Redirect handling in resource prefetch predictor

chrome/browser/predictors/resource_prefetch_predictor.cc

 // Copyright 2014 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/predictors/resource_prefetch_predictor.h"
 
 #include <map>
 #include <set>
 #include <utility>
 
@@ skipping 54 matching lines @@
 
 namespace predictors {
 
 ////////////////////////////////////////////////////////////////////////////////
 // History lookup task.
 
 // Used to fetch the visit count for a URL from the History database.
 class GetUrlVisitCountTask : public history::HistoryDBTask {
  public:
   typedef ResourcePrefetchPredictor::URLRequestSummary URLRequestSummary;
+  typedef ResourcePrefetchPredictor::Navigation Navigation;
   typedef base::Callback<void(
       size_t,   // Visit count.
       const NavigationID&,
-      const std::vector<URLRequestSummary>&)> VisitInfoCallback;
+      const Navigation&)> VisitInfoCallback;
 
   GetUrlVisitCountTask(
       const NavigationID& navigation_id,
-      std::vector<URLRequestSummary>* requests,
+      std::unique_ptr<Navigation> navigation,
       VisitInfoCallback callback)
       : visit_count_(0),
         navigation_id_(navigation_id),
-        requests_(requests),
+        navigation_(std::move(navigation)),
         callback_(callback) {
-    DCHECK(requests_.get());
+    DCHECK(navigation_.get());
   }
 
   bool RunOnDBThread(history::HistoryBackend* backend,
                      history::HistoryDatabase* db) override {
     history::URLRow url_row;
     if (db->GetRowForURL(navigation_id_.main_frame_url, &url_row))
       visit_count_ = url_row.visit_count();
     return true;
   }
 
   void DoneRunOnMainThread() override {
-    callback_.Run(visit_count_, navigation_id_, *requests_);
+    callback_.Run(visit_count_, navigation_id_, *navigation_);
   }
 
  private:
   ~GetUrlVisitCountTask() override {}
 
   int visit_count_;
   NavigationID navigation_id_;
-  std::unique_ptr<std::vector<URLRequestSummary>> requests_;
+  std::unique_ptr<Navigation> navigation_;
   VisitInfoCallback callback_;
 
   DISALLOW_COPY_AND_ASSIGN(GetUrlVisitCountTask);
 };
 
 ////////////////////////////////////////////////////////////////////////////////
 // ResourcePrefetchPredictor static functions.
 
 // static
 bool ResourcePrefetchPredictor::ShouldRecordRequest(
@@ skipping 155 matching lines @@
       priority(other.priority),
       mime_type(other.mime_type),
       was_cached(other.was_cached),
       redirect_url(other.redirect_url),
       has_validators(other.has_validators),
       always_revalidate(other.always_revalidate) {}
 
 ResourcePrefetchPredictor::URLRequestSummary::~URLRequestSummary() {
 }
 
+ResourcePrefetchPredictor::Navigation::Navigation(
+    const GURL& i_initial_url)
+  : initial_url(i_initial_url),
+    requests(new std::vector<URLRequestSummary>) {
+}
+
+ResourcePrefetchPredictor::Navigation::~Navigation() {
+}
+
 // static
 bool ResourcePrefetchPredictor::URLRequestSummary::SummarizeResponse(
     const net::URLRequest& request,
     URLRequestSummary* summary) {
   const content::ResourceRequestInfo* info =
       content::ResourceRequestInfo::ForRequest(&request);
   if (!info)
     return false;
 
   int render_process_id, render_frame_id;
@@ skipping 136 matching lines @@
     const URLRequestSummary& request) {
   DCHECK_CURRENTLY_ON(BrowserThread::UI);
   DCHECK_EQ(INITIALIZED, initialization_state_);
 
   StartPrefetching(request.navigation_id);
 
   // Cleanup older navigations.
   CleanupAbandonedNavigations(request.navigation_id);
 
   // New empty navigation entry.
+  const GURL& initial_url = request.navigation_id.main_frame_url;
   inflight_navigations_.insert(std::make_pair(
       request.navigation_id,
-      make_linked_ptr(new std::vector<URLRequestSummary>())));
+      std::unique_ptr<Navigation>(new Navigation(initial_url))));
 }
 
 void ResourcePrefetchPredictor::OnMainFrameResponse(
     const URLRequestSummary& response) {
   DCHECK_CURRENTLY_ON(BrowserThread::UI);
   if (initialization_state_ != INITIALIZED)
     return;
 
