history_url_provider_unittest.cc revision 5d1f7b1de12d16ceb2c938c56701a3e8bfa558f7 
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::ASCIIToUTF16;
32using base::Time;
33using base::TimeDelta;
34
35using content::TestBrowserThreadBundle;
36
37struct TestURLInfo {
38  const char* url;
39  const char* title;
40  int visit_count;
41  int typed_count;
42  int age_in_days;
43} test_db[] = {
44  {"http://www.google.com/", "Google", 3, 3, 80},
45
46  // High-quality pages should get a host synthesized as a lower-quality match.
47  {"http://slashdot.org/favorite_page.html", "Favorite page", 200, 100, 80},
48
49  // Less popular pages should have hosts synthesized as higher-quality
50  // matches.
51  {"http://kerneltrap.org/not_very_popular.html", "Less popular", 4, 0, 80},
52
53  // Unpopular pages should not appear in the results at all.
54  {"http://freshmeat.net/unpopular.html", "Unpopular", 1, 0, 80},
55
56  // If a host has a match, we should pick it up during host synthesis.
57  {"http://news.google.com/?ned=us&topic=n", "Google News - U.S.", 2, 2, 80},
58  {"http://news.google.com/", "Google News", 1, 1, 80},
59
60  // Matches that are normally not inline-autocompletable should be
61  // autocompleted if they are shorter substitutes for longer matches that would
62  // have been inline autocompleted.
63  {"http://synthesisatest.com/foo/", "Test A", 1, 1, 80},
64  {"http://synthesisbtest.com/foo/", "Test B", 1, 1, 80},
65  {"http://synthesisbtest.com/foo/bar.html", "Test B Bar", 2, 2, 80},
66
67  // Suggested short URLs must be "good enough" and must match user input.
68  {"http://foo.com/", "Dir", 5, 5, 80},
69  {"http://foo.com/dir/", "Dir", 2, 2, 80},
70  {"http://foo.com/dir/another/", "Dir", 5, 1, 80},
71  {"http://foo.com/dir/another/again/", "Dir", 10, 0, 80},
72  {"http://foo.com/dir/another/again/myfile.html", "File", 10, 2, 80},
73
74  // We throw in a lot of extra URLs here to make sure we're testing the
75  // history database's query, not just the autocomplete provider.
76  {"http://startest.com/y/a", "A", 2, 2, 80},
77  {"http://startest.com/y/b", "B", 5, 2, 80},
78  {"http://startest.com/x/c", "C", 5, 2, 80},
79  {"http://startest.com/x/d", "D", 5, 5, 80},
80  {"http://startest.com/y/e", "E", 4, 2, 80},
81  {"http://startest.com/y/f", "F", 3, 2, 80},
82  {"http://startest.com/y/g", "G", 3, 2, 80},
83  {"http://startest.com/y/h", "H", 3, 2, 80},
84  {"http://startest.com/y/i", "I", 3, 2, 80},
85  {"http://startest.com/y/j", "J", 3, 2, 80},
86  {"http://startest.com/y/k", "K", 3, 2, 80},
87  {"http://startest.com/y/l", "L", 3, 2, 80},
88  {"http://startest.com/y/m", "M", 3, 2, 80},
89
90  // A file: URL is useful for testing that fixup does the right thing w.r.t.
91  // the number of trailing slashes on the user's input.
92  {"file:///C:/foo.txt", "", 2, 2, 80},
93
94  // Results with absurdly high typed_counts so that very generic queries like
95  // "http" will give consistent results even if more data is added above.
96  {"http://bogussite.com/a", "Bogus A", 10002, 10000, 80},
97  {"http://bogussite.com/b", "Bogus B", 10001, 10000, 80},
98  {"http://bogussite.com/c", "Bogus C", 10000, 10000, 80},
99
100  // Domain name with number.
101  {"http://www.17173.com/", "Domain with number", 3, 3, 80},
102
103  // URLs to test exact-matching behavior.
104  {"http://go/", "Intranet URL", 1, 1, 80},
105  {"http://gooey/", "Intranet URL 2", 5, 5, 80},
106
107  // URLs for testing offset adjustment.
108  {"http://www.\xEA\xB5\x90\xEC\x9C\xA1.kr/", "Korean", 2, 2, 80},
109  {"http://spaces.com/path%20with%20spaces/foo.html", "Spaces", 2, 2, 80},
110  {"http://ms/c++%20style%20guide", "Style guide", 2, 2, 80},
111
112  // URLs for testing ctrl-enter behavior.
113  {"http://binky/", "Intranet binky", 2, 2, 80},
114  {"http://winky/", "Intranet winky", 2, 2, 80},
115  {"http://www.winky.com/", "Internet winky", 5, 0, 80},
116
117  // URLs used by EmptyVisits.
118  {"http://pandora.com/", "Pandora", 2, 2, 80},
119  // This entry is explicitly added more recently than
120  // history::kLowQualityMatchAgeLimitInDays.
121  // {"http://p/", "p", 0, 0, 80},
122
123  // For intranet based tests.
124  {"http://intra/one", "Intranet", 2, 2, 80},
125  {"http://intra/two", "Intranet two", 1, 1, 80},
126  {"http://intra/three", "Intranet three", 2, 2, 80},
127  {"http://moo/bar", "Intranet moo", 1, 1, 80},
128  {"http://typedhost/typedpath", "Intranet typed", 1, 1, 80},
129  {"http://typedhost/untypedpath", "Intranet untyped", 1, 0, 80},
130
131  {"http://x.com/one", "Internet", 2, 2, 80},
132  {"http://x.com/two", "Internet two", 1, 1, 80},
133  {"http://x.com/three", "Internet three", 2, 2, 80},
134
135  // For experimental HUP scoring test.
