Changed the precache config settings proto to not include the whitelist of starting URLs.

The precache fetcher will now attempt to fetch manifests for the first |top_sites_count| starting URLs. Fetches of manifests for non-whitelisted URLs will fail and be skipped.

The reason for this change is so that the server side component can have a large whitelist of starting URLs, without forcing the client to download the entire list every time precaching is done.

BUG=306185

Review URL: https://codereview.chromium.org/83283004

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@239077 0039d316-1c4b-4281-b951-d872f2087c98

4 files changed, 24 insertions, 44 deletions

diff --git a/components/precache/core/precache_fetcher.cc b/components/precache/core/precache_fetcher.cc
index 42b281f..d5ef31e 100644
--- a/components/precache/core/precache_fetcher.cc
+++ b/components/precache/core/precache_fetcher.cc
@@ -10,7 +10,6 @@
 #include "base/callback.h"
 #include "base/command_line.h"
 #include "base/compiler_specific.h"
-#include "base/containers/hash_tables.h"
 #include "components/precache/core/precache_switches.h"
 #include "components/precache/core/proto/precache.pb.h"
 #include "net/base/escape.h"
@@ -193,25 +192,14 @@ void PrecacheFetcher::OnConfigFetchComplete(const URLFetcher& source) {
   PrecacheConfigurationSettings config;
 
   if (ParseProtoFromFetchResponse(source, &config)) {
-    // A hash set of strings is used instead of GURLs because there is no
-    // standard hash function defined for GURLs.
-    base::hash_set<std::string> whitelisted_urls;
-    for (int i = 0; i < config.whitelisted_starting_url_size(); ++i) {
-      // Instead of using the raw URL string, construct a GURL and take the spec
-      // of it so that the URL string is canonicalized.
-      whitelisted_urls.insert(GURL(config.whitelisted_starting_url(i)).spec());
-    }
-
-    // Only fetch manifests for starting URLs up to the maximum rank that are in
-    // the whitelist.
+    // Attempt to fetch manifests for starting URLs up to the maximum top sites
+    // count. If a manifest does not exist for a particular starting URL, then
+    // the fetch will fail, and that starting URL will be ignored.
     int64 rank = 0;
     for (std::list<GURL>::const_iterator it = starting_urls_.begin();
-         it != starting_urls_.end() &&
-             rank < config.maximum_rank_starting_url();
+         it != starting_urls_.end() && rank < config.top_sites_count();
          ++it, ++rank) {
-      if (whitelisted_urls.find(it->spec()) != whitelisted_urls.end()) {
-        manifest_urls_to_fetch_.push_back(ConstructManifestURL(*it));
-      }
+      manifest_urls_to_fetch_.push_back(ConstructManifestURL(*it));
     }
   }
 
diff --git a/components/precache/core/precache_fetcher.h b/components/precache/core/precache_fetcher.h
index a3dde62..1fc56d7 100644
--- a/components/precache/core/precache_fetcher.h
+++ b/components/precache/core/precache_fetcher.h
@@ -91,10 +91,9 @@ class PrecacheFetcher {
   void StartNextFetch();
 
   // Called when the precache configuration settings have been fetched.
-  // Determines the list of manifest URLs to fetch according to the URLs that
-  // are present in both the list of |starting_urls_| and the whitelist
-  // contained in the precache configuration settings. If the fetch of the
-  // configuration settings fails, then precaching ends.
+  // Determines the list of manifest URLs to fetch according to the list of
+  // |starting_urls_| and information from the precache configuration settings.
+  // If the fetch of the configuration settings fails, then precaching ends.
   void OnConfigFetchComplete(const net::URLFetcher& source);
 
   // Called when a precache manifest has been fetched. Builds the list of
diff --git a/components/precache/core/precache_fetcher_unittest.cc b/components/precache/core/precache_fetcher_unittest.cc
index 2552586..227fc27 100644
--- a/components/precache/core/precache_fetcher_unittest.cc
+++ b/components/precache/core/precache_fetcher_unittest.cc
@@ -107,19 +107,13 @@ TEST_F(PrecacheFetcherTest, FullPrecache) {
       switches::kPrecacheManifestURLPrefix, kManfiestURLPrefix);
 
   std::list<GURL> starting_urls;
-  starting_urls.push_back(GURL("http://not-whitelisted.com"));
   starting_urls.push_back(GURL("http://manifest-fetch-failure.com"));
   starting_urls.push_back(GURL("http://bad-manifest.com"));
   starting_urls.push_back(GURL("http://good-manifest.com"));
-  starting_urls.push_back(GURL("http://not-in-top-4.com"));
+  starting_urls.push_back(GURL("http://not-in-top-3.com"));
 
