chrome/renderer/safe_browsing/scorer.h

   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
   3 // found in the LICENSE file.
   4 //
   5 // This class loads a client-side model and lets you compute a phishing score
   6 // for a set of previously extracted features.  The phishing score corresponds
   7 // to the probability that the features are indicative of a phishing site.
   8 //
   9 // For more details on how the score is actually computed for a given model
  10 // and a given set of features read the comments in client_model.proto file.
  11 //
  12 // See features.h for a list of features that are currently used.
  13
  14 #ifndef CHROME_RENDERER_SAFE_BROWSING_SCORER_H_
  15 #define CHROME_RENDERER_SAFE_BROWSING_SCORER_H_
  16
  17 #include <string>
  18
  19 #include "base/basictypes.h"
  20 #include "base/containers/hash_tables.h"
  21 #include "base/strings/string_piece.h"
  22 #include "chrome/common/safe_browsing/client_model.pb.h"
  23
  24 namespace safe_browsing {
  25 class FeatureMap;
  26
  27 // Scorer methods are virtual to simplify mocking of this class.
  28 class Scorer {
  29  public:
  30   virtual ~Scorer();
  31
  32   // Factory method which creates a new Scorer object by parsing the given
  33   // model.  If parsing fails this method returns NULL.
  34   static Scorer* Create(const base::StringPiece& model_str);
  35
  36   // This method computes the probability that the given features are indicative
  37   // of phishing.  It returns a score value that falls in the range [0.0,1.0]
  38   // (range is inclusive on both ends).
  39   virtual double ComputeScore(const FeatureMap& features) const;
  40
  41   // Returns the version number of the loaded client model.
  42   int model_version() const;
  43
  44   // -- Accessors used by the page feature extractor ---------------------------
  45
  46   // Returns a set of hashed page terms that appear in the model in binary
  47   // format.
  48   const base::hash_set<std::string>& page_terms() const;
  49
  50   // Returns a set of hashed page words that appear in the model in binary
  51   // format.
  52   const base::hash_set<uint32>& page_words() const;
  53
  54   // Return the maximum number of words per term for the loaded model.
  55   size_t max_words_per_term() const;
  56
  57   // Returns the murmurhash3 seed for the loaded model.
  58   uint32 murmurhash3_seed() const;
  59
  60   // Return the maximum number of unique shingle hashes per page.
  61   size_t max_shingles_per_page() const;
  62
  63   // Return the number of words in a shingle.
  64   size_t shingle_size() const;
  65
  66  protected:
  67   // Most clients should use the factory method.  This constructor is public
  68   // to allow for mock implementations.
  69   Scorer();
  70
  71  private:
  72   friend class PhishingScorerTest;
  73
  74   // Computes the score for a given rule and feature map.  The score is computed
  75   // by multiplying the rule weight with the product of feature weights for the
  76   // given rule.  The feature weights are stored in the feature map.  If a
  77   // particular feature does not exist in the feature map we set its weight to
  78   // zero.
  79   double ComputeRuleScore(const ClientSideModel::Rule& rule,
  80                           const FeatureMap& features) const;
  81
  82   ClientSideModel model_;
  83   base::hash_set<std::string> page_terms_;
  84   base::hash_set<uint32> page_words_;
  85
  86   DISALLOW_COPY_AND_ASSIGN(Scorer);
  87 };
  88 }  // namespace safe_browsing
  89
  90 #endif  // CHROME_RENDERER_SAFE_BROWSING_SCORER_H_