Google search subdomains included for Safesearch

components/google/core/browser/google_util.cc

Index: components/google/core/browser/google_util.cc
diff --git a/components/google/core/browser/google_util.cc b/components/google/core/browser/google_util.cc
index a16d55eaafdd0daa4d89de8523ad121be7cc5f84..2c779ee13023b55bb4028b95bdca7e545518cf29 100644
--- a/components/google/core/browser/google_util.cc
+++ b/components/google/core/browser/google_util.cc
@@ -41,10 +41,19 @@ namespace {
 
 bool gUseMockLinkDoctorBaseURLForTesting = false;
 
+const char* const gGoogleSearchSubdomains[] = {"ipv4.google.com",

[Ilya Sherman, 2017/05/11 00:33:35]

nit: s/g/k, since this is a constant.

[Peter Kasting, 2017/05/11 00:48:08]

Also, prefer to define this locally in the function that uses it.

Could probably be constexpr as well.

[igorcov, 2017/05/11 11:30:05]

Included all this locally inside a static set.

+                                               "ipv6.google.com"};
+
 bool IsPathHomePageBase(base::StringPiece path) {
   return (path == "/") || (path == "/webhp");
 }
 
+// Removes a single trailing dot if present in |host|.
+void StripTrailingDot(base::StringPiece* host) {
+  if (host->ends_with("."))
+    host->remove_suffix(1);
+}
+
 // True if the given canonical |host| is "[www.]<domain_in_lower_case>.<TLD>"
 // with a valid TLD. If |subdomain_permission| is ALLOW_SUBDOMAIN, we check
 // against host "*.<domain_in_lower_case>.<TLD>" instead. Will return the TLD
@@ -104,6 +113,10 @@ bool IsCanonicalHostGoogleHostname(base::StringPiece canonical_host,
   if (!IsValidHostName(canonical_host, "google", subdomain_permission, &tld))
     return false;
 
+  // Remove the trailing dot from tld if present, as for google domain it's the
+  // same page.
+  StripTrailingDot(&tld);
+
   CR_DEFINE_STATIC_LOCAL(std::set<std::string>, google_tlds,
                          ({GOOGLE_TLD_LIST}));
   return base::ContainsKey(google_tlds, tld.as_string());
@@ -148,6 +161,8 @@ GURL AppendGoogleLocaleParam(const GURL& url,
 
 std::string GetGoogleCountryCode(const GURL& google_homepage_url) {
   base::StringPiece google_hostname = google_homepage_url.host_piece();
+  // TODO(igorcov): This needs a fix for case when the host has a trailing dot,
+  // like "google.com./". https://crbug.com/720295.
   const size_t last_dot = google_hostname.find_last_of('.');
   if (last_dot == std::string::npos) {
     NOTREACHED();
@@ -215,10 +230,32 @@ bool IsGoogleDomainUrl(const GURL& url,
          IsCanonicalHostGoogleHostname(url.host_piece(), subdomain_permission);
 }
 
+bool IsGoogleSearchSubdomainUrl(const GURL& url,
+                                PortPermission port_permission) {

[Peter Kasting, 2017/05/11 00:48:08]

Nit: Callers all seem to pass DISALLOW_NON_STANDARD_PORTS here, so I'd remove
this second argument and hardcode that.  (That goes for other functions above
too as applicable.)

[igorcov, 2017/05/11 11:30:05]

Done.

+  if (!IsValidURL(url, port_permission))
+    return false;
+
+  base::StringPiece host(url.host_piece());
+  StripTrailingDot(&host);
+
+  // If the size of |gGoogleSearchSubdomains| becomes larger, it might make
+  // sense to std::binary_search this.
+  for (const char* subdomain : gGoogleSearchSubdomains) {

[Peter Kasting, 2017/05/11 00:48:08]

Nit: It might make sense to CR_DEFINE_STATIC_LOCAL a string array, instead of
the char* array, so you can easily just use std::find() here.  Or possibly a
flat_map, which could be allocated to the right size and would have even simpler
syntax to use.  But maybe that's overkill.

[igorcov, 2017/05/11 11:30:05]

Done.

+    if (host == subdomain) {

[Peter Kasting, 2017/05/11 00:48:08]

Nit: {} not necessary

+      return true;
+    }
+  }
+
+  return false;
+}
+
 bool IsGoogleHomePageUrl(const GURL& url) {
   // First check to see if this has a Google domain.
-  if (!IsGoogleDomainUrl(url, DISALLOW_SUBDOMAIN, DISALLOW_NON_STANDARD_PORTS))
+  if (!IsGoogleDomainUrl(url, DISALLOW_SUBDOMAIN,
+                         DISALLOW_NON_STANDARD_PORTS) &&
+      !IsGoogleSearchSubdomainUrl(url, DISALLOW_NON_STANDARD_PORTS)) {
     return false;
+  }
 
   // Make sure the path is a known home page path.
   base::StringPiece path(url.path_piece());
@@ -228,8 +265,11 @@ bool IsGoogleHomePageUrl(const GURL& url) {
 
 bool IsGoogleSearchUrl(const GURL& url) {
   // First check to see if this has a Google domain.
-  if (!IsGoogleDomainUrl(url, DISALLOW_SUBDOMAIN, DISALLOW_NON_STANDARD_PORTS))
+  if (!IsGoogleDomainUrl(url, DISALLOW_SUBDOMAIN,
+                         DISALLOW_NON_STANDARD_PORTS) &&
+      !IsGoogleSearchSubdomainUrl(url, DISALLOW_NON_STANDARD_PORTS)) {
     return false;
+  }
 
   // Make sure the path is a known search path.
   base::StringPiece path(url.path_piece());