history_url_provider_unittest.cc revision 1e9bf3e0803691d0a228da41fc608347b6db4340 
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/message_loop.h"
10#include "base/path_service.h"
11#include "base/prefs/pref_service.h"
12#include "base/strings/string_util.h"
13#include "base/strings/utf_string_conversions.h"
14#include "base/time/time.h"
15#include "chrome/browser/autocomplete/autocomplete_match.h"
16#include "chrome/browser/autocomplete/autocomplete_provider.h"
17#include "chrome/browser/autocomplete/autocomplete_provider_listener.h"
18#include "chrome/browser/autocomplete/history_quick_provider.h"
19#include "chrome/browser/history/history_service.h"
20#include "chrome/browser/history/history_service_factory.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/common/net/url_fixer_upper.h"
25#include "chrome/common/pref_names.h"
26#include "chrome/test/base/testing_browser_process.h"
27#include "chrome/test/base/testing_profile.h"
28#include "content/public/test/test_browser_thread_bundle.h"
29#include "testing/gtest/include/gtest/gtest.h"
30
31using base::Time;
32using base::TimeDelta;
33
34using content::TestBrowserThreadBundle;
35
36struct TestURLInfo {
37  const char* url;
38  const char* title;
39  int visit_count;
40  int typed_count;
41} test_db[] = {
42  {"http://www.google.com/", "Google", 3, 3},
43
44  // High-quality pages should get a host synthesized as a lower-quality match.
45  {"http://slashdot.org/favorite_page.html", "Favorite page", 200, 100},
46
47  // Less popular pages should have hosts synthesized as higher-quality
48  // matches.
49  {"http://kerneltrap.org/not_very_popular.html", "Less popular", 4, 0},
50
51  // Unpopular pages should not appear in the results at all.
52  {"http://freshmeat.net/unpopular.html", "Unpopular", 1, 0},
53
54  // If a host has a match, we should pick it up during host synthesis.
55  {"http://news.google.com/?ned=us&topic=n", "Google News - U.S.", 2, 2},
56  {"http://news.google.com/", "Google News", 1, 1},
57
58  // Matches that are normally not inline-autocompletable should be
59  // autocompleted if they are shorter substitutes for longer matches that would
60  // have been inline autocompleted.
61  {"http://synthesisatest.com/foo/", "Test A", 1, 1},
62  {"http://synthesisbtest.com/foo/", "Test B", 1, 1},
63  {"http://synthesisbtest.com/foo/bar.html", "Test B Bar", 2, 2},
64
65  // Suggested short URLs must be "good enough" and must match user input.
66  {"http://foo.com/", "Dir", 5, 5},
67  {"http://foo.com/dir/", "Dir", 2, 2},
68  {"http://foo.com/dir/another/", "Dir", 5, 1},
69  {"http://foo.com/dir/another/again/", "Dir", 10, 0},
70  {"http://foo.com/dir/another/again/myfile.html", "File", 10, 2},
71
72  // We throw in a lot of extra URLs here to make sure we're testing the
73  // history database's query, not just the autocomplete provider.
74  {"http://startest.com/y/a", "A", 2, 2},
75  {"http://startest.com/y/b", "B", 5, 2},
76  {"http://startest.com/x/c", "C", 5, 2},
77  {"http://startest.com/x/d", "D", 5, 5},
78  {"http://startest.com/y/e", "E", 4, 2},
79  {"http://startest.com/y/f", "F", 3, 2},
80  {"http://startest.com/y/g", "G", 3, 2},
81  {"http://startest.com/y/h", "H", 3, 2},
82  {"http://startest.com/y/i", "I", 3, 2},
83  {"http://startest.com/y/j", "J", 3, 2},
84  {"http://startest.com/y/k", "K", 3, 2},
85  {"http://startest.com/y/l", "L", 3, 2},
86  {"http://startest.com/y/m", "M", 3, 2},
87
88  // A file: URL is useful for testing that fixup does the right thing w.r.t.
89  // the number of trailing slashes on the user's input.
90  {"file:///C:/foo.txt", "", 2, 2},
91
92  // Results with absurdly high typed_counts so that very generic queries like
93  // "http" will give consistent results even if more data is added above.
94  {"http://bogussite.com/a", "Bogus A", 10002, 10000},
95  {"http://bogussite.com/b", "Bogus B", 10001, 10000},
96  {"http://bogussite.com/c", "Bogus C", 10000, 10000},
97
98  // Domain name with number.
99  {"http://www.17173.com/", "Domain with number", 3, 3},
100
101  // URLs to test exact-matching behavior.
102  {"http://go/", "Intranet URL", 1, 1},
103  {"http://gooey/", "Intranet URL 2", 5, 5},
104
105  // URLs for testing offset adjustment.
106  {"http://www.\xEA\xB5\x90\xEC\x9C\xA1.kr/", "Korean", 2, 2},
107  {"http://spaces.com/path%20with%20spaces/foo.html", "Spaces", 2, 2},
108  {"http://ms/c++%20style%20guide", "Style guide", 2, 2},
109
110  // URLs for testing ctrl-enter behavior.