136  {"http://7.com/1a", "One", 8, 4, 4},
137  {"http://7.com/2a", "Two A", 4, 2, 8},
138  {"http://7.com/2b", "Two B", 4, 1, 8},
139  {"http://7.com/3a", "Three", 2, 1, 16},
140  {"http://7.com/4a", "Four A", 1, 1, 32},
141  {"http://7.com/4b", "Four B", 1, 1, 64},
142  {"http://7.com/5a", "Five A", 8, 0, 64},  // never typed.
143};
144
145class HistoryURLProviderTest : public testing::Test,
146                               public AutocompleteProviderListener {
147 public:
148  struct UrlAndLegalDefault {
149    std::string url;
150    bool allowed_to_be_default_match;
151  };
152
153  HistoryURLProviderTest()
154      : sort_matches_(false) {
155    HistoryQuickProvider::set_disabled(true);
156  }
157
158  virtual ~HistoryURLProviderTest() {
159    HistoryQuickProvider::set_disabled(false);
160  }
161
162  // AutocompleteProviderListener:
163  virtual void OnProviderUpdate(bool updated_matches) OVERRIDE;
164
165 protected:
166  static BrowserContextKeyedService* CreateTemplateURLService(
167      content::BrowserContext* profile) {
168    return new TemplateURLService(static_cast<Profile*>(profile));
169  }
170
171  // testing::Test
172  virtual void SetUp() {
173    ASSERT_TRUE(SetUpImpl(false));
174  }
175  virtual void TearDown();
176
177  // Does the real setup.
178  bool SetUpImpl(bool no_db) WARN_UNUSED_RESULT;
179
180  // Fills test data into the history system.
181  void FillData();
182
183  // Runs an autocomplete query on |text| and checks to see that the returned
184  // results' destination URLs match those provided.  Also allows checking
185  // that the input type was identified correctly.
186  void RunTest(const base::string16 text,
187               const base::string16& desired_tld,
188               bool prevent_inline_autocomplete,
189               const UrlAndLegalDefault* expected_urls,
190               size_t num_results,
191               AutocompleteInput::Type* identified_input_type);
192
193  // A version of the above without the final |type| output parameter.
194  void RunTest(const base::string16 text,
195               const base::string16& desired_tld,
196               bool prevent_inline_autocomplete,
197               const UrlAndLegalDefault* expected_urls,
198               size_t num_results) {
199    AutocompleteInput::Type type;
200    return RunTest(text, desired_tld, prevent_inline_autocomplete,
201                   expected_urls, num_results, &type);
202  }
203
204  content::TestBrowserThreadBundle thread_bundle_;
205  ACMatches matches_;
206  scoped_ptr<TestingProfile> profile_;
207  HistoryService* history_service_;
208  scoped_refptr<HistoryURLProvider> autocomplete_;
209  // Should the matches be sorted and duplicates removed?
210  bool sort_matches_;
211};
212
213class HistoryURLProviderTestNoDB : public HistoryURLProviderTest {
214 protected:
215  virtual void SetUp() {
216    ASSERT_TRUE(SetUpImpl(true));
217  }
218};
219
220void HistoryURLProviderTest::OnProviderUpdate(bool updated_matches) {
221  if (autocomplete_->done())
222    base::MessageLoop::current()->Quit();
223}
224
225bool HistoryURLProviderTest::SetUpImpl(bool no_db) {
226  profile_.reset(new TestingProfile());
227  if (!(profile_->CreateHistoryService(true, no_db)))
228    return false;
229  if (!no_db) {
230    profile_->BlockUntilHistoryProcessesPendingRequests();
231    profile_->BlockUntilHistoryIndexIsRefreshed();
232  }
233  profile_->GetPrefs()->SetString(prefs::kAcceptLanguages, "en-US,en,ko");
234  history_service_ = HistoryServiceFactory::GetForProfile(
235      profile_.get(), Profile::EXPLICIT_ACCESS);
236
237  autocomplete_ = new HistoryURLProvider(this, profile_.get());
238  TemplateURLServiceFactory::GetInstance()->SetTestingFactoryAndUse(
239      profile_.get(), &HistoryURLProviderTest::CreateTemplateURLService);
240  FillData();
241  return true;
242}
243
244void HistoryURLProviderTest::TearDown() {
245  autocomplete_ = NULL;
246}
247
248void HistoryURLProviderTest::FillData() {
249  // Most visits are a long time ago (some tests require this since we do some
250  // special logic for things visited very recently). Note that this time must
251  // be more recent than the "archived history" threshold for the data to go
252  // into the main database.
253  //
254  // TODO(brettw) It would be nice if we could test this behavior, in which
255  // case the time would be specifed in the test_db structure.
256  const Time now = Time::Now();
257
258  for (size_t i = 0; i < arraysize(test_db); ++i) {
259    const TestURLInfo& cur = test_db[i];
260    const GURL current_url(cur.url);
261    history_service_->AddPageWithDetails(
262        current_url, base::UTF8ToUTF16(cur.title), cur.visit_count,
263        cur.typed_count, now - TimeDelta::FromDays(cur.age_in_days), false,
264        history::SOURCE_BROWSED);
265  }
266
267  history_service_->AddPageWithDetails(
268      GURL("http://p/"), base::UTF8ToUTF16("p"), 0, 0,
269      Time::Now() -
270      TimeDelta::FromDays(history::kLowQualityMatchAgeLimitInDays - 1),
271      false, history::SOURCE_BROWSED);
272}
273
274void HistoryURLProviderTest::RunTest(
275    const base::string16 text,
276    const base::string16& desired_tld,
277    bool prevent_inline_autocomplete,
278    const UrlAndLegalDefault* expected_urls,
279    size_t num_results,
280    AutocompleteInput::Type* identified_input_type) {
281  AutocompleteInput input(text, base::string16::npos, desired_tld, GURL(),
282                          AutocompleteInput::INVALID_SPEC,
283                          prevent_inline_autocomplete, false, true,
284                          AutocompleteInput::ALL_MATCHES);
285  *identified_input_type = input.type();
286  autocomplete_->Start(input, false);
287  if (!autocomplete_->done())
288    base::MessageLoop::current()->Run();
289
290  matches_ = autocomplete_->matches();
291  if (sort_matches_) {
292    for (ACMatches::iterator i = matches_.begin(); i != matches_.end(); ++i)
293      i->ComputeStrippedDestinationURL(profile_.get());
294    std::sort(matches_.begin(), matches_.end(),
295              &AutocompleteMatch::DestinationSortFunc);
296    matches_.erase(std::unique(matches_.begin(), matches_.end(),
297                               &AutocompleteMatch::DestinationsEqual),
298                   matches_.end());
299    std::sort(matches_.begin(), matches_.end(),
300              &AutocompleteMatch::MoreRelevant);
301  }
302  ASSERT_EQ(num_results, matches_.size()) << "Input text: " << text
303                                          << "\nTLD: \"" << desired_tld << "\"";
304  for (size_t i = 0; i < num_results; ++i) {
305    EXPECT_EQ(expected_urls[i].url, matches_[i].destination_url.spec());
306    EXPECT_EQ(expected_urls[i].allowed_to_be_default_match,
307              matches_[i].allowed_to_be_default_match);
308  }
309}
310
311TEST_F(HistoryURLProviderTest, PromoteShorterURLs) {
312  // Test that hosts get synthesized below popular pages.
