OLD | NEW |
1 // Copyright (c) 2010 The Chromium Authors. All rights reserved. | 1 // Copyright (c) 2010 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/renderer/safe_browsing/phishing_classifier_delegate.h" | 5 #include "chrome/renderer/safe_browsing/phishing_classifier_delegate.h" |
6 | 6 |
| 7 #include <set> |
| 8 |
7 #include "base/callback.h" | 9 #include "base/callback.h" |
| 10 #include "base/lazy_instance.h" |
8 #include "base/logging.h" | 11 #include "base/logging.h" |
| 12 #include "base/scoped_callback_factory.h" |
9 #include "chrome/common/render_messages.h" | 13 #include "chrome/common/render_messages.h" |
10 #include "chrome/renderer/navigation_state.h" | 14 #include "chrome/renderer/navigation_state.h" |
| 15 #include "chrome/renderer/render_thread.h" |
11 #include "chrome/renderer/render_view.h" | 16 #include "chrome/renderer/render_view.h" |
12 #include "chrome/renderer/safe_browsing/feature_extractor_clock.h" | 17 #include "chrome/renderer/safe_browsing/feature_extractor_clock.h" |
13 #include "chrome/renderer/safe_browsing/phishing_classifier.h" | 18 #include "chrome/renderer/safe_browsing/phishing_classifier.h" |
| 19 #include "chrome/renderer/safe_browsing/scorer.h" |
14 #include "third_party/WebKit/Source/WebKit/chromium/public/WebFrame.h" | 20 #include "third_party/WebKit/Source/WebKit/chromium/public/WebFrame.h" |
15 #include "third_party/WebKit/Source/WebKit/chromium/public/WebURL.h" | 21 #include "third_party/WebKit/Source/WebKit/chromium/public/WebURL.h" |
16 #include "third_party/WebKit/Source/WebKit/chromium/public/WebView.h" | 22 #include "third_party/WebKit/Source/WebKit/chromium/public/WebView.h" |
17 | 23 |
18 namespace safe_browsing { | 24 namespace safe_browsing { |
19 | 25 |
| 26 typedef std::set<PhishingClassifierDelegate*> PhishingClassifierDelegates; |
| 27 static base::LazyInstance<PhishingClassifierDelegates> |
| 28 g_delegates(base::LINKER_INITIALIZED); |
| 29 |
| 30 static base::LazyInstance<scoped_ptr<const safe_browsing::Scorer> > |
| 31 g_phishing_scorer(base::LINKER_INITIALIZED); |
| 32 |
| 33 class ScorerCallback { |
| 34 public: |
| 35 static Scorer::CreationCallback* CreateCallback() { |
| 36 ScorerCallback* scorer_callback = new ScorerCallback(); |
| 37 return scorer_callback->callback_factory_->NewCallback( |
| 38 &ScorerCallback::PhishingScorerCreated); |
| 39 } |
| 40 |
| 41 private: |
| 42 ScorerCallback() { |
| 43 callback_factory_.reset( |
| 44 new base::ScopedCallbackFactory<ScorerCallback>(this)); |
| 45 } |
| 46 |
| 47 // Callback to be run once the phishing Scorer has been created. |
| 48 void PhishingScorerCreated(safe_browsing::Scorer* scorer) { |
| 49 if (!scorer) { |
| 50 DLOG(ERROR) << "Unable to create a PhishingScorer - corrupt model?"; |
| 51 return; |
| 52 } |
| 53 |
| 54 g_phishing_scorer.Get().reset(scorer); |
| 55 |
| 56 PhishingClassifierDelegates::iterator i; |
| 57 for (i = g_delegates.Get().begin(); i != g_delegates.Get().end(); ++i) |
| 58 (*i)->SetPhishingScorer(scorer); |
| 59 |
| 60 delete this; |
| 61 } |
| 62 |
| 63 scoped_ptr<base::ScopedCallbackFactory<ScorerCallback> > callback_factory_; |
| 64 }; |
| 65 |
| 66 void PhishingClassifierDelegate::SetPhishingModel( |
| 67 IPC::PlatformFileForTransit model_file) { |
| 68 safe_browsing::Scorer::CreateFromFile( |
| 69 IPC::PlatformFileForTransitToPlatformFile(model_file), |
| 70 RenderThread::current()->GetFileThreadMessageLoopProxy(), |
| 71 ScorerCallback::CreateCallback()); |
| 72 } |
| 73 |
20 PhishingClassifierDelegate::PhishingClassifierDelegate( | 74 PhishingClassifierDelegate::PhishingClassifierDelegate( |
21 RenderView* render_view, | 75 RenderView* render_view, |
22 PhishingClassifier* classifier) | 76 PhishingClassifier* classifier) |
23 : render_view_(render_view), | 77 : RenderViewObserver(render_view), |
24 last_page_id_sent_to_classifier_(-1), | 78 last_page_id_sent_to_classifier_(-1), |
25 pending_classification_(false) { | 79 pending_classification_(false) { |
| 80 g_delegates.Get().insert(this); |
26 if (!classifier) { | 81 if (!classifier) { |
27 classifier = new PhishingClassifier(render_view_, | 82 classifier = new PhishingClassifier(render_view, |
28 new FeatureExtractorClock()); | 83 new FeatureExtractorClock()); |