   StopPrefetching(response.navigation_id);
 }
 
 void ResourcePrefetchPredictor::OnMainFrameRedirect(
     const URLRequestSummary& response) {
   DCHECK_CURRENTLY_ON(BrowserThread::UI);
 
   // TODO(shishir): There are significant gains to be had here if we can use the
   // start URL in a redirect chain as the key to start prefetching. We can save
   // of redirect times considerably assuming that the redirect chains do not
   // change.
 
   // Stop any inflight prefetching. Remove the older navigation.
   StopPrefetching(response.navigation_id);
-  inflight_navigations_.erase(response.navigation_id);
+
+  std::unique_ptr<Navigation> navigation;
+  NavigationMap::iterator nav_it =
+      inflight_navigations_.find(response.navigation_id);
+  if (nav_it != inflight_navigations_.end())

[mattcary, 2016/09/14 15:37:08]

Run git cl format; the brace should be on the same line as the if statement.

[alexilin, 2016/09/14 18:59:55]

Oops... Done.

+  {
+    navigation.reset(nav_it->second.release());
+    inflight_navigations_.erase(nav_it);
+  }
 
   // A redirect will not lead to another OnMainFrameRequest call, so record the
   // redirect url as a new navigation.
 
   // The redirect url may be empty if the URL was invalid.
   if (response.redirect_url.is_empty())
     return;
 
+  if (!navigation) {
+    navigation.reset(new Navigation(response.navigation_id.main_frame_url));
+  }
+
   NavigationID navigation_id(response.navigation_id);
   navigation_id.main_frame_url = response.redirect_url;
   inflight_navigations_.insert(std::make_pair(
       navigation_id,
-      make_linked_ptr(new std::vector<URLRequestSummary>())));
+      std::move(navigation)));
 }
 
 void ResourcePrefetchPredictor::OnSubresourceResponse(
     const URLRequestSummary& response) {
   DCHECK_CURRENTLY_ON(BrowserThread::UI);
 
   NavigationMap::const_iterator nav_it =
         inflight_navigations_.find(response.navigation_id);
   if (nav_it == inflight_navigations_.end()) {
     return;
   }
 
-  nav_it->second->push_back(response);
+  nav_it->second->requests->push_back(response);
 }
 
 void ResourcePrefetchPredictor::OnNavigationComplete(
     const NavigationID& nav_id_without_timing_info) {
   DCHECK_CURRENTLY_ON(BrowserThread::UI);
 
   NavigationMap::iterator nav_it =
       inflight_navigations_.find(nav_id_without_timing_info);
   if (nav_it == inflight_navigations_.end())
     return;
 
   const NavigationID navigation_id(nav_it->first);
 
   // Remove the navigation from the inflight navigations.
-  std::vector<URLRequestSummary>* requests = (nav_it->second).release();
+  std::unique_ptr<Navigation> navigation = std::move(nav_it->second);
   inflight_navigations_.erase(nav_it);
 
   // Kick off history lookup to determine if we should record the URL.
   history::HistoryService* history_service =
       HistoryServiceFactory::GetForProfile(profile_,
                                            ServiceAccessType::EXPLICIT_ACCESS);
   DCHECK(history_service);
   history_service->ScheduleDBTask(
       std::unique_ptr<history::HistoryDBTask>(new GetUrlVisitCountTask(
-          navigation_id, requests,
+          navigation_id,
+          std::move(navigation),
           base::Bind(&ResourcePrefetchPredictor::OnVisitCountLookup,
                      AsWeakPtr()))),
       &history_lookup_consumer_);
 }
 
 bool ResourcePrefetchPredictor::GetPrefetchData(
     const NavigationID& navigation_id,
     ResourcePrefetcher::RequestVector* prefetch_requests,
     PrefetchKeyType* key_type) {
   DCHECK(prefetch_requests);
@@ skipping 82 matching lines @@
 
 void ResourcePrefetchPredictor::StartInitialization() {
   DCHECK_CURRENTLY_ON(BrowserThread::UI);
 
