OLD | NEW |
---|---|
1 // Copyright 2017 The Chromium Authors. All rights reserved. | 1 // Copyright 2017 The Chromium Authors. All rights reserved. |
2 // Use of this source code is governed by a BSD-style license that can be | 2 // Use of this source code is governed by a BSD-style license that can be |
3 // found in the LICENSE file. | 3 // found in the LICENSE file. |
4 | 4 |
5 #include "chrome/browser/page_load_metrics/observers/ads_page_load_metrics_obser ver.h" | 5 #include "chrome/browser/page_load_metrics/observers/ads_page_load_metrics_obser ver.h" |
6 | 6 |
7 #include <string> | 7 #include <string> |
8 #include <utility> | 8 #include <utility> |
9 | 9 |
10 #include "base/feature_list.h" | 10 #include "base/feature_list.h" |
11 #include "base/logging.h" | 11 #include "base/logging.h" |
12 #include "base/memory/ptr_util.h" | 12 #include "base/memory/ptr_util.h" |
13 #include "base/strings/string_util.h" | 13 #include "base/strings/string_util.h" |
14 #include "chrome/browser/page_load_metrics/page_load_metrics_util.h" | 14 #include "chrome/browser/page_load_metrics/page_load_metrics_util.h" |
15 #include "content/public/browser/navigation_handle.h" | 15 #include "content/public/browser/navigation_handle.h" |
16 #include "content/public/browser/render_frame_host.h" | 16 #include "content/public/browser/render_frame_host.h" |
17 #include "content/public/browser/web_contents.h" | 17 #include "content/public/browser/web_contents.h" |
18 #include "url/gurl.h" | 18 #include "url/gurl.h" |
19 | 19 |
20 namespace { | 20 namespace { |
21 | 21 |
22 const base::Feature kAdsFeature{"AdsMetrics", | 22 const base::Feature kAdsFeature{"AdsMetrics", base::FEATURE_ENABLED_BY_DEFAULT}; |
23 base::FEATURE_DISABLED_BY_DEFAULT}; | |
24 | 23 |
25 bool FrameIsAd(content::NavigationHandle* navigation_handle) { | 24 bool FrameIsAd(content::NavigationHandle* navigation_handle) { |
26 content::RenderFrameHost* current_frame_host = | |
27 navigation_handle->GetRenderFrameHost(); | |
28 DCHECK(current_frame_host); | |
29 const std::string& name = current_frame_host->GetFrameName(); | |
30 const GURL& url = navigation_handle->GetURL(); | |
31 | |
32 // Because sub-resource filtering isn't always enabled, and doesn't work | 25 // Because sub-resource filtering isn't always enabled, and doesn't work |
33 // well in monitoring mode (no CSS enforcement), it's difficult to identify | 26 // well in monitoring mode (no CSS enforcement), it's difficult to identify |
34 // ads. Google ads are prevalent and easy to track, so we'll start by | 27 // ads. Google ads are prevalent and easy to track, so we'll start by |
35 // tracking those. Note that the frame name can be very large, so be careful | 28 // tracking those. Note that the frame name can be very large, so be careful |
36 // to avoid full string searches if possible. | 29 // to avoid full string searches if possible. |
37 // TODO(jkarlin): Track other ad networks that are easy to identify. | 30 // TODO(jkarlin): Track other ad networks that are easy to identify. |
38 return base::StartsWith(name, "google_ads_iframe", | 31 |
39 base::CompareCase::SENSITIVE) || | 32 // In case the navigation aborted, look up the RFH by the Frame Tree Node |
40 base::StartsWith(name, "google_ads_frame", | 33 // ID. If there is no RFH remaining, then the frame is gone. |
41 base::CompareCase::SENSITIVE) || | 34 content::RenderFrameHost* current_frame_host = |
42 (url.host_piece() == "tpc.googlesyndication.com" && | 35 navigation_handle->GetWebContents()->UnsafeFindFrameByFrameTreeNodeId( |
Charlie Harrison
2017/05/10 16:47:21
Can you comment why the "Unsafe" is necessary?
jkarlin
2017/05/10 18:19:24
I commented on way it's safe to use in this case.
