Fix Accept-Language detection

Bug: ether#6129
Follows-Up: ether#3882
Change-Id: Ie3fcec84948fde36fa12448d5eabaf05f79f1283

src/static/js/vendors/html10n.js

@@ -181,21 +181,60 @@ window.html10n = (function(window, document, undefined) {
       return
     }
 
+    // Issue #6129: Fix exceptions caused by browsers
+    // Also for fallback, see BCP 47 RFC 4647 section 3.4
+    function getBcp47LangTag(browserLang) {
+      var lang = browserLang.toLowerCase();
+      // Browser => BCP 47
+      var langCodeMap = {
+        'zh-cn': 'zh-Hans-CN',
+        'zh-hk': 'zh-Hant-HK',
+        'zh-mo': 'zh-Hant-MO',
+        'zh-my': 'zh-Hans-MY',
+        'zh-sg': 'zh-Hans-SG',
+        'zh-tw': 'zh-Hant-TW',
+      };
+
+      return langCodeMap[lang] ?? browserLang;
+    }
+
+    // Issue #6129: Fix exceptions
+    function getJsonLangCode(bcp47Lang) {
+      var lang = bcp47Lang.toLowerCase();
+      // BCP 47 => JSON
+      var langCodeMap = {
+        'sr-cyrl': 'sr-ec',
+        'sr-latn': 'sr-el',
+        'zh-hant-hk': 'zh-hk',
+      };
+
+      return langCodeMap[lang] ?? lang;
+    }
+
+    // translatewiki.net use all lowercase form by default ('en-gb' insted of 'en-GB')
+    bcp47LangTag = getBcp47LangTag(lang).toLowerCase();
+
     // Check if lang exists
-    if (!data[lang]) {
+    if (!data[getJsonLangCode(bcp47LangTag)]) {
       // lang not found
       // This may be due to formatting (expected 'ru' but browser sent 'ru-RU')
       // Set err msg before mutating lang (we may need this later)
-      var msg = 'Couldn\'t find translations for ' + lang;
-
-      // Check for '-' ('ROOT-VARIANT')
-      if (lang.indexOf('-') > -1) {
-        // ROOT-VARIANT formatting detected
-        lang = lang.split('-')[0]; // set lang to ROOT lang
+      var msg = 'Couldn\'t find translations for ' + lang +
+        '(BCP 47 lang tag ' + bcp47LangTag +
+        ', JSON lang code ' + getJsonLangCode(bcp47LangTag) + ')';
+
+      // Check for '-' (BCP 47 'ROOT-SCRIPT-REGION-VARIANT') and fallback until found data or ROOT
+      // - 'ROOT-SCRIPT-REGION': 'zh-Hans-CN'
+      // - 'ROOT-SCRIPT': 'zh-Hans'
+      // - 'ROOT-REGION': 'en-GB'
+      // - 'ROOT-VARIANT': 'be-tarask'
+      while (!data[getJsonLangCode(lang)] && lang.lastIndexOf('-') > -1) {
+        // ROOT-SCRIPT-REGION-VARIANT formatting detected
+        lang = lang.substring(0, lang.lastIndexOf('-')); // set lang to ROOT lang
       }
 
-      // Check if ROOT lang exists (e.g 'ru')
-      if (!data[lang]) {
+      // Check if already found data or ROOT lang exists (e.g 'ru')
+      if (!data[getJsonLangCode(lang)]) {
         // ROOT lang not found. (e.g 'zh')
         // Loop through langs data. Maybe we have a variant? e.g (zh-hans)
         var l; // langs item. Declare outside of loop
@@ -204,8 +243,8 @@ window.html10n = (function(window, document, undefined) {
           // Is not ROOT?
           // And index of ROOT equals 0?
           // And is known lang?
-          if (lang != l && l.indexOf(lang) === 0 && data[l]) {
-            lang = l; // set lang to ROOT-VARIANT (e.g 'zh-hans')
+          if (lang != l && l.indexOf(lang) === 0 && data[getJsonLangCode(l)]) {
+            lang = l; // set lang to ROOT-SCRIPT (e.g 'zh-hans')
             break;
           }
         }