src/chrome/renderer/safe_browsing/scorer_unittest.cc

   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 #include "chrome/renderer/safe_browsing/scorer.h"
   6
   7 #include "base/containers/hash_tables.h"
   8 #include "base/files/file_path.h"
   9 #include "base/files/scoped_temp_dir.h"
  10 #include "base/format_macros.h"
  11 #include "base/memory/scoped_ptr.h"
  12 #include "base/message_loop/message_loop.h"
  13 #include "base/threading/thread.h"
  14 #include "chrome/common/safe_browsing/client_model.pb.h"
  15 #include "chrome/renderer/safe_browsing/features.h"
  16 #include "testing/gmock/include/gmock/gmock.h"
  17 #include "testing/gtest/include/gtest/gtest.h"
  18
  19 namespace safe_browsing {
  20
  21 class PhishingScorerTest : public ::testing::Test {
  22  protected:
  23   virtual void SetUp() {
  24     // Setup a simple model.  Note that the scorer does not care about
  25     // how features are encoded so we use readable strings here to make
  26     // the test simpler to follow.
  27     model_.Clear();
  28     model_.add_hashes("feature1");
  29     model_.add_hashes("feature2");
  30     model_.add_hashes("feature3");
  31     model_.add_hashes("token one");
  32     model_.add_hashes("token two");
  33
  34     ClientSideModel::Rule* rule;
  35     rule = model_.add_rule();
  36     rule->set_weight(0.5);
  37
  38     rule = model_.add_rule();
  39     rule->add_feature(0);  // feature1
  40     rule->set_weight(2.0);
  41
  42     rule = model_.add_rule();
  43     rule->add_feature(0);  // feature1
  44     rule->add_feature(1);  // feature2
  45     rule->set_weight(3.0);
  46
  47     model_.add_page_term(3);  // token one
  48     model_.add_page_term(4);  // token two
  49
  50     // These will be murmur3 hashes, but for this test it's not necessary
  51     // that the hashes correspond to actual words.
  52     model_.add_page_word(1000U);
  53     model_.add_page_word(2000U);
  54     model_.add_page_word(3000U);
  55
  56     model_.set_max_words_per_term(2);
  57     model_.set_murmur_hash_seed(12345U);
  58     model_.set_max_shingles_per_page(10);
  59     model_.set_shingle_size(3);
  60   }
  61
  62   ClientSideModel model_;
  63 };
  64
  65 TEST_F(PhishingScorerTest, HasValidModel) {
  66   scoped_ptr<Scorer> scorer;
  67   scorer.reset(Scorer::Create(model_.SerializeAsString()));
  68   EXPECT_TRUE(scorer.get() != NULL);
  69
  70   // Invalid model string.
  71   scorer.reset(Scorer::Create("bogus string"));
  72   EXPECT_FALSE(scorer.get());
  73
  74   // Mode is missing a required field.
  75   model_.clear_max_words_per_term();
  76   scorer.reset(Scorer::Create(model_.SerializePartialAsString()));
  77   EXPECT_FALSE(scorer.get());
  78 }
  79
  80 TEST_F(PhishingScorerTest, PageTerms) {
  81   scoped_ptr<Scorer> scorer(Scorer::Create(model_.SerializeAsString()));
  82   ASSERT_TRUE(scorer.get());
  83
  84   // Use std::vector instead of base::hash_set for comparison.
  85   // On Android, EXPECT_THAT(..., ContainerEq(...)) doesn't support
  86   // std::hash_set, but std::vector works fine.
  87   std::vector<std::string> expected_page_terms;
  88   expected_page_terms.push_back("token one");
  89   expected_page_terms.push_back("token two");
  90   std::sort(expected_page_terms.begin(), expected_page_terms.end());
  91
  92   base::hash_set<std::string> page_terms = scorer->page_terms();
  93   std::vector<std::string> page_terms_v(page_terms.begin(), page_terms.end());
  94   std::sort(page_terms_v.begin(), page_terms_v.end());
  95
  96   EXPECT_THAT(page_terms_v, ::testing::ContainerEq(expected_page_terms));
  97 }
  98
  99 TEST_F(PhishingScorerTest, PageWords) {
 100   scoped_ptr<Scorer> scorer(Scorer::Create(model_.SerializeAsString()));
 101   ASSERT_TRUE(scorer.get());
 102   std::vector<uint32> expected_page_words;
 103   expected_page_words.push_back(1000U);
 104   expected_page_words.push_back(2000U);
 105   expected_page_words.push_back(3000U);
 106   std::sort(expected_page_words.begin(), expected_page_words.end());
 107
 108   base::hash_set<uint32> page_words = scorer->page_words();
 109   std::vector<uint32> page_words_v(page_words.begin(), page_words.end());
 110   std::sort(page_words_v.begin(), page_words_v.end());
 111
 112   EXPECT_THAT(page_words_v, ::testing::ContainerEq(expected_page_words));
 113
 114   EXPECT_EQ(2U, scorer->max_words_per_term());
 115   EXPECT_EQ(12345U, scorer->murmurhash3_seed());
 116   EXPECT_EQ(10U, scorer->max_shingles_per_page());
 117   EXPECT_EQ(3U, scorer->shingle_size());
 118 }
 119
 120 TEST_F(PhishingScorerTest, ComputeScore) {
 121   scoped_ptr<Scorer> scorer(Scorer::Create(model_.SerializeAsString()));
 122   ASSERT_TRUE(scorer.get());
 123
 124   // An empty feature map should match the empty rule.
 125   FeatureMap features;
 126   // The expected logodds is 0.5 (empty rule) => p = exp(0.5) / (exp(0.5) + 1)
 127   // => 0.62245933120185459
 128   EXPECT_DOUBLE_EQ(0.62245933120185459, scorer->ComputeScore(features));
 129   // Same if the feature does not match any rule.
 130   EXPECT_TRUE(features.AddBooleanFeature("not existing feature"));
 131   EXPECT_DOUBLE_EQ(0.62245933120185459, scorer->ComputeScore(features));
 132
 133   // Feature 1 matches which means that the logodds will be:
 134   //   0.5 (empty rule) + 2.0 (rule weight) * 0.15 (feature weight) = 0.8
 135   //   => p = 0.6899744811276125
 136   EXPECT_TRUE(features.AddRealFeature("feature1", 0.15));
 137   EXPECT_DOUBLE_EQ(0.6899744811276125, scorer->ComputeScore(features));
 138
 139   // Now, both feature 1 and feature 2 match.  Expected logodds:
 140   //   0.5 (empty rule) + 2.0 (rule weight) * 0.15 (feature weight) +
 141   //   3.0 (rule weight) * 0.15 (feature1 weight) * 1.0 (feature2) weight = 9.8
 142   //   => p = 0.99999627336071584
 143   EXPECT_TRUE(features.AddBooleanFeature("feature2"));
 144   EXPECT_DOUBLE_EQ(0.77729986117469119, scorer->ComputeScore(features));
 145 }
 146 }  // namespace safe_browsing