src/chrome/browser/autocomplete/history_quick_provider_unittest.cc

   1 // Copyright (c) 2012 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/browser/autocomplete/history_quick_provider.h"
   6
   7 #include <algorithm>
   8 #include <functional>
   9 #include <set>
  10 #include <string>
  11 #include <vector>
  12
  13 #include "base/format_macros.h"
  14 #include "base/memory/scoped_ptr.h"
  15 #include "base/message_loop/message_loop.h"
  16 #include "base/prefs/pref_service.h"
  17 #include "base/strings/stringprintf.h"
  18 #include "base/strings/utf_string_conversions.h"
  19 #include "chrome/browser/autocomplete/chrome_autocomplete_scheme_classifier.h"
  20 #include "chrome/browser/autocomplete/history_url_provider.h"
  21 #include "chrome/browser/bookmarks/bookmark_model_factory.h"
  22 #include "chrome/browser/chrome_notification_types.h"
  23 #include "chrome/browser/history/history_backend.h"
  24 #include "chrome/browser/history/history_database.h"
  25 #include "chrome/browser/history/history_service.h"
  26 #include "chrome/browser/history/history_service_factory.h"
  27 #include "chrome/browser/history/in_memory_url_index.h"
  28 #include "chrome/browser/history/url_index_private_data.h"
  29 #include "chrome/browser/search_engines/chrome_template_url_service_client.h"
  30 #include "chrome/browser/search_engines/template_url_service_factory.h"
  31 #include "chrome/common/pref_names.h"
  32 #include "chrome/test/base/testing_browser_process.h"
  33 #include "chrome/test/base/testing_profile.h"
  34 #include "components/bookmarks/test/bookmark_test_helpers.h"
  35 #include "components/history/core/browser/url_database.h"
  36 #include "components/metrics/proto/omnibox_event.pb.h"
  37 #include "components/omnibox/autocomplete_match.h"
  38 #include "components/omnibox/autocomplete_result.h"
  39 #include "components/search_engines/search_terms_data.h"
  40 #include "components/search_engines/template_url.h"
  41 #include "components/search_engines/template_url_service.h"
  42 #include "content/public/browser/notification_service.h"
  43 #include "content/public/test/test_browser_thread.h"
  44 #include "content/public/test/test_utils.h"
  45 #include "sql/transaction.h"
  46 #include "testing/gtest/include/gtest/gtest.h"
  47
  48 using base::ASCIIToUTF16;
  49 using base::Time;
  50 using base::TimeDelta;
  51
  52 using content::BrowserThread;
  53
  54 struct TestURLInfo {
  55   std::string url;
  56   std::string title;
  57   int visit_count;
  58   int typed_count;
  59   int days_from_now;
  60 } quick_test_db[] = {
  61   {"http://www.google.com/", "Google", 3, 3, 0},
  62   {"http://slashdot.org/favorite_page.html", "Favorite page", 200, 100, 0},
  63   {"http://kerneltrap.org/not_very_popular.html", "Less popular", 4, 0, 0},
  64   {"http://freshmeat.net/unpopular.html", "Unpopular", 1, 1, 0},
  65   {"http://news.google.com/?ned=us&topic=n", "Google News - U.S.", 2, 2, 0},
  66   {"http://news.google.com/", "Google News", 1, 1, 0},
  67   {"http://foo.com/", "Dir", 200, 100, 0},
  68   {"http://foo.com/dir/", "Dir", 2, 1, 10},
  69   {"http://foo.com/dir/another/", "Dir", 10, 5, 0},
  70   {"http://foo.com/dir/another/again/", "Dir", 5, 1, 0},
  71   {"http://foo.com/dir/another/again/myfile.html", "File", 3, 1, 0},
  72   {"http://visitedest.com/y/a", "VA", 10, 1, 20},
  73   {"http://visitedest.com/y/b", "VB", 9, 1, 20},
  74   {"http://visitedest.com/x/c", "VC", 8, 1, 20},
  75   {"http://visitedest.com/x/d", "VD", 7, 1, 20},
  76   {"http://visitedest.com/y/e", "VE", 6, 1, 20},
  77   {"http://typeredest.com/y/a", "TA", 5, 5, 0},
  78   {"http://typeredest.com/y/b", "TB", 5, 4, 0},
  79   {"http://typeredest.com/x/c", "TC", 5, 3, 0},
  80   {"http://typeredest.com/x/d", "TD", 5, 2, 0},
  81   {"http://typeredest.com/y/e", "TE", 5, 1, 0},
  82   {"http://daysagoest.com/y/a", "DA", 1, 1, 0},
  83   {"http://daysagoest.com/y/b", "DB", 1, 1, 1},
  84   {"http://daysagoest.com/x/c", "DC", 1, 1, 2},
  85   {"http://daysagoest.com/x/d", "DD", 1, 1, 3},
  86   {"http://daysagoest.com/y/e", "DE", 1, 1, 4},
  87   {"http://abcdefghixyzjklmnopqrstuvw.com/a", "", 3, 1, 0},
  88   {"http://spaces.com/path%20with%20spaces/foo.html", "Spaces", 2, 2, 0},
  89   {"http://abcdefghijklxyzmnopqrstuvw.com/a", "", 3, 1, 0},
  90   {"http://abcdefxyzghijklmnopqrstuvw.com/a", "", 3, 1, 0},
  91   {"http://abcxyzdefghijklmnopqrstuvw.com/a", "", 3, 1, 0},
  92   {"http://xyzabcdefghijklmnopqrstuvw.com/a", "", 3, 1, 0},
  93   {"http://cda.com/Dogs%20Cats%20Gorillas%20Sea%20Slugs%20and%20Mice",
  94    "Dogs & Cats & Mice & Other Animals", 1, 1, 0},
  95   {"https://monkeytrap.org/", "", 3, 1, 0},
  96   {"http://popularsitewithpathonly.com/moo",
  97    "popularsitewithpathonly.com/moo", 50, 50, 0},
  98   {"http://popularsitewithroot.com/", "popularsitewithroot.com", 50, 50, 0},
  99   {"http://testsearch.com/?q=thequery", "Test Search Engine", 10, 10, 0},
 100   {"http://testsearch.com/", "Test Search Engine", 9, 9, 0},
 101   {"http://anotherengine.com/?q=thequery", "Another Search Engine", 8, 8, 0},
 102   // The encoded stuff between /wiki/ and the # is 第二次世界大戦
 103   {"http://ja.wikipedia.org/wiki/%E7%AC%AC%E4%BA%8C%E6%AC%A1%E4%B8%96%E7%95"
 104    "%8C%E5%A4%A7%E6%88%A6#.E3.83.B4.E3.82.A7.E3.83.AB.E3.82.B5.E3.82.A4.E3."
