history_url_provider_unittest.cc revision eb525c5499e34cc9c4b825d6d9e75bb07cc06ace 
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_url_provider.h"
6
7#include <algorithm>
8
9#include "base/message_loop.h"
10#include "base/path_service.h"
11#include "base/strings/string_util.h"
12#include "base/strings/utf_string_conversions.h"
13#include "base/time/time.h"
14#include "chrome/browser/autocomplete/autocomplete_match.h"
15#include "chrome/browser/autocomplete/autocomplete_provider.h"
16#include "chrome/browser/autocomplete/autocomplete_provider_listener.h"
17#include "chrome/browser/autocomplete/history_quick_provider.h"
18#include "chrome/browser/history/history_service.h"
19#include "chrome/browser/history/history_service_factory.h"
20#include "chrome/browser/net/url_fixer_upper.h"
21#include "chrome/browser/search_engines/template_url.h"
22#include "chrome/browser/search_engines/template_url_service.h"
23#include "chrome/browser/search_engines/template_url_service_factory.h"
24#include "chrome/test/base/testing_browser_process.h"
25#include "chrome/test/base/testing_profile.h"
26#include "content/public/test/test_browser_thread_bundle.h"
27#include "testing/gtest/include/gtest/gtest.h"
28
29using base::Time;
30using base::TimeDelta;
31
32using content::TestBrowserThreadBundle;
33
34struct TestURLInfo {
35  const char* url;
36  const char* title;
37  int visit_count;
38  int typed_count;
39} test_db[] = {
40  {"http://www.google.com/", "Google", 3, 3},
41
42  // High-quality pages should get a host synthesized as a lower-quality match.
43  {"http://slashdot.org/favorite_page.html", "Favorite page", 200, 100},
44
45  // Less popular pages should have hosts synthesized as higher-quality
46  // matches.
47  {"http://kerneltrap.org/not_very_popular.html", "Less popular", 4, 0},
48
49  // Unpopular pages should not appear in the results at all.
50  {"http://freshmeat.net/unpopular.html", "Unpopular", 1, 0},
51
52  // If a host has a match, we should pick it up during host synthesis.
53  {"http://news.google.com/?ned=us&topic=n", "Google News - U.S.", 2, 2},
54  {"http://news.google.com/", "Google News", 1, 1},
55
56  // Matches that are normally not inline-autocompletable should be
57  // autocompleted if they are shorter substitutes for longer matches that would
58  // have been inline autocompleted.
59  {"http://synthesisatest.com/foo/", "Test A", 1, 1},
60  {"http://synthesisbtest.com/foo/", "Test B", 1, 1},
61  {"http://synthesisbtest.com/foo/bar.html", "Test B Bar", 2, 2},
62
63  // Suggested short URLs must be "good enough" and must match user input.
64  {"http://foo.com/", "Dir", 5, 5},
65  {"http://foo.com/dir/", "Dir", 2, 2},
66  {"http://foo.com/dir/another/", "Dir", 5, 1},
67  {"http://foo.com/dir/another/again/", "Dir", 10, 0},
68  {"http://foo.com/dir/another/again/myfile.html", "File", 10, 2},
69
70  // We throw in a lot of extra URLs here to make sure we're testing the
71  // history database's query, not just the autocomplete provider.
72  {"http://startest.com/y/a", "A", 2, 2},
73  {"http://startest.com/y/b", "B", 5, 2},
74  {"http://startest.com/x/c", "C", 5, 2},
75  {"http://startest.com/x/d", "D", 5, 5},
76  {"http://startest.com/y/e", "E", 4, 2},
77  {"http://startest.com/y/f", "F", 3, 2},
78  {"http://startest.com/y/g", "G", 3, 2},
79  {"http://startest.com/y/h", "H", 3, 2},
80  {"http://startest.com/y/i", "I", 3, 2},
81  {"http://startest.com/y/j", "J", 3, 2},
82  {"http://startest.com/y/k", "K", 3, 2},
83  {"http://startest.com/y/l", "L", 3, 2},
84  {"http://startest.com/y/m", "M", 3, 2},
85
86  // A file: URL is useful for testing that fixup does the right thing w.r.t.
87  // the number of trailing slashes on the user's input.
88  {"file:///C:/foo.txt", "", 2, 2},
89
90  // Results with absurdly high typed_counts so that very generic queries like
91  // "http" will give consistent results even if more data is added above.
92  {"http://bogussite.com/a", "Bogus A", 10002, 10000},
93  {"http://bogussite.com/b", "Bogus B", 10001, 10000},
94  {"http://bogussite.com/c", "Bogus C", 10000, 10000},
95
96  // Domain name with number.
97  {"http://www.17173.com/", "Domain with number", 3, 3},
98
99  // URLs to test exact-matching behavior.
100  {"http://go/", "Intranet URL", 1, 1},
101  {"http://gooey/", "Intranet URL 2", 5, 5},
102
103  // URLs for testing offset adjustment.