111  {"http://binky/", "Intranet binky", 2, 2},
112  {"http://winky/", "Intranet winky", 2, 2},
113  {"http://www.winky.com/", "Internet winky", 5, 0},
114
115  // URLs used by EmptyVisits.
116  {"http://pandora.com/", "Pandora", 2, 2},
117  // This entry is explicitly added more recently than
118  // history::kLowQualityMatchAgeLimitInDays.
119  // {"http://p/", "p", 0, 0},
120
121  // For intranet based tests.
122  {"http://intra/one", "Intranet", 2, 2},
123  {"http://intra/two", "Intranet two", 1, 1},
124  {"http://intra/three", "Intranet three", 2, 2},
125  {"http://moo/bar", "Intranet moo", 1, 1},
126  {"http://typedhost/typedpath", "Intranet typed", 1, 1},
127  {"http://typedhost/untypedpath", "Intranet untyped", 1, 0},
128
129  {"http://x.com/one", "Internet", 2, 2},
130  {"http://x.com/two", "Internet two", 1, 1},
131  {"http://x.com/three", "Internet three", 2, 2},
132};
133
134class HistoryURLProviderTest : public testing::Test,
135                               public AutocompleteProviderListener {
136 public:
137  struct UrlAndLegalDefault {
138    std::string url;
139    bool allowed_to_be_default_match;
140  };
141
142  HistoryURLProviderTest()
143      : sort_matches_(false) {
144    HistoryQuickProvider::set_disabled(true);
145  }
146
147  virtual ~HistoryURLProviderTest() {
148    HistoryQuickProvider::set_disabled(false);
149  }
150
151  // AutocompleteProviderListener:
152  virtual void OnProviderUpdate(bool updated_matches) OVERRIDE;
153
154 protected:
155  static BrowserContextKeyedService* CreateTemplateURLService(
156      content::BrowserContext* profile) {
157    return new TemplateURLService(static_cast<Profile*>(profile));
158  }
159
160  // testing::Test
161  virtual void SetUp() {
162    ASSERT_TRUE(SetUpImpl(false));
163  }
164  virtual void TearDown();
165
166  // Does the real setup.
167  bool SetUpImpl(bool no_db) WARN_UNUSED_RESULT;
168
169  // Fills test data into the history system.
170  void FillData();
171
172  // Runs an autocomplete query on |text| and checks to see that the returned
173  // results' destination URLs match those provided.  Also allows checking
174  // that the input type was identified correctly.
175  void RunTest(const string16 text,
176               const string16& desired_tld,
177               bool prevent_inline_autocomplete,
178               const UrlAndLegalDefault* expected_urls,
179               size_t num_results,
180               AutocompleteInput::Type* identified_input_type);
181
182  // A version of the above without the final |type| output parameter.
183  void RunTest(const string16 text,
184               const string16& desired_tld,
185               bool prevent_inline_autocomplete,
186               const UrlAndLegalDefault* expected_urls,
187               size_t num_results) {
188    AutocompleteInput::Type type;
189    return RunTest(text, desired_tld, prevent_inline_autocomplete,
190                   expected_urls, num_results, &type);
191  }
192
193  content::TestBrowserThreadBundle thread_bundle_;
194  ACMatches matches_;
195  scoped_ptr<TestingProfile> profile_;
196  HistoryService* history_service_;
197  scoped_refptr<HistoryURLProvider> autocomplete_;
198  // Should the matches be sorted and duplicates removed?
199  bool sort_matches_;
200};
201
202class HistoryURLProviderTestNoDB : public HistoryURLProviderTest {
203 protected:
204  virtual void SetUp() {
205    ASSERT_TRUE(SetUpImpl(true));
206  }
207};
208
209void HistoryURLProviderTest::OnProviderUpdate(bool updated_matches) {
210  if (autocomplete_->done())
211    base::MessageLoop::current()->Quit();
212}
213
214bool HistoryURLProviderTest::SetUpImpl(bool no_db) {
215  profile_.reset(new TestingProfile());
216  if (!(profile_->CreateHistoryService(true, no_db)))
217    return false;
218  if (!no_db) {
219    profile_->BlockUntilHistoryProcessesPendingRequests();
220    profile_->BlockUntilHistoryIndexIsRefreshed();
221  }
222  profile_->GetPrefs()->SetString(prefs::kAcceptLanguages, "en-US,en,ko");
223  history_service_ = HistoryServiceFactory::GetForProfile(
224      profile_.get(), Profile::EXPLICIT_ACCESS);
225
226  autocomplete_ = new HistoryURLProvider(this, profile_.get());
227  TemplateURLServiceFactory::GetInstance()->SetTestingFactoryAndUse(
228      profile_.get(), &HistoryURLProviderTest::CreateTemplateURLService);
229  FillData();
230  return true;
231}
232
233void HistoryURLProviderTest::TearDown() {
234  autocomplete_ = NULL;
235}
236
237void HistoryURLProviderTest::FillData() {
238  // All visits are a long time ago (some tests require this since we do some
239  // special logic for things visited very recently). Note that this time must
240  // be more recent than the "archived history" threshold for the data to go
241  // into the main database.