 105    "83.A6.E4.BD.93.E5.88.B6", "Title Unimportant", 2, 2, 0}
 106 };
 107
 108 class HistoryQuickProviderTest : public testing::Test {
 109  public:
 110   HistoryQuickProviderTest()
 111       : ui_thread_(BrowserThread::UI, &message_loop_),
 112         file_thread_(BrowserThread::FILE, &message_loop_) {}
 113
 114  protected:
 115   class SetShouldContain : public std::unary_function<const std::string&,
 116                                                       std::set<std::string> > {
 117    public:
 118     explicit SetShouldContain(const ACMatches& matched_urls);
 119
 120     void operator()(const std::string& expected);
 121
 122     std::set<std::string> LeftOvers() const { return matches_; }
 123
 124    private:
 125     std::set<std::string> matches_;
 126   };
 127
 128   static KeyedService* CreateTemplateURLService(
 129       content::BrowserContext* context) {
 130     Profile* profile = static_cast<Profile*>(context);
 131     return new TemplateURLService(
 132         profile->GetPrefs(), make_scoped_ptr(new SearchTermsData), NULL,
 133         scoped_ptr<TemplateURLServiceClient>(
 134             new ChromeTemplateURLServiceClient(profile)),
 135         NULL, NULL, base::Closure());
 136   }
 137
 138   virtual void SetUp();
 139   virtual void TearDown();
 140
 141   virtual void GetTestData(size_t* data_count, TestURLInfo** test_data);
 142
 143   // Fills test data into the history system.
 144   void FillData();
 145
 146   // Runs an autocomplete query on |text| and checks to see that the returned
 147   // results' destination URLs match those provided. |expected_urls| does not
 148   // need to be in sorted order.
 149   void RunTest(const base::string16 text,
 150                bool prevent_inline_autocomplete,
 151                std::vector<std::string> expected_urls,
 152                bool can_inline_top_result,
 153                base::string16 expected_fill_into_edit,
 154                base::string16 autocompletion);
 155
 156   // As above, simply with a cursor position specified.
 157   void RunTestWithCursor(const base::string16 text,
 158                          const size_t cursor_position,
 159                          bool prevent_inline_autocomplete,
 160                          std::vector<std::string> expected_urls,
 161                          bool can_inline_top_result,
 162                          base::string16 expected_fill_into_edit,
 163                          base::string16 autocompletion);
 164
 165   history::HistoryBackend* history_backend() {
 166     return history_service_->history_backend_;
 167   }
 168
 169   base::MessageLoopForUI message_loop_;
 170   content::TestBrowserThread ui_thread_;
 171   content::TestBrowserThread file_thread_;
 172
 173   scoped_ptr<TestingProfile> profile_;
 174   HistoryService* history_service_;
 175
 176   ACMatches ac_matches_;  // The resulting matches after running RunTest.
 177
 178   scoped_refptr<HistoryQuickProvider> provider_;
 179 };
 180
 181 void HistoryQuickProviderTest::SetUp() {
 182   profile_.reset(new TestingProfile());
 183   ASSERT_TRUE(profile_->CreateHistoryService(true, false));
 184   profile_->CreateBookmarkModel(true);
 185   test::WaitForBookmarkModelToLoad(
 186       BookmarkModelFactory::GetForProfile(profile_.get()));
 187   profile_->BlockUntilHistoryIndexIsRefreshed();
 188   history_service_ =
 189       HistoryServiceFactory::GetForProfile(profile_.get(),
 190                                            Profile::EXPLICIT_ACCESS);
 191   EXPECT_TRUE(history_service_);
 192   provider_ = new HistoryQuickProvider(profile_.get());
 193   TemplateURLServiceFactory::GetInstance()->SetTestingFactoryAndUse(
 194       profile_.get(), &HistoryQuickProviderTest::CreateTemplateURLService);
 195   FillData();
 196   provider_->GetIndex()->RebuildFromHistory(history_backend()->db());
 197 }
 198
 199 void HistoryQuickProviderTest::TearDown() {
 200   provider_ = NULL;
 201 }
 202
 203 void HistoryQuickProviderTest::GetTestData(size_t* data_count,
 204                                            TestURLInfo** test_data) {
 205   DCHECK(data_count);
 206   DCHECK(test_data);
 207   *data_count = arraysize(quick_test_db);
 208   *test_data = &quick_test_db[0];
 209 }
 210
 211 void HistoryQuickProviderTest::FillData() {
 212   sql::Connection& db(history_backend()->db()->GetDB());
 213   ASSERT_TRUE(db.is_open());
 214
 215   size_t data_count = 0;
 216   TestURLInfo* test_data = NULL;
 217   GetTestData(&data_count, &test_data);
 218   size_t visit_id = 1;
 219   for (size_t i = 0; i < data_count; ++i) {
 220     const TestURLInfo& cur(test_data[i]);
 221     Time visit_time = Time::Now() - TimeDelta::FromDays(cur.days_from_now);
 222     sql::Transaction transaction(&db);
 223
 224     // Add URL.
