[Autofill] Fix the dedupe of verified profiles.

components/autofill/core/browser/personal_data_manager_unittest.cc

Index: components/autofill/core/browser/personal_data_manager_unittest.cc
diff --git a/components/autofill/core/browser/personal_data_manager_unittest.cc b/components/autofill/core/browser/personal_data_manager_unittest.cc
index 6c2b79aced15ba2f692a05a6247a8c44bc48e60f..ad101329292cca6db5c20a1eb4096782ae9ee7f8 100644
--- a/components/autofill/core/browser/personal_data_manager_unittest.cc
+++ b/components/autofill/core/browser/personal_data_manager_unittest.cc
@@ -2735,7 +2735,7 @@ TEST_F(PersonalDataManagerTest, SaveImportedProfileWithVerifiedData) {
 }
 
 // Ensure that verified profiles can be saved via SaveImportedProfile,
-// overwriting existing verified profiles as well.
+// but do not merge with existing verified profiles.
 TEST_F(PersonalDataManagerTest, SaveImportedProfileWithExistingVerifiedData) {
   // Start with a verified profile.
   AutofillProfile profile(base::GenerateGUID(), kSettingsOrigin);
@@ -2766,10 +2766,10 @@ TEST_F(PersonalDataManagerTest, SaveImportedProfileWithExistingVerifiedData) {
       .WillOnce(QuitMainMessageLoop());
   base::MessageLoop::current()->Run();
 
-  // The new profile should be merged into the existing one.
+  // The new profile should not be merged into the existing one. It should be
+  // saved as a separate profile.
   const std::vector<AutofillProfile*>& results = personal_data_->GetProfiles();
-  ASSERT_EQ(1U, results.size());
-  EXPECT_EQ(0, new_verified_profile.Compare(*results[0]));
+  ASSERT_EQ(2U, results.size());
 }
 
 // Ensure that verified credit cards can be saved via SaveImportedCreditCard.
@@ -4668,4 +4668,252 @@ TEST_F(PersonalDataManagerTest,
             profiles[0]->use_date());
 }
 
+// Tests that FindAndMergeDuplicateProfiles only keeps the verified profile with
+// it's original data when deduping with similar profiles, even if it has a

[Mathieu, 2016/06/17 14:15:09]

*its

[sebsg, 2016/06/17 17:09:01]

Done.

+// higher frecency score.
+TEST_F(PersonalDataManagerTest,
+       FindAndMergeDuplicateProfiles_VerifiedProfileFirst) {
+  EnableAutofillProfileCleanup();
+
+  // Create a verified profile with a higher frecency score.
+  AutofillProfile profile1(base::GenerateGUID(), kSettingsOrigin);
+  test::SetProfileInfo(&profile1, "Homer", "Jay", "Simpson",
+                       "homer.simpson@abc.com", "", "742 Evergreen Terrace", "",
+                       "Springfield", "IL", "91601", "", "12345678910");
+  profile1.set_use_count(5);
+  profile1.set_use_date(base::Time::Now() - base::TimeDelta::FromDays(3));
+
+  // Create a similar non verified profile with a lower frecency score.
+  AutofillProfile profile2(base::GenerateGUID(), "https://www.example.com");
+  test::SetProfileInfo(&profile2, "Homer", "J", "Simpson",
+                       "homer.simpson@abc.com", "Fox", "742 Evergreen Terrace.",
+                       "", "Springfield", "IL", "91601", "", "");
+  profile2.set_use_count(3);
+  profile2.set_use_date(base::Time::Now() - base::TimeDelta::FromDays(5));
+
+  personal_data_->AddProfile(profile1);
+  personal_data_->AddProfile(profile2);
+
+  EXPECT_CALL(personal_data_observer_, OnPersonalDataChanged())
+      .WillOnce(QuitMainMessageLoop());
+  base::MessageLoop::current()->Run();
+
+  EXPECT_EQ(2U, personal_data_->GetProfiles().size());
+
+  // Create a similar new non verified imported profile to be merged with the
+  // saved profiles.
+  AutofillProfile imported_profile(base::GenerateGUID(),
+                                   "https://www.example.com");
+  test::SetProfileInfo(&imported_profile, "Homer", "J", "Simpson",
+                       "homer.simpson@abc.com", "", "742. Evergreen Terrace",
+                       "", "Springfield", "IL", "91601", "US", "");
+
+  base::HistogramTester histogram_tester;
+  personal_data_->SaveImportedProfile(imported_profile);
+
+  EXPECT_CALL(personal_data_observer_, OnPersonalDataChanged())
+      .WillOnce(QuitMainMessageLoop());
+  base::MessageLoop::current()->Run();
+
+  std::vector<AutofillProfile*> profiles = personal_data_->GetProfiles();
+
+  // The |imported_profile| should have merged with |profile2|. |profile2|
+  // should then have been discarded because it is similar to the verified
+  // |profile1|.
+  ASSERT_EQ(1U, profiles.size());
+  // 2 profiles were considered for dedupe (profiles 1 and 2).
+  histogram_tester.ExpectUniqueSample(
+      "Autofill.NumberOfProfilesConsideredForDedupe", 2, 1);
+  // 1 profile was removed (|profile2|).
+  histogram_tester.ExpectUniqueSample(
+      "Autofill.NumberOfProfilesRemovedDuringDedupe", 1, 1);
+
+  // Only the verified |profile2| with it's original data should have been kept.

