Include all server redirects in referrer chain

chrome/browser/safe_browsing/safe_browsing_navigation_observer_manager.cc

 // Copyright 2016 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/safe_browsing/safe_browsing_navigation_observer_manager.h"
 
 #include "base/memory/ptr_util.h"
 #include "base/metrics/histogram_macros.h"
 #include "base/strings/stringprintf.h"
 #include "base/time/time.h"
@@ skipping 147 matching lines @@
 
 void SafeBrowsingNavigationObserverManager::RecordHostToIpMapping(
     const std::string& host,
     const std::string& ip) {
   auto insert_result = host_to_ip_map_.insert(
       std::make_pair(host, std::vector<ResolvedIPAddress>()));
   if (!insert_result.second) {
     // host_to_ip_map already contains this key.
     // If this IP is already in the vector, we update its timestamp.
     for (auto& vector_entry : insert_result.first->second) {
-      if (vector_entry.ip == host) {
+      if (vector_entry.ip == ip) {
         vector_entry.timestamp = base::Time::Now();
         return;
       }
     }
   }
   // If this is a new IP of this host, and we added to the end of the vector.
   insert_result.first->second.push_back(
       ResolvedIPAddress(base::Time::Now(), ip));
 }
 
@@ skipping 22 matching lines @@
   NavigationEvent* nav_event = FindNavigationEvent(
       target_url,
       GURL(),
       target_tab_id);
   if (!nav_event) {
     // We cannot find a single navigation event related to this download.
     return NAVIGATION_EVENT_NOT_FOUND;
   }
   AttributionResult result = SUCCESS;
   AddToReferrerChain(out_referrer_chain, nav_event,
+                     GURL(),
                      ReferrerChainEntry::DOWNLOAD_URL);
   int user_gesture_count = 0;
   GetRemainingReferrerChain(
       nav_event,
       user_gesture_count,
       user_gesture_count_limit,
       out_referrer_chain,
       &result);
   return result;
 }
 
 SafeBrowsingNavigationObserverManager::AttributionResult
 SafeBrowsingNavigationObserverManager::IdentifyReferrerChainForPPAPIDownload(
     const GURL& initiating_frame_url,
+    const GURL& initiating_main_frame_url,
     int tab_id,
     bool has_user_gesture,
     int user_gesture_count_limit,
     ReferrerChain* out_referrer_chain) {
   if (!initiating_frame_url.is_valid())
     return INVALID_URL;
 
   NavigationEvent* nav_event =
       FindNavigationEvent(initiating_frame_url, GURL(), tab_id);
   if (!nav_event) {
     // We cannot find a single navigation event related to this download.
     return NAVIGATION_EVENT_NOT_FOUND;
   }
 
   AttributionResult result = SUCCESS;
 
   int user_gesture_count = 0;
   // If this initiating_frame has user gesture, we consider this as the landing
   // page of the PPAPI download.
   if (has_user_gesture) {
     user_gesture_count = 1;
-    AddToReferrerChain(out_referrer_chain, nav_event,
+    AddToReferrerChain(out_referrer_chain, nav_event, initiating_main_frame_url,
                        GetURLTypeAndAdjustAttributionResult(
                           user_gesture_count == user_gesture_count_limit,
                           &result));
   } else {
-    AddToReferrerChain(out_referrer_chain, nav_event,
-                       nav_event->has_server_redirect
-                           ? ReferrerChainEntry::SERVER_REDIRECT
-                           : ReferrerChainEntry::CLIENT_REDIRECT);
+    AddToReferrerChain(out_referrer_chain, nav_event, initiating_main_frame_url,
+                       ReferrerChainEntry::CLIENT_REDIRECT);
   }
 
