history_url_provider_unittest.cc revision 558790d6acca3451cf3a6b497803a5f07d0bec58 
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/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/search_engines/template_url.h"
21#include "chrome/browser/search_engines/template_url_service.h"
22#include "chrome/browser/search_engines/template_url_service_factory.h"
23#include "chrome/common/net/url_fixer_upper.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    ASSERT_TRUE(SetUpImpl(false));
156  }
157  virtual void TearDown();
158
159  // Does the real setup.
160  bool SetUpImpl(bool no_db) WARN_UNUSED_RESULT;
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  content::TestBrowserThreadBundle thread_bundle_;
187  ACMatches matches_;
188  scoped_ptr<TestingProfile> profile_;
189  HistoryService* history_service_;
190  scoped_refptr<HistoryURLProvider> autocomplete_;
191  // Should the matches be sorted and duplicates removed?
192  bool sort_matches_;
193};
194
195class HistoryURLProviderTestNoDB : public HistoryURLProviderTest {
196 protected:
197  virtual void SetUp() {
198    ASSERT_TRUE(SetUpImpl(true));
199  }
200};
201
202void HistoryURLProviderTest::OnProviderUpdate(bool updated_matches) {
203  if (autocomplete_->done())
204    base::MessageLoop::current()->Quit();
205}
206
207bool HistoryURLProviderTest::SetUpImpl(bool no_db) {
208  profile_.reset(new TestingProfile());
209  if (!(profile_->CreateHistoryService(true, no_db)))
210    return false;
211  if (!no_db) {
212    profile_->BlockUntilHistoryProcessesPendingRequests();
213    profile_->BlockUntilHistoryIndexIsRefreshed();
214  }
215  history_service_ =
216      HistoryServiceFactory::GetForProfile(profile_.get(),
217                                           Profile::EXPLICIT_ACCESS);
218
219  autocomplete_ = new HistoryURLProvider(this, profile_.get(), "en-US,en,ko");
220  TemplateURLServiceFactory::GetInstance()->SetTestingFactoryAndUse(
221      profile_.get(), &HistoryURLProviderTest::CreateTemplateURLService);
222  FillData();
223  return true;
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
289TEST_F(HistoryURLProviderTest, PromoteShorterURLs) {
290  // Test that hosts get synthesized below popular pages.
291  const std::string expected_nonsynth[] = {
292    "http://slashdot.org/favorite_page.html",
293    "http://slashdot.org/",
294  };
295  RunTest(ASCIIToUTF16("slash"), string16(), true, expected_nonsynth,
296          arraysize(expected_nonsynth));
297
298  // Test that hosts get synthesized above less popular pages.
299  const std::string expected_synth[] = {
300    "http://kerneltrap.org/",
301    "http://kerneltrap.org/not_very_popular.html",
302  };
303  RunTest(ASCIIToUTF16("kernel"), string16(), true, expected_synth,
304          arraysize(expected_synth));
305
306  // Test that unpopular pages are ignored completely.
307  RunTest(ASCIIToUTF16("fresh"), string16(), true, NULL, 0);
308
309  // Test that if we create or promote shorter suggestions that would not
310  // normally be inline autocompletable, we make them inline autocompletable if
311  // the original suggestion (that we replaced as "top") was inline
312  // autocompletable.
313  const std::string expected_synthesisa[] = {
314    "http://synthesisatest.com/",
315    "http://synthesisatest.com/foo/",
316  };
317  RunTest(ASCIIToUTF16("synthesisa"), string16(), false, expected_synthesisa,
318          arraysize(expected_synthesisa));
319  EXPECT_LT(matches_.front().relevance, 1200);
320  const std::string expected_synthesisb[] = {
321    "http://synthesisbtest.com/foo/",
322    "http://synthesisbtest.com/foo/bar.html",
323  };
324  RunTest(ASCIIToUTF16("synthesisb"), string16(), false, expected_synthesisb,
325          arraysize(expected_synthesisb));
326  EXPECT_GE(matches_.front().relevance, 1410);
327
328  // Test that if we have a synthesized host that matches a suggestion, they
329  // get combined into one.