104  {"http://www.\xEA\xB5\x90\xEC\x9C\xA1.kr/", "Korean", 2, 2},
105  {"http://spaces.com/path%20with%20spaces/foo.html", "Spaces", 2, 2},
106  {"http://ms/c++%20style%20guide", "Style guide", 2, 2},
107
108  // URLs for testing ctrl-enter behavior.
109  {"http://binky/", "Intranet binky", 2, 2},
110  {"http://winky/", "Intranet winky", 2, 2},
111  {"http://www.winky.com/", "Internet winky", 5, 0},
112
113  // URLs used by EmptyVisits.
114  {"http://pandora.com/", "Pandora", 2, 2},
115  // This entry is explicitly added more recently than
116  // history::kLowQualityMatchAgeLimitInDays.
117  // {"http://p/", "p", 0, 0},
118
119  // For intranet based tests.
120  {"http://intra/one", "Intranet", 2, 2},
121  {"http://intra/two", "Intranet two", 1, 1},
122  {"http://intra/three", "Intranet three", 2, 2},
123  {"http://moo/bar", "Intranet moo", 1, 1},
124  {"http://typedhost/typedpath", "Intranet typed", 1, 1},
125  {"http://typedhost/untypedpath", "Intranet untyped", 1, 0},
126
127  {"http://x.com/one", "Internet", 2, 2},
128  {"http://x.com/two", "Internet two", 1, 1},
129  {"http://x.com/three", "Internet three", 2, 2},
130};
131
132class HistoryURLProviderTest : public testing::Test,
133                               public AutocompleteProviderListener {
134 public:
135  HistoryURLProviderTest()
136      : sort_matches_(false) {
137    HistoryQuickProvider::set_disabled(true);
138  }
139
140  virtual ~HistoryURLProviderTest() {
141    HistoryQuickProvider::set_disabled(false);
142  }
143
144  // AutocompleteProviderListener:
145  virtual void OnProviderUpdate(bool updated_matches) OVERRIDE;
146
147 protected:
148  static BrowserContextKeyedService* CreateTemplateURLService(
149      content::BrowserContext* profile) {
150    return new TemplateURLService(static_cast<Profile*>(profile));
151  }
152
153  // testing::Test
154  virtual void SetUp() {
155    SetUpImpl(false);
156  }
157  virtual void TearDown();
158
159  // Does the real setup.
160  void SetUpImpl(bool no_db);
161
162  // Fills test data into the history system.
163  void FillData();
164
165  // Runs an autocomplete query on |text| and checks to see that the returned
166  // results' destination URLs match those provided.  Also allows checking
167  // that the input type was identified correctly.
168  void RunTest(const string16 text,
169               const string16& desired_tld,
170               bool prevent_inline_autocomplete,
171               const std::string* expected_urls,
172               size_t num_results,
173               AutocompleteInput::Type* identified_input_type);
174
175  // A version of the above without the final |type| output parameter.
176  void RunTest(const string16 text,
177               const string16& desired_tld,
178               bool prevent_inline_autocomplete,
179               const std::string* expected_urls,
180               size_t num_results) {
181    AutocompleteInput::Type type;
182    return RunTest(text, desired_tld, prevent_inline_autocomplete,
183                   expected_urls, num_results, &type);
184  }
185
186  void RunAdjustOffsetTest(const string16 text, size_t expected_offset);
187
188  content::TestBrowserThreadBundle thread_bundle_;
189  ACMatches matches_;
190  scoped_ptr<TestingProfile> profile_;
191  HistoryService* history_service_;
192  scoped_refptr<HistoryURLProvider> autocomplete_;
193  // Should the matches be sorted and duplicates removed?
194  bool sort_matches_;
195};
196
197class HistoryURLProviderTestNoDB : public HistoryURLProviderTest {
198 protected:
199  virtual void SetUp() {
200    SetUpImpl(true);
201  }
202};
203
204void HistoryURLProviderTest::OnProviderUpdate(bool updated_matches) {
205  if (autocomplete_->done())
206    base::MessageLoop::current()->Quit();
207}
208
209void HistoryURLProviderTest::SetUpImpl(bool no_db) {
210  profile_.reset(new TestingProfile());
211  profile_->CreateHistoryService(true, no_db);
212  if (!no_db) {
213    profile_->BlockUntilHistoryProcessesPendingRequests();
214    profile_->BlockUntilHistoryIndexIsRefreshed();
215  }
216  history_service_ =
217      HistoryServiceFactory::GetForProfile(profile_.get(),
218                                           Profile::EXPLICIT_ACCESS);
219
220  autocomplete_ = new HistoryURLProvider(this, profile_.get(), "en-US,en,ko");
221  TemplateURLServiceFactory::GetInstance()->SetTestingFactoryAndUse(
222      profile_.get(), &HistoryURLProviderTest::CreateTemplateURLService);
223  FillData();
224}
225
226void HistoryURLProviderTest::TearDown() {
227  autocomplete_ = NULL;
228}
229
230void HistoryURLProviderTest::FillData() {
231  // All visits are a long time ago (some tests require this since we do some
232  // special logic for things visited very recently). Note that this time must
233  // be more recent than the "archived history" threshold for the data to go
234  // into the main database.