313  const UrlAndLegalDefault expected_nonsynth[] = {
314    { "http://slashdot.org/favorite_page.html", false },
315    { "http://slashdot.org/", false }
316  };
317  RunTest(ASCIIToUTF16("slash"), base::string16(), true, expected_nonsynth,
318          arraysize(expected_nonsynth));
319
320  // Test that hosts get synthesized above less popular pages.
321  const UrlAndLegalDefault expected_synth[] = {
322    { "http://kerneltrap.org/", false },
323    { "http://kerneltrap.org/not_very_popular.html", false }
324  };
325  RunTest(ASCIIToUTF16("kernel"), base::string16(), true, expected_synth,
326          arraysize(expected_synth));
327
328  // Test that unpopular pages are ignored completely.
329  RunTest(ASCIIToUTF16("fresh"), base::string16(), true, NULL, 0);
330
331  // Test that if we create or promote shorter suggestions that would not
332  // normally be inline autocompletable, we make them inline autocompletable if
333  // the original suggestion (that we replaced as "top") was inline
334  // autocompletable.
335  const UrlAndLegalDefault expected_synthesisa[] = {
336    { "http://synthesisatest.com/", true },
337    { "http://synthesisatest.com/foo/", true }
338  };
339  RunTest(ASCIIToUTF16("synthesisa"), base::string16(), false,
340          expected_synthesisa, arraysize(expected_synthesisa));
341  EXPECT_LT(matches_.front().relevance, 1200);
342  const UrlAndLegalDefault expected_synthesisb[] = {
343    { "http://synthesisbtest.com/foo/", true },
344    { "http://synthesisbtest.com/foo/bar.html", true }
345  };
346  RunTest(ASCIIToUTF16("synthesisb"), base::string16(), false,
347          expected_synthesisb, arraysize(expected_synthesisb));
348  EXPECT_GE(matches_.front().relevance, 1410);
349
350  // Test that if we have a synthesized host that matches a suggestion, they
351  // get combined into one.
352  const UrlAndLegalDefault expected_combine[] = {
353    { "http://news.google.com/", false },
354    { "http://news.google.com/?ned=us&topic=n", false },
355  };
356  ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("news"), base::string16(), true,
357      expected_combine, arraysize(expected_combine)));
358  // The title should also have gotten set properly on the host for the
359  // synthesized one, since it was also in the results.
360  EXPECT_EQ(ASCIIToUTF16("Google News"), matches_.front().description);
361
362  // Test that short URL matching works correctly as the user types more
363  // (several tests):
364  // The entry for foo.com is the best of all five foo.com* entries.
365  const UrlAndLegalDefault short_1[] = {
366    { "http://foo.com/", false },
367    { "http://foo.com/dir/another/again/myfile.html", false },
368    { "http://foo.com/dir/", false }
369  };
370  RunTest(ASCIIToUTF16("foo"), base::string16(), true,
371          short_1, arraysize(short_1));
372
373  // When the user types the whole host, make sure we don't get two results for
374  // it.
375  const UrlAndLegalDefault short_2[] = {
376    { "http://foo.com/", true },
377    { "http://foo.com/dir/another/again/myfile.html", false },
378    { "http://foo.com/dir/", false },
379    { "http://foo.com/dir/another/", false }
380  };
381  RunTest(ASCIIToUTF16("foo.com"), base::string16(), true, short_2,
382          arraysize(short_2));
383  RunTest(ASCIIToUTF16("foo.com/"), base::string16(), true, short_2,
384          arraysize(short_2));
385
386  // The filename is the second best of the foo.com* entries, but there is a
387  // shorter URL that's "good enough".  The host doesn't match the user input
388  // and so should not appear.
389  const UrlAndLegalDefault short_3[] = {
390    { "http://foo.com/d", true },
391    { "http://foo.com/dir/another/", false },
392    { "http://foo.com/dir/another/again/myfile.html", false },
393    { "http://foo.com/dir/", false }
394  };
395  RunTest(ASCIIToUTF16("foo.com/d"), base::string16(), true, short_3,
396          arraysize(short_3));
397
398  // We shouldn't promote shorter URLs than the best if they're not good
399  // enough.
400  const UrlAndLegalDefault short_4[] = {
401    { "http://foo.com/dir/another/a", true },
402    { "http://foo.com/dir/another/again/myfile.html", false },
403    { "http://foo.com/dir/another/again/", false }
404  };
405  RunTest(ASCIIToUTF16("foo.com/dir/another/a"), base::string16(), true,
406          short_4, arraysize(short_4));
407
408  // Exact matches should always be best no matter how much more another match
409  // has been typed.