 225     transaction.Begin();
 226     std::string sql_cmd_line = base::StringPrintf(
 227         "INSERT INTO \"urls\" VALUES(%" PRIuS ", \'%s\', \'%s\', %d, %d, %"
 228         PRId64 ", 0, 0)",
 229         i + 1, cur.url.c_str(), cur.title.c_str(), cur.visit_count,
 230         cur.typed_count, visit_time.ToInternalValue());
 231     sql::Statement sql_stmt(db.GetUniqueStatement(sql_cmd_line.c_str()));
 232     EXPECT_TRUE(sql_stmt.Run());
 233     transaction.Commit();
 234
 235     // Add visits.
 236     for (int j = 0; j < cur.visit_count; ++j) {
 237       // Assume earlier visits are at one-day intervals.
 238       visit_time -= TimeDelta::FromDays(1);
 239       transaction.Begin();
 240       // Mark the most recent |cur.typed_count| visits as typed.
 241       std::string sql_cmd_line = base::StringPrintf(
 242           "INSERT INTO \"visits\" VALUES(%" PRIuS ", %" PRIuS ", %" PRId64
 243           ", 0, %d, 0, 1)",
 244           visit_id++, i + 1, visit_time.ToInternalValue(),
 245           (j < cur.typed_count) ? content::PAGE_TRANSITION_TYPED :
 246                                   content::PAGE_TRANSITION_LINK);
 247
 248       sql::Statement sql_stmt(db.GetUniqueStatement(sql_cmd_line.c_str()));
 249       EXPECT_TRUE(sql_stmt.Run());
 250       transaction.Commit();
 251     }
 252   }
 253 }
 254
 255 HistoryQuickProviderTest::SetShouldContain::SetShouldContain(
 256     const ACMatches& matched_urls) {
 257   for (ACMatches::const_iterator iter = matched_urls.begin();
 258        iter != matched_urls.end(); ++iter)
 259     matches_.insert(iter->destination_url.spec());
 260 }
 261
 262 void HistoryQuickProviderTest::SetShouldContain::operator()(
 263     const std::string& expected) {
 264   EXPECT_EQ(1U, matches_.erase(expected))
 265       << "Results did not contain '" << expected << "' but should have.";
 266 }
 267
 268 void HistoryQuickProviderTest::RunTest(
 269     const base::string16 text,
 270     bool prevent_inline_autocomplete,
 271     std::vector<std::string> expected_urls,
 272     bool can_inline_top_result,
 273     base::string16 expected_fill_into_edit,
 274     base::string16 expected_autocompletion) {
 275   RunTestWithCursor(text, base::string16::npos, prevent_inline_autocomplete,
 276                     expected_urls, can_inline_top_result,
 277                     expected_fill_into_edit, expected_autocompletion);
 278 }
 279
 280 void HistoryQuickProviderTest::RunTestWithCursor(
 281     const base::string16 text,
 282     const size_t cursor_position,
 283     bool prevent_inline_autocomplete,
 284     std::vector<std::string> expected_urls,
 285     bool can_inline_top_result,
 286     base::string16 expected_fill_into_edit,
 287     base::string16 expected_autocompletion) {
 288   SCOPED_TRACE(text);  // Minimal hint to query being run.
 289   base::MessageLoop::current()->RunUntilIdle();
 290   AutocompleteInput input(text, cursor_position, base::string16(),
 291                           GURL(), metrics::OmniboxEventProto::INVALID_SPEC,
 292                           prevent_inline_autocomplete, false, true, true,
 293                           ChromeAutocompleteSchemeClassifier(profile_.get()));
 294   provider_->Start(input, false);
 295   EXPECT_TRUE(provider_->done());
 296
 297   ac_matches_ = provider_->matches();
 298
 299   // We should have gotten back at most AutocompleteProvider::kMaxMatches.
 300   EXPECT_LE(ac_matches_.size(), AutocompleteProvider::kMaxMatches);
 301
 302   // If the number of expected and actual matches aren't equal then we need
 303   // test no further, but let's do anyway so that we know which URLs failed.
 304   EXPECT_EQ(expected_urls.size(), ac_matches_.size());
 305
 306   // Verify that all expected URLs were found and that all found URLs
 307   // were expected.
 308   std::set<std::string> leftovers =
 309       for_each(expected_urls.begin(), expected_urls.end(),
 310                SetShouldContain(ac_matches_)).LeftOvers();
 311   EXPECT_EQ(0U, leftovers.size()) << "There were " << leftovers.size()
 312       << " unexpected results, one of which was: '"
 313       << *(leftovers.begin()) << "'.";
 314
 315   if (expected_urls.empty())
 316     return;
 317
 318   // Verify that we got the results in the order expected.
