nso',
 				'gogleddol — tutchone gogleddol' => 'ttm',
+				'gondi — adilabad gondi' => 'wsg',
 				'gondi — tiếng gondi' => 'gon',
 				'gorllewin — ffriseg y gorllewin' => 'fy',
 				'gorllewinol — balochi gorllewinol' => 'bgn',
@@ -10326,6 +10335,7 @@ class LanguageNameSearchData {
 				'guarani — afaan guarani' => 'gn',
 				'guarani — tiếng guarani' => 'gn',
 				'guarani — èdè guarani' => 'gn',
+				'guinea — upper guinea crioulo' => 'pov',
 				'gujarati — afaan gujarati' => 'gu',
 				'gujarati — tiếng gujarati' => 'gu',
 				'gujarati — èdè gujarati' => 'gu',
@@ -10538,6 +10548,7 @@ class LanguageNameSearchData {
 				'harýanwi dili' => 'bgc',
 				'harāna tongo' => 'srn',
 				'hassaniya-arabisk' => 'mey',
+				'hassaniyya' => 'mey',
 				'hatāri' => 'sat',
 				'hauajahnesch' => 'haw',
 				'hauha' => 'ha',
@@ -10872,6 +10883,8 @@ class LanguageNameSearchData {
 				'hizkuntza ezezaguna' => 'und',
 				'hmong' => 'hmn',
 				'hmong (latinsk alfabet)' => 'mww-latn',
+				'hmong daw' => 'mww',
+				'hmong daw (latin script)' => 'mww-latn',
 				'hmong daw simi' => 'hmn',
 				'hmong dili' => 'hmn',
 				'hmong njua' => 'hnj',
@@ -10954,6 +10967,7 @@ class LanguageNameSearchData {
 				'hoto ki te raki' => 'nso',
 				'hrvatski' => 'hr',
 				'hrvaščina' => 'hr',
+				'huaylas ancash quechua' => 'qwh',
 				'huaylas-quechua' => 'qwh',
 				'huitene' => 'sv',
 				'hukuma' => 'suk',
@@ -15304,6 +15318,7 @@ class LanguageNameSearchData {
 				'kom' => 'bkm',
 				'kome' => 'kv',
 				'komering' => 'kge',
+				'komering (arabic script)' => 'kge-arab',
 				'komgbe' => 'bkm',
 				'komi' => 'kv',
 				'komi (aafrika)' => 'bkm',
@@ -16434,6 +16449,7 @@ class LanguageNameSearchData {
 				'ketesi — lea faka-tonga ‘o ketesi' => 'sdh',
 				'kevneşopî — çîniya kevneşopî' => 'zh-hant',
 				'kečujų — čimboraso aukštumų kečujų' => 'qug',
+				'kham — sheshi kham' => 'kip',
 				'khasi — tiếng khasi' => 'kha',
 				'khmer — tiếng khmer' => 'km',
 				'khoa — tiếng khoa học quốc tế' => 'ia',
@@ -16522,6 +16538,7 @@ class LanguageNameSearchData {
 				'klériku — slavu klériku' => 'cu',
 				'kodha — changma kodha' => 'ccp',
 				'kojrabore — senishte kojrabore' => 'ses',
+				'kolami — southeastern kolami' => 'nit',
 				'kolimi — akukho okuqukethwe kolimi' => 'zxx',
 				'kolońska — gwara kolońska' => 'ksh',
 				'kolsch — dialek kolsch' => 'ksh',
@@ -19357,6 +19374,7 @@ class LanguageNameSearchData {
 				'makame' => 'jmc',
 				'makasa' => 'mak',
 				'makasar' => 'mak',
+				'makasar (buginese script)' => 'mak',
 				'makasar (buginesisk alfabet)' => 'mak-bugi',
 				'makasar dili' => 'mak',
 				'makasar simi' => 'mak',
@@ -19742,6 +19760,7 @@ class LanguageNameSearchData {
 				'mandžuština' => 'mnc',
 				'mandžūru' => 'mnc',
 				'maneze' => 'gv',
+				'mangareva' => 'mrv',
