Added list of forced starting URLs to precache configuration settings.

BUG=306185

Review URL: https://codereview.chromium.org/103373006

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@241174 0039d316-1c4b-4281-b951-d872f2087c98

3 files changed, 39 insertions, 1 deletions

diff --git a/components/precache/core/precache_fetcher.cc b/components/precache/core/precache_fetcher.cc
index d5ef31e..78ae7ac 100644
--- a/components/precache/core/precache_fetcher.cc
+++ b/components/precache/core/precache_fetcher.cc
@@ -10,6 +10,7 @@
 #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"
@@ -192,6 +193,11 @@ void PrecacheFetcher::OnConfigFetchComplete(const URLFetcher& source) {
   PrecacheConfigurationSettings config;
 
   if (ParseProtoFromFetchResponse(source, &config)) {
+    // Keep track of starting URLs that manifests are being fetched for, in
+    // order to remove duplicates. This is a hash set on strings, and not GURLs,
+    // because there is no hash function defined for GURL.
+    base::hash_set<std::string> unique_starting_urls;
+
     // 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.
@@ -199,7 +205,24 @@ void PrecacheFetcher::OnConfigFetchComplete(const URLFetcher& source) {
     for (std::list<GURL>::const_iterator it = starting_urls_.begin();
          it != starting_urls_.end() && rank < config.top_sites_count();
          ++it, ++rank) {
-      manifest_urls_to_fetch_.push_back(ConstructManifestURL(*it));
+      if (unique_starting_urls.find(it->spec()) == unique_starting_urls.end()) {
+        // Only add a fetch for the manifest URL if this manifest isn't already
+        // going to be fetched.
+        manifest_urls_to_fetch_.push_back(ConstructManifestURL(*it));
+        unique_starting_urls.insert(it->spec());
+      }
+    }
+
+    for (int i = 0; i < config.forced_starting_url_size(); ++i) {
+      // Convert the string URL into a GURL and take the spec() of it so that
+      // the URL string gets canonicalized.
+      GURL url(config.forced_starting_url(i));
+      if (unique_starting_urls.find(url.spec()) == unique_starting_urls.end()) {
+        // Only add a fetch for the manifest URL if this manifest isn't already
+        // going to be fetched.
+        manifest_urls_to_fetch_.push_back(ConstructManifestURL(url));
+        unique_starting_urls.insert(url.spec());
+      }
     }
   }
 
diff --git a/components/precache/core/precache_fetcher_unittest.cc b/components/precache/core/precache_fetcher_unittest.cc
index 227fc27..acdd59d 100644
--- a/components/precache/core/precache_fetcher_unittest.cc
+++ b/components/precache/core/precache_fetcher_unittest.cc
@@ -99,6 +99,9 @@ const char kGoodManifestURL[] =
     "http://manifest-url-prefix.com/http%253A%252F%252Fgood-manifest.com%252F";
 const char kResourceFetchFailureURL[] = "http://resource-fetch-failure.com";
 const char kGoodResourceURL[] = "http://good-resource.com";
+const char kForcedStartingURLManifestURL[] =
+    "http://manifest-url-prefix.com/"
+    "http%253A%252F%252Fforced-starting-url.com%252F";
 
 TEST_F(PrecacheFetcherTest, FullPrecache) {
   CommandLine::ForCurrentProcess()->AppendSwitchASCII(
@@ -114,6 +117,9 @@ TEST_F(PrecacheFetcherTest, FullPrecache) {
 
   PrecacheConfigurationSettings config;
   config.set_top_sites_count(3);
+  config.add_forced_starting_url("http://forced-starting-url.com");
+  // Duplicate starting URL, the manifest for this should only be fetched once.
+  config.add_forced_starting_url("http://good-manifest.com");
 
   PrecacheManifest good_manifest;
   good_manifest.add_resource()->set_url(kResourceFetchFailureURL);
@@ -135,6 +141,9 @@ TEST_F(PrecacheFetcherTest, FullPrecache) {
                            net::URLRequestStatus::FAILED);
   factory_.SetFakeResponse(GURL(kGoodResourceURL), "good", net::HTTP_OK,
                            net::URLRequestStatus::SUCCESS);
+  factory_.SetFakeResponse(GURL(kForcedStartingURLManifestURL),
+                           PrecacheManifest().SerializeAsString(), net::HTTP_OK,
+                           net::URLRequestStatus::SUCCESS);
 
   PrecacheFetcher precache_fetcher(starting_urls, request_context_.get(),
                                    &precache_delegate_);
@@ -149,6 +158,7 @@ TEST_F(PrecacheFetcherTest, FullPrecache) {
   expected_requested_urls.insert(GURL(kGoodManifestURL));
   expected_requested_urls.insert(GURL(kResourceFetchFailureURL));
   expected_requested_urls.insert(GURL(kGoodResourceURL));
+  expected_requested_urls.insert(GURL(kForcedStartingURLManifestURL));
 
   EXPECT_EQ(expected_requested_urls, url_callback_.requested_urls());
 
diff --git a/components/precache/core/proto/precache.proto b/components/precache/core/proto/precache.proto
index ee0193f..ace5384 100644
--- a/components/precache/core/proto/precache.proto
+++ b/components/precache/core/proto/precache.proto
@@ -33,4 +33,9 @@ message PrecacheConfigurationSettings {
   // 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];
+
+  // List of additional starting URLs that resources will be precached for.
+  // These are URLs that the server predicts that the user will visit, as a
+  // result of server-side analytics.
+  repeated string forced_starting_url = 2;
 };