235  //
236  // TODO(brettw) It would be nice if we could test this behavior, in which
237  // case the time would be specifed in the test_db structure.
238  Time visit_time = Time::Now() - TimeDelta::FromDays(80);
239
240  for (size_t i = 0; i < arraysize(test_db); ++i) {
241    const TestURLInfo& cur = test_db[i];
242    const GURL current_url(cur.url);
243    history_service_->AddPageWithDetails(current_url, UTF8ToUTF16(cur.title),
244                                         cur.visit_count, cur.typed_count,
245                                         visit_time, false,
246                                         history::SOURCE_BROWSED);
247  }
248
249  history_service_->AddPageWithDetails(
250      GURL("http://p/"), UTF8ToUTF16("p"), 0, 0,
251      Time::Now() -
252      TimeDelta::FromDays(history::kLowQualityMatchAgeLimitInDays - 1),
253      false, history::SOURCE_BROWSED);
254}
255
256void HistoryURLProviderTest::RunTest(
257    const string16 text,
258    const string16& desired_tld,
259    bool prevent_inline_autocomplete,
260    const std::string* expected_urls,
261    size_t num_results,
262    AutocompleteInput::Type* identified_input_type) {
263  AutocompleteInput input(text, string16::npos, desired_tld, GURL(),
264                          prevent_inline_autocomplete, false, true,
265                          AutocompleteInput::ALL_MATCHES);
266  *identified_input_type = input.type();
267  autocomplete_->Start(input, false);
268  if (!autocomplete_->done())
269    base::MessageLoop::current()->Run();
270
271  matches_ = autocomplete_->matches();
272  if (sort_matches_) {
273    for (ACMatches::iterator i = matches_.begin(); i != matches_.end(); ++i)
274      i->ComputeStrippedDestinationURL(profile_.get());
275    std::sort(matches_.begin(), matches_.end(),
276              &AutocompleteMatch::DestinationSortFunc);
277    matches_.erase(std::unique(matches_.begin(), matches_.end(),
278                               &AutocompleteMatch::DestinationsEqual),
279                   matches_.end());
280    std::sort(matches_.begin(), matches_.end(),
281              &AutocompleteMatch::MoreRelevant);
282  }
283  ASSERT_EQ(num_results, matches_.size()) << "Input text: " << text
284                                          << "\nTLD: \"" << desired_tld << "\"";
285  for (size_t i = 0; i < num_results; ++i)
286    EXPECT_EQ(expected_urls[i], matches_[i].destination_url.spec());
287}
288
289void HistoryURLProviderTest::RunAdjustOffsetTest(const string16 text,
290                                                 size_t expected_offset) {
291  AutocompleteInput input(text, string16::npos, string16(), GURL(), false,
292                          false, true, AutocompleteInput::ALL_MATCHES);
293  autocomplete_->Start(input, false);
294  if (!autocomplete_->done())
295    base::MessageLoop::current()->Run();
296
297  matches_ = autocomplete_->matches();
298  ASSERT_GE(matches_.size(), 1U) << "Input text: " << text;
299  EXPECT_EQ(expected_offset, matches_[0].inline_autocomplete_offset);
300}
301
302TEST_F(HistoryURLProviderTest, PromoteShorterURLs) {
303  // Test that hosts get synthesized below popular pages.
304  const std::string expected_nonsynth[] = {
305    "http://slashdot.org/favorite_page.html",
306    "http://slashdot.org/",
307  };
308  RunTest(ASCIIToUTF16("slash"), string16(), true, expected_nonsynth,
309          arraysize(expected_nonsynth));
310
311  // Test that hosts get synthesized above less popular pages.
312  const std::string expected_synth[] = {
313    "http://kerneltrap.org/",
314    "http://kerneltrap.org/not_very_popular.html",
315  };
316  RunTest(ASCIIToUTF16("kernel"), string16(), true, expected_synth,
317          arraysize(expected_synth));
318
319  // Test that unpopular pages are ignored completely.
320  RunTest(ASCIIToUTF16("fresh"), string16(), true, NULL, 0);
321
322  // Test that if we create or promote shorter suggestions that would not
323  // normally be inline autocompletable, we make them inline autocompletable if
324  // the original suggestion (that we replaced as "top") was inline
325  // autocompletable.
326  const std::string expected_synthesisa[] = {
327    "http://synthesisatest.com/",
328    "http://synthesisatest.com/foo/",
329  };
330  RunTest(ASCIIToUTF16("synthesisa"), string16(), false, expected_synthesisa,
331          arraysize(expected_synthesisa));
332  EXPECT_LT(matches_.front().relevance, 1200);
333  const std::string expected_synthesisb[] = {
334    "http://synthesisbtest.com/foo/",
335    "http://synthesisbtest.com/foo/bar.html",
336  };
337  RunTest(ASCIIToUTF16("synthesisb"), string16(), false, expected_synthesisb,
338          arraysize(expected_synthesisb));
339  EXPECT_GE(matches_.front().relevance, 1410);
340
341  // Test that if we have a synthesized host that matches a suggestion, they
342  // get combined into one.
