Extend with a language code in http accept languages

net/http/http_util_icu.cc

Index: net/http/http_util_icu.cc
diff --git a/net/http/http_util_icu.cc b/net/http/http_util_icu.cc
new file mode 100644
index 0000000000000000000000000000000000000000..b63b616bd801ff1424d5fbaefd8f356b86726e72
--- /dev/null
+++ b/net/http/http_util_icu.cc
@@ -0,0 +1,84 @@
+// Copyright (c) 2016 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 "net/http/http_util_icu.h"
+
+#include <algorithm>
+#include <set>
+#include <vector>
+
+#include "base/logging.h"
+#include "base/strings/string_split.h"
+#include "base/strings/stringprintf.h"
+#include "third_party/icu/source/common/unicode/uloc.h"
+
+namespace net {
+
+// The input is a comma separated languages list, this function allows
+// whitespace between each languages. It will come from the preference and a
+// user does not manually edit the preference file. The accept languages that
+// customixed from developers are not processed here.

[Seigo Nonaka, 2016/12/12 07:29:26]

nit: /customixed/customized/
And it may be good to mention about Android WebView.

How about

Similarly, the accept languages on Android WebView can be overridden by
developers, but they are not processed here.

[Yirui Huang, 2016/12/12 07:41:35]

Modified with Android Webview. "+ including those from Android Webview,"

+std::string HttpUtilIcu::GenerateAcceptLanguageHeader(
+    const std::string& raw_language_list) {
+  std::vector<std::string> locales_list = base::SplitString(
+      raw_language_list, ",", base::TRIM_WHITESPACE, base::SPLIT_WANT_NONEMPTY);
+
+  // 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.
+  std::set<std::string> seen_languages;
+  std::vector<std::string> output_list;
+  for (auto it = locales_list.rbegin(); it != locales_list.rend(); ++it) {
+    char locale_ID[ULOC_FULLNAME_CAPACITY] = {};
+    char language_code[ULOC_LANG_CAPACITY] = {};
+
+    UErrorCode error = U_ZERO_ERROR;
+    std::string locale_string = *it;
+    uloc_forLanguageTag(locale_string.c_str(), locale_ID,
+                        ULOC_FULLNAME_CAPACITY, nullptr, &error);
+    if (U_FAILURE(error)) {
+      LOG(ERROR) << "Ignoring invalid locale representation " << locale_string;
+      continue;
+    }
+
+    error = U_ZERO_ERROR;
+    uloc_getLanguage(locale_ID, language_code, ULOC_LANG_CAPACITY, &error);
+    if (U_FAILURE(error)) {
+      LOG(ERROR) << "Ignoring invalid locale representation " << locale_string;
+      continue;
+    }
+
+    if (seen_languages.find(language_code) == seen_languages.end()) {
+      output_list.push_back(language_code);
+      seen_languages.insert(language_code);
+    }
+
+    if (language_code != *it)
+      output_list.push_back(locale_string);
+  }
+
+  std::reverse(output_list.begin(), output_list.end());
+
+  // We use integers for q-value and q-value decrement that are 10 times larger
+  // than actual values to avoid a problem with comparing two floating point
+  // numbers.
+  const unsigned int kQvalueDecrement10 = 2;
+  unsigned int qvalue10 = 10;
+  std::string lang_list_with_q;
+  for (unsigned i = 0; i < output_list.size(); i++) {
+    if (qvalue10 == 10) {
+      // q=1.0 is implicit.
+      lang_list_with_q = output_list[i];
+    } else {
+      DCHECK_LT(qvalue10, 10U);
+      base::StringAppendF(&lang_list_with_q, ",%s;q=0.%d",
+                          output_list[i].c_str(), qvalue10);
+    }
+    // It does not make sense to have 'q=0'.
+    if (qvalue10 > kQvalueDecrement10)
+      qvalue10 -= kQvalueDecrement10;
+  }
+  return lang_list_with_q;
+}
+}