 				'mangarevansk' => 'mrv',
 				'mangkasarak' => 'mak',
 				'mangooli' => 'mn',
@@ -20420,6 +20439,7 @@ class LanguageNameSearchData {
 				'minnan (han script)' => 'nan-hani',
 				'minnan (kinesisk skrift)' => 'nan-hani',
 				'minnan (pe̍h-ōe-jī)' => 'nan-latn-pehoeji',
+				'minnan (simplified han script)' => 'nan-hans',
 				'minnan (traditional han script)' => 'nan-hant',
 				'minnan (tâi-lô)' => 'nan-latn-tailo',
 				'minnanyu' => 'nan',
@@ -21359,6 +21379,7 @@ class LanguageNameSearchData {
 				'moroccan — daidaitaccen moroccan tamazight' => 'zgh',
 				'moroccan — standard moroccan tamazait' => 'zgh',
 				'moroccan — standard moroccan tamazight' => 'zgh',
+				'moroccan — standard moroccan tamazight (latin script)' => 'zgh-latn',
 				'moroco — arabeg moroco' => 'ary',
 				'moroco — tamaisich stannardach moroco' => 'zgh',
 				'moroco — tamaseit moroco safonol' => 'zgh',
@@ -21729,6 +21750,7 @@ class LanguageNameSearchData {
 				'nemčina (švajčiarska)' => 'gsw',
 				'nemščina' => 'de',
 				'nemščina (švica)' => 'gsw',
+				'nenets' => 'yrk',
 				'nenetsisk' => 'yrk',
 				'nengatoueg' => 'yrl',
 				'nengatu' => 'yrl',
@@ -22253,6 +22275,7 @@ class LanguageNameSearchData {
 				'northern sami (sweden)' => 'se-se',
 				'northern sotho' => 'nso',
 				'northern thai' => 'nod',
+				'northern thai (thai script)' => 'nod-thai',
 				'northern tujia' => 'tji',
 				'northern tutchone' => 'ttm',
 				'northwestern ojibwa' => 'ojb',
@@ -23791,6 +23814,7 @@ class LanguageNameSearchData {
 				'paali (siddhami kiri)' => 'pi-sidd',
 				'pachto' => 'ps',
 				'pagrindinė frankonų' => 'vmf',
+				'pahari-potwari' => 'phr',
 				'pahlafi' => 'pal',
 				'pahlavani' => 'sdh',
 				'pahlavi' => 'pal',
@@ -25215,6 +25239,7 @@ class LanguageNameSearchData {
 				'qendrore — kurdishte qendrore' => 'ckb',
 				'qendrore — oxhibuaishte qendrore' => 'ojc',
 				'qichwa — anqash qichwa' => 'qwh',
+				'quechua — huaylas ancash quechua' => 'qwh',
 				'quechua — tiếng quechua' => 'qu',
 				'quechua — tiếng quechua ở cao nguyên chimborazo' => 'qug',
 				'quichua — chimborazo highland quichua' => 'qug',
@@ -25725,6 +25750,7 @@ class LanguageNameSearchData {
 				'rétoromán' => 'rm',
 				'rétorománčina' => 'rm',
 				'rétorománština' => 'rm',
+				'réunion creole french' => 'rcf',
 				'rífština' => 'rif',
 				'rîpwarî' => 'ksh',
 				'róihinis' => 'rhg',
@@ -26867,6 +26893,7 @@ class LanguageNameSearchData {
 				'shawiya (tifinagh script)' => 'shy-tfng',
 				'shawiya (tifinagh-schrift)' => 'shy-tfng',
 				'shelta' => 'sth',
+				'sheshi kham' => 'kip',
 				'shiinaha mandarin' => 'zh',
 				'shiinaha rasmiga ah' => 'zh-hans',
 				'shiinahii hore' => 'zh-hant',
@@ -27622,6 +27649,7 @@ class LanguageNameSearchData {
 				'soussou' => 'sus',
 				'south azerbaijani' => 'azb',
 				'south ndebele' => 'nr',