343  const std::string expected_combine[] = {
344    "http://news.google.com/",
345    "http://news.google.com/?ned=us&topic=n",
346  };
347  ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("news"), string16(), true,
348      expected_combine, arraysize(expected_combine)));
349  // The title should also have gotten set properly on the host for the
350  // synthesized one, since it was also in the results.
351  EXPECT_EQ(ASCIIToUTF16("Google News"), matches_.front().description);
352
353  // Test that short URL matching works correctly as the user types more
354  // (several tests):
355  // The entry for foo.com is the best of all five foo.com* entries.
356  const std::string short_1[] = {
357    "http://foo.com/",
358    "http://foo.com/dir/another/again/myfile.html",
359    "http://foo.com/dir/",
360  };
361  RunTest(ASCIIToUTF16("foo"), string16(), true, short_1, arraysize(short_1));
362
363  // When the user types the whole host, make sure we don't get two results for
364  // it.
365  const std::string short_2[] = {
366    "http://foo.com/",
367    "http://foo.com/dir/another/again/myfile.html",
368    "http://foo.com/dir/",
369    "http://foo.com/dir/another/",
370  };
371  RunTest(ASCIIToUTF16("foo.com"), string16(), true, short_2,
372          arraysize(short_2));
373  RunTest(ASCIIToUTF16("foo.com/"), string16(), true, short_2,
374          arraysize(short_2));
375
376  // The filename is the second best of the foo.com* entries, but there is a
377  // shorter URL that's "good enough".  The host doesn't match the user input
378  // and so should not appear.
379  const std::string short_3[] = {
380    "http://foo.com/d",
381    "http://foo.com/dir/another/",
382    "http://foo.com/dir/another/again/myfile.html",
383    "http://foo.com/dir/",
384  };
385  RunTest(ASCIIToUTF16("foo.com/d"), string16(), true, short_3,
386          arraysize(short_3));
387
388  // We shouldn't promote shorter URLs than the best if they're not good
389  // enough.
390  const std::string short_4[] = {
391    "http://foo.com/dir/another/a",
392    "http://foo.com/dir/another/again/myfile.html",
393    "http://foo.com/dir/another/again/",
394  };
395  RunTest(ASCIIToUTF16("foo.com/dir/another/a"), string16(), true, short_4,
396          arraysize(short_4));
397
398  // Exact matches should always be best no matter how much more another match
399  // has been typed.
400  const std::string short_5a[] = {
401    "http://gooey/",
402    "http://www.google.com/",
403    "http://go/",
404  };
405  const std::string short_5b[] = {
406    "http://go/",
407    "http://gooey/",
408    "http://www.google.com/",
409  };
410  RunTest(ASCIIToUTF16("g"), string16(), false, short_5a, arraysize(short_5a));
411  RunTest(ASCIIToUTF16("go"), string16(), false, short_5b, arraysize(short_5b));
412}
413
414TEST_F(HistoryURLProviderTest, CullRedirects) {
415  // URLs we will be using, plus the visit counts they will initially get
416  // (the redirect set below will also increment the visit counts). We want
417  // the results to be in A,B,C order. Note also that our visit counts are
418  // all high enough so that domain synthesizing won't get triggered.
419  struct TestCase {
420    const char* url;
421    int count;
422  } test_cases[] = {
423    {"http://redirects/A", 30},
424    {"http://redirects/B", 20},
425    {"http://redirects/C", 10}
426  };
427  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(test_cases); ++i) {
428    history_service_->AddPageWithDetails(GURL(test_cases[i].url),
429        UTF8ToUTF16("Title"), test_cases[i].count, test_cases[i].count,
430        Time::Now(), false, history::SOURCE_BROWSED);
431  }
432
433  // Create a B->C->A redirect chain, but set the visit counts such that they
434  // will appear in A,B,C order in the results. The autocomplete query will
435  // search for the most recent visit when looking for redirects, so this will
436  // be found even though the previous visits had no redirects.
437  history::RedirectList redirects_to_a;
438  redirects_to_a.push_back(GURL(test_cases[1].url));
439  redirects_to_a.push_back(GURL(test_cases[2].url));
440  redirects_to_a.push_back(GURL(test_cases[0].url));
441  history_service_->AddPage(GURL(test_cases[0].url), base::Time::Now(),
442      NULL, 0, GURL(), redirects_to_a, content::PAGE_TRANSITION_TYPED,
443      history::SOURCE_BROWSED, true);
444
445  // Because all the results are part of a redirect chain with other results,
446  // all but the first one (A) should be culled. We should get the default
447  // "what you typed" result, plus this one.