   DCHECK_EQ(NOT_INITIALIZED, initialization_state_);
   initialization_state_ = INITIALIZING;
 
   // Create local caches using the database as loaded.
   std::unique_ptr<PrefetchDataMap> url_data_map(new PrefetchDataMap());
   std::unique_ptr<PrefetchDataMap> host_data_map(new PrefetchDataMap());
+  std::unique_ptr<RedirectDataMap> url_redirect_data_map(
+      new RedirectDataMap());
+  std::unique_ptr<RedirectDataMap> host_redirect_data_map(
+      new RedirectDataMap());
   PrefetchDataMap* url_data_ptr = url_data_map.get();
   PrefetchDataMap* host_data_ptr = host_data_map.get();
+  RedirectDataMap* url_redirect_data_ptr = url_redirect_data_map.get();
+  RedirectDataMap* host_redirect_data_ptr = host_redirect_data_map.get();
 
   BrowserThread::PostTaskAndReply(
       BrowserThread::DB, FROM_HERE,
-      base::Bind(&ResourcePrefetchPredictorTables::GetAllData,
-                 tables_, url_data_ptr, host_data_ptr),
+      base::Bind(&ResourcePrefetchPredictorTables::GetAllData, tables_,
+                 url_data_ptr,
+                 host_data_ptr,
+                 url_redirect_data_ptr,
+                 host_redirect_data_ptr),
       base::Bind(&ResourcePrefetchPredictor::CreateCaches, AsWeakPtr(),
-                 base::Passed(&url_data_map), base::Passed(&host_data_map)));
+                 base::Passed(&url_data_map),
+                 base::Passed(&host_data_map),
+                 base::Passed(&url_redirect_data_map),
+                 base::Passed(&host_redirect_data_map)));
 }
 
 void ResourcePrefetchPredictor::CreateCaches(
     std::unique_ptr<PrefetchDataMap> url_data_map,
-    std::unique_ptr<PrefetchDataMap> host_data_map) {
+    std::unique_ptr<PrefetchDataMap> host_data_map,
+    std::unique_ptr<RedirectDataMap> url_redirect_data_map,
+    std::unique_ptr<RedirectDataMap> host_redirect_data_map) {
   DCHECK_CURRENTLY_ON(BrowserThread::UI);
 
   DCHECK_EQ(INITIALIZING, initialization_state_);
   DCHECK(!url_table_cache_);
   DCHECK(!host_table_cache_);
+  DCHECK(!url_redirect_table_cache_);
+  DCHECK(!host_redirect_table_cache_);
   DCHECK(inflight_navigations_.empty());
 
   url_table_cache_.reset(url_data_map.release());
   host_table_cache_.reset(host_data_map.release());
+  url_redirect_table_cache_.reset(url_redirect_data_map.release());
+  host_redirect_table_cache_.reset(host_redirect_data_map.release());
 
   UMA_HISTOGRAM_COUNTS("ResourcePrefetchPredictor.UrlTableMainFrameUrlCount",
                        url_table_cache_->size());
   UMA_HISTOGRAM_COUNTS("ResourcePrefetchPredictor.HostTableHostCount",
                        host_table_cache_->size());
 
   ConnectToHistoryService();
 }
 
 void ResourcePrefetchPredictor::OnHistoryAndCacheLoaded() {
@@ skipping 57 matching lines @@
 
     const std::string& host = it.url().host();
     if (host_table_cache_->find(host) != host_table_cache_->end()) {
       hosts_to_delete.push_back(host);
       host_table_cache_->erase(host);
     }
   }
 
   if (!urls_to_delete.empty() || !hosts_to_delete.empty()) {
     BrowserThread::PostTask(BrowserThread::DB, FROM_HERE,
-        base::Bind(&ResourcePrefetchPredictorTables::DeleteData,
+        base::Bind(&ResourcePrefetchPredictorTables::DeleteResourceData,
                    tables_,
                    urls_to_delete,
                    hosts_to_delete));
   }
 }
 
 void ResourcePrefetchPredictor::RemoveOldestEntryInPrefetchDataMap(
     PrefetchKeyType key_type,
     PrefetchDataMap* data_map) {
   if (data_map->empty())
     return;
 