+				'southeastern kolami' => 'nit',
 				'southern altai' => 'alt',
 				'southern balochi' => 'bcc',
 				'southern east cree' => 'crj',
@@ -27792,6 +27820,7 @@ class LanguageNameSearchData {
 				'standard marrokansk tamazight' => 'zgh',
 				'standard moroccan tamazait' => 'zgh',
 				'standard moroccan tamazight' => 'zgh',
+				'standard moroccan tamazight (latin script)' => 'zgh-latn',
 				'standard tamazight' => 'zgh',
 				'standardaraabsch' => 'ar-001',
 				'standardni marokanski tamazigt' => 'zgh',
@@ -31996,6 +32025,7 @@ class LanguageNameSearchData {
 				'tolışi' => 'tly',
 				'tonehsesch arahbesch' => 'aeb',
 				'tonga' => 'to',
+				'tonga (botatwe)' => 'toi',
 				'tonga (niasa)' => 'tog',
 				'tonga (nyasa)' => 'tog',
 				'tonga del nyasa' => 'tog',
@@ -32818,6 +32848,7 @@ class LanguageNameSearchData {
 				'tamazight — standard marokkansk tamazight' => 'zgh',
 				'tamazight — standard marrokansk tamazight' => 'zgh',
 				'tamazight — standard moroccan tamazight' => 'zgh',
+				'tamazight — standard moroccan tamazight (latin script)' => 'zgh-latn',
 				'tamazight — standard tamazight' => 'zgh',
 				'tamazight — standardny marokkański tamazight' => 'zgh',
 				'tamazight — standardowy marokański tamazight' => 'zgh',
@@ -32935,6 +32966,7 @@ class LanguageNameSearchData {
 				'tetum — èdè tetum' => 'tet',
 				'teu — basa teu dikenal' => 'und',
 				'thai — northern thai' => 'nod',
+				'thai — northern thai (thai script)' => 'nod-thai',
 				'theas — sáimis theas' => 'sma',
 				'thjeshtuar — kinezishte e thjeshtuar' => 'zh-hans',
 				'thoir — craís an deiscirt thoir' => 'crj',
@@ -33484,6 +33516,7 @@ class LanguageNameSearchData {
 				'uolof' => 'wo',
 				'uolofe' => 'wo',
 				'uolofisht' => 'wo',
+				'upper guinea crioulo' => 'pov',
 				'upper saxon' => 'sxu',
 				'upper sorbian' => 'hsb',
 				'uqarit' => 'uga',
@@ -34242,6 +34275,7 @@ class LanguageNameSearchData {
 				'walesi' => 'cy',
 				'walesiska' => 'cy',
 				'waleština' => 'cy',
+				'wali' => 'wlx',
 				'waliesch' => 'cy',
 				'waliisk' => 'cy',
 				'walijski' => 'cy',

tests/phpunit/LanguageSearchTest.php

@@ -102,6 +102,21 @@ class LanguageSearchTest extends PHPUnit\Framework\TestCase {
 				'ka' => 'qartuli',
 			]
 			],
+			[ 'tonga', [
+				'to' => 'tonga',
+				'toi' => 'tonga (botatwe)',
+				'tog' => 'tonga (niasa)',
+				'ts' => 'tsonga',
+				'nr' => 'tonga — enetepēra ki te tonga',
+				'hax' => 'tonga — haira ki te tonga',
+				'st' => 'tonga — hōto ki te tonga',
+				'es-419' => 'tonga — pāniora amerikana ki te tonga',
+				'slh' => 'tonga — ratūti ki te tonga',
+				'tce' => 'tonga — tatōne ki te tonga',
+				'alt' => 'tonga — ātai ki te tonga',
+				'crj' => 'tonga-mā-rāwhiti — kirī tonga-mā-rāwhiti',
+			]
+			],
 			[ 'valencia', [
 				'ca' => 'valencia',
 			]