448  const string16 typing(ASCIIToUTF16("http://redirects/"));
449  const std::string expected_results[] = {
450    UTF16ToUTF8(typing),
451    test_cases[0].url,
452  };
453  RunTest(typing, string16(), true, expected_results,
454          arraysize(expected_results));
455}
456
457TEST_F(HistoryURLProviderTest, WhatYouTyped) {
458  // Make sure we suggest a What You Typed match at the right times.
459  RunTest(ASCIIToUTF16("wytmatch"), string16(), false, NULL, 0);
460  RunTest(ASCIIToUTF16("wytmatch foo bar"), string16(), false, NULL, 0);
461  RunTest(ASCIIToUTF16("wytmatch+foo+bar"), string16(), false, NULL, 0);
462  RunTest(ASCIIToUTF16("wytmatch+foo+bar.com"), string16(), false, NULL, 0);
463
464  const std::string results_1[] = {"http://www.wytmatch.com/"};
465  RunTest(ASCIIToUTF16("wytmatch"), ASCIIToUTF16("com"), false, results_1,
466          arraysize(results_1));
467
468  const std::string results_2[] = {"http://wytmatch%20foo%20bar/"};
469  RunTest(ASCIIToUTF16("http://wytmatch foo bar"), string16(), false, results_2,
470          arraysize(results_2));
471
472  const std::string results_3[] = {"https://wytmatch%20foo%20bar/"};
473  RunTest(ASCIIToUTF16("https://wytmatch foo bar"), string16(), false,
474          results_3, arraysize(results_3));
475}
476
477TEST_F(HistoryURLProviderTest, Fixup) {
478  // Test for various past crashes we've had.
479  RunTest(ASCIIToUTF16("\\"), string16(), false, NULL, 0);
480  RunTest(ASCIIToUTF16("#"), string16(), false, NULL, 0);
481  RunTest(ASCIIToUTF16("%20"), string16(), false, NULL, 0);
482  const std::string fixup_crash[] = {"http://%EF%BD%A5@s/"};
483  RunTest(WideToUTF16(L"\uff65@s"), string16(), false, fixup_crash,
484          arraysize(fixup_crash));
485  RunTest(WideToUTF16(L"\u2015\u2015@ \uff7c"), string16(), false, NULL, 0);
486
487  // Fixing up "file:" should result in an inline autocomplete offset of just
488  // after "file:", not just after "file://".
489  const string16 input_1(ASCIIToUTF16("file:"));
490  const std::string fixup_1[] = {"file:///C:/foo.txt"};
491  ASSERT_NO_FATAL_FAILURE(RunTest(input_1, string16(), false, fixup_1,
492                                  arraysize(fixup_1)));
493  EXPECT_EQ(input_1.length(), matches_.front().inline_autocomplete_offset);
494
495  // Fixing up "http:/" should result in an inline autocomplete offset of just
496  // after "http:/", not just after "http:".
497  const string16 input_2(ASCIIToUTF16("http:/"));
498  const std::string fixup_2[] = {
499    "http://bogussite.com/a",
500    "http://bogussite.com/b",
501    "http://bogussite.com/c",
502  };
503  ASSERT_NO_FATAL_FAILURE(RunTest(input_2, string16(), false, fixup_2,
504                                  arraysize(fixup_2)));
505  EXPECT_EQ(input_2.length(), matches_.front().inline_autocomplete_offset);
506
507  // Adding a TLD to a small number like "56" should result in "www.56.com"
508  // rather than "0.0.0.56.com".
509  const std::string fixup_3[] = {"http://www.56.com/"};
510  RunTest(ASCIIToUTF16("56"), ASCIIToUTF16("com"), true, fixup_3,
511          arraysize(fixup_3));
512
513  // An input looks like a IP address like "127.0.0.1" should result in
514  // "http://127.0.0.1/".
515  const std::string fixup_4[] = {"http://127.0.0.1/"};
516  RunTest(ASCIIToUTF16("127.0.0.1"), string16(), false, fixup_4,
517          arraysize(fixup_4));
518
519  // An number "17173" should result in "http://www.17173.com/" in db.
520  const std::string fixup_5[] = {"http://www.17173.com/"};
521  RunTest(ASCIIToUTF16("17173"), string16(), false, fixup_5,
522          arraysize(fixup_5));
523}
524
525TEST_F(HistoryURLProviderTest, AdjustOffset) {
526  RunAdjustOffsetTest(WideToUTF16(L"http://www.\uAD50\uC721"), 13);
527  RunAdjustOffsetTest(ASCIIToUTF16("http://spaces.com/path%20with%20spa"), 31);
528  RunAdjustOffsetTest(ASCIIToUTF16("http://ms/c++ s"), 15);
529}
530
531// Make sure the results for the input 'p' don't change between the first and
532// second passes.
533TEST_F(HistoryURLProviderTest, EmptyVisits) {
534  // Wait for history to create the in memory DB.
535  profile_->BlockUntilHistoryProcessesPendingRequests();
536
537  AutocompleteInput input(ASCIIToUTF16("p"), string16::npos, string16(), GURL(),
538                          false, false, true, AutocompleteInput::ALL_MATCHES);
539  autocomplete_->Start(input, false);
540  // HistoryURLProvider shouldn't be done (waiting on async results).