242  //
243  // TODO(brettw) It would be nice if we could test this behavior, in which
244  // case the time would be specifed in the test_db structure.
245  Time visit_time = Time::Now() - TimeDelta::FromDays(80);
246
247  for (size_t i = 0; i < arraysize(test_db); ++i) {
248    const TestURLInfo& cur = test_db[i];
249    const GURL current_url(cur.url);
250    history_service_->AddPageWithDetails(current_url, UTF8ToUTF16(cur.title),
251                                         cur.visit_count, cur.typed_count,
252                                         visit_time, false,
253                                         history::SOURCE_BROWSED);
254  }
255
256  history_service_->AddPageWithDetails(
257      GURL("http://p/"), UTF8ToUTF16("p"), 0, 0,
258      Time::Now() -
259      TimeDelta::FromDays(history::kLowQualityMatchAgeLimitInDays - 1),
260      false, history::SOURCE_BROWSED);
261}
262
263void HistoryURLProviderTest::RunTest(
264    const string16 text,
265    const string16& desired_tld,
266    bool prevent_inline_autocomplete,
267    const UrlAndLegalDefault* expected_urls,
268    size_t num_results,
269    AutocompleteInput::Type* identified_input_type) {
270  AutocompleteInput input(text, string16::npos, desired_tld, GURL(),
271                          AutocompleteInput::INVALID_SPEC,
272                          prevent_inline_autocomplete, false, true,
273                          AutocompleteInput::ALL_MATCHES);
274  *identified_input_type = input.type();
275  autocomplete_->Start(input, false);
276  if (!autocomplete_->done())
277    base::MessageLoop::current()->Run();
278
279  matches_ = autocomplete_->matches();
280  if (sort_matches_) {
281    for (ACMatches::iterator i = matches_.begin(); i != matches_.end(); ++i)
282      i->ComputeStrippedDestinationURL(profile_.get());
283    std::sort(matches_.begin(), matches_.end(),
284              &AutocompleteMatch::DestinationSortFunc);
285    matches_.erase(std::unique(matches_.begin(), matches_.end(),
286                               &AutocompleteMatch::DestinationsEqual),
287                   matches_.end());
288    std::sort(matches_.begin(), matches_.end(),
289              &AutocompleteMatch::MoreRelevant);
290  }
291  ASSERT_EQ(num_results, matches_.size()) << "Input text: " << text
292                                          << "\nTLD: \"" << desired_tld << "\"";
293  for (size_t i = 0; i < num_results; ++i) {
294    EXPECT_EQ(expected_urls[i].url, matches_[i].destination_url.spec());
295    EXPECT_EQ(expected_urls[i].allowed_to_be_default_match,
296              matches_[i].allowed_to_be_default_match);
297  }
298}
299
300TEST_F(HistoryURLProviderTest, PromoteShorterURLs) {
301  // Test that hosts get synthesized below popular pages.
302  const UrlAndLegalDefault expected_nonsynth[] = {
303    { "http://slashdot.org/favorite_page.html", false },
304    { "http://slashdot.org/", false }
305  };
306  RunTest(ASCIIToUTF16("slash"), string16(), true, expected_nonsynth,
307          arraysize(expected_nonsynth));
308
309  // Test that hosts get synthesized above less popular pages.
310  const UrlAndLegalDefault expected_synth[] = {
311    { "http://kerneltrap.org/", false },
312    { "http://kerneltrap.org/not_very_popular.html", false }
313  };
314  RunTest(ASCIIToUTF16("kernel"), string16(), true, expected_synth,
315          arraysize(expected_synth));
316
317  // Test that unpopular pages are ignored completely.
318  RunTest(ASCIIToUTF16("fresh"), string16(), true, NULL, 0);
319
320  // Test that if we create or promote shorter suggestions that would not
321  // normally be inline autocompletable, we make them inline autocompletable if
322  // the original suggestion (that we replaced as "top") was inline
323  // autocompletable.
324  const UrlAndLegalDefault expected_synthesisa[] = {
325    { "http://synthesisatest.com/", true },
326    { "http://synthesisatest.com/foo/", true }
327  };
328  RunTest(ASCIIToUTF16("synthesisa"), string16(), false, expected_synthesisa,
329          arraysize(expected_synthesisa));
330  EXPECT_LT(matches_.front().relevance, 1200);
331  const UrlAndLegalDefault expected_synthesisb[] = {
332    { "http://synthesisbtest.com/foo/", true },
333    { "http://synthesisbtest.com/foo/bar.html", true }
334  };
335  RunTest(ASCIIToUTF16("synthesisb"), string16(), false, expected_synthesisb,
336          arraysize(expected_synthesisb));
337  EXPECT_GE(matches_.front().relevance, 1410);
338
339  // Test that if we have a synthesized host that matches a suggestion, they
340  // get combined into one.
341  const UrlAndLegalDefault expected_combine[] = {
342    { "http://news.google.com/", false },
343    { "http://news.google.com/?ned=us&topic=n", false },
344  };
345  ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("news"), string16(), true,
346      expected_combine, arraysize(expected_combine)));
347  // The title should also have gotten set properly on the host for the
348  // synthesized one, since it was also in the results.