   base::Time oldest_time;
   std::string key_to_delete;
   for (PrefetchDataMap::iterator it = data_map->begin();
        it != data_map->end(); ++it) {
     if (key_to_delete.empty() || it->second.last_visit < oldest_time) {
       key_to_delete = it->first;
       oldest_time = it->second.last_visit;
     }
   }
 
   data_map->erase(key_to_delete);
-    BrowserThread::PostTask(BrowserThread::DB, FROM_HERE,
-        base::Bind(&ResourcePrefetchPredictorTables::DeleteSingleDataPoint,
-                   tables_,
-                   key_to_delete,
-                   key_type));
+  BrowserThread::PostTask(BrowserThread::DB, FROM_HERE, base::Bind(
+      &ResourcePrefetchPredictorTables::DeleteSingleResourceDataPoint,
+      tables_,
+      key_to_delete,
+      key_type));
+}
+
+void ResourcePrefetchPredictor::RemoveOldestEntryInRedirectDataMap(
+    PrefetchKeyType key_type,
+    RedirectDataMap* data_map) {
+  if (data_map->empty())
+    return;
+
+  base::Time oldest_time;
+  std::string key_to_delete;
+  for (RedirectDataMap::iterator it = data_map->begin();
+       it != data_map->end(); ++it) {
+    if (key_to_delete.empty() || it->second.last_visit < oldest_time) {
+      key_to_delete = it->first;
+      oldest_time = it->second.last_visit;
+    }
+  }
+
+  data_map->erase(key_to_delete);
+  BrowserThread::PostTask(BrowserThread::DB, FROM_HERE, base::Bind(
+      &ResourcePrefetchPredictorTables::DeleteSingleRedirectDataPoint,
+      tables_,
+      key_to_delete,
+      key_type));
 }
 
 void ResourcePrefetchPredictor::OnVisitCountLookup(
     size_t visit_count,
     const NavigationID& navigation_id,
-    const std::vector<URLRequestSummary>& requests) {
+    const Navigation& navigation) {
   DCHECK_CURRENTLY_ON(BrowserThread::UI);
 
   UMA_HISTOGRAM_COUNTS("ResourcePrefetchPredictor.HistoryVisitCountForUrl",
                        visit_count);
 
+  // TODO(alexilin) make only one request to DB thread
+
   // URL level data - merge only if we are already saving the data, or we it
   // meets the cutoff requirement.
   const std::string url_spec = navigation_id.main_frame_url.spec();
   bool already_tracking = url_table_cache_->find(url_spec) !=
       url_table_cache_->end();
   bool should_track_url = already_tracking ||
       (visit_count >= config_.min_url_visit_count);
 
   if (should_track_url && config_.IsURLLearningEnabled()) {
-    LearnNavigation(url_spec, PREFETCH_KEY_TYPE_URL, requests,
+    LearnNavigation(url_spec, PREFETCH_KEY_TYPE_URL, *navigation.requests,
                     config_.max_urls_to_track, url_table_cache_.get());
+
+    const std::string initial_url_spec = navigation.initial_url.spec();
+    if (url_spec != initial_url_spec) {
+      LearnRedirect(initial_url_spec, PREFETCH_KEY_TYPE_URL, url_spec,
+                    config_.max_urls_to_track,
+                    url_redirect_table_cache_.get());
+    }
   }
 
   // Host level data - no cutoff, always learn the navigation if enabled.
   if (config_.IsHostLearningEnabled()) {
-    LearnNavigation(navigation_id.main_frame_url.host(),
-                    PREFETCH_KEY_TYPE_HOST,
-                    requests,
+    const std::string host = navigation_id.main_frame_url.host();
+    LearnNavigation(host, PREFETCH_KEY_TYPE_HOST, *navigation.requests,
+                    config_.max_hosts_to_track, host_table_cache_.get());
+
+    const std::string initial_host = navigation.initial_url.host();
+    if (host != initial_host) {
+      LearnRedirect(initial_host, PREFETCH_KEY_TYPE_HOST, host,
                     config_.max_hosts_to_track,
-                    host_table_cache_.get());
+                    host_redirect_table_cache_.get());
+    }
   }
 
