Use BCP47 compliant format for locale representation

chrome/browser/android/preferences/pref_service_bridge.cc

 // Copyright 2014 The Chromium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 #include "chrome/browser/android/preferences/pref_service_bridge.h"
 
 #include <jni.h>
 #include <stddef.h>
 
 #include <memory>
@@ skipping 1129 matching lines @@
 
 // static
 bool PrefServiceBridge::RegisterPrefServiceBridge(JNIEnv* env) {
   return RegisterNativesImpl(env);
 }
 
 // static
 // 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) {
   std::vector<std::string> locale_list =
       base::SplitString(locales + "," + *accept_languages, ",",
                         base::TRIM_WHITESPACE, base::SPLIT_WANT_NONEMPTY);
 
   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] == '-') {

[ksk1, 2016/10/12 05:52:50]

Please have length check before accessing elements.

[Yirui Huang, 2016/10/12 09:02:04]

Please see patch 6 for another implementing method.

+      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.
     if (lang_code == "iw")
       lang_code = "he";
     else if (lang_code == "ji")
       lang_code = "yi";
     else if (lang_code == "in")
       lang_code = "id";

[ksk1, 2016/10/12 05:52:50]

Is it OK not to have tl -> fil conversion?

[Yirui Huang, 2016/10/12 09:02:04]

Done in patch 6 :) 

 
     std::string language_tag(lang_code + "-" + country_code);
 
     if (seen_tags.find(language_tag) != seen_tags.end())
       continue;
+    if (!country_code.empty())
+      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
   // code. A language code should only be inserted after the last
   // languageTag that contains that language.
   // This will work with the IDS_ACCEPT_LANGUAGE localized strings bundled
   // with Chrome but may fail on arbitrary lists of language tags due to
   // differences in case and whitespace.
   std::set<std::string> seen_languages;
   std::vector<std::string> output_list;
   for (auto it = unique_locale_list.rbegin(); it != unique_locale_list.rend();
        ++it) {
     if (seen_languages.find(it->first) == seen_languages.end()) {
       output_list.push_back(it->first);
       seen_languages.insert(it->first);
     }
-    output_list.push_back(it->first + "-" + it->second);
+    if (!it->second.empty())
+      output_list.push_back(it->first + "-" + it->second);
   }
 
   std::reverse(output_list.begin(), output_list.end());
   *accept_languages = base::JoinString(output_list, ",");
 }
 
 // static
 std::string PrefServiceBridge::GetAndroidPermissionForContentSetting(
     ContentSettingsType content_type) {
   JNIEnv* env = AttachCurrentThread();
   base::android::ScopedJavaLocalRef<jstring> android_permission =
       Java_PrefServiceBridge_getAndroidPermissionForContentSetting(
           env, content_type);
   if (android_permission.is_null())
     return std::string();
 
   return ConvertJavaStringToUTF8(android_permission);
 }
 
 static void SetSupervisedUserId(JNIEnv* env,
                                 const JavaParamRef<jobject>& obj,
                                 const JavaParamRef<jstring>& pref) {
   GetPrefService()->SetString(prefs::kSupervisedUserId,
                               ConvertJavaStringToUTF8(env, pref));
 }