349  EXPECT_EQ(ASCIIToUTF16("Google News"), matches_.front().description);
350
351  // Test that short URL matching works correctly as the user types more
352  // (several tests):
353  // The entry for foo.com is the best of all five foo.com* entries.
354  const UrlAndLegalDefault short_1[] = {
355    { "http://foo.com/", false },
356    { "http://foo.com/dir/another/again/myfile.html", false },
357    { "http://foo.com/dir/", false }
358  };
359  RunTest(ASCIIToUTF16("foo"), string16(), true, short_1, arraysize(short_1));
360
361  // When the user types the whole host, make sure we don't get two results for
362  // it.
363  const UrlAndLegalDefault short_2[] = {
364    { "http://foo.com/", true },
365    { "http://foo.com/dir/another/again/myfile.html", false },
366    { "http://foo.com/dir/", false },
367    { "http://foo.com/dir/another/", false }
368  };
369  RunTest(ASCIIToUTF16("foo.com"), string16(), true, short_2,
370          arraysize(short_2));
371  RunTest(ASCIIToUTF16("foo.com/"), string16(), true, short_2,
372          arraysize(short_2));
373
374  // The filename is the second best of the foo.com* entries, but there is a
375  // shorter URL that's "good enough".  The host doesn't match the user input
376  // and so should not appear.
377  const UrlAndLegalDefault short_3[] = {
378    { "http://foo.com/d", true },
379    { "http://foo.com/dir/another/", false },
380    { "http://foo.com/dir/another/again/myfile.html", false },
381    { "http://foo.com/dir/", false }
382  };
383  RunTest(ASCIIToUTF16("foo.com/d"), string16(), true, short_3,
384          arraysize(short_3));
385
386  // We shouldn't promote shorter URLs than the best if they're not good
387  // enough.
388  const UrlAndLegalDefault short_4[] = {
389    { "http://foo.com/dir/another/a", true },
390    { "http://foo.com/dir/another/again/myfile.html", false },
391    { "http://foo.com/dir/another/again/", false }
392  };
393  RunTest(ASCIIToUTF16("foo.com/dir/another/a"), string16(), true, short_4,
394          arraysize(short_4));
395
396  // Exact matches should always be best no matter how much more another match
397  // has been typed.
398  const UrlAndLegalDefault short_5a[] = {
399    { "http://gooey/", true },
400    { "http://www.google.com/", true },
401    { "http://go/", true }
402  };
403  const UrlAndLegalDefault short_5b[] = {
404    { "http://go/", true },
405    { "http://gooey/", true },
406    { "http://www.google.com/", true }
407  };
408  RunTest(ASCIIToUTF16("g"), string16(), false, short_5a, arraysize(short_5a));
409  RunTest(ASCIIToUTF16("go"), string16(), false, short_5b, arraysize(short_5b));
410}
411
412TEST_F(HistoryURLProviderTest, CullRedirects) {
413  // URLs we will be using, plus the visit counts they will initially get
414  // (the redirect set below will also increment the visit counts). We want
415  // the results to be in A,B,C order. Note also that our visit counts are
416  // all high enough so that domain synthesizing won't get triggered.
417  struct TestCase {
418    const char* url;
419    int count;
420  } test_cases[] = {
421    {"http://redirects/A", 30},
422    {"http://redirects/B", 20},
423    {"http://redirects/C", 10}
424  };
425  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(test_cases); ++i) {
426    history_service_->AddPageWithDetails(GURL(test_cases[i].url),
427        UTF8ToUTF16("Title"), test_cases[i].count, test_cases[i].count,
428        Time::Now(), false, history::SOURCE_BROWSED);
429  }
430
431  // Create a B->C->A redirect chain, but set the visit counts such that they
432  // will appear in A,B,C order in the results. The autocomplete query will
433  // search for the most recent visit when looking for redirects, so this will
434  // be found even though the previous visits had no redirects.
435  history::RedirectList redirects_to_a;
436  redirects_to_a.push_back(GURL(test_cases[1].url));
437  redirects_to_a.push_back(GURL(test_cases[2].url));
438  redirects_to_a.push_back(GURL(test_cases[0].url));
439  history_service_->AddPage(GURL(test_cases[0].url), base::Time::Now(),
440      NULL, 0, GURL(), redirects_to_a, content::PAGE_TRANSITION_TYPED,
441      history::SOURCE_BROWSED, true);
442
443  // Because all the results are part of a redirect chain with other results,
444  // all but the first one (A) should be culled. We should get the default
445  // "what you typed" result, plus this one.
446  const string16 typing(ASCIIToUTF16("http://redirects/"));
447  const UrlAndLegalDefault expected_results[] = {
448    { UTF16ToUTF8(typing), true },
449    { test_cases[0].url, false }
450  };
451  RunTest(typing, string16(), true, expected_results,
452          arraysize(expected_results));
453}
454
455TEST_F(HistoryURLProviderTest, WhatYouTyped) {
456  // Make sure we suggest a What You Typed match at the right times.