   // Remove the navigation from the results map.
   results_map_.erase(navigation_id);
 }
 
 void ResourcePrefetchPredictor::LearnNavigation(
     const std::string& key,
     PrefetchKeyType key_type,
     const std::vector<URLRequestSummary>& new_resources,
@@ skipping 27 matching lines @@
       row_to_add.resource_type = new_resources[i].resource_type;
       row_to_add.number_of_hits = 1;
       row_to_add.average_position = i + 1;
       row_to_add.priority = new_resources[i].priority;
       row_to_add.has_validators = new_resources[i].has_validators;
       row_to_add.always_revalidate = new_resources[i].always_revalidate;
       cache_entry->second.resources.push_back(row_to_add);
       resources_seen.insert(new_resources[i].resource_url);
     }
   } else {
-    ResourceRows& old_resources = cache_entry->second.resources;
+    std::vector<ResourceRow>& old_resources = cache_entry->second.resources;
     cache_entry->second.last_visit = base::Time::Now();
 
     // Build indices over the data.
     std::map<GURL, int> new_index, old_index;
     int new_resources_size = static_cast<int>(new_resources.size());
     for (int i = 0; i < new_resources_size; ++i) {
       const URLRequestSummary& summary = new_resources[i];
       // Take the first occurence of every url.
       if (new_index.find(summary.resource_url) == new_index.end())
         new_index[summary.resource_url] = i;
@@ skipping 46 matching lines @@
       row_to_add.has_validators = new_resources[i].has_validators;
       row_to_add.always_revalidate = new_resources[i].always_revalidate;
       old_resources.push_back(row_to_add);
 
       // To ensure we dont add the same url twice.
       old_index[summary.resource_url] = 0;
     }
   }
 
   // Trim and sort the resources after the update.
-  ResourceRows& resources = cache_entry->second.resources;
-  for (ResourceRows::iterator it = resources.begin();
+  std::vector<ResourceRow>& resources = cache_entry->second.resources;
+  for (std::vector<ResourceRow>::iterator it = resources.begin();
        it != resources.end();) {
     it->UpdateScore();
     if (it->consecutive_misses >= config_.max_consecutive_misses)
       it = resources.erase(it);
     else
       ++it;
   }
   ResourcePrefetchPredictorTables::SortResourceRows(&resources);
   if (resources.size() > config_.max_resources_per_entry)
     resources.resize(config_.max_resources_per_entry);
 
   // If the row has no resources, remove it from the cache and delete the
   // entry in the database. Else update the database.
   if (resources.empty()) {
     data_map->erase(key);
-    BrowserThread::PostTask(
-        BrowserThread::DB, FROM_HERE,
-        base::Bind(&ResourcePrefetchPredictorTables::DeleteSingleDataPoint,
-                   tables_,
-                   key,
-                   key_type));
+    BrowserThread::PostTask(BrowserThread::DB, FROM_HERE, base::Bind(
+        &ResourcePrefetchPredictorTables::DeleteSingleResourceDataPoint,
+        tables_,
+        key,
+        key_type));
   } else {
     bool is_host = key_type == PREFETCH_KEY_TYPE_HOST;
     PrefetchData empty_data(
         !is_host ? PREFETCH_KEY_TYPE_HOST : PREFETCH_KEY_TYPE_URL,
         std::string());
+    RedirectData empty_url_redirect_data(PREFETCH_KEY_TYPE_URL, std::string());
+    RedirectData empty_host_redirect_data(
+        PREFETCH_KEY_TYPE_HOST, std::string());
     const PrefetchData& host_data = is_host ? cache_entry->second : empty_data;
     const PrefetchData& url_data = is_host ? empty_data : cache_entry->second;
     BrowserThread::PostTask(
         BrowserThread::DB, FROM_HERE,
         base::Bind(&ResourcePrefetchPredictorTables::UpdateData,
                    tables_,
                    url_data,
-                   host_data));
+                   host_data,
+                   empty_url_redirect_data,
+                   empty_host_redirect_data));
   }
 }
 