410  const UrlAndLegalDefault short_5a[] = {
411    { "http://gooey/", true },
412    { "http://www.google.com/", true },
413    { "http://go/", true }
414  };
415  const UrlAndLegalDefault short_5b[] = {
416    { "http://go/", true },
417    { "http://gooey/", true },
418    { "http://www.google.com/", true }
419  };
420  RunTest(ASCIIToUTF16("g"), base::string16(), false,
421          short_5a, arraysize(short_5a));
422  RunTest(ASCIIToUTF16("go"), base::string16(), false,
423          short_5b, arraysize(short_5b));
424}
425
426TEST_F(HistoryURLProviderTest, CullRedirects) {
427  // URLs we will be using, plus the visit counts they will initially get
428  // (the redirect set below will also increment the visit counts). We want
429  // the results to be in A,B,C order. Note also that our visit counts are
430  // all high enough so that domain synthesizing won't get triggered.
431  struct TestCase {
432    const char* url;
433    int count;
434  } test_cases[] = {
435    {"http://redirects/A", 30},
436    {"http://redirects/B", 20},
437    {"http://redirects/C", 10}
438  };
439  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(test_cases); ++i) {
440    history_service_->AddPageWithDetails(GURL(test_cases[i].url),
441        ASCIIToUTF16("Title"), test_cases[i].count, test_cases[i].count,
442        Time::Now(), false, history::SOURCE_BROWSED);
443  }
444
445  // Create a B->C->A redirect chain, but set the visit counts such that they
446  // will appear in A,B,C order in the results. The autocomplete query will
447  // search for the most recent visit when looking for redirects, so this will
448  // be found even though the previous visits had no redirects.
449  history::RedirectList redirects_to_a;
450  redirects_to_a.push_back(GURL(test_cases[1].url));
451  redirects_to_a.push_back(GURL(test_cases[2].url));
452  redirects_to_a.push_back(GURL(test_cases[0].url));
453  history_service_->AddPage(GURL(test_cases[0].url), base::Time::Now(),
454      NULL, 0, GURL(), redirects_to_a, content::PAGE_TRANSITION_TYPED,
455      history::SOURCE_BROWSED, true);
456
457  // Because all the results are part of a redirect chain with other results,
458  // all but the first one (A) should be culled. We should get the default
459  // "what you typed" result, plus this one.
460  const base::string16 typing(ASCIIToUTF16("http://redirects/"));
461  const UrlAndLegalDefault expected_results[] = {
462    { base::UTF16ToUTF8(typing), true },
463    { test_cases[0].url, false }
464  };
465  RunTest(typing, base::string16(), true, expected_results,
466          arraysize(expected_results));
467}
468
469TEST_F(HistoryURLProviderTest, WhatYouTyped) {
470  // Make sure we suggest a What You Typed match at the right times.
471  RunTest(ASCIIToUTF16("wytmatch"), base::string16(), false, NULL, 0);
472  RunTest(ASCIIToUTF16("wytmatch foo bar"), base::string16(), false, NULL, 0);
473  RunTest(ASCIIToUTF16("wytmatch+foo+bar"), base::string16(), false, NULL, 0);
474  RunTest(ASCIIToUTF16("wytmatch+foo+bar.com"), base::string16(), false,
475          NULL, 0);
476
477  const UrlAndLegalDefault results_1[] = {
478    { "http://www.wytmatch.com/", true }
479  };
480  RunTest(ASCIIToUTF16("wytmatch"), ASCIIToUTF16("com"), false, results_1,
481          arraysize(results_1));
482
483  const UrlAndLegalDefault results_2[] = {
484    { "http://wytmatch%20foo%20bar/", true }
485  };
486  RunTest(ASCIIToUTF16("http://wytmatch foo bar"), base::string16(), false,
487          results_2, arraysize(results_2));
488
489  const UrlAndLegalDefault results_3[] = {
490    { "https://wytmatch%20foo%20bar/", true }
491  };
492  RunTest(ASCIIToUTF16("https://wytmatch foo bar"), base::string16(), false,
493          results_3, arraysize(results_3));
494}
495
496TEST_F(HistoryURLProviderTest, Fixup) {
497  // Test for various past crashes we've had.
498  RunTest(ASCIIToUTF16("\\"), base::string16(), false, NULL, 0);
499  RunTest(ASCIIToUTF16("#"), base::string16(), false, NULL, 0);
500  RunTest(ASCIIToUTF16("%20"), base::string16(), false, NULL, 0);
501  const UrlAndLegalDefault fixup_crash[] = {
502    { "http://%EF%BD%A5@s/", true }
503  };
504  RunTest(base::WideToUTF16(L"\uff65@s"), base::string16(), false, fixup_crash,
505          arraysize(fixup_crash));
506  RunTest(base::WideToUTF16(L"\u2015\u2015@ \uff7c"), base::string16(), false,
507          NULL, 0);
508
509  // Fixing up "file:" should result in an inline autocomplete offset of just
510  // after "file:", not just after "file://".
511  const base::string16 input_1(ASCIIToUTF16("file:"));
512  const UrlAndLegalDefault fixup_1[] = {
513    { "file:///C:/foo.txt", true }
514  };
515  ASSERT_NO_FATAL_FAILURE(RunTest(input_1, base::string16(), false, fixup_1,
516                                  arraysize(fixup_1)));
517  EXPECT_EQ(ASCIIToUTF16("///C:/foo.txt"),
518            matches_.front().inline_autocompletion);
519
520  // Fixing up "http:/" should result in an inline autocomplete offset of just
521  // after "http:/", not just after "http:".