29 } | 84 } |
| 85 |
30 classifier_.reset(classifier); | 86 classifier_.reset(classifier); |
| 87 |
| 88 if (g_phishing_scorer.Get().get()) |
| 89 SetPhishingScorer(g_phishing_scorer.Get().get()); |
31 } | 90 } |
32 | 91 |
33 PhishingClassifierDelegate::~PhishingClassifierDelegate() { | 92 PhishingClassifierDelegate::~PhishingClassifierDelegate() { |
34 CancelPendingClassification(); | 93 CancelPendingClassification(); |
| 94 g_delegates.Get().erase(this); |
35 } | 95 } |
36 | 96 |
37 void PhishingClassifierDelegate::SetPhishingScorer( | 97 void PhishingClassifierDelegate::SetPhishingScorer( |
38 const safe_browsing::Scorer* scorer) { | 98 const safe_browsing::Scorer* scorer) { |
| 99 if (!render_view()->webview()) |
| 100 return; // RenderView is tearing down. |
| 101 |
39 classifier_->set_phishing_scorer(scorer); | 102 classifier_->set_phishing_scorer(scorer); |
40 | 103 |
41 if (pending_classification_) { | 104 if (pending_classification_) { |
42 pending_classification_ = false; | 105 pending_classification_ = false; |
43 // If we have a pending classificaton, it should always be true that the | 106 // If we have a pending classificaton, it should always be true that the |
44 // main frame URL and page id have not changed since we queued the | 107 // main frame URL and page id have not changed since we queued the |
45 // classification. This is because we stop any pending classification on | 108 // classification. This is because we stop any pending classification on |
46 // main frame loads in RenderView::didCommitProvisionalLoad(). | 109 // main frame loads in RenderView::didCommitProvisionalLoad(). |
47 DCHECK_EQ(StripToplevelUrl(), last_url_sent_to_classifier_); | 110 DCHECK_EQ(StripToplevelUrl(), last_url_sent_to_classifier_); |
48 DCHECK_EQ(render_view_->page_id(), last_page_id_sent_to_classifier_); | 111 DCHECK_EQ(render_view()->page_id(), last_page_id_sent_to_classifier_); |
49 classifier_->BeginClassification( | 112 classifier_->BeginClassification( |
50 &classifier_page_text_, | 113 &classifier_page_text_, |
51 NewCallback(this, &PhishingClassifierDelegate::ClassificationDone)); | 114 NewCallback(this, &PhishingClassifierDelegate::ClassificationDone)); |
52 } | 115 } |
53 } | 116 } |
54 | 117 |
55 void PhishingClassifierDelegate::CommittedLoadInFrame( | 118 void PhishingClassifierDelegate::DidCommitProvisionalLoad( |
56 WebKit::WebFrame* frame) { | 119 WebKit::WebFrame* frame, bool is_new_navigation) { |
| 120 if (!is_new_navigation) |
| 121 return; |
| 122 |
57 // A new page is starting to load. Unless the load is a navigation within | 123 // A new page is starting to load. Unless the load is a navigation within |
58 // the same page, we need to cancel classification since the content will | 124 // the same page, we need to cancel classification since the content will |
59 // now be inconsistent with the phishing model. | 125 // now be inconsistent with the phishing model. |
60 NavigationState* state = NavigationState::FromDataSource( | 126 NavigationState* state = NavigationState::FromDataSource( |
61 frame->dataSource()); | 127 frame->dataSource()); |
62 if (!state->was_within_same_page()) { | 128 if (!state->was_within_same_page()) { |
63 CancelPendingClassification(); | 129 CancelPendingClassification(); |
64 } | 130 } |
65 } | 131 } |
66 | 132 |
67 void PhishingClassifierDelegate::FinishedLoad(string16* page_text) { | 133 void PhishingClassifierDelegate::PageCaptured(const string16& page_text) { |
68 // We check that the page id has incremented so that we don't reclassify | 134 // We check that the page id has incremented so that we don't reclassify |
69 // pages as the user moves back and forward in session history. Note: we | 135 // pages as the user moves back and forward in session history. Note: we |
70 // don't send every page id to the classifier, only those where the toplevel | 136 // don't send every page id to the classifier, only those where the toplevel |
71 // URL changed. | 137 // URL changed. |
72 int load_id = render_view_->page_id(); | 138 int load_id = render_view()->page_id(); |
73 if (load_id <= last_page_id_sent_to_classifier_) { | 139 if (load_id <= last_page_id_sent_to_classifier_) { |
74 return; | 140 return; |