 319   int best_score = ac_matches_.begin()->relevance + 1;
 320   int i = 0;
 321   std::vector<std::string>::const_iterator expected = expected_urls.begin();
 322   for (ACMatches::const_iterator actual = ac_matches_.begin();
 323        actual != ac_matches_.end() && expected != expected_urls.end();
 324        ++actual, ++expected, ++i) {
 325     EXPECT_EQ(*expected, actual->destination_url.spec())
 326         << "For result #" << i << " we got '" << actual->destination_url.spec()
 327         << "' but expected '" << *expected << "'.";
 328     EXPECT_LT(actual->relevance, best_score)
 329       << "At result #" << i << " (url=" << actual->destination_url.spec()
 330       << "), we noticed scores are not monotonically decreasing.";
 331     best_score = actual->relevance;
 332   }
 333
 334   EXPECT_EQ(can_inline_top_result, ac_matches_[0].allowed_to_be_default_match);
 335   if (can_inline_top_result)
 336     EXPECT_EQ(expected_autocompletion, ac_matches_[0].inline_autocompletion);
 337   EXPECT_EQ(expected_fill_into_edit, ac_matches_[0].fill_into_edit);
 338 }
 339
 340 TEST_F(HistoryQuickProviderTest, SimpleSingleMatch) {
 341   std::vector<std::string> expected_urls;
 342   expected_urls.push_back("http://slashdot.org/favorite_page.html");
 343   RunTest(ASCIIToUTF16("slashdot"), false, expected_urls, true,
 344           ASCIIToUTF16("slashdot.org/favorite_page.html"),
 345                   ASCIIToUTF16(".org/favorite_page.html"));
 346 }
 347
 348 TEST_F(HistoryQuickProviderTest, SingleMatchWithCursor) {
 349   std::vector<std::string> expected_urls;
 350   expected_urls.push_back("http://slashdot.org/favorite_page.html");
 351   // With cursor after "slash", we should retrieve the desired result but it
 352   // should not be allowed to be the default match.
 353   RunTestWithCursor(ASCIIToUTF16("slashfavorite_page.html"), 5, false,
 354                     expected_urls, false,
 355                     ASCIIToUTF16("slashdot.org/favorite_page.html"),
 356                     base::string16());
 357   // If the cursor is in the middle of a valid URL suggestion, it should be
 358   // allowed to be the default match.  The inline completion will be empty
 359   // though as no completion is necessary.
 360   RunTestWithCursor(ASCIIToUTF16("slashdot.org/favorite_page.html"), 5, false,
 361                     expected_urls, true,
 362                     ASCIIToUTF16("slashdot.org/favorite_page.html"),
 363                     base::string16());
 364 }
 365
 366 TEST_F(HistoryQuickProviderTest, WordBoundariesWithPunctuationMatch) {
 367   std::vector<std::string> expected_urls;
 368   expected_urls.push_back("http://popularsitewithpathonly.com/moo");
 369   RunTest(ASCIIToUTF16("/moo"), false, expected_urls, false,
 370           ASCIIToUTF16("popularsitewithpathonly.com/moo"), base::string16());
 371 }
 372
 373 TEST_F(HistoryQuickProviderTest, MultiTermTitleMatch) {
 374   std::vector<std::string> expected_urls;
 375   expected_urls.push_back(
 376       "http://cda.com/Dogs%20Cats%20Gorillas%20Sea%20Slugs%20and%20Mice");
 377   RunTest(ASCIIToUTF16("mice other animals"), false, expected_urls, false,
 378           ASCIIToUTF16("cda.com/Dogs Cats Gorillas Sea Slugs and Mice"),
 379           base::string16());
 380 }
 381
 382 TEST_F(HistoryQuickProviderTest, NonWordLastCharacterMatch) {
 383   std::string expected_url("http://slashdot.org/favorite_page.html");
 384   std::vector<std::string> expected_urls;
 385   expected_urls.push_back(expected_url);
 386   RunTest(ASCIIToUTF16("slashdot.org/"), false, expected_urls, true,
 387           ASCIIToUTF16("slashdot.org/favorite_page.html"),
 388                        ASCIIToUTF16("favorite_page.html"));
 389 }
 390
 391 TEST_F(HistoryQuickProviderTest, MultiMatch) {
 392   std::vector<std::string> expected_urls;
 393   // Scores high because of typed_count.
 394   expected_urls.push_back("http://foo.com/");
 395   // Scores high because of visit count.
 396   expected_urls.push_back("http://foo.com/dir/another/");
 397   // Scores high because of high visit count.
 398   expected_urls.push_back("http://foo.com/dir/another/again/");
 399   RunTest(ASCIIToUTF16("foo"), false, expected_urls, true,
 400           ASCIIToUTF16("foo.com"), ASCIIToUTF16(".com"));
 401 }
 402
 403 TEST_F(HistoryQuickProviderTest, StartRelativeMatch) {
 404   std::vector<std::string> expected_urls;
 405   expected_urls.push_back("http://xyzabcdefghijklmnopqrstuvw.com/a");
 406   RunTest(ASCIIToUTF16("xyza"), false, expected_urls, true,
 407           ASCIIToUTF16("xyzabcdefghijklmnopqrstuvw.com/a"),
 408               ASCIIToUTF16("bcdefghijklmnopqrstuvw.com/a"));
 409 }
 410
 411 TEST_F(HistoryQuickProviderTest, EncodingMatch) {
 412   std::vector<std::string> expected_urls;
 413   expected_urls.push_back("http://spaces.com/path%20with%20spaces/foo.html");
 414   RunTest(ASCIIToUTF16("path with spaces"), false, expected_urls, false,
 415           ASCIIToUTF16("spaces.com/path with spaces/foo.html"),
 416           base::string16());
 417 }
 418
 419 TEST_F(HistoryQuickProviderTest, ContentsClass) {
 420   std::vector<std::string> expected_urls;
 421   expected_urls.push_back(
 422       "http://ja.wikipedia.org/wiki/%E7%AC%AC%E4%BA%8C%E6%AC%A1%E4%B8%96%E7"
 423       "%95%8C%E5%A4%A7%E6%88%A6#.E3.83.B4.E3.82.A7.E3.83.AB.E3.82.B5.E3.82."