330  const std::string expected_combine[] = {
331    "http://news.google.com/",
332    "http://news.google.com/?ned=us&topic=n",
333  };
334  ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("news"), string16(), true,
335      expected_combine, arraysize(expected_combine)));
336  // The title should also have gotten set properly on the host for the
337  // synthesized one, since it was also in the results.
338  EXPECT_EQ(ASCIIToUTF16("Google News"), matches_.front().description);
339
340  // Test that short URL matching works correctly as the user types more
341  // (several tests):
342  // The entry for foo.com is the best of all five foo.com* entries.
343  const std::string short_1[] = {
344    "http://foo.com/",
345    "http://foo.com/dir/another/again/myfile.html",
346    "http://foo.com/dir/",
347  };
348  RunTest(ASCIIToUTF16("foo"), string16(), true, short_1, arraysize(short_1));
349
350  // When the user types the whole host, make sure we don't get two results for
351  // it.
352  const std::string short_2[] = {
353    "http://foo.com/",
354    "http://foo.com/dir/another/again/myfile.html",
355    "http://foo.com/dir/",
356    "http://foo.com/dir/another/",
357  };
358  RunTest(ASCIIToUTF16("foo.com"), string16(), true, short_2,
359          arraysize(short_2));
360  RunTest(ASCIIToUTF16("foo.com/"), string16(), true, short_2,
361          arraysize(short_2));
362
363  // The filename is the second best of the foo.com* entries, but there is a
364  // shorter URL that's "good enough".  The host doesn't match the user input
365  // and so should not appear.
366  const std::string short_3[] = {
367    "http://foo.com/d",
368    "http://foo.com/dir/another/",
369    "http://foo.com/dir/another/again/myfile.html",
370    "http://foo.com/dir/",
371  };
372  RunTest(ASCIIToUTF16("foo.com/d"), string16(), true, short_3,
373          arraysize(short_3));
374
375  // We shouldn't promote shorter URLs than the best if they're not good
376  // enough.
377  const std::string short_4[] = {
378    "http://foo.com/dir/another/a",
379    "http://foo.com/dir/another/again/myfile.html",
380    "http://foo.com/dir/another/again/",
381  };
382  RunTest(ASCIIToUTF16("foo.com/dir/another/a"), string16(), true, short_4,
383          arraysize(short_4));
384
385  // Exact matches should always be best no matter how much more another match
386  // has been typed.
387  const std::string short_5a[] = {
388    "http://gooey/",
389    "http://www.google.com/",
390    "http://go/",
391  };
392  const std::string short_5b[] = {
393    "http://go/",
394    "http://gooey/",
395    "http://www.google.com/",
396  };
397  RunTest(ASCIIToUTF16("g"), string16(), false, short_5a, arraysize(short_5a));
398  RunTest(ASCIIToUTF16("go"), string16(), false, short_5b, arraysize(short_5b));
399}
400
401TEST_F(HistoryURLProviderTest, CullRedirects) {
402  // URLs we will be using, plus the visit counts they will initially get
403  // (the redirect set below will also increment the visit counts). We want
404  // the results to be in A,B,C order. Note also that our visit counts are
405  // all high enough so that domain synthesizing won't get triggered.
406  struct TestCase {
407    const char* url;
408    int count;
409  } test_cases[] = {
410    {"http://redirects/A", 30},
411    {"http://redirects/B", 20},
412    {"http://redirects/C", 10}
413  };
414  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(test_cases); ++i) {
415    history_service_->AddPageWithDetails(GURL(test_cases[i].url),
416        UTF8ToUTF16("Title"), test_cases[i].count, test_cases[i].count,
417        Time::Now(), false, history::SOURCE_BROWSED);
418  }
419
420  // Create a B->C->A redirect chain, but set the visit counts such that they
421  // will appear in A,B,C order in the results. The autocomplete query will
422  // search for the most recent visit when looking for redirects, so this will
423  // be found even though the previous visits had no redirects.
