feat(intl): ignore script subtags when canonicalizing locale strings

pkgs/intl/CHANGELOG.md

@@ -2,6 +2,9 @@
  * Updated the Turkish Lira (TRY) currency symbol in `simpleCurrencySymbols`  
    from "TL" to "₺" (U+20BA). This ensures accuracy and alignment with the  
    official symbol introduced in 2012.
+ * Fix parsing of locale strings containing script codes in `verifiedLocale`.
+   For example, `zh-Hans-CN` would have been previously parsed as `zh`, but is
+   now parsed as `zh_CN`.
 
 ## 0.20.2
  * Remove the dependency on `package:http`.

pkgs/intl/lib/intl.dart

@@ -208,7 +208,8 @@ class Intl {
       helpers.verifiedLocale(newLocale, localeExists, onFailure);
 
   /// Return the short version of a locale name, e.g. 'en_US' => 'en'
-  static String shortLocale(String aLocale) => helpers.shortLocale(aLocale);
+  static String shortLocale(String aLocale) =>
+      helpers.languageOnlyLocale(aLocale);
 
   /// Return the name [aLocale] turned into xx_YY where it might possibly be
   /// in the wrong case or with a hyphen instead of an underscore. If

pkgs/intl/lib/src/intl_helpers.dart

@@ -128,10 +128,10 @@ String? computeMessageName(String? name, String? text, String? meaning) {
   return meaning == null ? text : '${text}_$meaning';
 }
 
-/// Returns an index of a separator between language and region.
+/// Returns an index of a separator between language and other subtags.
 ///
 /// Assumes that language length can be only 2 or 3.
-int _separatorIndex(String locale) {
+int _languageSeparatorIndex(String locale) {
   if (locale.length < 3) {
     return -1;
   }
@@ -147,6 +147,19 @@ int _separatorIndex(String locale) {
   return -1;
 }
 
+/// Returns an index of a separator between script and region.
+///
+/// Assumes that script contains exactly 4 characters.
+int _scriptSeparatorIndex(String region) {
+  if (region.length < 5) {
+    return -1;
+  }
+  if (region[4] == '-' || region[4] == '_') {
+    return 4;
+  }
+  return -1;
+}
+
 String canonicalizedLocale(String? aLocale) {
 // Locales of length < 5 are presumably two-letter forms, or else malformed.
 // We return them unmodified and if correct they will be found.
@@ -159,7 +172,7 @@ String canonicalizedLocale(String? aLocale) {
   if (aLocale == 'C') return 'en_ISO';
   if (aLocale.length < 5) return aLocale;
 
-  var separatorIndex = _separatorIndex(aLocale);
+  var separatorIndex = _languageSeparatorIndex(aLocale);
   if (separatorIndex == -1) {
     return aLocale;
   }
@@ -186,9 +199,10 @@ String? verifiedLocale(String? newLocale, bool Function(String) localeExists,
   }
   final fallbackOptions = [
     canonicalizedLocale,
-    shortLocale,
+    languageRegionOnlyLocale,
+    languageOnlyLocale,
     deprecatedLocale,
-    (locale) => deprecatedLocale(shortLocale(locale)),
+    (locale) => deprecatedLocale(languageOnlyLocale(locale)),
     (locale) => deprecatedLocale(canonicalizedLocale(locale)),
     (_) => 'fallback'
   ];
@@ -233,15 +247,15 @@ String deprecatedLocale(String aLocale) {
 }
 
 /// Return the short version of a locale name, e.g. 'en_US' => 'en'
-String shortLocale(String aLocale) {
+String languageOnlyLocale(String aLocale) {
   // TODO(b/241094372): Remove this check.
   if (aLocale == 'invalid') {
     return 'in';
   }
   if (aLocale.length < 2) {
     return aLocale;
   }
-  var separatorIndex = _separatorIndex(aLocale);
+  var separatorIndex = _languageSeparatorIndex(aLocale);
   if (separatorIndex == -1) {
     if (aLocale.length < 4) {
       // aLocale is already only a language code.
@@ -253,3 +267,19 @@ String shortLocale(String aLocale) {
   }
   return aLocale.substring(0, separatorIndex).toLowerCase();
 }
+
+String languageRegionOnlyLocale(String aLocale) {
+  if (aLocale.length < 10) return aLocale;
+
+  var separatorIndex = _languageSeparatorIndex(aLocale);
+  if (separatorIndex == -1) {
+    return aLocale;
+  }
+  var language = aLocale.substring(0, separatorIndex);
+  var subtags = aLocale.substring(separatorIndex + 1);
+  separatorIndex = _scriptSeparatorIndex(subtags);
+  var region = subtags.substring(separatorIndex + 1);
+  // If it's longer than three it's something odd, so don't touch it.
+  if (region.length <= 3) region = region.toUpperCase();
+  return '${language}_$region';
+}

pkgs/intl/test/intl_test.dart

@@ -51,6 +51,10 @@ void main() {
     expect(Intl.verifiedLocale('en-ZZ', NumberFormat.localeExists), 'en');
     expect(Intl.verifiedLocale('es-999', NumberFormat.localeExists), 'es');
     expect(Intl.verifiedLocale('gsw-CH', NumberFormat.localeExists), 'gsw');
+    expect(
+      Intl.verifiedLocale('zh-Hant-CN', NumberFormat.localeExists),
+      'zh_CN',
+    );
 
     void checkAsNumberDefault(String locale, String expected) {
       var oldDefault = Intl.defaultLocale;
@@ -74,6 +78,7 @@ void main() {
     expect(Intl.verifiedLocale('en-ZZ', DateFormat.localeExists), 'en');
     expect(Intl.verifiedLocale('es-999', DateFormat.localeExists), 'es');
     expect(Intl.verifiedLocale('gsw-CH', DateFormat.localeExists), 'gsw');
+    expect(Intl.verifiedLocale('zh-Hant-CN', DateFormat.localeExists), 'zh_CN');
     // TODO(b/241094372): Remove this check.
     expect(Intl.verifiedLocale('invalid', DateFormat.localeExists), 'in');
 

pkgs/intl/test/number_format_test_core.dart

@@ -650,6 +650,17 @@ void runTests(Map<String, num> allTestNumbers) {
       });
     }
   });
+
+  test('Use script', () {
+    expect(
+      NumberFormat.currency(locale: 'zh_Hant_TW').format(12),
+      'TWD12.00',
+    );
+    expect(
+      NumberFormat.currency(locale: 'zh_Hant_CN').format(12),
+      'CNY12.00',
+    );
+  });
 }
 
 String stripExtras(String input) {