75 } | 141 } |
76 | 142 |
77 GURL url_without_ref = StripToplevelUrl(); | 143 GURL url_without_ref = StripToplevelUrl(); |
78 if (url_without_ref == last_url_sent_to_classifier_) { | 144 if (url_without_ref == last_url_sent_to_classifier_) { |
79 // The toplevle URL is the same, except for the ref. | 145 // The toplevle URL is the same, except for the ref. |
80 // Update the last page id we sent, but don't trigger a new classification. | 146 // Update the last page id we sent, but don't trigger a new classification. |
81 last_page_id_sent_to_classifier_ = load_id; | 147 last_page_id_sent_to_classifier_ = load_id; |
82 return; | 148 return; |
83 } | 149 } |
84 | 150 |
85 last_url_sent_to_classifier_ = url_without_ref; | 151 last_url_sent_to_classifier_ = url_without_ref; |
86 last_page_id_sent_to_classifier_ = load_id; | 152 last_page_id_sent_to_classifier_ = load_id; |
87 classifier_page_text_.swap(*page_text); | 153 classifier_page_text_ = page_text; |
88 | 154 |
89 if (classifier_->is_ready()) { | 155 if (classifier_->is_ready()) { |
90 classifier_->BeginClassification( | 156 classifier_->BeginClassification( |
91 &classifier_page_text_, | 157 &classifier_page_text_, |
92 NewCallback(this, &PhishingClassifierDelegate::ClassificationDone)); | 158 NewCallback(this, &PhishingClassifierDelegate::ClassificationDone)); |
93 } else { | 159 } else { |
94 // If there is no phishing classifier yet, we'll begin classification once | 160 // If there is no phishing classifier yet, we'll begin classification once |
95 // SetPhishingScorer() is called by the RenderView. | 161 // SetPhishingScorer() is called by the RenderView. |
96 pending_classification_ = true; | 162 pending_classification_ = true; |
97 } | 163 } |
98 } | 164 } |
99 | 165 |
100 void PhishingClassifierDelegate::CancelPendingClassification() { | 166 void PhishingClassifierDelegate::CancelPendingClassification() { |
101 if (classifier_->is_ready()) { | 167 if (classifier_->is_ready()) { |
102 classifier_->CancelPendingClassification(); | 168 classifier_->CancelPendingClassification(); |
103 } | 169 } |
104 classifier_page_text_.clear(); | 170 classifier_page_text_.clear(); |
105 pending_classification_ = false; | 171 pending_classification_ = false; |
106 } | 172 } |
107 | 173 |
| 174 bool PhishingClassifierDelegate::OnMessageReceived( |
| 175 const IPC::Message& message) { |
| 176 /* |
| 177 bool handled = true; |
| 178 IPC_BEGIN_MESSAGE_MAP(PhishingClassifierDelegate, message) |
| 179 IPC_MESSAGE_UNHANDLED(handled = false) |
| 180 IPC_END_MESSAGE_MAP() |
| 181 return handled; |
| 182 */ |
| 183 return false; |
| 184 } |
| 185 |
108 void PhishingClassifierDelegate::ClassificationDone(bool is_phishy, | 186 void PhishingClassifierDelegate::ClassificationDone(bool is_phishy, |
109 double phishy_score) { | 187 double phishy_score) { |
110 // We no longer need the page text. | 188 // We no longer need the page text. |
111 classifier_page_text_.clear(); | 189 classifier_page_text_.clear(); |
112 VLOG(2) << "Phishy verdict = " << is_phishy << " score = " << phishy_score; | 190 VLOG(2) << "Phishy verdict = " << is_phishy << " score = " << phishy_score; |
113 if (!is_phishy) { | 191 if (!is_phishy) { |
114 return; | 192 return; |
115 } | 193 } |
116 | 194 |
117 render_view_->Send(new ViewHostMsg_DetectedPhishingSite( | 195 render_view()->Send(new ViewHostMsg_DetectedPhishingSite( |
118 render_view_->routing_id(), | 196 render_view()->routing_id(), |
119 last_url_sent_to_classifier_, | 197 last_url_sent_to_classifier_, |
120 phishy_score)); | 198 phishy_score)); |
121 } | 199 } |
122 | 200 |
123 GURL PhishingClassifierDelegate::StripToplevelUrl() { | 201 GURL PhishingClassifierDelegate::StripToplevelUrl() { |
124 GURL toplevel_url = render_view_->webview()->mainFrame()->url(); | 202 GURL toplevel_url = render_view()->webview()->mainFrame()->url(); |
125 GURL::Replacements replacements; | 203 GURL::Replacements replacements; |
126 replacements.ClearRef(); | 204 replacements.ClearRef(); |
127 return toplevel_url.ReplaceComponents(replacements); | 205 return toplevel_url.ReplaceComponents(replacements); |
128 } | 206 } |
129 | 207 |
130 } // namespace safe_browsing | 208 } // namespace safe_browsing |
OLD | NEW |