bryce/mediawiki-extensions-UniversalLanguageSelector
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge "LanguageNameIndexer: Remove directionality chars that cannot be typed"

Browse Source
This commit is contained in:
Nikerabbit
2016-08-08 13:18:08 +00:00
committed by Gerrit Code Review
parent f8c94b2cb4 42f4f9650b
commit 760fe1ad39
2 changed files with 26 additions and 24 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
data/LanguageNameIndexer.php
View File

@@ -38,6 +38,9 @@ class LanguageNameIndexer extends Maintenance {
$translations = LanguageNames::getNames( $sourceLanguage, 0, 2 ); $translations = LanguageNames::getNames( $sourceLanguage, 0, 2 );
foreach ( $translations as $targetLanguage => $translation ) { foreach ( $translations as $targetLanguage => $translation ) {
$translation = mb_strtolower( $translation ); $translation = mb_strtolower( $translation );
// Remove directionality markers used in Names.php: users are not
// going to type these.
$translation = str_replace( "\xE2\x80\x8E", '', $translation );
$bucket = LanguageNameSearch::getIndex( $translation ); $bucket = LanguageNameSearch::getIndex( $translation );
$buckets[$bucket][$translation] = $targetLanguage; $buckets[$bucket][$translation] = $targetLanguage;
} }