   GetRemainingReferrerChain(
       nav_event,
       user_gesture_count,
       user_gesture_count_limit,
       out_referrer_chain,
       &result);
   return result;
 }
@@ skipping 32 matching lines @@
   NavigationEvent nav_event;
   if (rfh) {
     nav_event.source_url = SafeBrowsingNavigationObserverManager::ClearEmptyRef(
         rfh->GetLastCommittedURL());
   }
   nav_event.source_tab_id = SessionTabHelper::IdForTab(source_contents);
   nav_event.source_main_frame_url =
       SafeBrowsingNavigationObserverManager::ClearEmptyRef(
           source_contents->GetLastCommittedURL());
   nav_event.original_request_url = target_url;
-  nav_event.destination_url = target_url;
   nav_event.target_tab_id = SessionTabHelper::IdForTab(target_contents);
   nav_event.frame_id = rfh ? rfh->GetFrameTreeNodeId() : -1;
   auto it = user_gesture_map_.find(source_contents);
   if (it != user_gesture_map_.end() &&
       !SafeBrowsingNavigationObserverManager::IsUserGestureExpired(
           it->second)) {
     nav_event.is_user_initiated = true;
     OnUserGestureConsumed(it->first, it->second);
   } else {
     nav_event.is_user_initiated = false;
   }
 
   auto insertion_result = navigation_map_.insert(
       std::make_pair(target_url, std::vector<NavigationEvent>()));
   insertion_result.first->second.push_back(std::move(nav_event));
 }
 
 void SafeBrowsingNavigationObserverManager::CleanUpNavigationEvents() {
   // Remove any stale NavigationEnvent, if it is older than
   // kNavigationFootprintTTLInSecond.
+  std::size_t remove_count = 0;
   for (auto it = navigation_map_.begin(); it != navigation_map_.end();) {
+    std::size_t size_before_removal = it->second.size();
     it->second.erase(std::remove_if(it->second.begin(), it->second.end(),
                                     [](const NavigationEvent& nav_event) {
                                       return IsEventExpired(
                                           nav_event.last_updated,
                                           kNavigationFootprintTTLInSecond);
                                     }),
                      it->second.end());
-    if (it->second.size() == 0)
+    std::size_t size_after_removal = it->second.size();
+    remove_count += (size_before_removal-size_after_removal);

[Nathan Parker, 2017/01/27 22:53:38]

nit: spaces around -
here and below

[Jialiu Lin, 2017/01/27 23:25:06]

Yes. Did git cl format.

+    if (size_after_removal == 0)
       it = navigation_map_.erase(it);
     else
       ++it;
   }
+  UMA_HISTOGRAM_COUNTS_10000(
+    "SafeBrowsing.NavigationObserver.NavigationEventCleanUpCount",
+    remove_count);
 }
 
 void SafeBrowsingNavigationObserverManager::CleanUpUserGestures() {
   for (auto it = user_gesture_map_.begin(); it != user_gesture_map_.end();) {
     if (IsEventExpired(it->second, kUserGestureTTLInSecond))
       it = user_gesture_map_.erase(it);
     else
       ++it;
   }
 }
 
 void SafeBrowsingNavigationObserverManager::CleanUpIpAddresses() {
+  std::size_t remove_count = 0;
   for (auto it = host_to_ip_map_.begin(); it != host_to_ip_map_.end();) {
+    std::size_t size_before_removal = it->second.size();
     it->second.erase(std::remove_if(it->second.begin(), it->second.end(),
                                     [](const ResolvedIPAddress& resolved_ip) {
                                       return IsEventExpired(
                                           resolved_ip.timestamp,
                                           kNavigationFootprintTTLInSecond);
                                     }),
                      it->second.end());
-    if (it->second.size() == 0)
+    std::size_t size_after_removal = it->second.size();
+    remove_count += (size_before_removal-size_after_removal);
+    if (size_after_removal == 0)
       it = host_to_ip_map_.erase(it);
     else
       ++it;
   }
+  UMA_HISTOGRAM_COUNTS_10000(
+    "SafeBrowsing.NavigationObserver.IPAddressCleanUpCount",
+    remove_count);
 }
 
 bool SafeBrowsingNavigationObserverManager::IsCleanUpScheduled() const {
   return cleanup_timer_.IsRunning();
 }
 
 void SafeBrowsingNavigationObserverManager::ScheduleNextCleanUpAfterInterval(
     base::TimeDelta interval) {
   DCHECK_GT(interval, base::TimeDelta());
   cleanup_timer_.Stop();
@@ skipping 14 matching lines @@
   GURL search_url =
       target_url.is_empty() ? target_main_frame_url : target_url;
   auto it = navigation_map_.find(search_url);
   if (it == navigation_map_.end())
     return nullptr;
 