[Mathieu, 2016/06/17 14:15:09]

*its

[sebsg, 2016/06/17 17:09:01]

Done.

+  EXPECT_EQ(profile1.guid(), profiles[0]->guid());
+  EXPECT_EQ(UTF8ToUTF16("742 Evergreen Terrace"),
+            profiles[0]->GetRawInfo(ADDRESS_HOME_LINE1));
+  EXPECT_EQ(UTF8ToUTF16("Jay"), profiles[0]->GetRawInfo(NAME_MIDDLE));
+  EXPECT_EQ(UTF8ToUTF16("12345678910"),
+            profiles[0]->GetRawInfo(PHONE_HOME_WHOLE_NUMBER));
+  EXPECT_EQ(UTF8ToUTF16(""), profiles[0]->GetRawInfo(COMPANY_NAME));
+  EXPECT_EQ(UTF8ToUTF16(""), profiles[0]->GetRawInfo(ADDRESS_HOME_COUNTRY));
+  EXPECT_EQ(profile1.use_count(), profiles[0]->use_count());
+  EXPECT_LT(profile1.use_date() - TimeDelta::FromSeconds(2),
+            profiles[0]->use_date());
+  EXPECT_GT(profile1.use_date() + TimeDelta::FromSeconds(2),
+            profiles[0]->use_date());
+}
+
+// Tests that FindAndMergeDuplicateProfiles only keeps the verified profile with
+// it's original data when deduping with similar profiles, even if it has a
+// lower frecency score.
+TEST_F(PersonalDataManagerTest,
+       FindAndMergeDuplicateProfiles_VerifiedProfileLast) {
+  EnableAutofillProfileCleanup();
+
+  // Create a non verified profile with a higher frecency score.
+  AutofillProfile profile1(base::GenerateGUID(), "https://www.example.com");
+  test::SetProfileInfo(&profile1, "Homer", "J", "Simpson",
+                       "homer.simpson@abc.com", "Fox", "742 Evergreen Terrace.",
+                       "", "Springfield", "IL", "91601", "", "");
+  profile1.set_use_count(5);
+  profile1.set_use_date(base::Time::Now() - base::TimeDelta::FromDays(3));
+
+  // Create a similar verified profile with a lower frecency score.
+  AutofillProfile profile2(base::GenerateGUID(), kSettingsOrigin);
+  test::SetProfileInfo(&profile2, "Homer", "Jay", "Simpson",
+                       "homer.simpson@abc.com", "", "742 Evergreen Terrace", "",
+                       "Springfield", "IL", "91601", "", "12345678910");
+  profile2.set_use_count(3);
+  profile2.set_use_date(base::Time::Now() - base::TimeDelta::FromDays(5));
+
+  personal_data_->AddProfile(profile1);
+  personal_data_->AddProfile(profile2);
+
+  EXPECT_CALL(personal_data_observer_, OnPersonalDataChanged())
+      .WillOnce(QuitMainMessageLoop());
+  base::MessageLoop::current()->Run();
+
+  EXPECT_EQ(2U, personal_data_->GetProfiles().size());
+
+  // Create a similar non verified imported profile to be merged with the saved
+  // profiles.
+  AutofillProfile imported_profile(base::GenerateGUID(),
+                                   "https://www.example.com");
+  test::SetProfileInfo(&imported_profile, "Homer", "J", "Simpson",
+                       "homer.simpson@abc.com", "", "742. Evergreen Terrace",
+                       "", "Springfield", "IL", "91601", "US", "");
+
+  base::HistogramTester histogram_tester;
+  personal_data_->SaveImportedProfile(imported_profile);
+
+  EXPECT_CALL(personal_data_observer_, OnPersonalDataChanged())
+      .WillOnce(QuitMainMessageLoop());
+  base::MessageLoop::current()->Run();
+
+  std::vector<AutofillProfile*> profiles = personal_data_->GetProfiles();
+
+  // The |imported_profile| should have merged with |profile1|. |profile1|
+  // should then have been discarded because it is similar to the verified
+  // |profile2|.
+  ASSERT_EQ(1U, profiles.size());
+  // 2 profiles were considered for dedupe (profiles 1 and 2).
+  histogram_tester.ExpectUniqueSample(
+      "Autofill.NumberOfProfilesConsideredForDedupe", 2, 1);
+  // 1 profile was removed (|profile1|).
+  histogram_tester.ExpectUniqueSample(
+      "Autofill.NumberOfProfilesRemovedDuringDedupe", 1, 1);
+
+  // Only the verified |profile2| with it's original data should have been kept.
+  EXPECT_EQ(profile2.guid(), profiles[0]->guid());
+  EXPECT_EQ(UTF8ToUTF16("742 Evergreen Terrace"),
+            profiles[0]->GetRawInfo(ADDRESS_HOME_LINE1));
+  EXPECT_EQ(UTF8ToUTF16("Jay"), profiles[0]->GetRawInfo(NAME_MIDDLE));
+  EXPECT_EQ(UTF8ToUTF16("12345678910"),
+            profiles[0]->GetRawInfo(PHONE_HOME_WHOLE_NUMBER));
+  EXPECT_EQ(UTF8ToUTF16(""), profiles[0]->GetRawInfo(COMPANY_NAME));
+  EXPECT_EQ(UTF8ToUTF16(""), profiles[0]->GetRawInfo(ADDRESS_HOME_COUNTRY));
+  EXPECT_EQ(profile2.use_count(), profiles[0]->use_count());
+  EXPECT_LT(profile2.use_date() - TimeDelta::FromSeconds(2),
+            profiles[0]->use_date());
+  EXPECT_GT(profile2.use_date() + TimeDelta::FromSeconds(2),
+            profiles[0]->use_date());
+}
+
+// Tests that FindAndMergeDuplicateProfiles does not merge unverfied data into a