| |
43 base::StartsWith(url.path_piece(), "/safeframe", | 36 navigation_handle->GetFrameTreeNodeId()); |
44 base::CompareCase::SENSITIVE)); | 37 if (current_frame_host) { |
38 const std::string& frame_name = current_frame_host->GetFrameName(); | |
39 if (base::StartsWith(frame_name, "google_ads_iframe", | |
40 base::CompareCase::SENSITIVE) || | |
41 base::StartsWith(frame_name, "google_ads_frame", | |
42 base::CompareCase::SENSITIVE)) { | |
43 return true; | |
44 } | |
45 } | |
46 | |
47 const GURL& url = navigation_handle->GetURL(); | |
48 return url.host_piece() == "tpc.googlesyndication.com" && | |
49 base::StartsWith(url.path_piece(), "/safeframe", | |
50 base::CompareCase::SENSITIVE); | |
45 } | 51 } |
46 | 52 |
47 } // namespace | 53 } // namespace |
48 | 54 |
49 AdsPageLoadMetricsObserver::AdFrameData::AdFrameData( | 55 AdsPageLoadMetricsObserver::AdFrameData::AdFrameData( |
50 FrameTreeNodeId frame_tree_node_id) | 56 FrameTreeNodeId frame_tree_node_id) |
51 : frame_bytes(0u), | 57 : frame_bytes(0u), |
52 frame_bytes_uncached(0u), | 58 frame_bytes_uncached(0u), |
53 frame_tree_node_id(frame_tree_node_id) {} | 59 frame_tree_node_id(frame_tree_node_id) {} |
54 | 60 |
(...skipping 15 matching lines...) Expand all Loading... | |
70 | 76 |
71 // The main frame is never considered an ad. | 77 // The main frame is never considered an ad. |
72 ad_frames_data_[navigation_handle->GetFrameTreeNodeId()] = nullptr; | 78 ad_frames_data_[navigation_handle->GetFrameTreeNodeId()] = nullptr; |
73 ProcessOngoingNavigationResource(navigation_handle->GetFrameTreeNodeId()); | 79 ProcessOngoingNavigationResource(navigation_handle->GetFrameTreeNodeId()); |
74 return CONTINUE_OBSERVING; | 80 return CONTINUE_OBSERVING; |
75 } | 81 } |
76 | 82 |
77 page_load_metrics::PageLoadMetricsObserver::ObservePolicy | 83 page_load_metrics::PageLoadMetricsObserver::ObservePolicy |
78 AdsPageLoadMetricsObserver::OnDidFinishSubFrameNavigation( | 84 AdsPageLoadMetricsObserver::OnDidFinishSubFrameNavigation( |
79 content::NavigationHandle* navigation_handle) { | 85 content::NavigationHandle* navigation_handle) { |
86 // Determine if the frame is part of an existing ad, the root of a new ad, | |
87 // or a non-ad frame. Once a frame is labled as an ad, it is always | |
88 // considered an ad, even if it navigates to a non-ad page. This function | |
89 // labels all of a page's frames, even those that fail to navigate. | |
Charlie Harrison
2017/05/10 16:47:21
s/navigate/commit
jkarlin
2017/05/10 18:19:24
Done.
| |
80 FrameTreeNodeId frame_tree_node_id = navigation_handle->GetFrameTreeNodeId(); | 90 FrameTreeNodeId frame_tree_node_id = navigation_handle->GetFrameTreeNodeId(); |
81 | |
82 if (!navigation_handle->HasCommitted() || | |
83 navigation_handle->IsSameDocument() || navigation_handle->IsErrorPage()) { | |
84 // We're not interested in tracking this navigation. In case we've seen a | |
85 // resource for the navigation before this message, clear it from | |
86 // ongoing_navigation_resources_. | |
87 ongoing_navigation_resources_.erase(frame_tree_node_id); | |
88 return CONTINUE_OBSERVING; | |
89 } | |
90 | |
91 content::RenderFrameHost* parent_frame_host = | 91 content::RenderFrameHost* parent_frame_host = |
92 navigation_handle->GetRenderFrameHost()->GetParent(); | 92 navigation_handle->GetParentFrame(); |
93 DCHECK(parent_frame_host); | |
94 | |
95 bool top_level_subframe = !parent_frame_host->GetParent(); | |
96 | 93 |
97 const auto& id_and_data = ad_frames_data_.find(frame_tree_node_id); | 94 const auto& id_and_data = ad_frames_data_.find(frame_tree_node_id); |
98 if (id_and_data != ad_frames_data_.end()) { | 95 if (id_and_data != ad_frames_data_.end()) { |
99 // An existing subframe is navigating again. | 96 // An existing subframe is navigating again. |
100 if (id_and_data->second) { | 97 if (id_and_data->second) { |
101 // The subframe was an ad to begin with, keep tracking it as an ad. | 98 // The subframe was an ad to begin with, keep tracking it as an ad. |