 424       "A4.E3.83.A6.E4.BD.93.E5.88.B6");
 425   RunTest(base::UTF8ToUTF16("第二 e3"), false, expected_urls, false,
 426           base::UTF8ToUTF16("ja.wikipedia.org/wiki/第二次世界大戦#.E3.83.B4.E3."
 427                             "82.A7.E3.83.AB.E3.82.B5.E3.82.A4.E3.83.A6.E4.BD."
 428                             "93.E5.88.B6"),
 429           base::string16());
 430 #ifndef NDEBUG
 431   ac_matches_[0].Validate();
 432 #endif
 433   // Verify that contents_class divides the string in the right places.
 434   // [22, 24) is the "第二".  All the other pairs are the "e3".
 435   ACMatchClassifications contents_class(ac_matches_[0].contents_class);
 436   size_t expected_offsets[] = { 0, 22, 24, 31, 33, 40, 42, 49, 51, 58, 60, 67,
 437                                 69, 76, 78 };
 438   // ScoredHistoryMatch may not highlight all the occurrences of these terms
 439   // because it only highlights terms at word breaks, and it only stores word
 440   // breaks up to some specified number of characters (50 at the time of this
 441   // comment).  This test is written flexibly so it still will pass if we
 442   // increase that number in the future.  Regardless, we require the first
 443   // five offsets to be correct--in this example these cover at least one
 444   // occurrence of each term.
 445   EXPECT_LE(contents_class.size(), arraysize(expected_offsets));
 446   EXPECT_GE(contents_class.size(), 5u);
 447   for (size_t i = 0; i < contents_class.size(); ++i)
 448     EXPECT_EQ(expected_offsets[i], contents_class[i].offset);
 449 }
 450
 451 TEST_F(HistoryQuickProviderTest, VisitCountMatches) {
 452   std::vector<std::string> expected_urls;
 453   expected_urls.push_back("http://visitedest.com/y/a");
 454   expected_urls.push_back("http://visitedest.com/y/b");
 455   expected_urls.push_back("http://visitedest.com/x/c");
 456   RunTest(ASCIIToUTF16("visitedest"), false, expected_urls, true,
 457           ASCIIToUTF16("visitedest.com/y/a"),
 458                     ASCIIToUTF16(".com/y/a"));
 459 }
 460
 461 TEST_F(HistoryQuickProviderTest, TypedCountMatches) {
 462   std::vector<std::string> expected_urls;
 463   expected_urls.push_back("http://typeredest.com/y/a");
 464   expected_urls.push_back("http://typeredest.com/y/b");
 465   expected_urls.push_back("http://typeredest.com/x/c");
 466   RunTest(ASCIIToUTF16("typeredest"), false, expected_urls, true,
 467           ASCIIToUTF16("typeredest.com/y/a"),
 468                     ASCIIToUTF16(".com/y/a"));
 469 }
 470
 471 TEST_F(HistoryQuickProviderTest, DaysAgoMatches) {
 472   std::vector<std::string> expected_urls;
 473   expected_urls.push_back("http://daysagoest.com/y/a");
 474   expected_urls.push_back("http://daysagoest.com/y/b");
 475   expected_urls.push_back("http://daysagoest.com/x/c");
 476   RunTest(ASCIIToUTF16("daysagoest"), false, expected_urls, true,
 477           ASCIIToUTF16("daysagoest.com/y/a"),
 478                     ASCIIToUTF16(".com/y/a"));
 479 }
 480
 481 TEST_F(HistoryQuickProviderTest, EncodingLimitMatch) {
 482   std::vector<std::string> expected_urls;
 483   std::string url(
 484       "http://cda.com/Dogs%20Cats%20Gorillas%20Sea%20Slugs%20and%20Mice");
 485   // First check that a mid-word match yield no results.
 486   RunTest(ASCIIToUTF16("ice"), false, expected_urls, false,
 487           ASCIIToUTF16("cda.com/Dogs Cats Gorillas Sea Slugs and Mice"),
 488           base::string16());
 489   // Then check that we get results when the match is at a word start
 490   // that is present because of an encoded separate (%20 = space).
 491   expected_urls.push_back(url);
 492   RunTest(ASCIIToUTF16("Mice"), false, expected_urls, false,
 493           ASCIIToUTF16("cda.com/Dogs Cats Gorillas Sea Slugs and Mice"),
 494           base::string16());
 495   // Verify that the matches' ACMatchClassifications offsets are in range.
 496   ACMatchClassifications content(ac_matches_[0].contents_class);
 497   // The max offset accounts for 6 occurrences of '%20' plus the 'http://'.
 498   const size_t max_offset = url.length() - ((6 * 2) + 7);
 499   for (ACMatchClassifications::const_iterator citer = content.begin();
 500        citer != content.end(); ++citer)
 501     EXPECT_LT(citer->offset, max_offset);
 502   ACMatchClassifications description(ac_matches_[0].description_class);
 503   std::string page_title("Dogs & Cats & Mice & Other Animals");
 504   for (ACMatchClassifications::const_iterator diter = description.begin();
 505        diter != description.end(); ++diter)
 506     EXPECT_LT(diter->offset, page_title.length());
 507 }
 508
 509 TEST_F(HistoryQuickProviderTest, Spans) {
 510   // Test SpansFromTermMatch
 511   history::TermMatches matches_a;
 512   // Simulates matches: '.xx.xxx..xx...xxxxx..' which will test no match at
 513   // either beginning or end as well as adjacent matches.