[Mathieu, 2016/06/17 14:15:09]

nit: unverified

[sebsg, 2016/06/17 17:09:02]

Done.

+// verified profile. Also tests that two verified profiles don't get merged.
+TEST_F(PersonalDataManagerTest,
+       FindAndMergeDuplicateProfiles_MultipleVerifiedProfiles) {
+  EnableAutofillProfileCleanup();
+
+  // Create a verified profile with a higher frecency score.
+  AutofillProfile profile1(base::GenerateGUID(), kSettingsOrigin);
+  test::SetProfileInfo(&profile1, "Homer", "J", "Simpson",
+                       "homer.simpson@abc.com", "Fox", "742 Evergreen Terrace.",
+                       "", "Springfield", "IL", "91601", "", "");
+  profile1.set_use_count(5);
+  profile1.set_use_date(base::Time::Now() - base::TimeDelta::FromDays(3));
+
+  // Create a similar verified profile with a lower frecency score.
+  AutofillProfile profile2(base::GenerateGUID(), kSettingsOrigin);
+  test::SetProfileInfo(&profile2, "Homer", "Jay", "Simpson",
+                       "homer.simpson@abc.com", "", "742 Evergreen Terrace", "",
+                       "Springfield", "IL", "91601", "", "12345678910");
+  profile2.set_use_count(3);
+  profile2.set_use_date(base::Time::Now() - base::TimeDelta::FromDays(5));
+
+  personal_data_->AddProfile(profile1);
+  personal_data_->AddProfile(profile2);
+
+  EXPECT_CALL(personal_data_observer_, OnPersonalDataChanged())
+      .WillOnce(QuitMainMessageLoop());
+  base::MessageLoop::current()->Run();
+
+  EXPECT_EQ(2U, personal_data_->GetProfiles().size());
+
+  // Create a non verified imported profile to be merged with the saved
+  // profiles.
+  AutofillProfile imported_profile(base::GenerateGUID(),
+                                   "https://www.example.com");
+  test::SetProfileInfo(&imported_profile, "Homer", "J", "Simpson",
+                       "homer.simpson@abc.com", "", "742. Evergreen Terrace",
+                       "", "Springfield", "IL", "91601", "US", "");
+
+  base::HistogramTester histogram_tester;
+  personal_data_->SaveImportedProfile(imported_profile);
+
+  EXPECT_CALL(personal_data_observer_, OnPersonalDataChanged())
+      .WillOnce(QuitMainMessageLoop());
+  base::MessageLoop::current()->Run();
+
+  std::vector<AutofillProfile*> profiles = personal_data_->GetProfiles();
+
+  // Sort the profiles by frecency to have a deterministic order.

