Use BCP47 compliant format for locale representation

chrome/browser/android/preferences/pref_service_bridge.cc

Index: chrome/browser/android/preferences/pref_service_bridge.cc
diff --git a/chrome/browser/android/preferences/pref_service_bridge.cc b/chrome/browser/android/preferences/pref_service_bridge.cc
index 4a3de8f5e1d8bb5088ed41c613fbdc95f601ea8a..99284ba07425c50d1fd76c466c685a8188bd8d2b 100644
--- a/chrome/browser/android/preferences/pref_service_bridge.cc
+++ b/chrome/browser/android/preferences/pref_service_bridge.cc
@@ -1147,10 +1147,9 @@ bool PrefServiceBridge::RegisterPrefServiceBridge(JNIEnv* env) {
 // This logic should be kept in sync with prependToAcceptLanguagesIfNecessary in
 // chrome/android/java/src/org/chromium/chrome/browser/
 //     physicalweb/PwsClientImpl.java
-// Input |locales| is a comma separated locale representaion. Each locale
-// representation should be xx_XX style, where xx is a 2-letter
-// ISO 639-1 compliant language code and XX is a 2-letter
-// ISO 3166-1 compliant country code.
+// Input |locales| is a comma separated locale representation that consists of
+// language tags(BCP47 compliant format). Each language tag contains
+// a language code and a country code or a language code only.
 void PrefServiceBridge::PrependToAcceptLanguagesIfNecessary(
     const std::string& locales,
     std::string* accept_languages) {
@@ -1161,14 +1160,21 @@ void PrefServiceBridge::PrependToAcceptLanguagesIfNecessary(
   std::set<std::string> seen_tags;
   std::vector<std::pair<std::string, std::string>> unique_locale_list;
   for (const std::string& locale_str : locale_list) {
-    // TODO(yirui): Support BCP47 compliant format including 3-letter
-    // country code, '-' separator and missing country case.
-    if (locale_str.size() != 5u ||
-        (locale_str[2] != '_' && locale_str[2] != '-'))
-      continue;  // Skip not well formed locale.
-
-    std::string lang_code(locale_str.substr(0, 2));
-    std::string country_code(locale_str.substr(3, 2));
+    std::string lang_code;
+    std::string country_code;
+    if (locale_str.size() == 2u || locale_str.size() == 3u) {
+      // Language tag consists of language only
+      lang_code = locale_str;
+      country_code = "";
+    } else if (locale_str[2] == '-') {
+      lang_code = locale_str.substr(0, 2);
+      country_code = locale_str.substr(3);
+    } else if (locale_str[3] == '-') {
+      lang_code = locale_str.substr(0, 3);
+      country_code = locale_str.substr(4);
+    } else {
+      continue;
+    }
 
     // Java mostly follows ISO-639-1 and ICU, except for the following three.
     // See documentation on java.util.Locale constructor for more.
@@ -1183,8 +1189,9 @@ void PrefServiceBridge::PrependToAcceptLanguagesIfNecessary(
 
     if (seen_tags.find(language_tag) != seen_tags.end())
       continue;
+    if (country_code != "")

[Seigo Nonaka, 2016/10/11 11:29:28]

Please use !country_code.empty() here.

[Yirui Huang, 2016/10/11 12:17:32]

Done.

+      seen_tags.insert(language_tag);
     unique_locale_list.push_back(std::make_pair(lang_code, country_code));
-    seen_tags.insert(language_tag);
   }
 
   // If language is not in the accept languages list, also add language
@@ -1201,7 +1208,8 @@ void PrefServiceBridge::PrependToAcceptLanguagesIfNecessary(
       output_list.push_back(it->first);
       seen_languages.insert(it->first);
     }
-    output_list.push_back(it->first + "-" + it->second);
+    if (it->second != "")

[Seigo Nonaka, 2016/10/11 11:29:28]

ditto.

[Yirui Huang, 2016/10/11 12:17:32]

Done.

+      output_list.push_back(it->first + "-" + it->second);
   }
 
   std::reverse(output_list.begin(), output_list.end());