   // Since navigation events are recorded in chronological order, we traverse
   // the vector in reverse order to get the latest match.
   for (auto rit = it->second.rbegin(); rit != it->second.rend(); ++rit) {
     // If tab id is not valid, we only compare url, otherwise we compare both.
-    if (rit->destination_url == search_url &&
+    if (rit->GetDestinationUrl() == search_url &&
         (target_tab_id == -1 || rit->target_tab_id == target_tab_id)) {
       // If both source_url and source_main_frame_url are empty, and this
       // navigation is not triggered by user, a retargeting navigation probably
       // causes this navigation. In this case, we skip this navigation event and
       // looks for the retargeting navigation event.
       if (rit->source_url.is_empty() && rit->source_main_frame_url.is_empty() &&
           !rit->is_user_initiated) {
         // If there is a server redirection immediately after retargeting, we
         // need to adjust our search url to the original request.
-        if (rit->has_server_redirect){
+        if (!rit->server_redirect_urls.empty()){
           NavigationEvent* retargeting_nav_event =
               FindNavigationEvent(rit->original_request_url,
                                   GURL(),
                                   rit->target_tab_id);
           if (!retargeting_nav_event)
             return nullptr;
           // Adjust retargeting navigation event's attributes.
-          retargeting_nav_event->has_server_redirect = true;
-          retargeting_nav_event->destination_url = search_url;
+          retargeting_nav_event->server_redirect_urls.push_back(
+              std::move(search_url));
           return retargeting_nav_event;
         } else {
           continue;
         }
       } else {
         return &*rit;
       }
     }
   }
   return nullptr;
 }
 
 void SafeBrowsingNavigationObserverManager::AddToReferrerChain(
     ReferrerChain* referrer_chain,
     NavigationEvent* nav_event,
+    const GURL& destination_main_frame_url,
     ReferrerChainEntry::URLType type) {
   std::unique_ptr<ReferrerChainEntry> referrer_chain_entry =
       base::MakeUnique<ReferrerChainEntry>();
-  referrer_chain_entry->set_url(nav_event->destination_url.spec());
+  const GURL destination_url = nav_event->GetDestinationUrl();
+  referrer_chain_entry->set_url(destination_url.spec());
+  if (destination_main_frame_url.is_valid() &&
+      destination_url != destination_main_frame_url)
+    referrer_chain_entry->set_main_frame_url(destination_main_frame_url.spec());
   referrer_chain_entry->set_type(type);
-  auto ip_it = host_to_ip_map_.find(nav_event->destination_url.host());
+  auto ip_it = host_to_ip_map_.find(destination_url.host());
   if (ip_it != host_to_ip_map_.end()) {
     for (ResolvedIPAddress entry : ip_it->second) {
       referrer_chain_entry->add_ip_addresses(entry.ip);
     }
   }
   // Since we only track navigation to landing referrer, we will not log the
   // referrer of the landing referrer page.
   if (type != ReferrerChainEntry::LANDING_REFERRER) {
     referrer_chain_entry->set_referrer_url(nav_event->source_url.spec());
-    referrer_chain_entry->set_referrer_main_frame_url(
-        nav_event->source_main_frame_url.spec());
+    // Only set |referrer_main_frame_url| if it is diff from |referrer_url|.
+    if (nav_event->source_main_frame_url.is_valid() &&
+        nav_event->source_url != nav_event->source_main_frame_url) {
+      referrer_chain_entry->set_referrer_main_frame_url(
+          nav_event->source_main_frame_url.spec());
+    }
   }
   referrer_chain_entry->set_is_retargeting(nav_event->source_tab_id !=
                                           nav_event->target_tab_id);
   referrer_chain_entry->set_navigation_time_msec(
       nav_event->last_updated.ToJavaTime());
+  if (!nav_event->server_redirect_urls.empty()) {
+    // The first entry in |server_redirect_chain| should be the original request
+    // url.
+    ReferrerChainEntry::ServerRedirect* server_redirect =
+        referrer_chain_entry->add_server_redirect_chain();
+    server_redirect->set_url(nav_event->original_request_url.spec());
+    for (const GURL& redirect: nav_event->server_redirect_urls) {
+      server_redirect = referrer_chain_entry->add_server_redirect_chain();
+      server_redirect->set_url(redirect.spec());
+    }
+  }
   referrer_chain->push_back(std::move(referrer_chain_entry));
 }
 