102 ProcessOngoingNavigationResource(frame_tree_node_id); | 99 ProcessOngoingNavigationResource(frame_tree_node_id); |
103 | 100 |
104 if (frame_tree_node_id == id_and_data->second->frame_tree_node_id) { | 101 if (frame_tree_node_id == id_and_data->second->frame_tree_node_id) { |
105 // This is the top-most frame in the ad. | 102 // This is the top-most frame in the ad. |
106 UMA_HISTOGRAM_BOOLEAN( | 103 UMA_HISTOGRAM_BOOLEAN( |
107 "PageLoad.Clients.Ads.Google.Navigations.AdFrameRenavigatedToAd", | 104 "PageLoad.Clients.Ads.Google.Navigations.AdFrameRenavigatedToAd", |
108 FrameIsAd(navigation_handle)); | 105 FrameIsAd(navigation_handle)); |
109 } | 106 } |
110 | |
111 return CONTINUE_OBSERVING; | 107 return CONTINUE_OBSERVING; |
112 } | 108 } |
113 // This frame was previously not an ad, process it as usual. If it had | 109 // This frame was previously not an ad, process it as usual. If it had |
114 // any child frames that were ads, those will still be recorded. | 110 // any child frames that were ads, those will still be recorded. |
115 UMA_HISTOGRAM_BOOLEAN( | 111 UMA_HISTOGRAM_BOOLEAN( |
116 "PageLoad.Clients.Ads.Google.Navigations.NonAdFrameRenavigatedToAd", | 112 "PageLoad.Clients.Ads.Google.Navigations.NonAdFrameRenavigatedToAd", |
117 FrameIsAd(navigation_handle)); | 113 FrameIsAd(navigation_handle)); |
118 } else if (top_level_subframe) { | 114 } else if (navigation_handle->IsParentMainFrame()) { |
119 top_level_subframe_count_ += 1; | 115 top_level_subframe_count_ += 1; |
120 } | 116 } |
121 | 117 |
122 // Determine who the parent frame's ad ancestor is. | 118 // Determine who the parent frame's ad ancestor is. |
123 const auto& parent_id_and_data = | 119 const auto& parent_id_and_data = |
124 ad_frames_data_.find(parent_frame_host->GetFrameTreeNodeId()); | 120 ad_frames_data_.find(parent_frame_host->GetFrameTreeNodeId()); |
125 DCHECK(parent_id_and_data != ad_frames_data_.end()); | 121 DCHECK(parent_id_and_data != ad_frames_data_.end()); |
126 AdFrameData* ad_data = parent_id_and_data->second; | 122 AdFrameData* ad_data = parent_id_and_data->second; |
127 | 123 |
128 if (!ad_data && FrameIsAd(navigation_handle)) { | 124 if (!ad_data && FrameIsAd(navigation_handle)) { |
129 // This frame is not nested within an ad frame but is itself an ad. | 125 // This frame is not nested within an ad frame but is itself an ad. |
130 ad_frames_data_storage_.emplace_back(frame_tree_node_id); | 126 ad_frames_data_storage_.emplace_back(frame_tree_node_id); |
131 ad_data = &ad_frames_data_storage_.back(); | 127 ad_data = &ad_frames_data_storage_.back(); |
132 } | 128 } |
133 | 129 |
134 ad_frames_data_[frame_tree_node_id] = ad_data; | 130 ad_frames_data_[frame_tree_node_id] = ad_data; |
135 | 131 |
136 if (top_level_subframe && ad_data) | 132 if (navigation_handle->IsParentMainFrame() && ad_data) |
137 top_level_ad_frame_count_ += 1; | 133 top_level_ad_frame_count_ += 1; |
138 | 134 |
139 ProcessOngoingNavigationResource(frame_tree_node_id); | 135 ProcessOngoingNavigationResource(frame_tree_node_id); |
140 return CONTINUE_OBSERVING; | 136 return CONTINUE_OBSERVING; |
141 } | 137 } |
142 | 138 |
143 page_load_metrics::PageLoadMetricsObserver::ObservePolicy | 139 page_load_metrics::PageLoadMetricsObserver::ObservePolicy |
144 AdsPageLoadMetricsObserver::FlushMetricsOnAppEnterBackground( | 140 AdsPageLoadMetricsObserver::FlushMetricsOnAppEnterBackground( |
145 const page_load_metrics::PageLoadTiming& timing, | 141 const page_load_metrics::PageLoadTiming& timing, |
146 const page_load_metrics::PageLoadExtraInfo& extra_info) { | 142 const page_load_metrics::PageLoadExtraInfo& extra_info) { |
(...skipping 22 matching lines...) Expand all Loading... | |
169 // Data uris should be accounted for in the generating resource, not | 165 // Data uris should be accounted for in the generating resource, not |
170 // here. Blobs for PlzNavigate shouldn't be counted as the http resource | 166 // here. Blobs for PlzNavigate shouldn't be counted as the http resource |
