Extracting page shingle hashes for similarity detection.

chrome/renderer/safe_browsing/scorer_unittest.cc

 // 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.
 
 #include "chrome/renderer/safe_browsing/scorer.h"
 
 #include "base/containers/hash_tables.h"
 #include "base/files/file_path.h"
 #include "base/files/scoped_temp_dir.h"
 #include "base/format_macros.h"
@@ skipping 37 matching lines @@
     model_.add_page_term(4);  // token two
 
     // These will be murmur3 hashes, but for this test it's not necessary
     // that the hashes correspond to actual words.
     model_.add_page_word(1000U);
     model_.add_page_word(2000U);
     model_.add_page_word(3000U);
 
     model_.set_max_words_per_term(2);
     model_.set_murmur_hash_seed(12345U);
+    model_.set_max_shingles_per_page(10);
+    model_.set_shingle_size(3);
   }
 
   ClientSideModel model_;
 };
 
 TEST_F(PhishingScorerTest, HasValidModel) {
   scoped_ptr<Scorer> scorer;
   scorer.reset(Scorer::Create(model_.SerializeAsString()));
   EXPECT_TRUE(scorer.get() != NULL);
 
@@ skipping 21 matching lines @@
   scoped_ptr<Scorer> scorer(Scorer::Create(model_.SerializeAsString()));
   ASSERT_TRUE(scorer.get());
   base::hash_set<uint32> expected_page_words;
   expected_page_words.insert(1000U);
   expected_page_words.insert(2000U);
   expected_page_words.insert(3000U);
   EXPECT_THAT(scorer->page_words(),
               ::testing::ContainerEq(expected_page_words));
   EXPECT_EQ(2U, scorer->max_words_per_term());
   EXPECT_EQ(12345U, scorer->murmurhash3_seed());
+  EXPECT_EQ(10U, scorer->max_shingles_per_page());
+  EXPECT_EQ(3U, scorer->shingle_size());
 }
 
 TEST_F(PhishingScorerTest, ComputeScore) {
   scoped_ptr<Scorer> scorer(Scorer::Create(model_.SerializeAsString()));
   ASSERT_TRUE(scorer.get());
 
   // An empty feature map should match the empty rule.
   FeatureMap features;
   // The expected logodds is 0.5 (empty rule) => p = exp(0.5) / (exp(0.5) + 1)
   // => 0.62245933120185459
   EXPECT_DOUBLE_EQ(0.62245933120185459, scorer->ComputeScore(features));
   // Same if the feature does not match any rule.
   EXPECT_TRUE(features.AddBooleanFeature("not existing feature"));
   EXPECT_DOUBLE_EQ(0.62245933120185459, scorer->ComputeScore(features));
 
   // Feature 1 matches which means that the logodds will be:
   //   0.5 (empty rule) + 2.0 (rule weight) * 0.15 (feature weight) = 0.8
   //   => p = 0.6899744811276125
   EXPECT_TRUE(features.AddRealFeature("feature1", 0.15));
   EXPECT_DOUBLE_EQ(0.6899744811276125, scorer->ComputeScore(features));
 
   // Now, both feature 1 and feature 2 match.  Expected logodds:
   //   0.5 (empty rule) + 2.0 (rule weight) * 0.15 (feature weight) +
   //   3.0 (rule weight) * 0.15 (feature1 weight) * 1.0 (feature2) weight = 9.8
   //   => p = 0.99999627336071584
   EXPECT_TRUE(features.AddBooleanFeature("feature2"));
   EXPECT_DOUBLE_EQ(0.77729986117469119, scorer->ComputeScore(features));
 }
 }  // namespace safe_browsing