522  const base::string16 input_2(ASCIIToUTF16("http:/"));
523  const UrlAndLegalDefault fixup_2[] = {
524    { "http://bogussite.com/a", true },
525    { "http://bogussite.com/b", true },
526    { "http://bogussite.com/c", true }
527  };
528  ASSERT_NO_FATAL_FAILURE(RunTest(input_2, base::string16(), false, fixup_2,
529                                  arraysize(fixup_2)));
530  EXPECT_EQ(ASCIIToUTF16("/bogussite.com/a"),
531            matches_.front().inline_autocompletion);
532
533  // Adding a TLD to a small number like "56" should result in "www.56.com"
534  // rather than "0.0.0.56.com".
535  const UrlAndLegalDefault fixup_3[] = {
536    { "http://www.56.com/", true }
537  };
538  RunTest(ASCIIToUTF16("56"), ASCIIToUTF16("com"), true, fixup_3,
539          arraysize(fixup_3));
540
541  // An input looks like a IP address like "127.0.0.1" should result in
542  // "http://127.0.0.1/".
543  const UrlAndLegalDefault fixup_4[] = {
544    { "http://127.0.0.1/", true }
545  };
546  RunTest(ASCIIToUTF16("127.0.0.1"), base::string16(), false, fixup_4,
547          arraysize(fixup_4));
548
549  // An number "17173" should result in "http://www.17173.com/" in db.
550  const UrlAndLegalDefault fixup_5[] = {
551    { "http://www.17173.com/", true }
552  };
553  RunTest(ASCIIToUTF16("17173"), base::string16(), false, fixup_5,
554          arraysize(fixup_5));
555}
556
557// Make sure the results for the input 'p' don't change between the first and
558// second passes.
559TEST_F(HistoryURLProviderTest, EmptyVisits) {
560  // Wait for history to create the in memory DB.
561  profile_->BlockUntilHistoryProcessesPendingRequests();
562
563  AutocompleteInput input(ASCIIToUTF16("p"), base::string16::npos,
564                          base::string16(), GURL(),
565                          AutocompleteInput::INVALID_SPEC, false, false, true,
566                          AutocompleteInput::ALL_MATCHES);
567  autocomplete_->Start(input, false);
568  // HistoryURLProvider shouldn't be done (waiting on async results).
569  EXPECT_FALSE(autocomplete_->done());
570
571  // We should get back an entry for pandora.
572  matches_ = autocomplete_->matches();
573  ASSERT_GT(matches_.size(), 0u);
574  EXPECT_EQ(GURL("http://pandora.com/"), matches_[0].destination_url);
575  int pandora_relevance = matches_[0].relevance;
576
577  // Run the message loop. When |autocomplete_| finishes the loop is quit.
578  base::MessageLoop::current()->Run();
579  EXPECT_TRUE(autocomplete_->done());
580  matches_ = autocomplete_->matches();
581  ASSERT_GT(matches_.size(), 0u);
582  EXPECT_EQ(GURL("http://pandora.com/"), matches_[0].destination_url);
583  EXPECT_EQ(pandora_relevance, matches_[0].relevance);
584}
585
586TEST_F(HistoryURLProviderTestNoDB, NavigateWithoutDB) {
587  // Ensure that we will still produce matches for navigation when there is no
588  // database.
589  UrlAndLegalDefault navigation_1[] = {
590    { "http://test.com/", true }
591  };
592  RunTest(ASCIIToUTF16("test.com"), base::string16(), false, navigation_1,
593          arraysize(navigation_1));
594
595  UrlAndLegalDefault navigation_2[] = {
596    { "http://slash/", true }
597  };
598  RunTest(ASCIIToUTF16("slash"), base::string16(), false, navigation_2,
599          arraysize(navigation_2));
600
601  RunTest(ASCIIToUTF16("this is a query"), base::string16(), false, NULL, 0);
602}
603
604TEST_F(HistoryURLProviderTest, DontAutocompleteOnTrailingWhitespace) {
605  AutocompleteInput input(ASCIIToUTF16("slash "), base::string16::npos,
606                          base::string16(), GURL(),
607                          AutocompleteInput::INVALID_SPEC, false, false,
608                          true, AutocompleteInput::ALL_MATCHES);
609  autocomplete_->Start(input, false);
610  if (!autocomplete_->done())
611    base::MessageLoop::current()->Run();
612
613  // None of the matches should attempt to autocomplete.
614  matches_ = autocomplete_->matches();
615  for (size_t i = 0; i < matches_.size(); ++i) {
616    EXPECT_TRUE(matches_[i].inline_autocompletion.empty());
617    EXPECT_FALSE(matches_[i].allowed_to_be_default_match);
618  }
619}
620
621TEST_F(HistoryURLProviderTest, TreatEmailsAsSearches) {
622  // Visiting foo.com should not make this string be treated as a navigation.
623  // That means the result should be scored around 1200 ("what you typed")
624  // and not 1400+.
625  const UrlAndLegalDefault expected[] = {
626    { "http://user@foo.com/", true }
627  };
628  ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("user@foo.com"),
629                                  base::string16(), false, expected,
630                                  arraysize(expected)));
631  EXPECT_LE(1200, matches_[0].relevance);
632  EXPECT_LT(matches_[0].relevance, 1210);
633}
634
635TEST_F(HistoryURLProviderTest, IntranetURLsWithPaths) {
636  struct TestCase {
637    const char* input;
638    int relevance;
639  } test_cases[] = {
640    { "fooey", 0 },
641    { "fooey/", 1200 },     // 1200 for URL would still navigate by default.
642    { "fooey/a", 1200 },    // 1200 for UNKNOWN would not.
643    { "fooey/a b", 1200 },  // Also UNKNOWN.
644    { "gooey", 1410 },
645    { "gooey/", 1410 },
646    { "gooey/a", 1400 },
647    { "gooey/a b", 1400 },
648  };
649  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(test_cases); ++i) {
650    SCOPED_TRACE(test_cases[i].input);
651    if (test_cases[i].relevance == 0) {
652      RunTest(ASCIIToUTF16(test_cases[i].input), base::string16(), false,
653              NULL, 0);
654    } else {
655      const UrlAndLegalDefault output[] = {
656        { URLFixerUpper::FixupURL(test_cases[i].input, std::string()).spec(),
657          true }
658      };
659      ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16(test_cases[i].input),
660                              base::string16(), false,
661                              output, arraysize(output)));
662      // Actual relevance should be at least what test_cases expects and
663      // and no more than 10 more.
