Add more flexible handling of what is considered a google home page.

BUG=http:b/5609053
TEST=See unit tests for valid and invalid URLs.

Review URL: http://codereview.chromium.org/8728004

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@112726 0039d316-1c4b-4281-b951-d872f2087c98

chrome/browser/chrome_browser_main.cc

@@ -1739,8 +1739,7 @@ int ChromeBrowserMainParts::PreMainMessageLoopRunImpl() {
   PrefService* pref_service = profile_->GetPrefs();
   if (pref_service) {
     std::string homepage = pref_service->GetString(prefs::kHomePage);
-    google_search_homepage =
-        homepage == GoogleURLTracker::kDefaultGoogleHomepage;
+    google_search_homepage = google_util::IsGoogleHomePageUrl(homepage);
   }
 
   RLZTracker::InitRlzDelayed(is_first_run_, master_prefs_->ping_delay,

chrome/browser/google/google_util.cc

@@ -128,6 +128,43 @@ bool GetReactivationBrand(std::string* brand) {
 
 #endif
 
+bool IsGoogleHomePageUrl(const std::string& url) {
+  GURL original_url(url);
+  if (!original_url.is_valid())
+    return false;
+
+  // Make sure the scheme is valid.
+  if (!original_url.SchemeIs("http") && !original_url.SchemeIs("https"))
+    return false;
+
+  // Make sure port is default for the respective scheme.
+  if (!original_url.port().empty())
+    return false;
+
+  // Accept only valid TLD.
+  size_t tld_length = net::RegistryControlledDomainService::GetRegistryLength(
+      original_url, false);
+  if (tld_length == 0 || tld_length == std::string::npos)
+    return false;
+
+  // We only accept "www.google." in front of the TLD.
+  std::string host = original_url.host();
+  host = host.substr(0, host.length() - tld_length);
+  if (!LowerCaseEqualsASCII(host, "www.google.") &&
+      !LowerCaseEqualsASCII(host, "google."))
+    return false;
+
+  // Make sure the path is a known home page path.
+  std::string path(original_url.path());
+  if (!LowerCaseEqualsASCII(path, "/") &&
+      !LowerCaseEqualsASCII(path, "/webhp") &&
+      !StartsWithASCII(path, "/ig", false)) {
+    return false;
+  }
+
+  return true;
+}
+
 bool IsOrganic(const std::string& brand) {
   const CommandLine& command_line = *CommandLine::ForCurrentProcess();
   if (command_line.HasSwitch(switches::kOrganicInstall))

chrome/browser/google/google_util.h

@@ -38,6 +38,9 @@ bool GetBrand(std::string* brand);
 // install. Returns false if the information is not available.
 bool GetReactivationBrand(std::string* brand);
 
+// True if |url| represents a valid Google home page URL.
+bool IsGoogleHomePageUrl(const std::string& url);
+
 // True if a build is strictly organic, according to its brand code.
 bool IsOrganic(const std::string& brand);
 

