Wireup SafeBrowsingNavigationObserverManager to help PPAPI download attribution

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 24 matching lines @@
 // Note, if for some reason this event's timestamp is in the future, this
 // event's timestamp is invalid, hence we treat it as expired.
 bool IsEventExpired(const base::Time& event_time, double ttl_in_second) {
   double current_time_in_second = base::Time::Now().ToDoubleT();
   double event_time_in_second = event_time.ToDoubleT();
   if (current_time_in_second <= event_time_in_second)
     return true;
   return current_time_in_second - event_time_in_second > ttl_in_second;
 }
 
+// Helper function to determine if the URL type should be LANDING_REFERRER or
+// LANDING_PAGE, and modify AttributionResult accordingly.
+ReferrerChainEntry::URLType GetURLTypeAndAdjustAttributionResult(
+    int current_user_gesture_count,

[Nathan Parker, 2017/01/04 19:49:12]

What is current_user_gesture_count? Is that how many we've seen so far stepping
backward through navigations?

Would it be simpler to have one arg, as
  bool at_user_gesture_limit
and have the callers compute it? Not sure.

[Jialiu Lin, 2017/01/04 21:53:35]

Good suggestion. Thanks! 

+    int user_gesture_count_limit,
+    SafeBrowsingNavigationObserverManager::AttributionResult* out_result) {
+  // Landing page of a download refers to the page user directly interacts
+  // with to trigger this download (e.g. clicking on download button). Landing
+  // referrer page is the one user interacts with right before navigating to
+  // the landing page.
+  // Since we are tracing navigations backwards, if we've reached
+  // |user_gesture_count_limit| number of user gesture before this navigation
+  // event, this is a navigation leading to the landing referrer page,
+  // otherwise it leads to landing page.
+  if (current_user_gesture_count == user_gesture_count_limit) {
+    *out_result =
+        SafeBrowsingNavigationObserverManager::SUCCESS_LANDING_REFERRER;
+    return ReferrerChainEntry::LANDING_REFERRER;
+  } else {
+    *out_result = SafeBrowsingNavigationObserverManager::SUCCESS_LANDING_PAGE;
+    return ReferrerChainEntry::LANDING_PAGE;
+  }
+}
+
 }  // namespace
 
 // The expiration period of a user gesture. Any user gesture that happened 1.0
 // second ago is considered as expired and not relevant to upcoming navigation
 // events.
 static const double kUserGestureTTLInSecond = 1.0;
 // The expiration period of navigation events and resolved IP addresses. Any
 // navigation related records that happened 2 minutes ago are considered as
 // expired. So we clean up these navigation footprints every 2 minutes.
 static const double kNavigationFootprintTTLInSecond = 120.0;
@@ skipping 94 matching lines @@
 
 void SafeBrowsingNavigationObserverManager::CleanUpStaleNavigationFootprints() {
   CleanUpNavigationEvents();
   CleanUpUserGestures();
   CleanUpIpAddresses();
   ScheduleNextCleanUpAfterInterval(
       base::TimeDelta::FromSecondsD(kNavigationFootprintTTLInSecond));
 }
 
 SafeBrowsingNavigationObserverManager::AttributionResult
-SafeBrowsingNavigationObserverManager::IdentifyReferrerChain(
+SafeBrowsingNavigationObserverManager::IdentifyReferrerChainForDownload(
     const GURL& target_url,
     int target_tab_id,
     int user_gesture_count_limit,
     std::vector<ReferrerChainEntry>* out_referrer_chain) {
   if (!target_url.is_valid())
     return INVALID_URL;
 
   NavigationEvent* nav_event = FindNavigationEvent(target_url, 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,
+                     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,
+    content::WebContents* web_contents,
+    int user_gesture_count_limit,
+    std::vector<ReferrerChainEntry>* out_referrer_chain) {
+  if (!initiating_frame_url.is_valid())
+    return INVALID_URL;
+
+  int tab_id = SessionTabHelper::IdForTab(web_contents);
+  UMA_HISTOGRAM_BOOLEAN(
+      "SafeBrowsing.ReferrerHasInvalidTabID.PPAPIDownloadAttribution",
+      tab_id == -1);
+
+  NavigationEvent* nav_event = FindNavigationEvent(initiating_frame_url,
+                                                   tab_id);
+  if (!nav_event) {
+    // We cannot find a single navigation event related to this download.
+    return NAVIGATION_EVENT_NOT_FOUND;
+  }
 