457  RunTest(ASCIIToUTF16("wytmatch"), string16(), false, NULL, 0);
458  RunTest(ASCIIToUTF16("wytmatch foo bar"), string16(), false, NULL, 0);
459  RunTest(ASCIIToUTF16("wytmatch+foo+bar"), string16(), false, NULL, 0);
460  RunTest(ASCIIToUTF16("wytmatch+foo+bar.com"), string16(), false, NULL, 0);
461
462  const UrlAndLegalDefault results_1[] = {
463    { "http://www.wytmatch.com/", true }
464  };
465  RunTest(ASCIIToUTF16("wytmatch"), ASCIIToUTF16("com"), false, results_1,
466          arraysize(results_1));
467
468  const UrlAndLegalDefault results_2[] = {
469    { "http://wytmatch%20foo%20bar/", true }
470  };
471  RunTest(ASCIIToUTF16("http://wytmatch foo bar"), string16(), false, results_2,
472          arraysize(results_2));
473
474  const UrlAndLegalDefault results_3[] = {
475    { "https://wytmatch%20foo%20bar/", true }
476  };
477  RunTest(ASCIIToUTF16("https://wytmatch foo bar"), string16(), false,
478          results_3, arraysize(results_3));
479}
480
481TEST_F(HistoryURLProviderTest, Fixup) {
482  // Test for various past crashes we've had.
483  RunTest(ASCIIToUTF16("\\"), string16(), false, NULL, 0);
484  RunTest(ASCIIToUTF16("#"), string16(), false, NULL, 0);
485  RunTest(ASCIIToUTF16("%20"), string16(), false, NULL, 0);
486  const UrlAndLegalDefault fixup_crash[] = {
487    { "http://%EF%BD%A5@s/", true }
488  };
489  RunTest(WideToUTF16(L"\uff65@s"), string16(), false, fixup_crash,
490          arraysize(fixup_crash));
491  RunTest(WideToUTF16(L"\u2015\u2015@ \uff7c"), string16(), false, NULL, 0);
492
493  // Fixing up "file:" should result in an inline autocomplete offset of just
494  // after "file:", not just after "file://".
495  const string16 input_1(ASCIIToUTF16("file:"));
496  const UrlAndLegalDefault fixup_1[] = {
497    { "file:///C:/foo.txt", true }
498  };
499  ASSERT_NO_FATAL_FAILURE(RunTest(input_1, string16(), false, fixup_1,
500                                  arraysize(fixup_1)));
501  EXPECT_EQ(ASCIIToUTF16("///C:/foo.txt"),
502            matches_.front().inline_autocompletion);
503
504  // Fixing up "http:/" should result in an inline autocomplete offset of just
505  // after "http:/", not just after "http:".
506  const string16 input_2(ASCIIToUTF16("http:/"));
507  const UrlAndLegalDefault fixup_2[] = {
508    { "http://bogussite.com/a", true },
509    { "http://bogussite.com/b", true },
510    { "http://bogussite.com/c", true }
511  };
512  ASSERT_NO_FATAL_FAILURE(RunTest(input_2, string16(), false, fixup_2,
513                                  arraysize(fixup_2)));
514  EXPECT_EQ(ASCIIToUTF16("/bogussite.com/a"),
515            matches_.front().inline_autocompletion);
516
517  // Adding a TLD to a small number like "56" should result in "www.56.com"
518  // rather than "0.0.0.56.com".
519  const UrlAndLegalDefault fixup_3[] = {
520    { "http://www.56.com/", true }
521  };
522  RunTest(ASCIIToUTF16("56"), ASCIIToUTF16("com"), true, fixup_3,
523          arraysize(fixup_3));
524
525  // An input looks like a IP address like "127.0.0.1" should result in
526  // "http://127.0.0.1/".
527  const UrlAndLegalDefault fixup_4[] = {
528    { "http://127.0.0.1/", true }
529  };
530  RunTest(ASCIIToUTF16("127.0.0.1"), string16(), false, fixup_4,
531          arraysize(fixup_4));
532
533  // An number "17173" should result in "http://www.17173.com/" in db.
534  const UrlAndLegalDefault fixup_5[] = {
535    { "http://www.17173.com/", true }
536  };
537  RunTest(ASCIIToUTF16("17173"), string16(), false, fixup_5,
538          arraysize(fixup_5));
539}
540
541// Make sure the results for the input 'p' don't change between the first and
542// second passes.
543TEST_F(HistoryURLProviderTest, EmptyVisits) {
544  // Wait for history to create the in memory DB.
545  profile_->BlockUntilHistoryProcessesPendingRequests();
546
547  AutocompleteInput input(ASCIIToUTF16("p"), string16::npos, string16(), GURL(),
548                          AutocompleteInput::INVALID_SPEC, false, false, true,
549                          AutocompleteInput::ALL_MATCHES);
550  autocomplete_->Start(input, false);
551  // HistoryURLProvider shouldn't be done (waiting on async results).
552  EXPECT_FALSE(autocomplete_->done());
553
554  // We should get back an entry for pandora.
555  matches_ = autocomplete_->matches();
556  ASSERT_GT(matches_.size(), 0u);
557  EXPECT_EQ(GURL("http://pandora.com/"), matches_[0].destination_url);
558  int pandora_relevance = matches_[0].relevance;
559
560  // Run the message loop. When |autocomplete_| finishes the loop is quit.
