VPN-6622: Fix broken device language for Chinese (#9933) (#9968)

* VPN-6622 fix for iOS

* make it generalizable to help with Android

* make it generalizable to help with Android

* better iOS explanation

* make linter happy

* update tests, improve logic for non-iOS

src/localizer.cpp

@@ -77,6 +77,14 @@ QList<QPair<QString, QString>> Localizer::parseBCP47Languages(
  continue;
  }
 
+ // Chinese is often reported as `zh-Hant-HK`, and we want the middle
+ // section.
+ QString script = parts[1];
+ if (script == "Hans" || script == "Hant") {
+ codes.append(QPair<QString, QString>{parts[0], script});
+ continue;
+ }
+
  QString country = parts.last();
  if (country.length() == 2) {
  codes.append(QPair<QString, QString>{parts[0], country});
@@ -96,11 +104,23 @@ QList<QPair<QString, QString>> Localizer::parseIOSLanguages(
  QList<QPair<QString, QString>> codes;
 
  for (const QString& language : languages) {
- // By documentation, the language code should be in the format
- // <language>-<script>_<country>. And we don't care about the script.
+ // The language code comes in the format <language>-<country> or
+ // <language>-<script>-<country>.
+ // For an iOS device set to these 6 languages: Chinese Traditional (Macao),
+ // Chinese Traditional, Chinese Simplified, English, Mexican Spanish,
+ // Spanish
+ // ...we get these language codes:
+ // [zh-Hant-MO,zh-Hant-US,zh-Hans-US,en-US,es-MX,es-US]
+ // iOS shows the 2 part versions for nearly all languages, and shows the
+ // 3 part version only when there is a script - like for Chinese variants.
+ // Thus, we can pull the second chunk of the locale string in all cases to
+ // get the translations flow required by our app. That second chunk gives
+ // the country for all languages except Chinese, where it gives the script -
+ // and in both these situations, this is what we want (script for Chinese,
+ // country for all others).
  QString countryCode;
 
- QStringList parts = language.split('_');
+ QStringList parts = language.split('-');
  if (parts.length() > 1) {
  countryCode = parts[1];
  }
@@ -125,8 +145,27 @@ QString Localizer::systemLanguageCode() const {
 #endif
 
  for (const QPair<QString, QString>& language : uiLanguages) {
- QString selectedLanguage =
- findLanguageCode(language.first, language.second);
+ QString languagePart = language.first;
+ QString localePart = language.second;
+
+ // iOS reports Chinese as "zh-Hans" and "zh-Hant". Android reports
+ // them as "zh-CN" and "zh-Hans". However the app uses "zh-CN" and
+ // "zh-TW". We need to manually adjust these. More info:
+ // https://stackoverflow.com/questions/4892372/language-codes-for-simplified-chinese-and-traditional-chinese
+ // (For country-specific variants, it reports it as "zh-Hant-MO", etc. Since
+ // the app only looks at the strings before the 1st and 2nd dashes, we
+ // ignore the country - which is what we want to do, as we don't have
+ // translations for Chinese beyond simplified and traditional.)
+ if (languagePart == "zh") {
+ if (localePart == "Hans") {
+ localePart = "CN";
+ }
+ if (localePart == "Hant") {
+ localePart = "TW";
+ }
+ }
+
+ QString selectedLanguage = findLanguageCode(languagePart, localePart);
  if (!selectedLanguage.isEmpty()) {
  return selectedLanguage;
  }

tests/unit_tests/testlocalizer.cpp

@@ -158,17 +158,17 @@ void TestLocalizer::parseIOSLanguages_data() {
  {
  LanguageList a;
  a.append(QPair<QString, QString>{"aa", "bb"});
- QTest::addRow("language_country") << QStringList("aa_bb") << a;
+ QTest::addRow("language-country") << QStringList("aa-bb") << a;
  }
  {
  LanguageList a;
- a.append(QPair<QString, QString>{"aa", ""});
+ a.append(QPair<QString, QString>{"aa", "cc"});
  QTest::addRow("language-script") << QStringList("aa-cc") << a;
  }
  {
  LanguageList a;
- a.append(QPair<QString, QString>{"aa", "bb"});
- QTest::addRow("language-script_country") << QStringList("aa-cc_bb") << a;
+ a.append(QPair<QString, QString>{"aa", "cc"});
+ QTest::addRow("language-script-country") << QStringList("aa-cc-bb") << a;
  }
 }