 void SafeBrowsingNavigationObserverManager::GetRemainingReferrerChain(
     NavigationEvent* last_nav_event_traced,
     int current_user_gesture_count,
     int user_gesture_count_limit,
     ReferrerChain* out_referrer_chain,
     SafeBrowsingNavigationObserverManager::AttributionResult* out_result) {
-
+  GURL last_main_frame_url_traced(
+      last_nav_event_traced->source_main_frame_url);

[Nathan Parker, 2017/01/27 22:53:38]

What's the different between source_main_frame_url and source_url?  Is it that
the latter might be... an iframe?

[Jialiu Lin, 2017/01/27 23:25:07]

Yes, that's right. From the campaign example auk@ sent to me, actually a lot of
bad downloads are triggered by some navigation in subframes. 

   while (current_user_gesture_count < user_gesture_count_limit) {
     // Back trace to the next nav_event that was initiated by the user.
     while (!last_nav_event_traced->is_user_initiated) {
       last_nav_event_traced =
           FindNavigationEvent(last_nav_event_traced->source_url,
                               last_nav_event_traced->source_main_frame_url,
                               last_nav_event_traced->source_tab_id);
       if (!last_nav_event_traced)
         return;
       AddToReferrerChain(out_referrer_chain, last_nav_event_traced,
-                         last_nav_event_traced->has_server_redirect
-                             ? ReferrerChainEntry::SERVER_REDIRECT
-                             : ReferrerChainEntry::CLIENT_REDIRECT);
+                         last_main_frame_url_traced,

[Nathan Parker, 2017/01/27 22:53:38]

Do we no longer need SERVER_REDIRECT?

[Jialiu Lin, 2017/01/27 23:25:06]

We don't need SERVER_REDIRECT, since this can be deduced by the existence of
server_redirect_chain repeated field.

+                         ReferrerChainEntry::CLIENT_REDIRECT);
+      last_main_frame_url_traced = last_nav_event_traced->source_main_frame_url;
     }
 
     current_user_gesture_count++;
 
 
     // If the source_url and source_main_frame_url of current navigation event
     // are empty, and is_user_initiated is true, this is a browser initiated
     // navigation (e.g. trigged by typing in address bar, clicking on bookmark,
     // etc). We reached the end of the referrer chain.
     if (last_nav_event_traced->source_url.is_empty() &&
         last_nav_event_traced->source_main_frame_url.is_empty()) {
       DCHECK(last_nav_event_traced->is_user_initiated);
       return;
     }
 
     last_nav_event_traced =
         FindNavigationEvent(last_nav_event_traced->source_url,
                             last_nav_event_traced->source_main_frame_url,
                             last_nav_event_traced->source_tab_id);
     if (!last_nav_event_traced)
       return;
 
     AddToReferrerChain(out_referrer_chain, last_nav_event_traced,
+                       last_main_frame_url_traced,
                        GetURLTypeAndAdjustAttributionResult(
                           current_user_gesture_count ==
                               user_gesture_count_limit,
                           out_result));
+    last_main_frame_url_traced = last_nav_event_traced->source_main_frame_url;
   }
 }
 
 }  // namespace safe_browsing