 514   matches_a.push_back(history::TermMatch(1, 1, 2));
 515   matches_a.push_back(history::TermMatch(2, 4, 3));
 516   matches_a.push_back(history::TermMatch(3, 9, 1));
 517   matches_a.push_back(history::TermMatch(3, 10, 1));
 518   matches_a.push_back(history::TermMatch(4, 14, 5));
 519   ACMatchClassifications spans_a =
 520       HistoryQuickProvider::SpansFromTermMatch(matches_a, 20, false);
 521   // ACMatch spans should be: 'NM-NM---N-M-N--M----N-'
 522   ASSERT_EQ(9U, spans_a.size());
 523   EXPECT_EQ(0U, spans_a[0].offset);
 524   EXPECT_EQ(ACMatchClassification::NONE, spans_a[0].style);
 525   EXPECT_EQ(1U, spans_a[1].offset);
 526   EXPECT_EQ(ACMatchClassification::MATCH, spans_a[1].style);
 527   EXPECT_EQ(3U, spans_a[2].offset);
 528   EXPECT_EQ(ACMatchClassification::NONE, spans_a[2].style);
 529   EXPECT_EQ(4U, spans_a[3].offset);
 530   EXPECT_EQ(ACMatchClassification::MATCH, spans_a[3].style);
 531   EXPECT_EQ(7U, spans_a[4].offset);
 532   EXPECT_EQ(ACMatchClassification::NONE, spans_a[4].style);
 533   EXPECT_EQ(9U, spans_a[5].offset);
 534   EXPECT_EQ(ACMatchClassification::MATCH, spans_a[5].style);
 535   EXPECT_EQ(11U, spans_a[6].offset);
 536   EXPECT_EQ(ACMatchClassification::NONE, spans_a[6].style);
 537   EXPECT_EQ(14U, spans_a[7].offset);
 538   EXPECT_EQ(ACMatchClassification::MATCH, spans_a[7].style);
 539   EXPECT_EQ(19U, spans_a[8].offset);
 540   EXPECT_EQ(ACMatchClassification::NONE, spans_a[8].style);
 541   // Simulates matches: 'xx.xx' which will test matches at both beginning and
 542   // end.
 543   history::TermMatches matches_b;
 544   matches_b.push_back(history::TermMatch(1, 0, 2));
 545   matches_b.push_back(history::TermMatch(2, 3, 2));
 546   ACMatchClassifications spans_b =
 547       HistoryQuickProvider::SpansFromTermMatch(matches_b, 5, true);
 548   // ACMatch spans should be: 'M-NM-'
 549   ASSERT_EQ(3U, spans_b.size());
 550   EXPECT_EQ(0U, spans_b[0].offset);
 551   EXPECT_EQ(ACMatchClassification::MATCH | ACMatchClassification::URL,
 552             spans_b[0].style);
 553   EXPECT_EQ(2U, spans_b[1].offset);
 554   EXPECT_EQ(ACMatchClassification::URL, spans_b[1].style);
 555   EXPECT_EQ(3U, spans_b[2].offset);
 556   EXPECT_EQ(ACMatchClassification::MATCH | ACMatchClassification::URL,
 557             spans_b[2].style);
 558 }
 559
 560 TEST_F(HistoryQuickProviderTest, DeleteMatch) {
 561   GURL test_url("http://slashdot.org/favorite_page.html");
 562   std::vector<std::string> expected_urls;
 563   expected_urls.push_back(test_url.spec());
 564   // Fill up ac_matches_; we don't really care about the test yet.
 565   RunTest(ASCIIToUTF16("slashdot"), false, expected_urls, true,
 566           ASCIIToUTF16("slashdot.org/favorite_page.html"),
 567                   ASCIIToUTF16(".org/favorite_page.html"));
 568   EXPECT_EQ(1U, ac_matches_.size());
 569   EXPECT_TRUE(history_backend()->GetURL(test_url, NULL));
 570   provider_->DeleteMatch(ac_matches_[0]);
 571
 572   // Check that the underlying URL is deleted from the history DB (this implies
 573   // that all visits are gone as well). Also verify that a deletion notification
 574   // is sent, in response to which the secondary data stores (InMemoryDatabase,
 575   // InMemoryURLIndex) will drop any data they might have pertaining to the URL.
 576   // To ensure that the deletion has been propagated everywhere before we start
 577   // verifying post-deletion states, first wait until we see the notification.
 578   content::WindowedNotificationObserver observer(
 579         chrome::NOTIFICATION_HISTORY_URLS_DELETED,
 580         content::NotificationService::AllSources());
 581   observer.Wait();
 582   EXPECT_FALSE(history_backend()->GetURL(test_url, NULL));
 583
 584   // Just to be on the safe side, explicitly verify that we have deleted enough
 585   // data so that we will not be serving the same result again.
 586   expected_urls.clear();
 587   RunTest(ASCIIToUTF16("slashdot"), false, expected_urls, true,
 588           ASCIIToUTF16("NONE EXPECTED"), base::string16());
 589 }
 590
 591 TEST_F(HistoryQuickProviderTest, PreventBeatingURLWhatYouTypedMatch) {
 592   std::vector<std::string> expected_urls;
 593
 594   expected_urls.clear();
 595   expected_urls.push_back("http://popularsitewithroot.com/");
 596   // If the user enters a hostname (no path) that he/she has visited
 597   // before, we should make sure that all HistoryQuickProvider results
 598   // have scores less than what HistoryURLProvider will assign the
 599   // URL-what-you-typed match.