664      EXPECT_LE(test_cases[i].relevance, matches_[0].relevance);
665      EXPECT_LT(matches_[0].relevance, test_cases[i].relevance + 10);
666    }
667  }
668}
669
670TEST_F(HistoryURLProviderTest, IntranetURLsWithRefs) {
671  struct TestCase {
672    const char* input;
673    int relevance;
674    AutocompleteInput::Type type;
675  } test_cases[] = {
676    { "gooey", 1410, AutocompleteInput::UNKNOWN },
677    { "gooey/", 1410, AutocompleteInput::URL },
678    { "gooey#", 1200, AutocompleteInput::UNKNOWN },
679    { "gooey/#", 1200, AutocompleteInput::URL },
680    { "gooey#foo", 1200, AutocompleteInput::UNKNOWN },
681    { "gooey/#foo", 1200, AutocompleteInput::URL },
682    { "gooey# foo", 1200, AutocompleteInput::UNKNOWN },
683    { "gooey/# foo", 1200, AutocompleteInput::URL },
684  };
685  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(test_cases); ++i) {
686    SCOPED_TRACE(test_cases[i].input);
687    const UrlAndLegalDefault output[] = {
688      { URLFixerUpper::FixupURL(test_cases[i].input, std::string()).spec(),
689        true }
690    };
691    AutocompleteInput::Type type;
692    ASSERT_NO_FATAL_FAILURE(
693        RunTest(ASCIIToUTF16(test_cases[i].input),
694                base::string16(), false, output, arraysize(output), &type));
695    // Actual relevance should be at least what test_cases expects and
696    // and no more than 10 more.
697    EXPECT_LE(test_cases[i].relevance, matches_[0].relevance);
698    EXPECT_LT(matches_[0].relevance, test_cases[i].relevance + 10);
699    // Input type should be what we expect.  This is important because
700    // this provider counts on SearchProvider to give queries a relevance
701    // score >1200 for UNKNOWN inputs and <1200 for URL inputs.  (That's
702    // already tested in search_provider_unittest.cc.)  For this test
703    // here to test that the user sees the correct behavior, it needs
704    // to check that the input type was identified correctly.
705    EXPECT_EQ(test_cases[i].type, type);
706  }
707}
708
709// Makes sure autocompletion happens for intranet sites that have been
710// previoulsy visited.
711TEST_F(HistoryURLProviderTest, IntranetURLCompletion) {
712  sort_matches_ = true;
713
714  const UrlAndLegalDefault expected1[] = {
715    { "http://intra/three", true },
716    { "http://intra/two", true }
717  };
718  ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("intra/t"), base::string16(),
719                                  false, expected1, arraysize(expected1)));
720  EXPECT_LE(1410, matches_[0].relevance);
721  EXPECT_LT(matches_[0].relevance, 1420);
722  EXPECT_EQ(matches_[0].relevance - 1, matches_[1].relevance);
723
724  const UrlAndLegalDefault expected2[] = {
725    { "http://moo/b", true },
726    { "http://moo/bar", true }
727  };
728  ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("moo/b"), base::string16(),
729                                  false, expected2, arraysize(expected2)));
730  // The url what you typed match should be around 1400, otherwise the
731  // search what you typed match is going to be first.
732  EXPECT_LE(1400, matches_[0].relevance);
733  EXPECT_LT(matches_[0].relevance, 1410);
734
735  const UrlAndLegalDefault expected3[] = {
736    { "http://intra/one", true },
737    { "http://intra/three", true },
738    { "http://intra/two", true }
739  };
740  RunTest(ASCIIToUTF16("intra"), base::string16(), false, expected3,
741          arraysize(expected3));
742
743  const UrlAndLegalDefault expected4[] = {
744    { "http://intra/one", true },
745    { "http://intra/three", true },
746    { "http://intra/two", true }
747  };
748  RunTest(ASCIIToUTF16("intra/"), base::string16(), false, expected4,
749          arraysize(expected4));
750
751  const UrlAndLegalDefault expected5[] = {
752    { "http://intra/one", true }
753  };
754  ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("intra/o"), base::string16(),
755                                  false, expected5, arraysize(expected5)));
756  EXPECT_LE(1410, matches_[0].relevance);
757  EXPECT_LT(matches_[0].relevance, 1420);
758
759  const UrlAndLegalDefault expected6[] = {
760    { "http://intra/x", true }
761  };
762  ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("intra/x"), base::string16(),
763                                  false, expected6, arraysize(expected6)));
764  EXPECT_LE(1400, matches_[0].relevance);
765  EXPECT_LT(matches_[0].relevance, 1410);
766
767  const UrlAndLegalDefault expected7[] = {
768    { "http://typedhost/untypedpath", true }
769  };
770  ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("typedhost/untypedpath"),
771      base::string16(), false, expected7, arraysize(expected7)));
772  EXPECT_LE(1400, matches_[0].relevance);
773  EXPECT_LT(matches_[0].relevance, 1410);
774}
775
776TEST_F(HistoryURLProviderTest, CrashDueToFixup) {
777  // This test passes if we don't crash.  The results don't matter.
778  const char* const test_cases[] = {
779    "//c",
780    "\\@st",
781    "view-source:x",
782  };
783  for (size_t i = 0; i < arraysize(test_cases); ++i) {
784    AutocompleteInput input(ASCIIToUTF16(test_cases[i]), base::string16::npos,
785                            base::string16(), GURL(),
786                            AutocompleteInput::INVALID_SPEC,
787                            false, false, true, AutocompleteInput::ALL_MATCHES);
788    autocomplete_->Start(input, false);
789    if (!autocomplete_->done())
790      base::MessageLoop::current()->Run();
791  }
792}
793
794TEST_F(HistoryURLProviderTest, CullSearchResults) {
795  // Set up a default search engine.