[Mathieu, 2016/06/17 14:15:09]

not sure why this is here? I would rather have the tests fail if we ever change
the real ordering function than have this logic. What do you think?

[sebsg, 2016/06/17 17:09:01]

Otherwise the test fails half the time because the profiles are not in a
deterministic order.

Changed the call to GetProfilesToSuggest, which does the sorting.

+  base::Time comparison_time = base::Time::Now();
+  std::sort(profiles.begin(), profiles.end(),
+            [comparison_time](const AutofillDataModel* a,
+                              const AutofillDataModel* b) {
+              return a->CompareFrecency(b, comparison_time);
+            });
+
+  // The |imported_profile| should have been discarded because the saved profile
+  // with the highest frecency score is verified (|profile1|). Therefore, the
+  // |imported_profile|'s data should not have been merged with the
+  // |imported_profile|'s data. Then |profile1| should have been compared to

[Mathieu, 2016/06/17 14:15:09]

fix comment?

[sebsg, 2016/06/17 17:09:01]

Done.

+  // |profile2| but they should not have merged because both profiles are
+  // verified.
+  ASSERT_EQ(2U, profiles.size());
+  // 2 profiles were considered for dedupe (the 2 saved profiles).
+  histogram_tester.ExpectUniqueSample(
+      "Autofill.NumberOfProfilesConsideredForDedupe", 2, 1);
+  // No profile was removed.
+  histogram_tester.ExpectUniqueSample(
+      "Autofill.NumberOfProfilesRemovedDuringDedupe", 0, 1);
+
+  EXPECT_EQ(profile1.guid(), profiles[0]->guid());
+  EXPECT_EQ(profile2.guid(), profiles[1]->guid());
+  // The profiles should have kept their original data.
+  EXPECT_EQ(UTF8ToUTF16("742 Evergreen Terrace."),
+            profiles[0]->GetRawInfo(ADDRESS_HOME_LINE1));
+  EXPECT_EQ(UTF8ToUTF16("742 Evergreen Terrace"),
+            profiles[1]->GetRawInfo(ADDRESS_HOME_LINE1));
+  EXPECT_EQ(UTF8ToUTF16("J"), profiles[0]->GetRawInfo(NAME_MIDDLE));
+  EXPECT_EQ(UTF8ToUTF16("Jay"), profiles[1]->GetRawInfo(NAME_MIDDLE));
+  EXPECT_EQ(UTF8ToUTF16(""),
+            profiles[0]->GetRawInfo(PHONE_HOME_WHOLE_NUMBER));
+  EXPECT_EQ(UTF8ToUTF16("12345678910"),
+            profiles[1]->GetRawInfo(PHONE_HOME_WHOLE_NUMBER));
+  EXPECT_EQ(profile1.use_count(), profiles[0]->use_count());
+  EXPECT_EQ(profile2.use_count(), profiles[1]->use_count());
+  EXPECT_LT(profile1.use_date() - TimeDelta::FromSeconds(2),
+            profiles[0]->use_date());
+  EXPECT_GT(profile1.use_date() + TimeDelta::FromSeconds(2),
+            profiles[0]->use_date());
+  EXPECT_LT(profile2.use_date() - TimeDelta::FromSeconds(2),
+            profiles[1]->use_date());
+  EXPECT_GT(profile2.use_date() + TimeDelta::FromSeconds(2),
+            profiles[1]->use_date());
+
+}
+
 }  // namespace autofill