541  EXPECT_FALSE(autocomplete_->done());
542
543  // We should get back an entry for pandora.
544  matches_ = autocomplete_->matches();
545  ASSERT_GT(matches_.size(), 0u);
546  EXPECT_EQ(GURL("http://pandora.com/"), matches_[0].destination_url);
547  int pandora_relevance = matches_[0].relevance;
548
549  // Run the message loop. When |autocomplete_| finishes the loop is quit.
550  base::MessageLoop::current()->Run();
551  EXPECT_TRUE(autocomplete_->done());
552  matches_ = autocomplete_->matches();
553  ASSERT_GT(matches_.size(), 0u);
554  EXPECT_EQ(GURL("http://pandora.com/"), matches_[0].destination_url);
555  EXPECT_EQ(pandora_relevance, matches_[0].relevance);
556}
557
558TEST_F(HistoryURLProviderTestNoDB, NavigateWithoutDB) {
559  // Ensure that we will still produce matches for navigation when there is no
560  // database.
561  std::string navigation_1[] = {"http://test.com/"};
562  RunTest(ASCIIToUTF16("test.com"), string16(), false, navigation_1,
563          arraysize(navigation_1));
564
565  std::string navigation_2[] = {"http://slash/"};
566  RunTest(ASCIIToUTF16("slash"), string16(), false, navigation_2,
567          arraysize(navigation_2));
568
569  RunTest(ASCIIToUTF16("this is a query"), string16(), false, NULL, 0);
570}
571
572TEST_F(HistoryURLProviderTest, DontAutocompleteOnTrailingWhitespace) {
573  AutocompleteInput input(ASCIIToUTF16("slash "), string16::npos, string16(),
574                          GURL(), false, false, true,
575                          AutocompleteInput::ALL_MATCHES);
576  autocomplete_->Start(input, false);
577  if (!autocomplete_->done())
578    base::MessageLoop::current()->Run();
579
580  // None of the matches should attempt to autocomplete.
581  matches_ = autocomplete_->matches();
582  for (size_t i = 0; i < matches_.size(); ++i)
583    EXPECT_EQ(string16::npos, matches_[i].inline_autocomplete_offset);
584}
585
586TEST_F(HistoryURLProviderTest, TreatEmailsAsSearches) {
587  // Visiting foo.com should not make this string be treated as a navigation.
588  // That means the result should be scored around 1200 ("what you typed")
589  // and not 1400+.
590  const std::string expected[] = {"http://user@foo.com/"};
591  ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("user@foo.com"), string16(),
592                                  false, expected, arraysize(expected)));
593  EXPECT_LE(1200, matches_[0].relevance);
594  EXPECT_LT(matches_[0].relevance, 1210);
595}
596
597TEST_F(HistoryURLProviderTest, IntranetURLsWithPaths) {
598  struct TestCase {
599    const char* input;
600    int relevance;
601  } test_cases[] = {
602    { "fooey", 0 },
603    { "fooey/", 1200 },     // 1200 for URL would still navigate by default.
604    { "fooey/a", 1200 },    // 1200 for UNKNOWN would not.
605    { "fooey/a b", 1200 },  // Also UNKNOWN.
606    { "gooey", 1410 },
607    { "gooey/", 1410 },
608    { "gooey/a", 1400 },
609    { "gooey/a b", 1400 },
610  };
611  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(test_cases); ++i) {
612    SCOPED_TRACE(test_cases[i].input);
613    if (test_cases[i].relevance == 0) {
614      RunTest(ASCIIToUTF16(test_cases[i].input), string16(), false, NULL, 0);
615    } else {
616      const std::string output[] = {
617        URLFixerUpper::FixupURL(test_cases[i].input, std::string()).spec()
618      };
619      ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16(test_cases[i].input),
620                              string16(), false, output, arraysize(output)));
621      // Actual relevance should be at least what test_cases expects and
622      // and no more than 10 more.
623      EXPECT_LE(test_cases[i].relevance, matches_[0].relevance);
624      EXPECT_LT(matches_[0].relevance, test_cases[i].relevance + 10);
625    }
626  }
627}
628
629TEST_F(HistoryURLProviderTest, IntranetURLsWithRefs) {
630  struct TestCase {
631    const char* input;
632    int relevance;
633    AutocompleteInput::Type type;
634  } test_cases[] = {
635    { "gooey", 1410, AutocompleteInput::UNKNOWN },
636    { "gooey/", 1410, AutocompleteInput::URL },
637    { "gooey#", 1200, AutocompleteInput::UNKNOWN },
638    { "gooey/#", 1200, AutocompleteInput::URL },
639    { "gooey#foo", 1200, AutocompleteInput::UNKNOWN },
640    { "gooey/#foo", 1200, AutocompleteInput::URL },
641    { "gooey# foo", 1200, AutocompleteInput::UNKNOWN },
642    { "gooey/# foo", 1200, AutocompleteInput::URL },
643  };
644  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(test_cases); ++i) {
645    SCOPED_TRACE(test_cases[i].input);
646    const std::string output[] = {
647      URLFixerUpper::FixupURL(test_cases[i].input, std::string()).spec()
648    };
649    AutocompleteInput::Type type;
650    ASSERT_NO_FATAL_FAILURE(
651        RunTest(ASCIIToUTF16(test_cases[i].input),
652                string16(), false, output, arraysize(output), &type));
653    // Actual relevance should be at least what test_cases expects and
654    // and no more than 10 more.