-  AddToReferrerChain(out_referrer_chain, nav_event,
-                     ReferrerChainEntry::DOWNLOAD_URL);
   AttributionResult result = SUCCESS;
+  bool has_user_gesture = false;
+  auto it = user_gesture_map_.find(web_contents);
+  if (it != user_gesture_map_.end() &&
+      !IsEventExpired(it->second, kUserGestureTTLInSecond)) {
+    has_user_gesture = true;
+    user_gesture_map_.erase(it);
+  }
+
   int user_gesture_count = 0;
-  while (user_gesture_count < user_gesture_count_limit) {
-    // Back trace to the next nav_event that was initiated by the user.
-    while (!nav_event->is_user_initiated) {
-      nav_event =
-          FindNavigationEvent(nav_event->source_url, nav_event->source_tab_id);
-      if (!nav_event)
-        return result;
-      AddToReferrerChain(out_referrer_chain, nav_event,
-                         nav_event->has_server_redirect
-                             ? ReferrerChainEntry::SERVER_REDIRECT
-                             : ReferrerChainEntry::CLIENT_REDIRECT);
-    }
+  // 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,
+                       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);
+  }
 
-    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 (nav_event->source_url.is_empty() &&
-        nav_event->source_main_frame_url.is_empty()) {
-      DCHECK(nav_event->is_user_initiated);
-      return result;
-    }
-
-    nav_event =
-        FindNavigationEvent(nav_event->source_url, nav_event->source_tab_id);
-    if (!nav_event)
-      return result;
-
-    // Landing page of a download refers to the page user directly interacts
-    // with to trigger this download (e.g. clicking on download button). Landing
-    // referrer page is the one user interacts with right before navigating to
-    // the landing page.
-    // Since we are tracing navigations backwards, if we've encountered 1 user
-    // gesture before this navigation event, this is a navigation leading to the
-    // landing page. If we've encountered 2 user gestures, it leads to landing
-    // referrer page.
-    if (user_gesture_count == 1) {
-      AddToReferrerChain(out_referrer_chain, nav_event,
-                         ReferrerChainEntry::LANDING_PAGE);
-      result = SUCCESS_LANDING_PAGE;
-    } else if (user_gesture_count == 2) {
-      AddToReferrerChain(out_referrer_chain, nav_event,
-                         ReferrerChainEntry::LANDING_REFERRER);
-      result = SUCCESS_LANDING_REFERRER;
-    } else {
-      NOTREACHED();
-    }
-  }
+  GetRemainingReferrerChain(
+      nav_event,
+      user_gesture_count,
+      user_gesture_count_limit,
+      out_referrer_chain,
+      &result);
   return result;
 }
 
 SafeBrowsingNavigationObserverManager::
     ~SafeBrowsingNavigationObserverManager() {}
 
 void SafeBrowsingNavigationObserverManager::Observe(
     int type,
     const content::NotificationSource& source,
     const content::NotificationDetails& details) {
@@ skipping 153 matching lines @@
     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());
   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,
+    std::vector<ReferrerChainEntry>* out_referrer_chain,
+    SafeBrowsingNavigationObserverManager::AttributionResult* out_result) {
+
+  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_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);
+    }
+
+    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() &&

[Nathan Parker, 2017/01/04 19:49:12]

q: Up above in a few places is_valid() is used.  Is there a reason to use one vs
the other?

[Jialiu Lin, 2017/01/04 21:53:35]

invalid url could be empty or malformatted. Here I use is_empty() is because the
source url and source main frame url of browser initiated navigation can only be
empty.

BTW, I need to do a rebase here, since now FindNavigationEvent takes 3 args. 

+        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_tab_id);
+    if (!last_nav_event_traced)
+      return;
+
+    AddToReferrerChain(out_referrer_chain, last_nav_event_traced,
+                       GetURLTypeAndAdjustAttributionResult(
+                          current_user_gesture_count,
+                          user_gesture_count_limit,
+                          out_result));
+  }
+}
+
 }  // namespace safe_browsing