796  TemplateURLData data;
797  data.SetKeyword(ASCIIToUTF16("TestEngine"));
798  data.SetURL("http://testsearch.com/?q={searchTerms}");
799  TemplateURLService* template_url_service =
800      TemplateURLServiceFactory::GetForProfile(profile_.get());
801  TemplateURL* template_url = new TemplateURL(profile_.get(), data);
802  template_url_service->Add(template_url);
803  template_url_service->SetDefaultSearchProvider(template_url);
804  template_url_service->Load();
805
806  // URLs we will be using, plus the visit counts they will initially get
807  // (the redirect set below will also increment the visit counts). We want
808  // the results to be in A,B,C order. Note also that our visit counts are
809  // all high enough so that domain synthesizing won't get triggered.
810  struct TestCase {
811    const char* url;
812    int count;
813  } test_cases[] = {
814    {"https://testsearch.com/", 30},
815    {"https://testsearch.com/?q=foobar", 20},
816    {"http://foobar.com/", 10}
817  };
818  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(test_cases); ++i) {
819    history_service_->AddPageWithDetails(GURL(test_cases[i].url),
820        base::UTF8ToUTF16("Title"), test_cases[i].count, test_cases[i].count,
821        Time::Now(), false, history::SOURCE_BROWSED);
822  }
823
824  // We should not see search URLs when typing a previously used query.
825  const UrlAndLegalDefault expected_when_searching_query[] = {
826    { test_cases[2].url, false }
827  };
828  RunTest(ASCIIToUTF16("foobar"), base::string16(), true,
829      expected_when_searching_query, arraysize(expected_when_searching_query));
830
831  // We should not see search URLs when typing the search engine name.
832  const UrlAndLegalDefault expected_when_searching_site[] = {
833    { test_cases[0].url, false }
834  };
835  RunTest(ASCIIToUTF16("testsearch"), base::string16(), true,
836      expected_when_searching_site, arraysize(expected_when_searching_site));
837}
838
839TEST_F(HistoryURLProviderTest, SuggestExactInput) {
840  const size_t npos = std::string::npos;
841  struct TestCase {
842    // Inputs:
843    const char* input;
844    bool trim_http;
845    // Expected Outputs:
846    const char* contents;
847    // Offsets of the ACMatchClassifications, terminated by npos.
848    size_t offsets[3];
849    // The index of the ACMatchClassification that should have the MATCH bit
850    // set, npos if no ACMatchClassification should have the MATCH bit set.
851    size_t match_classification_index;
852  } test_cases[] = {
853    { "http://www.somesite.com", false,
854      "http://www.somesite.com", {0, npos, npos}, 0 },
855    { "www.somesite.com", true,
856      "www.somesite.com", {0, npos, npos}, 0 },
857    { "www.somesite.com", false,
858      "http://www.somesite.com", {0, 7, npos}, 1 },
859    { "somesite.com", true,
860      "somesite.com", {0, npos, npos}, 0 },
861    { "somesite.com", false,
862      "http://somesite.com", {0, 7, npos}, 1 },
863    { "w", true,
864      "w", {0, npos, npos}, 0 },
865    { "w", false,
866      "http://w", {0, 7, npos}, 1 },
867    { "w.com", true,
868      "w.com", {0, npos, npos}, 0 },
869    { "w.com", false,
870      "http://w.com", {0, 7, npos}, 1 },
871    { "www.w.com", true,
872      "www.w.com", {0, npos, npos}, 0 },
873    { "www.w.com", false,
874      "http://www.w.com", {0, 7, npos}, 1 },
875    { "view-source:w", true,
876      "view-source:w", {0, npos, npos}, 0 },
877    { "view-source:www.w.com/", true,
878      "view-source:www.w.com", {0, npos, npos}, npos },
879    { "view-source:www.w.com/", false,
880      "view-source:http://www.w.com", {0, npos, npos}, npos },
881    { "view-source:http://www.w.com/", false,
882      "view-source:http://www.w.com", {0, npos, npos}, 0 },
883    { "   view-source:", true,
884      "view-source:", {0, npos, npos}, 0 },
885    { "http:////////w.com", false,
886      "http://w.com", {0, npos, npos}, npos },
887    { "    http:////////www.w.com", false,
888      "http://www.w.com", {0, npos, npos}, npos },
889    { "http:a///www.w.com", false,
890      "http://a///www.w.com", {0, npos, npos}, npos },
891    { "mailto://a@b.com", true,
892      "mailto://a@b.com", {0, npos, npos}, 0 },
893    { "mailto://a@b.com", false,
894      "mailto://a@b.com", {0, npos, npos}, 0 },
895  };
896  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(test_cases); ++i) {
897    SCOPED_TRACE(testing::Message() << "Index " << i << " input: "
898                                    << test_cases[i].input << ", trim_http: "
899                                    << test_cases[i].trim_http);
900
901    AutocompleteInput input(ASCIIToUTF16(test_cases[i].input),
902                            base::string16::npos, base::string16(),
903                            GURL("about:blank"),
904                            AutocompleteInput::INVALID_SPEC, false, false, true,
905                            AutocompleteInput::ALL_MATCHES);
906    AutocompleteMatch match(autocomplete_->SuggestExactInput(
907        input.text(), input.canonicalized_url(), test_cases[i].trim_http));
908    EXPECT_EQ(ASCIIToUTF16(test_cases[i].contents), match.contents);
909    for (size_t match_index = 0; match_index < match.contents_class.size();
910         ++match_index) {
911      EXPECT_EQ(test_cases[i].offsets[match_index],
912                match.contents_class[match_index].offset);
913      EXPECT_EQ(ACMatchClassification::URL |
914                (match_index == test_cases[i].match_classification_index ?