   PrecacheConfigurationSettings config;
-  config.add_whitelisted_starting_url("http://whitelisted-unused.com");
-  config.add_whitelisted_starting_url("http://manifest-fetch-failure.com");
-  config.add_whitelisted_starting_url("http://bad-manifest.com");
-  config.add_whitelisted_starting_url("http://good-manifest.com");
-  config.add_whitelisted_starting_url("http://not-in-top-4.com");
-  config.set_maximum_rank_starting_url(4);
+  config.set_top_sites_count(3);
 
   PrecacheManifest good_manifest;
   good_manifest.add_resource()->set_url(kResourceFetchFailureURL);
@@ -213,8 +207,7 @@ TEST_F(PrecacheFetcherTest, Cancel) {
   std::list<GURL> starting_urls(1, GURL("http://starting-url.com"));
 
   PrecacheConfigurationSettings config;
-  config.add_whitelisted_starting_url("http://starting-url.com");
-  config.set_maximum_rank_starting_url(1);
+  config.set_top_sites_count(1);
 
   factory_.SetFakeResponse(GURL(kConfigURL), config.SerializeAsString(),
                            net::HTTP_OK, net::URLRequestStatus::SUCCESS);
@@ -243,9 +236,12 @@ TEST_F(PrecacheFetcherTest, Cancel) {
 TEST_F(PrecacheFetcherTest, PrecacheUsingDefaultConfigSettingsURL) {
   std::list<GURL> starting_urls(1, GURL("http://starting-url.com"));
 
+  PrecacheConfigurationSettings config;
+  config.set_top_sites_count(0);
+
   factory_.SetFakeResponse(GURL(PRECACHE_CONFIG_SETTINGS_URL),
-                           PrecacheConfigurationSettings().SerializeAsString(),
-                           net::HTTP_OK, net::URLRequestStatus::SUCCESS);
+                           config.SerializeAsString(), net::HTTP_OK,
+                           net::URLRequestStatus::SUCCESS);
 
   PrecacheFetcher precache_fetcher(starting_urls, request_context_.get(),
                                    &precache_delegate_);
@@ -273,8 +269,7 @@ TEST_F(PrecacheFetcherTest, PrecacheUsingDefaultManifestURLPrefix) {
   std::list<GURL> starting_urls(1, GURL("http://starting-url.com"));
 
   PrecacheConfigurationSettings config;
-  config.add_whitelisted_starting_url("http://starting-url.com");
-  config.set_maximum_rank_starting_url(1);
+  config.set_top_sites_count(1);
 
   GURL manifest_url(PRECACHE_MANIFEST_URL_PREFIX
                     "http%253A%252F%252Fstarting-url.com%252F");
diff --git a/components/precache/core/proto/precache.proto b/components/precache/core/proto/precache.proto
index 0429d58..ee0193f 100644
--- a/components/precache/core/proto/precache.proto
+++ b/components/precache/core/proto/precache.proto
@@ -24,15 +24,13 @@ message PrecacheManifest {
 };
 
 message PrecacheConfigurationSettings {
-  // The whitelist of starting URLs that are currently supported. Precaching
-  // should only be attempted for starting URLs that are in this list.
-  repeated string whitelisted_starting_url = 1;
-
   // The maximum rank of the user's most visited URLs to consider precaching
   // resources for, starting from 1. For example, a value of 10 means that only
-  // URLs that are both in the user's top 10 most visited URLs and in the
-  // whitelist will be considered as starting URLs for resource precaching.
-  // This is specified by the server for testing purposes, so that we can
-  // easily adjust how aggressively we precache resources.
-  optional int64 maximum_rank_starting_url = 2 [default = 10];
+  // URLs that are in the user's top 10 most visited URLs will be considered as
+  // starting URLs for resource precaching. This is specified by the server for
+  // testing purposes, so that it's easy to adjust how aggressively resources
+  // are precached.
+  // Values that are zero or lower indicate that none of the user's top sites
+  // will be used for precaching.
+  optional int64 top_sites_count = 1 [default = 10];
 };