561  base::MessageLoop::current()->Run();
562  EXPECT_TRUE(autocomplete_->done());
563  matches_ = autocomplete_->matches();
564  ASSERT_GT(matches_.size(), 0u);
565  EXPECT_EQ(GURL("http://pandora.com/"), matches_[0].destination_url);
566  EXPECT_EQ(pandora_relevance, matches_[0].relevance);
567}
568
569TEST_F(HistoryURLProviderTestNoDB, NavigateWithoutDB) {
570  // Ensure that we will still produce matches for navigation when there is no
571  // database.
572  UrlAndLegalDefault navigation_1[] = {
573    { "http://test.com/", true }
574  };
575  RunTest(ASCIIToUTF16("test.com"), string16(), false, navigation_1,
576          arraysize(navigation_1));
577
578  UrlAndLegalDefault navigation_2[] = {
579    { "http://slash/", true }
580  };
581  RunTest(ASCIIToUTF16("slash"), string16(), false, navigation_2,
582          arraysize(navigation_2));
583
584  RunTest(ASCIIToUTF16("this is a query"), string16(), false, NULL, 0);
585}
586
587TEST_F(HistoryURLProviderTest, DontAutocompleteOnTrailingWhitespace) {
588  AutocompleteInput input(ASCIIToUTF16("slash "), string16::npos, string16(),
589                          GURL(), AutocompleteInput::INVALID_SPEC, false, false,
590                          true, AutocompleteInput::ALL_MATCHES);
591  autocomplete_->Start(input, false);
592  if (!autocomplete_->done())
593    base::MessageLoop::current()->Run();
594
595  // None of the matches should attempt to autocomplete.
596  matches_ = autocomplete_->matches();
597  for (size_t i = 0; i < matches_.size(); ++i) {
598    EXPECT_TRUE(matches_[i].inline_autocompletion.empty());
599    EXPECT_FALSE(matches_[i].allowed_to_be_default_match);
600  }
601}
602
603TEST_F(HistoryURLProviderTest, TreatEmailsAsSearches) {
604  // Visiting foo.com should not make this string be treated as a navigation.
605  // That means the result should be scored around 1200 ("what you typed")
606  // and not 1400+.
607  const UrlAndLegalDefault expected[] = {
608    { "http://user@foo.com/", true }
609  };
610  ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("user@foo.com"), string16(),
611                                  false, expected, arraysize(expected)));
612  EXPECT_LE(1200, matches_[0].relevance);
613  EXPECT_LT(matches_[0].relevance, 1210);
614}
615
616TEST_F(HistoryURLProviderTest, IntranetURLsWithPaths) {
617  struct TestCase {
618    const char* input;
619    int relevance;
620  } test_cases[] = {
621    { "fooey", 0 },
622    { "fooey/", 1200 },     // 1200 for URL would still navigate by default.
623    { "fooey/a", 1200 },    // 1200 for UNKNOWN would not.
624    { "fooey/a b", 1200 },  // Also UNKNOWN.
625    { "gooey", 1410 },
626    { "gooey/", 1410 },
627    { "gooey/a", 1400 },
628    { "gooey/a b", 1400 },
629  };
630  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(test_cases); ++i) {
631    SCOPED_TRACE(test_cases[i].input);
632    if (test_cases[i].relevance == 0) {
633      RunTest(ASCIIToUTF16(test_cases[i].input), string16(), false, NULL, 0);
634    } else {
635      const UrlAndLegalDefault output[] = {
636        { URLFixerUpper::FixupURL(test_cases[i].input, std::string()).spec(),
637          true }
638      };
639      ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16(test_cases[i].input),
640                              string16(), false, output, arraysize(output)));
641      // Actual relevance should be at least what test_cases expects and
642      // and no more than 10 more.
643      EXPECT_LE(test_cases[i].relevance, matches_[0].relevance);
644      EXPECT_LT(matches_[0].relevance, test_cases[i].relevance + 10);
645    }
646  }
647}
648
649TEST_F(HistoryURLProviderTest, IntranetURLsWithRefs) {
650  struct TestCase {
651    const char* input;
652    int relevance;
653    AutocompleteInput::Type type;
654  } test_cases[] = {
655    { "gooey", 1410, AutocompleteInput::UNKNOWN },
656    { "gooey/", 1410, AutocompleteInput::URL },
657    { "gooey#", 1200, AutocompleteInput::UNKNOWN },
658    { "gooey/#", 1200, AutocompleteInput::URL },
659    { "gooey#foo", 1200, AutocompleteInput::UNKNOWN },
660    { "gooey/#foo", 1200, AutocompleteInput::URL },
661    { "gooey# foo", 1200, AutocompleteInput::UNKNOWN },
662    { "gooey/# foo", 1200, AutocompleteInput::URL },
663  };
664  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(test_cases); ++i) {
665    SCOPED_TRACE(test_cases[i].input);
666    const UrlAndLegalDefault output[] = {
667      { URLFixerUpper::FixupURL(test_cases[i].input, std::string()).spec(),
668        true }
669    };
670    AutocompleteInput::Type type;
671    ASSERT_NO_FATAL_FAILURE(
672        RunTest(ASCIIToUTF16(test_cases[i].input),
673                string16(), false, output, arraysize(output), &type));
674    // Actual relevance should be at least what test_cases expects and
675    // and no more than 10 more.
