OLD | NEW |
1 // Copyright (c) 2011 The Chromium Authors. All rights reserved. | 1 // Copyright (c) 2011 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 // PhishingTermFeatureExtractor handles computing term features from the text | 5 // PhishingTermFeatureExtractor handles computing term features from the text |
6 // of a web page for the client-side phishing detection model. To do this, it | 6 // of a web page for the client-side phishing detection model. To do this, it |
7 // takes a list of terms that appear in the model, and scans through the page | 7 // takes a list of terms that appear in the model, and scans through the page |
8 // text looking for them. Any terms that appear will cause a corresponding | 8 // text looking for them. Any terms that appear will cause a corresponding |
9 // features::kPageTerm feature to be added to the FeatureMap. | 9 // features::kPageTerm feature to be added to the FeatureMap. |
10 // | 10 // |
11 // To make it harder for a phisher to enumerate all of the relevant terms in | 11 // To make it harder for a phisher to enumerate all of the relevant terms in |
12 // the model, the terms are provided as SHA-256 hashes, rather than plain text. | 12 // the model, the terms are provided as SHA-256 hashes, rather than plain text. |
13 // | 13 // |
14 // There is one PhishingTermFeatureExtractor per RenderView. | 14 // There is one PhishingTermFeatureExtractor per RenderView. |
15 | 15 |
16 #ifndef CHROME_RENDERER_SAFE_BROWSING_PHISHING_TERM_FEATURE_EXTRACTOR_H_ | 16 #ifndef CHROME_RENDERER_SAFE_BROWSING_PHISHING_TERM_FEATURE_EXTRACTOR_H_ |
17 #define CHROME_RENDERER_SAFE_BROWSING_PHISHING_TERM_FEATURE_EXTRACTOR_H_ | 17 #define CHROME_RENDERER_SAFE_BROWSING_PHISHING_TERM_FEATURE_EXTRACTOR_H_ |
18 | 18 |
19 #include <string> | 19 #include <string> |
20 | 20 |
21 #include "base/basictypes.h" | 21 #include "base/basictypes.h" |
22 #include "base/callback_old.h" | 22 #include "base/callback.h" |
23 #include "base/hash_tables.h" | 23 #include "base/hash_tables.h" |
24 #include "base/memory/mru_cache.h" | 24 #include "base/memory/mru_cache.h" |
25 #include "base/memory/scoped_ptr.h" | 25 #include "base/memory/scoped_ptr.h" |
26 #include "base/string_piece.h" | 26 #include "base/string_piece.h" |
27 #include "base/string16.h" | 27 #include "base/string16.h" |
28 #include "base/task.h" | 28 #include "base/task.h" |
29 | 29 |
30 namespace safe_browsing { | 30 namespace safe_browsing { |
31 class FeatureExtractorClock; | 31 class FeatureExtractorClock; |
32 class FeatureMap; | 32 class FeatureMap; |
33 | 33 |
34 class PhishingTermFeatureExtractor { | 34 class PhishingTermFeatureExtractor { |
35 public: | 35 public: |
36 // Callback to be run when feature extraction finishes. The callback | 36 // Callback to be run when feature extraction finishes. The callback |
37 // argument is true if extraction was successful, false otherwise. | 37 // argument is true if extraction was successful, false otherwise. |
38 typedef Callback1<bool>::Type DoneCallback; | 38 typedef base::Callback<void(bool)> DoneCallback; |
39 | 39 |
40 // Creates a PhishingTermFeatureExtractor which will extract features for | 40 // Creates a PhishingTermFeatureExtractor which will extract features for |
41 // all of the terms whose SHA-256 hashes are in |page_term_hashes|. These | 41 // all of the terms whose SHA-256 hashes are in |page_term_hashes|. These |
42 // terms may be multi-word n-grams, with at most |max_words_per_term| words. | 42 // terms may be multi-word n-grams, with at most |max_words_per_term| words. |
43 // | 43 // |
44 // |page_word_hashes| contains the murmur3 hashes for all of the individual | 44 // |page_word_hashes| contains the murmur3 hashes for all of the individual |
45 // words that make up the terms. Both sets of strings are UTF-8 encoded and | 45 // words that make up the terms. Both sets of strings are UTF-8 encoded and |
46 // lowercased prior to hashing. The caller owns both sets of strings, and | 46 // lowercased prior to hashing. The caller owns both sets of strings, and |
47 // must ensure that they are valid until the PhishingTermFeatureExtractor is | 47 // must ensure that they are valid until the PhishingTermFeatureExtractor is |
48 // destroyed. | 48 // destroyed. |