171 // was already counted. Blobs for other things like CacheStorage or | 167 // was already counted. Blobs for other things like CacheStorage or |
172 // IndexedDB are also ignored for now, as they're not normal HTTP loads. | 168 // IndexedDB are also ignored for now, as they're not normal HTTP loads. |
173 return; | 169 return; |
174 } | 170 } |
175 | 171 |
176 const auto& id_and_data = | 172 const auto& id_and_data = |
177 ad_frames_data_.find(extra_request_info.frame_tree_node_id); | 173 ad_frames_data_.find(extra_request_info.frame_tree_node_id); |
178 if (id_and_data == ad_frames_data_.end()) { | 174 if (id_and_data == ad_frames_data_.end()) { |
179 // This resouce is for a frame that hasn't yet committed. It must be the | 175 // This resource is for a frame that hasn't ever finished a navigation. We |
180 // main document for the frame. Hold onto it and once it commits we'll run | 176 // expect it to be the frame's main resource but don't have enough |
181 // it in ProcessOngoingNavigationResource. | 177 // confidence in that to dcheck it. For example, there might be races |
182 // TODO(jkarlin): Plumb the resource type through and DCHECK that the type | 178 // between doc.written resources and navigation failure. |
183 // is document. | 179 // TODO(jkarlin): Add UMA to measure how often we see multiple resources |
184 auto it_and_success = ongoing_navigation_resources_.emplace( | 180 // for a frame before navigation finishes. |
181 ongoing_navigation_resources_.emplace( | |
185 std::piecewise_construct, | 182 std::piecewise_construct, |
186 std::forward_as_tuple(extra_request_info.frame_tree_node_id), | 183 std::forward_as_tuple(extra_request_info.frame_tree_node_id), |
187 std::forward_as_tuple( | 184 std::forward_as_tuple( |
188 extra_request_info.url, extra_request_info.frame_tree_node_id, | 185 extra_request_info.url, extra_request_info.frame_tree_node_id, |
189 extra_request_info.was_cached, extra_request_info.raw_body_bytes, | 186 extra_request_info.was_cached, extra_request_info.raw_body_bytes, |
190 extra_request_info.original_network_content_length, nullptr, | 187 extra_request_info.original_network_content_length, nullptr, |
191 extra_request_info.resource_type)); | 188 extra_request_info.resource_type)); |
192 DCHECK(it_and_success.second); | |
193 return; | 189 return; |
194 } | 190 } |
195 | 191 |
196 page_bytes_ += extra_request_info.raw_body_bytes; | 192 page_bytes_ += extra_request_info.raw_body_bytes; |
197 if (!extra_request_info.was_cached) | 193 if (!extra_request_info.was_cached) |
198 uncached_page_bytes_ += extra_request_info.raw_body_bytes; | 194 uncached_page_bytes_ += extra_request_info.raw_body_bytes; |
199 | 195 |
200 // Determine if the frame (or its ancestor) is an ad, if so attribute the | 196 // Determine if the frame (or its ancestor) is an ad, if so attribute the |
201 // bytes to the highest ad ancestor. | 197 // bytes to the highest ad ancestor. |
202 AdFrameData* ancestor_data = id_and_data->second; | 198 AdFrameData* ancestor_data = id_and_data->second; |
(...skipping 86 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... | |
289 void AdsPageLoadMetricsObserver::ProcessOngoingNavigationResource( | 285 void AdsPageLoadMetricsObserver::ProcessOngoingNavigationResource( |
290 FrameTreeNodeId frame_tree_node_id) { | 286 FrameTreeNodeId frame_tree_node_id) { |
291 const auto& frame_id_and_request = | 287 const auto& frame_id_and_request = |
292 ongoing_navigation_resources_.find(frame_tree_node_id); | 288 ongoing_navigation_resources_.find(frame_tree_node_id); |
293 if (frame_id_and_request == ongoing_navigation_resources_.end()) | 289 if (frame_id_and_request == ongoing_navigation_resources_.end()) |
294 return; | 290 return; |
295 | 291 |
296 ProcessLoadedResource(frame_id_and_request->second); | 292 ProcessLoadedResource(frame_id_and_request->second); |
297 ongoing_navigation_resources_.erase(frame_id_and_request); | 293 ongoing_navigation_resources_.erase(frame_id_and_request); |
298 } | 294 } |
OLD | NEW |