676    EXPECT_LE(test_cases[i].relevance, matches_[0].relevance);
677    EXPECT_LT(matches_[0].relevance, test_cases[i].relevance + 10);
678    // Input type should be what we expect.  This is important because
679    // this provider counts on SearchProvider to give queries a relevance
680    // score >1200 for UNKNOWN inputs and <1200 for URL inputs.  (That's
681    // already tested in search_provider_unittest.cc.)  For this test
682    // here to test that the user sees the correct behavior, it needs
683    // to check that the input type was identified correctly.
684    EXPECT_EQ(test_cases[i].type, type);
685  }
686}
687
688// Makes sure autocompletion happens for intranet sites that have been
689// previoulsy visited.
690TEST_F(HistoryURLProviderTest, IntranetURLCompletion) {
691  sort_matches_ = true;
692
693  const UrlAndLegalDefault expected1[] = {
694    { "http://intra/three", true },
695    { "http://intra/two", true }
696  };
697  ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("intra/t"), string16(), false,
698                                  expected1, arraysize(expected1)));
699  EXPECT_LE(1410, matches_[0].relevance);
700  EXPECT_LT(matches_[0].relevance, 1420);
701  EXPECT_EQ(matches_[0].relevance - 1, matches_[1].relevance);
702
703  const UrlAndLegalDefault expected2[] = {
704    { "http://moo/b", true },
705    { "http://moo/bar", true }
706  };
707  ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("moo/b"), string16(), false,
708                                  expected2, arraysize(expected2)));
709  // The url what you typed match should be around 1400, otherwise the
710  // search what you typed match is going to be first.
711  EXPECT_LE(1400, matches_[0].relevance);
712  EXPECT_LT(matches_[0].relevance, 1410);
713
714  const UrlAndLegalDefault expected3[] = {
715    { "http://intra/one", true },
716    { "http://intra/three", true },
717    { "http://intra/two", true }
718  };
719  RunTest(ASCIIToUTF16("intra"), string16(), false, expected3,
720          arraysize(expected3));
721
722  const UrlAndLegalDefault expected4[] = {
723    { "http://intra/one", true },
724    { "http://intra/three", true },
725    { "http://intra/two", true }
726  };
727  RunTest(ASCIIToUTF16("intra/"), string16(), false, expected4,
728          arraysize(expected4));
729
730  const UrlAndLegalDefault expected5[] = {
731    { "http://intra/one", true }
732  };
733  ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("intra/o"), string16(), false,
734                                  expected5, arraysize(expected5)));
735  EXPECT_LE(1410, matches_[0].relevance);
736  EXPECT_LT(matches_[0].relevance, 1420);
737
738  const UrlAndLegalDefault expected6[] = {
739    { "http://intra/x", true }
740  };
741  ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("intra/x"), string16(), false,
742                                  expected6, arraysize(expected6)));
743  EXPECT_LE(1400, matches_[0].relevance);
744  EXPECT_LT(matches_[0].relevance, 1410);
745
746  const UrlAndLegalDefault expected7[] = {
747    { "http://typedhost/untypedpath", true }
748  };
749  ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("typedhost/untypedpath"),
750      string16(), false, expected7, arraysize(expected7)));
751  EXPECT_LE(1400, matches_[0].relevance);
752  EXPECT_LT(matches_[0].relevance, 1410);
753}
754
755TEST_F(HistoryURLProviderTest, CrashDueToFixup) {
756  // This test passes if we don't crash.  The results don't matter.
757  const char* const test_cases[] = {
758    "//c",
759    "\\@st",
760    "view-source:x",
761  };
762  for (size_t i = 0; i < arraysize(test_cases); ++i) {
763    AutocompleteInput input(ASCIIToUTF16(test_cases[i]), string16::npos,
764                            string16(), GURL(), AutocompleteInput::INVALID_SPEC,
765                            false, false, true, AutocompleteInput::ALL_MATCHES);
766    autocomplete_->Start(input, false);
767    if (!autocomplete_->done())
768      base::MessageLoop::current()->Run();
769  }
770}
771
772TEST_F(HistoryURLProviderTest, CullSearchResults) {
773  // Set up a default search engine.
774  TemplateURLData data;
775  data.SetKeyword(ASCIIToUTF16("TestEngine"));
776  data.SetURL("http://testsearch.com/?q={searchTerms}");
777  TemplateURLService* template_url_service =
778      TemplateURLServiceFactory::GetForProfile(profile_.get());
779  TemplateURL* template_url = new TemplateURL(profile_.get(), data);
780  template_url_service->Add(template_url);
781  template_url_service->SetDefaultSearchProvider(template_url);
782  template_url_service->Load();
783
784  // URLs we will be using, plus the visit counts they will initially get
785  // (the redirect set below will also increment the visit counts). We want
786  // the results to be in A,B,C order. Note also that our visit counts are
787  // all high enough so that domain synthesizing won't get triggered.