915                 ACMatchClassification::MATCH : 0),
916                match.contents_class[match_index].style);
917    }
918    EXPECT_EQ(npos, test_cases[i].offsets[match.contents_class.size()]);
919  }
920}
921
922TEST_F(HistoryURLProviderTest, HUPScoringExperiment) {
923  HUPScoringParams max_2000_no_time_decay;
924  max_2000_no_time_decay.typed_count_buckets.buckets().push_back(
925      std::make_pair(0.0, 2000));
926  HUPScoringParams max_1250_no_time_decay;
927  max_1250_no_time_decay.typed_count_buckets.buckets().push_back(
928      std::make_pair(0.0, 1250));
929  HUPScoringParams max_1000_no_time_decay;
930  max_1000_no_time_decay.typed_count_buckets.buckets().push_back(
931      std::make_pair(0.0, 1000));
932
933  HUPScoringParams max_1100_with_time_decay_and_max_cap;
934  max_1100_with_time_decay_and_max_cap.typed_count_buckets.
935      set_relevance_cap(1400);
936  max_1100_with_time_decay_and_max_cap.typed_count_buckets.
937      set_half_life_days(16);
938  max_1100_with_time_decay_and_max_cap.typed_count_buckets.buckets().push_back(
939      std::make_pair(0.5, 1100));
940  max_1100_with_time_decay_and_max_cap.typed_count_buckets.buckets().push_back(
941      std::make_pair(0.24, 200));
942  max_1100_with_time_decay_and_max_cap.typed_count_buckets.buckets().push_back(
943      std::make_pair(0.0, 100));
944
945  HUPScoringParams max_1100_visit_typed_decays;
946  max_1100_visit_typed_decays.typed_count_buckets.set_half_life_days(16);
947  max_1100_visit_typed_decays.typed_count_buckets.buckets().push_back(
948      std::make_pair(0.5, 1100));
949  max_1100_visit_typed_decays.typed_count_buckets.buckets().push_back(
950      std::make_pair(0.0, 100));
951  max_1100_visit_typed_decays.visited_count_buckets.set_half_life_days(16);
952  max_1100_visit_typed_decays.visited_count_buckets.buckets().push_back(
953      std::make_pair(0.5, 550));
954  max_1100_visit_typed_decays.visited_count_buckets.buckets().push_back(
955      std::make_pair(0.0, 50));
956
957  const int kMaxMatches = 3;
958  struct TestCase {
959    const char* input;
960    HUPScoringParams scoring_params;
961    struct ExpectedMatch {
962      const char* url;
963      int control_relevance;
964      int experiment_relevance;
965    };
966    ExpectedMatch matches[kMaxMatches];
967  } test_cases[] = {
968    // Max score 2000 -> no demotion.
969    { "7.com/1", max_2000_no_time_decay,
970      {{"7.com/1a", 1413, 1413}, {NULL, 0, 0}, {NULL, 0, 0}} },
971
972    // Limit score to 1250/1000 and make sure that the top match is unchanged.
973    { "7.com/1", max_1250_no_time_decay,
974      {{"7.com/1a", 1413, 1413}, {NULL, 0, 0}, {NULL, 0, 0}} },
975    { "7.com/2", max_1250_no_time_decay,
976      {{"7.com/2a", 1413, 1413}, {"7.com/2b", 1412, 1250}, {NULL, 0, 0}} },
977    { "7.com/4", max_1000_no_time_decay,
978      {{"7.com/4", 1203, 1203}, {"7.com/4a", 1202, 1000},
979       {"7.com/4b", 1201, 999}} },
980
981    // Max relevance cap is 1400 and half-life is 16 days.
982    { "7.com/1", max_1100_with_time_decay_and_max_cap,
983      {{"7.com/1a", 1413, 1413}, {NULL, 0, 0}, {NULL, 0, 0}} },
984    { "7.com/4", max_1100_with_time_decay_and_max_cap,
985      {{"7.com/4", 1203, 1203}, {"7.com/4a", 1202, 200},
986       {"7.com/4b", 1201, 100}} },
987
988    // Max relevance cap is 1400 and half-life is 16 days for both visit/typed.
989    { "7.com/5", max_1100_visit_typed_decays,
990      {{"7.com/5", 1203, 1203}, {"7.com/5a", 1202, 50}, {NULL, 0, 0}} },
991  };
992  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(test_cases); ++i) {
993    SCOPED_TRACE(test_cases[i].input);
994    UrlAndLegalDefault output[kMaxMatches];
995    int max_matches;
996    for (max_matches = 0; max_matches < kMaxMatches; ++max_matches) {
997      if (test_cases[i].matches[max_matches].url == NULL)
998        break;
999      output[max_matches].url = URLFixerUpper::FixupURL(
1000          test_cases[i].matches[max_matches].url, std::string()).spec();
1001      output[max_matches].allowed_to_be_default_match = true;
1002    }
1003    autocomplete_->scoring_params_ = test_cases[i].scoring_params;
1004
1005    // Test the control (scoring disabled).
1006    autocomplete_->scoring_params_.experimental_scoring_enabled = false;
1007    ASSERT_NO_FATAL_FAILURE(
1008        RunTest(ASCIIToUTF16(test_cases[i].input),
1009                base::string16(), false, output, max_matches));
1010    for (int j = 0; j < max_matches; ++j) {
1011      EXPECT_EQ(test_cases[i].matches[j].control_relevance,
1012                matches_[j].relevance);
1013    }
1014
1015    // Test the experiment (scoring enabled).
1016    autocomplete_->scoring_params_.experimental_scoring_enabled = true;
1017    ASSERT_NO_FATAL_FAILURE(
1018        RunTest(ASCIIToUTF16(test_cases[i].input),
1019                base::string16(), false, output, max_matches));
1020    for (int j = 0; j < max_matches; ++j) {
1021      EXPECT_EQ(test_cases[i].matches[j].experiment_relevance,
1022                matches_[j].relevance);
1023    }
1024  }
1025}
1026