47
data/LanguageNameSearchData.php
View File

@@ -5,11 +5,11 @@ class LanguageNameSearchData {
113 => [ 113 => [
'qafar' => 'aa', 'qafar' => 'aa',
'qırımtatarca' => 'crh', 'qırımtatarca' => 'crh',
'qırımtatarca (latin)‎' => 'crh-latn', 'qırımtatarca (latin)' => 'crh-latn',
'qaraqalpaqsha' => 'kaa', 'qaraqalpaqsha' => 'kaa',
'qabardjajəbza' => 'kbd-latn', 'qabardjajəbza' => 'kbd-latn',
'qazaqşa (latın)‎' => 'kk-latn', 'qazaqşa (latın)' => 'kk-latn',
'qazaqşa (türkïya)‎' => 'kk-tr', 'qazaqşa (türkïya)' => 'kk-tr',
'qafár af' => 'aa', 'qafár af' => 'aa',
'quechua' => 'qu', 'quechua' => 'qu',
'quichua del altiplanu de chimborazo' => 'qug', 'quichua del altiplanu de chimborazo' => 'qug',
@@ -2665,7 +2665,7 @@ class LanguageNameSearchData {
'kinaray-a' => 'krj', 'kinaray-a' => 'krj',
'karjala' => 'krl', 'karjala' => 'krl',
'kurdî' => 'ku', 'kurdî' => 'ku',
'kurdî (latînî)‎' => 'ku-latn', 'kurdî (latînî)' => 'ku-latn',
'kernowek' => 'kw', 'kernowek' => 'kw',
'kapampangan' => 'pam', 'kapampangan' => 'pam',
'kreol réyoné' => 'rcf', 'kreol réyoné' => 'rcf',
@@ -9063,7 +9063,7 @@ class LanguageNameSearchData {
73 => [ 73 => [
'башҡортса' => 'ba', 'башҡортса' => 'ba',
'беларуская' => 'be', 'беларуская' => 'be',
'беларуская (тарашкевіца)‎' => 'be-x-old', 'беларуская (тарашкевіца)' => 'be-x-old',
'български' => 'bg', 'български' => 'bg',
'буряад' => 'bxr', 'буряад' => 'bxr',
'башкірская' => 'ba', 'башкірская' => 'ba',
@@ -9080,7 +9080,6 @@ class LanguageNameSearchData {
'бода' => 'brx', 'бода' => 'brx',
'бурацкая' => 'bxr', 'бурацкая' => 'bxr',
'баварская' => 'bar', 'баварская' => 'bar',
'беларуская (тарашкевіца)' => 'be-x-old',
'баўгарская' => 'bg', 'баўгарская' => 'bg',
'бэнгальская' => 'bn', 'бэнгальская' => 'bn',
'басьнійская' => 'bs', 'басьнійская' => 'bs',
@@ -13480,7 +13479,7 @@ class LanguageNameSearchData {
'niuē' => 'niu', 'niuē' => 'niu',
'nederlands' => 'nl', 'nederlands' => 'nl',
'nl-be' => 'nl-be', 'nl-be' => 'nl-be',
'nederlands (informeel)‎' => 'nl-informal', 'nederlands (informeel)' => 'nl-informal',
'norsk nynorsk' => 'nn', 'norsk nynorsk' => 'nn',
'novial' => 'nov', 'novial' => 'nov',
'nouormand' => 'nrm', 'nouormand' => 'nrm',
@@ -14187,7 +14186,7 @@ class LanguageNameSearchData {
'ngôn ngữ không xác định' => 'und', 'ngôn ngữ không xác định' => 'und',
], ],
82 => [ 82 => [
'къырымтатарджа (кирилл)‎' => 'crh-cyrl', 'къырымтатарджа (кирилл)' => 'crh-cyrl',
'къарачай-малкъар' => 'krc', 'къарачай-малкъар' => 'krc',
'коми' => 'kv', 'коми' => 'kv',
'кыргызча' => 'ky', 'кыргызча' => 'ky',
@@ -14662,7 +14661,7 @@ class LanguageNameSearchData {
'словѣньскъ / ⰔⰎⰑⰂⰡⰐⰠⰔⰍⰟ' => 'cu', 'словѣньскъ / ⰔⰎⰑⰂⰡⰐⰠⰔⰍⰟ' => 'cu',
'саха тыла' => 'sah', 'саха тыла' => 'sah',
'српски / srpski' => 'sr', 'српски / srpski' => 'sr',
'српски (ћирилица)‎' => 'sr-ec', 'српски (ћирилица)' => 'sr-ec',
'стараанглійская' => 'ang', 'стараанглійская' => 'ang',
'старагрэцкая' => 'grc', 'старагрэцкая' => 'grc',
'сычуань і' => 'ii', 'сычуань і' => 'ii',
@@ -15077,7 +15076,7 @@ class LanguageNameSearchData {
100 => [ 100 => [
'dansk' => 'da', 'dansk' => 'da',
'deutsch' => 'de', 'deutsch' => 'de',
'deutsch (sie-form)‎' => 'de-formal', 'deutsch (sie-form)' => 'de-formal',
'dolnoserbski' => 'dsb', 'dolnoserbski' => 'dsb',
'dusun bundu-liwan' => 'dtp', 'dusun bundu-liwan' => 'dtp',
'dari-e mazdeyasnā' => 'gbz', 'dari-e mazdeyasnā' => 'gbz',
@@ -15445,7 +15444,7 @@ class LanguageNameSearchData {
'salırça' => 'slr', 'salırça' => 'slr',
'soomaaliga' => 'so', 'soomaaliga' => 'so',
'shqip' => 'sq', 'shqip' => 'sq',
'srpski (latinica)‎' => 'sr-el', 'srpski (latinica)' => 'sr-el',
'sranantongo' => 'srn', 'sranantongo' => 'srn',
'siswati' => 'ss', 'siswati' => 'ss',
'sesotho' => 'st', 'sesotho' => 'st',
@@ -19101,7 +19100,7 @@ class LanguageNameSearchData {
], ],
123 => [ 123 => [
'贛語' => 'gan', '贛語' => 'gan',
'贛語（繁體）‎' => 'gan-hant', '贛語（繁體）' => 'gan-hant',
'ᐃᓄᒃᑎᑐᑦ/inuktitut' => 'ike', 'ᐃᓄᒃᑎᑐᑦ/inuktitut' => 'ike',
'ᐃᓄᒃᑎᑐᑦ' => 'ike-cans', 'ᐃᓄᒃᑎᑐᑦ' => 'ike-cans',
'ရခိုင်' => 'rki', 'ရခိုင်' => 'rki',
@@ -19120,7 +19119,7 @@ class LanguageNameSearchData {
'ရွမ်ဘို' => 'rof', 'ရွမ်ဘို' => 'rof',
], ],
195 => [ 195 => [
'赣语（简体）‎' => 'gan-hans', '赣语（简体）' => 'gan-hans',
'広東語' => 'zh-yue', '広東語' => 'zh-yue',
], ],
711 => [ 711 => [
@@ -21586,8 +21585,8 @@ class LanguageNameSearchData {
], ],
179 => [ 179 => [
'қазақша' => 'kk', 'қазақша' => 'kk',
'қазақша (кирил)‎' => 'kk-cyrl', 'қазақша (кирил)' => 'kk-cyrl',
'қазақша (қазақстан)‎' => 'kk-kz', 'қазақша (қазақстан)' => 'kk-kz',
'қазақ тілі' => 'kk', 'қазақ тілі' => 'kk',
'қырғыз тілі' => 'ky', 'қырғыз тілі' => 'ky',
'қытай тілі' => 'zh', 'қытай тілі' => 'zh',
@@ -22625,7 +22624,7 @@ class LanguageNameSearchData {
'剛果文' => 'kg', '剛果文' => 'kg',
], ],
604 => [ 604 => [
'لەکی‎' => 'lki', 'لەکی' => 'lki',
'لۊری شومالی' => 'lrc', 'لۊری شومالی' => 'lrc',
'لئری دوٙمینی' => 'luz', 'لئری دوٙمینی' => 'luz',
'لغة تتار القرم' => 'crh', 'لغة تتار القرم' => 'crh',
@@ -25190,14 +25189,14 @@ class LanguageNameSearchData {
], ],
13 => [ 13 => [
'中文' => 'zh', '中文' => 'zh',
'中文（中国大陆）‎' => 'zh-cn', '中文（中国大陆）' => 'zh-cn',
'中文（简体）‎' => 'zh-hans', '中文（简体）' => 'zh-hans',
'中文（繁體）‎' => 'zh-hant', '中文（繁體）' => 'zh-hant',
'中文（香港）‎' => 'zh-hk', '中文（香港）' => 'zh-hk',
'中文（澳門）‎' => 'zh-mo', '中文（澳門）' => 'zh-mo',
'中文（马来西亚）‎' => 'zh-my', '中文（马来西亚）' => 'zh-my',
'中文（新加坡）‎' => 'zh-sg', '中文（新加坡）' => 'zh-sg',
'中文（台灣）‎' => 'zh-tw', '中文（台灣）' => 'zh-tw',
'中央ドゥスン語' => 'dtp', '中央ドゥスン語' => 'dtp',
'中央アラスカ・ユピック語' => 'esu', '中央アラスカ・ユピック語' => 'esu',
'中央アトラス・タマジクト語' => 'tzm', '中央アトラス・タマジクト語' => 'tzm',