788  struct TestCase {
789    const char* url;
790    int count;
791  } test_cases[] = {
792    {"https://testsearch.com/", 30},
793    {"https://testsearch.com/?q=foobar", 20},
794    {"http://foobar.com/", 10}
795  };
796  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(test_cases); ++i) {
797    history_service_->AddPageWithDetails(GURL(test_cases[i].url),
798        UTF8ToUTF16("Title"), test_cases[i].count, test_cases[i].count,
799        Time::Now(), false, history::SOURCE_BROWSED);
800  }
801
802  // We should not see search URLs when typing a previously used query.
803  const UrlAndLegalDefault expected_when_searching_query[] = {
804    { test_cases[2].url, false }
805  };
806  RunTest(ASCIIToUTF16("foobar"), string16(), true,
807      expected_when_searching_query, arraysize(expected_when_searching_query));
808
809  // We should not see search URLs when typing the search engine name.
810  const UrlAndLegalDefault expected_when_searching_site[] = {
811    { test_cases[0].url, false }
812  };
813  RunTest(ASCIIToUTF16("testsearch"), string16(), true,
814      expected_when_searching_site, arraysize(expected_when_searching_site));
815}
816
817TEST_F(HistoryURLProviderTest, SuggestExactInput) {
818  const size_t npos = std::string::npos;
819  struct TestCase {
820    // Inputs:
821    const char* input;
822    bool trim_http;
823    // Expected Outputs:
824    const char* contents;
825    // Offsets of the ACMatchClassifications, terminated by npos.
826    size_t offsets[3];
827    // The index of the ACMatchClassification that should have the MATCH bit
828    // set, npos if no ACMatchClassification should have the MATCH bit set.
829    size_t match_classification_index;
830  } test_cases[] = {
831    { "http://www.somesite.com", false,
832      "http://www.somesite.com", {0, npos, npos}, 0 },
833    { "www.somesite.com", true,
834      "www.somesite.com", {0, npos, npos}, 0 },
835    { "www.somesite.com", false,
836      "http://www.somesite.com", {0, 7, npos}, 1 },
837    { "somesite.com", true,
838      "somesite.com", {0, npos, npos}, 0 },
839    { "somesite.com", false,
840      "http://somesite.com", {0, 7, npos}, 1 },
841    { "w", true,
842      "w", {0, npos, npos}, 0 },
843    { "w", false,
844      "http://w", {0, 7, npos}, 1 },
845    { "w.com", true,
846      "w.com", {0, npos, npos}, 0 },
847    { "w.com", false,
848      "http://w.com", {0, 7, npos}, 1 },
849    { "www.w.com", true,
850      "www.w.com", {0, npos, npos}, 0 },
851    { "www.w.com", false,
852      "http://www.w.com", {0, 7, npos}, 1 },
853    { "view-source:w", true,
854      "view-source:w", {0, npos, npos}, 0 },
855    { "view-source:www.w.com/", true,
856      "view-source:www.w.com", {0, npos, npos}, npos },
857    { "view-source:www.w.com/", false,
858      "view-source:http://www.w.com", {0, npos, npos}, npos },
859    { "view-source:http://www.w.com/", false,
860      "view-source:http://www.w.com", {0, npos, npos}, 0 },
861    { "   view-source:", true,
862      "view-source:", {0, npos, npos}, 0 },
863    { "http:////////w.com", false,
864      "http://w.com", {0, npos, npos}, npos },
865    { "    http:////////www.w.com", false,
866      "http://www.w.com", {0, npos, npos}, npos },
867    { "http:a///www.w.com", false,
868      "http://a///www.w.com", {0, npos, npos}, npos },
869    { "mailto://a@b.com", true,
870      "mailto://a@b.com", {0, npos, npos}, 0 },
871    { "mailto://a@b.com", false,
872      "mailto://a@b.com", {0, npos, npos}, 0 },
873  };
874  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(test_cases); ++i) {
875    SCOPED_TRACE(testing::Message() << "Index " << i << " input: "
876                                    << test_cases[i].input << ", trim_http: "
877                                    << test_cases[i].trim_http);
878
879    AutocompleteInput input(ASCIIToUTF16(test_cases[i].input), string16::npos,
880                            string16(), GURL("about:blank"),
881                            AutocompleteInput::INVALID_SPEC, false, false, true,
882                            AutocompleteInput::ALL_MATCHES);
883    AutocompleteMatch match = HistoryURLProvider::SuggestExactInput(
884        autocomplete_.get(), input, test_cases[i].trim_http);
885    EXPECT_EQ(ASCIIToUTF16(test_cases[i].contents), match.contents);
886    for (size_t match_index = 0; match_index < match.contents_class.size();
887         ++match_index) {
888      EXPECT_EQ(test_cases[i].offsets[match_index],
889                match.contents_class[match_index].offset);
890      EXPECT_EQ(ACMatchClassification::URL |
891                (match_index == test_cases[i].match_classification_index ?
892                 ACMatchClassification::MATCH : 0),
893                match.contents_class[match_index].style);
894    }
895    EXPECT_EQ(npos, test_cases[i].offsets[match.contents_class.size()]);
896  }
897}
898