655    EXPECT_LE(test_cases[i].relevance, matches_[0].relevance);
656    EXPECT_LT(matches_[0].relevance, test_cases[i].relevance + 10);
657    // Input type should be what we expect.  This is important because
658    // this provider counts on SearchProvider to give queries a relevance
659    // score >1200 for UNKNOWN inputs and <1200 for URL inputs.  (That's
660    // already tested in search_provider_unittest.cc.)  For this test
661    // here to test that the user sees the correct behavior, it needs
662    // to check that the input type was identified correctly.
663    EXPECT_EQ(test_cases[i].type, type);
664  }
665}
666
667// Makes sure autocompletion happens for intranet sites that have been
668// previoulsy visited.
669TEST_F(HistoryURLProviderTest, IntranetURLCompletion) {
670  sort_matches_ = true;
671
672  const std::string expected1[] = {
673    "http://intra/three",
674    "http://intra/two",
675  };
676  ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("intra/t"), string16(), false,
677                                  expected1, arraysize(expected1)));
678  EXPECT_LE(1410, matches_[0].relevance);
679  EXPECT_LT(matches_[0].relevance, 1420);
680  EXPECT_EQ(matches_[0].relevance - 1, matches_[1].relevance);
681
682  const std::string expected2[] = {
683    "http://moo/b",
684    "http://moo/bar",
685  };
686  ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("moo/b"), string16(), false,
687                                  expected2, arraysize(expected2)));
688  // The url what you typed match should be around 1400, otherwise the
689  // search what you typed match is going to be first.
690  EXPECT_LE(1400, matches_[0].relevance);
691  EXPECT_LT(matches_[0].relevance, 1410);
692
693  const std::string expected3[] = {
694    "http://intra/one",
695    "http://intra/three",
696    "http://intra/two",
697  };
698  RunTest(ASCIIToUTF16("intra"), string16(), false, expected3,
699          arraysize(expected3));
700
701  const std::string expected4[] = {
702    "http://intra/one",
703    "http://intra/three",
704    "http://intra/two",
705  };
706  RunTest(ASCIIToUTF16("intra/"), string16(), false, expected4,
707          arraysize(expected4));
708
709  const std::string expected5[] = {
710    "http://intra/one",
711  };
712  ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("intra/o"), string16(), false,
713                                  expected5, arraysize(expected5)));
714  EXPECT_LE(1410, matches_[0].relevance);
715  EXPECT_LT(matches_[0].relevance, 1420);
716
717  const std::string expected6[] = {
718    "http://intra/x",
719  };
720  ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("intra/x"), string16(), false,
721                                  expected6, arraysize(expected6)));
722  EXPECT_LE(1400, matches_[0].relevance);
723  EXPECT_LT(matches_[0].relevance, 1410);
724
725  const std::string expected7[] = {
726    "http://typedhost/untypedpath",
727  };
728  ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("typedhost/untypedpath"),
729      string16(), false, expected7, arraysize(expected7)));
730  EXPECT_LE(1400, matches_[0].relevance);
731  EXPECT_LT(matches_[0].relevance, 1410);
732}
733
734TEST_F(HistoryURLProviderTest, CrashDueToFixup) {
735  // This test passes if we don't crash.  The results don't matter.
736  const char* const test_cases[] = {
737    "//c",
738    "\\@st"
739  };
740  for (size_t i = 0; i < arraysize(test_cases); ++i) {
741    AutocompleteInput input(ASCIIToUTF16(test_cases[i]), string16::npos,
742                            string16(), GURL(), false, false, true,
743                            AutocompleteInput::ALL_MATCHES);
744    autocomplete_->Start(input, false);
745    if (!autocomplete_->done())
746      base::MessageLoop::current()->Run();
747  }
748}
749
750TEST_F(HistoryURLProviderTest, CullSearchResults) {
751  // Set up a default search engine.
752  TemplateURLData data;
753  data.SetKeyword(ASCIIToUTF16("TestEngine"));
754  data.SetURL("http://testsearch.com/?q={searchTerms}");
755  TemplateURLService* template_url_service =
756      TemplateURLServiceFactory::GetForProfile(profile_.get());
757  TemplateURL* template_url = new TemplateURL(profile_.get(), data);
758  template_url_service->Add(template_url);
759  template_url_service->SetDefaultSearchProvider(template_url);
760  template_url_service->Load();
761
762  // URLs we will be using, plus the visit counts they will initially get
763  // (the redirect set below will also increment the visit counts). We want
764  // the results to be in A,B,C order. Note also that our visit counts are
765  // all high enough so that domain synthesizing won't get triggered.