 600   RunTest(ASCIIToUTF16("popularsitewithroot.com"), false, expected_urls, true,
 601           ASCIIToUTF16("popularsitewithroot.com"), base::string16());
 602   EXPECT_LT(ac_matches_[0].relevance,
 603             HistoryURLProvider::kScoreForBestInlineableResult);
 604
 605   // Check that if the user didn't quite enter the full hostname, this
 606   // hostname would've normally scored above the URL-what-you-typed match.
 607   RunTest(ASCIIToUTF16("popularsitewithroot.c"), false, expected_urls, true,
 608           ASCIIToUTF16("popularsitewithroot.com"),
 609                                ASCIIToUTF16("om"));
 610   EXPECT_GE(ac_matches_[0].relevance,
 611             HistoryURLProvider::kScoreForWhatYouTypedResult);
 612
 613   expected_urls.clear();
 614   expected_urls.push_back("http://popularsitewithpathonly.com/moo");
 615   // If the user enters a hostname of a host that he/she has visited
 616   // but never visited the root page of, we should make sure that all
 617   // HistoryQuickProvider results have scores less than what the
 618   // HistoryURLProvider will assign the URL-what-you-typed match.
 619   RunTest(ASCIIToUTF16("popularsitewithpathonly.com"), false, expected_urls,
 620           true,
 621           ASCIIToUTF16("popularsitewithpathonly.com/moo"),
 622                                      ASCIIToUTF16("/moo"));
 623   EXPECT_LT(ac_matches_[0].relevance,
 624             HistoryURLProvider::kScoreForUnvisitedIntranetResult);
 625
 626   // Verify the same thing happens if the user adds a / to end of the
 627   // hostname.
 628   RunTest(ASCIIToUTF16("popularsitewithpathonly.com/"), false, expected_urls,
 629           true, ASCIIToUTF16("popularsitewithpathonly.com/moo"),
 630                                             ASCIIToUTF16("moo"));
 631   EXPECT_LT(ac_matches_[0].relevance,
 632             HistoryURLProvider::kScoreForUnvisitedIntranetResult);
 633
 634   // Check that if the user didn't quite enter the full hostname, this
 635   // page would've normally scored above the URL-what-you-typed match.
 636   RunTest(ASCIIToUTF16("popularsitewithpathonly.co"), false, expected_urls,
 637           true, ASCIIToUTF16("popularsitewithpathonly.com/moo"),
 638                                           ASCIIToUTF16("m/moo"));
 639   EXPECT_GE(ac_matches_[0].relevance,
 640             HistoryURLProvider::kScoreForWhatYouTypedResult);
 641
 642   // If the user enters a hostname + path that he/she has not visited
 643   // before (but visited other things on the host), we can allow
 644   // inline autocompletions.
 645   RunTest(ASCIIToUTF16("popularsitewithpathonly.com/mo"), false, expected_urls,
 646           true,
 647           ASCIIToUTF16("popularsitewithpathonly.com/moo"),
 648                                         ASCIIToUTF16("o"));
 649   EXPECT_GE(ac_matches_[0].relevance,
 650             HistoryURLProvider::kScoreForWhatYouTypedResult);
 651
 652   // If the user enters a hostname + path that he/she has visited
 653   // before, we should make sure that all HistoryQuickProvider results
 654   // have scores less than what the HistoryURLProvider will assign
 655   // the URL-what-you-typed match.
 656   RunTest(ASCIIToUTF16("popularsitewithpathonly.com/moo"), false,
 657           expected_urls, true,
 658           ASCIIToUTF16("popularsitewithpathonly.com/moo"), base::string16());
 659   EXPECT_LT(ac_matches_[0].relevance,
 660             HistoryURLProvider::kScoreForBestInlineableResult);
 661 }
 662
 663 TEST_F(HistoryQuickProviderTest, PreventInlineAutocomplete) {
 664   std::vector<std::string> expected_urls;
 665   expected_urls.push_back("http://popularsitewithroot.com/");
 666
 667   // Check that the desired URL is normally allowed to be the default match
 668   // against input that is a prefex of the URL.
 669   RunTest(ASCIIToUTF16("popularsitewithr"), false, expected_urls, true,
 670           ASCIIToUTF16("popularsitewithroot.com"),
 671                           ASCIIToUTF16("oot.com"));
 672
 673   // Check that it's not allowed to be the default match if
 674   // prevent_inline_autocomplete is true.
 675   RunTest(ASCIIToUTF16("popularsitewithr"), true, expected_urls, false,
 676           ASCIIToUTF16("popularsitewithroot.com"),
 677                           ASCIIToUTF16("oot.com"));
 678
 679   // But the exact hostname can still match even if prevent inline autocomplete
 680   // is true.  i.e., there's no autocompletion necessary; this is effectively
 681   // URL-what-you-typed.
 682   RunTest(ASCIIToUTF16("popularsitewithroot.com"), true, expected_urls, true,
 683           ASCIIToUTF16("popularsitewithroot.com"), base::string16());
 684
 685   // The above still holds even with an extra trailing slash.
 686   RunTest(ASCIIToUTF16("popularsitewithroot.com/"), true, expected_urls, true,
 687           ASCIIToUTF16("popularsitewithroot.com"), base::string16());
 688 }
 689
 690 TEST_F(HistoryQuickProviderTest, CullSearchResults) {
 691   // Set up a default search engine.