+void ResourcePrefetchPredictor::LearnRedirect(
+    const std::string& key,
+    PrefetchKeyType key_type,
+    const std::string& final_redirect,
+    size_t max_redirect_map_size,
+    RedirectDataMap* redirect_map) {
+  DCHECK_CURRENTLY_ON(BrowserThread::UI);
+
+  RedirectDataMap::iterator cache_entry = redirect_map->find(key);
+  if (cache_entry == redirect_map->end()) {
+    if (redirect_map->size() >= max_redirect_map_size) {
+      RemoveOldestEntryInRedirectDataMap(key_type, redirect_map);
+    }
+
+    cache_entry = redirect_map->insert(std::make_pair(
+        key, RedirectData(key_type, key))).first;
+    cache_entry->second.last_visit = base::Time::Now();
+    RedirectRow row_to_add;
+    row_to_add.final_redirect = final_redirect;
+    row_to_add.number_of_hits = 1;
+    cache_entry->second.redirects.push_back(row_to_add);
+  } else {
+    std::vector<RedirectRow>& redirects = cache_entry->second.redirects;
+    bool need_to_add = true;
+
+    cache_entry->second.last_visit = base::Time::Now();
+    for (RedirectRow& redirect : redirects) {
+      if (redirect.final_redirect == final_redirect) {
+        need_to_add = false;
+        ++redirect.number_of_hits;
+        redirect.consecutive_misses = 0;
+      } else {
+        ++redirect.number_of_misses;
+        ++redirect.consecutive_misses;
+      }
+    }
+
+    if (need_to_add) {
+      RedirectRow row_to_add;
+      row_to_add.final_redirect = final_redirect;
+      row_to_add.number_of_hits = 1;
+      redirects.push_back(row_to_add);
+    }
+  }
+
+  std::vector<RedirectRow>& redirects = cache_entry->second.redirects;
+  // Trim and sort redirects after update.
+  for (std::vector<RedirectRow>::iterator it = redirects.begin();
+       it != redirects.end();) {
+    it->UpdateScore();
+    if (it->consecutive_misses >= config_.max_consecutive_misses)
+      it = redirects.erase(it);
+    else
+      ++it;
+  }
+  ResourcePrefetchPredictorTables::SortRedirectRows(&redirects);
+
+  if (redirects.size() > config_.max_redirects_per_entry)
+    redirects.resize(config_.max_redirects_per_entry);
+
+  if (redirects.empty()) {
+    redirect_map->erase(cache_entry);
+    BrowserThread::PostTask(BrowserThread::DB, FROM_HERE, base::Bind(
+        &ResourcePrefetchPredictorTables::DeleteSingleRedirectDataPoint,
+        tables_,
+        key,
+        key_type));
+  } else {
+    bool is_host = key_type == PREFETCH_KEY_TYPE_HOST;
+    RedirectData empty_redirect_data(
+        !is_host ? PREFETCH_KEY_TYPE_HOST : PREFETCH_KEY_TYPE_URL,
+        std::string());
+    PrefetchData empty_url_data(PREFETCH_KEY_TYPE_URL, std::string());
+    PrefetchData empty_host_data(
+        PREFETCH_KEY_TYPE_HOST, std::string());
+    const RedirectData& host_redirect_data =
+        is_host ? cache_entry->second : empty_redirect_data;
+    const RedirectData& url_redirect_data =
+        is_host ? empty_redirect_data : cache_entry->second;
+    BrowserThread::PostTask(
+        BrowserThread::DB, FROM_HERE,
+        base::Bind(&ResourcePrefetchPredictorTables::UpdateData,
+                   tables_,
+                   empty_url_data,
+                   empty_host_data,
+                   url_redirect_data,
+                   host_redirect_data));
+  }
+}
+
 void ResourcePrefetchPredictor::OnURLsDeleted(
     history::HistoryService* history_service,
     bool all_history,
     bool expired,
     const history::URLRows& deleted_rows,
     const std::set<GURL>& favicon_urls) {
   DCHECK_CURRENTLY_ON(BrowserThread::UI);
   if (INITIALIZED != initialization_state_)
     return;
 
@@ skipping 27 matching lines @@
     // HistoryService is already loaded. Continue with Initialization.
     OnHistoryAndCacheLoaded();
     return;
   }
   DCHECK(!history_service_observer_.IsObserving(history_service));
   history_service_observer_.Add(history_service);
   return;
 }
 
 }  // namespace predictors