766  struct TestCase {
767    const char* url;
768    int count;
769  } test_cases[] = {
770    {"https://testsearch.com/", 30},
771    {"https://testsearch.com/?q=foobar", 20},
772    {"http://foobar.com/", 10}
773  };
774  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(test_cases); ++i) {
775    history_service_->AddPageWithDetails(GURL(test_cases[i].url),
776        UTF8ToUTF16("Title"), test_cases[i].count, test_cases[i].count,
777        Time::Now(), false, history::SOURCE_BROWSED);
778  }
779
780  // We should not see search URLs when typing a previously used query.
781  const std::string expected_when_searching_query[] = {
782    test_cases[2].url
783  };
784  RunTest(ASCIIToUTF16("foobar"), string16(), true,
785      expected_when_searching_query, arraysize(expected_when_searching_query));
786
787  // We should not see search URLs when typing the search engine name.
788  const std::string expected_when_searching_site[] = {
789    test_cases[0].url
790  };
791  RunTest(ASCIIToUTF16("testsearch"), string16(), true,
792      expected_when_searching_site, arraysize(expected_when_searching_site));
793}
794
795TEST_F(HistoryURLProviderTest, SuggestExactInput) {
796  const size_t npos = std::string::npos;
797  struct TestCase {
798    // Inputs:
799    const char* input;
800    bool trim_http;
801    // Expected Outputs:
802    const char* contents;
803    // Offsets of the ACMatchClassifications, terminated by npos.
804    size_t offsets[3];
805    // The index of the ACMatchClassification that should have the MATCH bit
806    // set, npos if no ACMatchClassification should have the MATCH bit set.
807    size_t match_classification_index;
808  } test_cases[] = {
809    { "http://www.somesite.com", false,
810      "http://www.somesite.com", {0, npos, npos}, 0 },
811    { "www.somesite.com", true,
812      "www.somesite.com", {0, npos, npos}, 0 },
813    { "www.somesite.com", false,
814      "http://www.somesite.com", {0, 7, npos}, 1 },
815    { "somesite.com", true,
816      "somesite.com", {0, npos, npos}, 0 },
817    { "somesite.com", false,
818      "http://somesite.com", {0, 7, npos}, 1 },
819    { "w", true,
820      "w", {0, npos, npos}, 0 },
821    { "w", false,
822      "http://w", {0, 7, npos}, 1 },
823    { "w.com", true,
824      "w.com", {0, npos, npos}, 0 },
825    { "w.com", false,
826      "http://w.com", {0, 7, npos}, 1 },
827    { "www.w.com", true,
828      "www.w.com", {0, npos, npos}, 0 },
829    { "www.w.com", false,
830      "http://www.w.com", {0, 7, npos}, 1 },
831    { "view-source:www.w.com/", true,
832      "view-source:www.w.com", {0, npos, npos}, npos },
833    { "view-source:www.w.com/", false,
834      "view-source:http://www.w.com", {0, npos, npos}, npos },
835    { "view-source:http://www.w.com/", false,
836      "view-source:http://www.w.com", {0, npos, npos}, 0 },
837    { "   view-source:", true,
838      "view-source:", {0, npos, npos}, 0 },
839    { "http:////////w.com", false,
840      "http://w.com", {0, npos, npos}, npos },
841    { "    http:////////www.w.com", false,
842      "http://www.w.com", {0, npos, npos}, npos },
843    { "http:a///www.w.com", false,
844      "http://a///www.w.com", {0, npos, npos}, npos },
845    { "mailto://a@b.com", true,
846      "mailto://a@b.com", {0, npos, npos}, 0 },
847    { "mailto://a@b.com", false,
848      "mailto://a@b.com", {0, npos, npos}, 0 },
849  };
850  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(test_cases); ++i) {
851    SCOPED_TRACE(testing::Message() << "Index " << i << " input: "
852                                    << test_cases[i].input << ", trim_http: "
853                                    << test_cases[i].trim_http);
854
855    AutocompleteInput input(ASCIIToUTF16(test_cases[i].input), string16::npos,
856                            string16(), GURL("about:blank"),
857                            false, false, true, AutocompleteInput::ALL_MATCHES);
858    AutocompleteMatch match = HistoryURLProvider::SuggestExactInput(
859        autocomplete_.get(), input, test_cases[i].trim_http);
860    EXPECT_EQ(ASCIIToUTF16(test_cases[i].contents), match.contents);
861    for (size_t match_index = 0; match_index < match.contents_class.size();
862         ++match_index) {
863      EXPECT_EQ(test_cases[i].offsets[match_index],
864                match.contents_class[match_index].offset);
865      EXPECT_EQ(ACMatchClassification::URL |
866                (match_index == test_cases[i].match_classification_index ?
867                 ACMatchClassification::MATCH : 0),
868                match.contents_class[match_index].style);
869    }
870    EXPECT_EQ(npos, test_cases[i].offsets[match.contents_class.size()]);
871  }
872}
873