 692   TemplateURLData data;
 693   data.SetKeyword(ASCIIToUTF16("TestEngine"));
 694   data.SetURL("http://testsearch.com/?q={searchTerms}");
 695   TemplateURLService* template_url_service =
 696       TemplateURLServiceFactory::GetForProfile(profile_.get());
 697   TemplateURL* template_url = new TemplateURL(data);
 698   template_url_service->Add(template_url);
 699   template_url_service->SetUserSelectedDefaultSearchProvider(template_url);
 700   template_url_service->Load();
 701
 702   // A search results page should not be returned when typing a query.
 703   std::vector<std::string> expected_urls;
 704   expected_urls.push_back("http://anotherengine.com/?q=thequery");
 705   RunTest(ASCIIToUTF16("thequery"), false, expected_urls, false,
 706           ASCIIToUTF16("anotherengine.com/?q=thequery"), base::string16());
 707
 708   // A search results page should not be returned when typing the engine URL.
 709   expected_urls.clear();
 710   expected_urls.push_back("http://testsearch.com/");
 711   RunTest(ASCIIToUTF16("testsearch"), false, expected_urls, true,
 712           ASCIIToUTF16("testsearch.com"),
 713                     ASCIIToUTF16(".com"));
 714 }
 715
 716 // HQPOrderingTest -------------------------------------------------------------
 717
 718 TestURLInfo ordering_test_db[] = {
 719   {"http://www.teamliquid.net/tlpd/korean/games/21648_bisu_vs_iris", "", 6, 3,
 720       256},
 721   {"http://www.amazon.com/", "amazon.com: online shopping for electronics, "
 722       "apparel, computers, books, dvds & more", 20, 20, 10},
 723   {"http://www.teamliquid.net/forum/viewmessage.php?topic_id=52045&"
 724       "currentpage=83", "google images", 6, 6, 0},
 725   {"http://www.tempurpedic.com/", "tempur-pedic", 7, 7, 0},
 726   {"http://www.teamfortress.com/", "", 5, 5, 6},
 727   {"http://www.rottentomatoes.com/", "", 3, 3, 7},
 728   {"http://music.google.com/music/listen?u=0#start_pl", "", 3, 3, 9},
 729   {"https://www.emigrantdirect.com/", "high interest savings account, high "
 730       "yield savings - emigrantdirect", 5, 5, 3},
 731   {"http://store.steampowered.com/", "", 6, 6, 1},
 732   {"http://techmeme.com/", "techmeme", 111, 110, 4},
 733   {"http://www.teamliquid.net/tlpd", "team liquid progaming database", 15, 15,
 734       2},
 735   {"http://store.steampowered.com/", "the steam summer camp sale", 6, 6, 1},
 736   {"http://www.teamliquid.net/tlpd/korean/players", "tlpd - bw korean - player "
 737       "index", 25, 7, 219},
 738   {"http://slashdot.org/", "slashdot: news for nerds, stuff that matters", 3, 3,
 739       6},
 740   {"http://translate.google.com/", "google translate", 3, 3, 0},
 741   {"http://arstechnica.com/", "ars technica", 3, 3, 3},
 742   {"http://www.rottentomatoes.com/", "movies | movie trailers | reviews - "
 743       "rotten tomatoes", 3, 3, 7},
 744   {"http://www.teamliquid.net/", "team liquid - starcraft 2 and brood war pro "
 745       "gaming news", 26, 25, 3},
 746   {"http://metaleater.com/", "metaleater", 4, 3, 8},
 747   {"http://half.com/", "half.com: textbooks , books , music , movies , games , "
 748       "video games", 4, 4, 6},
 749   {"http://teamliquid.net/", "team liquid - starcraft 2 and brood war pro "
 750       "gaming news", 8, 5, 9},
 751 };
 752
 753 class HQPOrderingTest : public HistoryQuickProviderTest {
 754  protected:
 755   virtual void GetTestData(size_t* data_count,
 756                            TestURLInfo** test_data) OVERRIDE;
 757 };
 758
 759 void HQPOrderingTest::GetTestData(size_t* data_count, TestURLInfo** test_data) {
 760   DCHECK(data_count);
 761   DCHECK(test_data);
 762   *data_count = arraysize(ordering_test_db);
 763   *test_data = &ordering_test_db[0];
 764 }
 765
 766 TEST_F(HQPOrderingTest, TEMatch) {
 767   std::vector<std::string> expected_urls;
 768   expected_urls.push_back("http://techmeme.com/");
 769   expected_urls.push_back("http://www.teamliquid.net/");
 770   expected_urls.push_back("http://www.teamliquid.net/tlpd");
 771   RunTest(ASCIIToUTF16("te"), false, expected_urls, true,
 772           ASCIIToUTF16("techmeme.com"),
 773             ASCIIToUTF16("chmeme.com"));
 774 }
 775
 776 TEST_F(HQPOrderingTest, TEAMatch) {
 777   std::vector<std::string> expected_urls;
 778   expected_urls.push_back("http://www.teamliquid.net/");
 779   expected_urls.push_back("http://www.teamliquid.net/tlpd");
 780   expected_urls.push_back("http://www.teamliquid.net/tlpd/korean/players");
 781   RunTest(ASCIIToUTF16("tea"), false, expected_urls, true,
 782           ASCIIToUTF16("www.teamliquid.net"),
 783                  ASCIIToUTF16("mliquid.net"));
 784 }