(...skipping 16 matching lines...) Expand all Loading... |
65 // may run in several chunks of work, posting a task to the current | 65 // may run in several chunks of work, posting a task to the current |
66 // MessageLoop to continue processing. Once feature extraction is complete, | 66 // MessageLoop to continue processing. Once feature extraction is complete, |
67 // |done_callback| is run on the current thread. | 67 // |done_callback| is run on the current thread. |
68 // PhishingTermFeatureExtractor takes ownership of the callback. | 68 // PhishingTermFeatureExtractor takes ownership of the callback. |
69 // | 69 // |
70 // |page_text| and |features| are owned by the caller, and must not be | 70 // |page_text| and |features| are owned by the caller, and must not be |
71 // destroyed until either |done_callback| is run or | 71 // destroyed until either |done_callback| is run or |
72 // CancelPendingExtraction() is called. | 72 // CancelPendingExtraction() is called. |
73 void ExtractFeatures(const string16* page_text, | 73 void ExtractFeatures(const string16* page_text, |
74 FeatureMap* features, | 74 FeatureMap* features, |
75 DoneCallback* done_callback); | 75 const DoneCallback& done_callback); |
76 | 76 |
77 // Cancels any pending feature extraction. The DoneCallback will not be run. | 77 // Cancels any pending feature extraction. The DoneCallback will not be run. |
78 // Must be called if there is a feature extraction in progress when the page | 78 // Must be called if there is a feature extraction in progress when the page |
79 // is unloaded or the PhishingTermFeatureExtractor is destroyed. | 79 // is unloaded or the PhishingTermFeatureExtractor is destroyed. |
80 void CancelPendingExtraction(); | 80 void CancelPendingExtraction(); |
81 | 81 |
82 private: | 82 private: |
83 struct ExtractionState; | 83 struct ExtractionState; |
84 | 84 |
85 // The maximum amount of wall time that we will spend on a single extraction | 85 // The maximum amount of wall time that we will spend on a single extraction |
(...skipping 55 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
141 // an LRU cache (i.e. it evicts the oldest accesses first). | 141 // an LRU cache (i.e. it evicts the oldest accesses first). |
142 typedef base::HashingMRUCache<base::StringPiece16, bool> WordCache; | 142 typedef base::HashingMRUCache<base::StringPiece16, bool> WordCache; |
143 WordCache negative_word_cache_; | 143 WordCache negative_word_cache_; |
144 | 144 |
145 // Non-owned pointer to our clock. | 145 // Non-owned pointer to our clock. |
146 FeatureExtractorClock* clock_; | 146 FeatureExtractorClock* clock_; |
147 | 147 |
148 // The output parameters from the most recent call to ExtractFeatures(). | 148 // The output parameters from the most recent call to ExtractFeatures(). |
149 const string16* page_text_; // The caller keeps ownership of this. | 149 const string16* page_text_; // The caller keeps ownership of this. |
150 FeatureMap* features_; // The caller keeps ownership of this. | 150 FeatureMap* features_; // The caller keeps ownership of this. |
151 scoped_ptr<DoneCallback> done_callback_; | 151 DoneCallback done_callback_; |
152 | 152 |
153 // Stores the current state of term extraction from |page_text_|. | 153 // Stores the current state of term extraction from |page_text_|. |
154 scoped_ptr<ExtractionState> state_; | 154 scoped_ptr<ExtractionState> state_; |
155 | 155 |
156 // Used in scheduling ExtractFeaturesWithTimeout tasks. | 156 // Used in scheduling ExtractFeaturesWithTimeout tasks. |
157 // These pointers are invalidated if extraction is cancelled. | 157 // These pointers are invalidated if extraction is cancelled. |
158 base::WeakPtrFactory<PhishingTermFeatureExtractor> weak_factory_; | 158 base::WeakPtrFactory<PhishingTermFeatureExtractor> weak_factory_; |
159 | 159 |
160 DISALLOW_COPY_AND_ASSIGN(PhishingTermFeatureExtractor); | 160 DISALLOW_COPY_AND_ASSIGN(PhishingTermFeatureExtractor); |
161 }; | 161 }; |
162 | 162 |
163 } // namespace safe_browsing | 163 } // namespace safe_browsing |
164 | 164 |
165 #endif // CHROME_RENDERER_SAFE_BROWSING_PHISHING_TERM_FEATURE_EXTRACTOR_H_ | 165 #endif // CHROME_RENDERER_SAFE_BROWSING_PHISHING_TERM_FEATURE_EXTRACTOR_H_ |
OLD | NEW |