Extracting page shingle hashes for similarity detection.

chrome/renderer/safe_browsing/scorer.h

 // Copyright (c) 2011 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.
 //
 // This class loads a client-side model and lets you compute a phishing score
 // for a set of previously extracted features.  The phishing score corresponds
 // to the probability that the features are indicative of a phishing site.
 //
 // For more details on how the score is actually computed for a given model
 // and a given set of features read the comments in client_model.proto file.
@@ skipping 39 matching lines @@
   // Returns a set of hashed page words that appear in the model in binary
   // format.
   const base::hash_set<uint32>& page_words() const;
 
   // Return the maximum number of words per term for the loaded model.
   size_t max_words_per_term() const;
 
   // Returns the murmurhash3 seed for the loaded model.
   uint32 murmurhash3_seed() const;
 
+  // Return the maximum number of unique shingle hashes per page.
+  size_t max_shingles_per_page() const;
+
+  // Return the number of words in a shingle.
+  size_t shingle_size() const;
+
  protected:
   // Most clients should use the factory method.  This constructor is public
   // to allow for mock implementations.
   Scorer();
 
  private:
   friend class PhishingScorerTest;
 
   // Computes the score for a given rule and feature map.  The score is computed
   // by multiplying the rule weight with the product of feature weights for the
   // given rule.  The feature weights are stored in the feature map.  If a
   // particular feature does not exist in the feature map we set its weight to
   // zero.
   double ComputeRuleScore(const ClientSideModel::Rule& rule,
                           const FeatureMap& features) const;
 
   ClientSideModel model_;
   base::hash_set<std::string> page_terms_;
   base::hash_set<uint32> page_words_;
 
   DISALLOW_COPY_AND_ASSIGN(Scorer);
 };
-}  // namepsace safe_browsing
+}  // namespace safe_browsing
 
 #endif  // CHROME_RENDERER_SAFE_BROWSING_SCORER_H_