chrome/browser/google/google_util_unittest.cc

+// Copyright (c) 2011 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/google/google_url_tracker.h"
+#include "chrome/browser/google/google_util.h"
+#include "testing/gtest/include/gtest/gtest.h"
+
+using google_util::IsGoogleHomePageUrl;
+
+TEST(GoogleUtilTest, GoodHomePagesNonSecure) {
+  // Valid home page hosts.
+  EXPECT_TRUE(IsGoogleHomePageUrl(GoogleURLTracker::kDefaultGoogleHomepage));
+  EXPECT_TRUE(IsGoogleHomePageUrl("http://google.com"));
+  EXPECT_TRUE(IsGoogleHomePageUrl("http://www.google.com"));
+  EXPECT_TRUE(IsGoogleHomePageUrl("http://www.google.ca"));
+  EXPECT_TRUE(IsGoogleHomePageUrl("http://www.google.co.uk"));
+  EXPECT_TRUE(IsGoogleHomePageUrl("http://www.google.com:80/"));
+
+  // Only the paths /, /webhp, and /ig.* are valid.  Query parameters are
+  // ignored.
+  EXPECT_TRUE(IsGoogleHomePageUrl("http://www.google.com/"));
+  EXPECT_TRUE(IsGoogleHomePageUrl("http://www.google.com/webhp"));
+  EXPECT_TRUE(IsGoogleHomePageUrl("http://www.google.com/webhp?rlz=TEST"));
+  EXPECT_TRUE(IsGoogleHomePageUrl("http://www.google.com/ig"));
+  EXPECT_TRUE(IsGoogleHomePageUrl("http://www.google.com/ig/foo"));
+  EXPECT_TRUE(IsGoogleHomePageUrl("http://www.google.com/ig?rlz=TEST"));
+  EXPECT_TRUE(IsGoogleHomePageUrl("http://www.google.com/ig/foo?rlz=TEST"));
+}
+
+TEST(GoogleUtilTest, GoodHomePagesSecure) {
+  // Valid home page hosts.
+  EXPECT_TRUE(IsGoogleHomePageUrl("https://google.com"));
+  EXPECT_TRUE(IsGoogleHomePageUrl("https://www.google.com"));
+  EXPECT_TRUE(IsGoogleHomePageUrl("https://www.google.ca"));
+  EXPECT_TRUE(IsGoogleHomePageUrl("https://www.google.co.uk"));
+  EXPECT_TRUE(IsGoogleHomePageUrl("https://www.google.com:443/"));
+
+  // Only the paths /, /webhp, and /ig.* are valid.  Query parameters are
+  // ignored.
+  EXPECT_TRUE(IsGoogleHomePageUrl("https://www.google.com/"));
+  EXPECT_TRUE(IsGoogleHomePageUrl("https://www.google.com/webhp"));
+  EXPECT_TRUE(IsGoogleHomePageUrl("https://www.google.com/webhp?rlz=TEST"));
+  EXPECT_TRUE(IsGoogleHomePageUrl("https://www.google.com/ig"));
+  EXPECT_TRUE(IsGoogleHomePageUrl("https://www.google.com/ig/foo"));
+  EXPECT_TRUE(IsGoogleHomePageUrl("https://www.google.com/ig?rlz=TEST"));
+  EXPECT_TRUE(IsGoogleHomePageUrl("https://www.google.com/ig/foo?rlz=TEST"));
+}
+
+TEST(GoogleUtilTest, BadHomePages) {
+  EXPECT_FALSE(IsGoogleHomePageUrl(""));
+
+  // If specified, only the "www" subdomain is OK.
+  EXPECT_FALSE(IsGoogleHomePageUrl("http://maps.google.com"));
+  EXPECT_FALSE(IsGoogleHomePageUrl("http://foo.google.com"));
+
+  // No non-standard port numbers.
+  EXPECT_FALSE(IsGoogleHomePageUrl("http://www.google.com:1234"));
+  EXPECT_FALSE(IsGoogleHomePageUrl("https://www.google.com:5678"));
+
+  // Invalid TLDs.
+  EXPECT_FALSE(IsGoogleHomePageUrl("http://www.google.abc"));
+  EXPECT_FALSE(IsGoogleHomePageUrl("http://www.google.com.abc"));
+  EXPECT_FALSE(IsGoogleHomePageUrl("http://www.google.abc.com"));
+  EXPECT_FALSE(IsGoogleHomePageUrl("http://www.google.ab.cd"));
+  EXPECT_FALSE(IsGoogleHomePageUrl("http://www.google.uk.qq"));
+
+  // Must be http or https.
+  EXPECT_FALSE(IsGoogleHomePageUrl("ftp://www.google.com"));
+  EXPECT_FALSE(IsGoogleHomePageUrl("file://does/not/exist"));
+  EXPECT_FALSE(IsGoogleHomePageUrl("bad://www.google.com"));
+  EXPECT_FALSE(IsGoogleHomePageUrl("www.google.com"));
+
+  // Only the paths /, /webhp, and /ig.* are valid.
+  EXPECT_FALSE(IsGoogleHomePageUrl("http://www.google.com/abc"));
+  EXPECT_FALSE(IsGoogleHomePageUrl("http://www.google.com/webhpabc"));
+  EXPECT_FALSE(IsGoogleHomePageUrl("http://www.google.com/webhp/abc"));
+  EXPECT_FALSE(IsGoogleHomePageUrl("http://www.google.com/abcig"));
+  EXPECT_FALSE(IsGoogleHomePageUrl("http://www.google.com/webhp/ig"));
+}

chrome/chrome_tests.gypi

@@ -1426,6 +1426,7 @@
         'browser/global_keyboard_shortcuts_mac_unittest.mm',
         'browser/google/google_update_settings_unittest.cc',
         'browser/google/google_url_tracker_unittest.cc',
+        'browser/google/google_util_unittest.cc',
         'browser/history/expire_history_backend_unittest.cc',
         'browser/history/history_backend_unittest.cc',
         'browser/history/history_querying_unittest.cc',