424  history::RedirectList redirects_to_a;
425  redirects_to_a.push_back(GURL(test_cases[1].url));
426  redirects_to_a.push_back(GURL(test_cases[2].url));
427  redirects_to_a.push_back(GURL(test_cases[0].url));
428  history_service_->AddPage(GURL(test_cases[0].url), base::Time::Now(),
429      NULL, 0, GURL(), redirects_to_a, content::PAGE_TRANSITION_TYPED,
430      history::SOURCE_BROWSED, true);
431
432  // Because all the results are part of a redirect chain with other results,
433  // all but the first one (A) should be culled. We should get the default
434  // "what you typed" result, plus this one.
435  const string16 typing(ASCIIToUTF16("http://redirects/"));
436  const std::string expected_results[] = {
437    UTF16ToUTF8(typing),
438    test_cases[0].url,
439  };
440  RunTest(typing, string16(), true, expected_results,
441          arraysize(expected_results));
442}
443
444TEST_F(HistoryURLProviderTest, WhatYouTyped) {
445  // Make sure we suggest a What You Typed match at the right times.
446  RunTest(ASCIIToUTF16("wytmatch"), string16(), false, NULL, 0);
447  RunTest(ASCIIToUTF16("wytmatch foo bar"), string16(), false, NULL, 0);
448  RunTest(ASCIIToUTF16("wytmatch+foo+bar"), string16(), false, NULL, 0);
449  RunTest(ASCIIToUTF16("wytmatch+foo+bar.com"), string16(), false, NULL, 0);
450
451  const std::string results_1[] = {"http://www.wytmatch.com/"};
452  RunTest(ASCIIToUTF16("wytmatch"), ASCIIToUTF16("com"), false, results_1,
453          arraysize(results_1));
454
455  const std::string results_2[] = {"http://wytmatch%20foo%20bar/"};
456  RunTest(ASCIIToUTF16("http://wytmatch foo bar"), string16(), false, results_2,
457          arraysize(results_2));
458
459  const std::string results_3[] = {"https://wytmatch%20foo%20bar/"};
460  RunTest(ASCIIToUTF16("https://wytmatch foo bar"), string16(), false,
461          results_3, arraysize(results_3));
462}
463
464TEST_F(HistoryURLProviderTest, Fixup) {
465  // Test for various past crashes we've had.
466  RunTest(ASCIIToUTF16("\\"), string16(), false, NULL, 0);
467  RunTest(ASCIIToUTF16("#"), string16(), false, NULL, 0);
468  RunTest(ASCIIToUTF16("%20"), string16(), false, NULL, 0);
469  const std::string fixup_crash[] = {"http://%EF%BD%A5@s/"};
470  RunTest(WideToUTF16(L"\uff65@s"), string16(), false, fixup_crash,
471          arraysize(fixup_crash));
472  RunTest(WideToUTF16(L"\u2015\u2015@ \uff7c"), string16(), false, NULL, 0);
473
474  // Fixing up "file:" should result in an inline autocomplete offset of just
475  // after "file:", not just after "file://".
476  const string16 input_1(ASCIIToUTF16("file:"));
477  const std::string fixup_1[] = {"file:///C:/foo.txt"};
478  ASSERT_NO_FATAL_FAILURE(RunTest(input_1, string16(), false, fixup_1,
479                                  arraysize(fixup_1)));
480  EXPECT_EQ(ASCIIToUTF16("///C:/foo.txt"),
481            matches_.front().inline_autocompletion);
482
483  // Fixing up "http:/" should result in an inline autocomplete offset of just
484  // after "http:/", not just after "http:".
485  const string16 input_2(ASCIIToUTF16("http:/"));
486  const std::string fixup_2[] = {
487    "http://bogussite.com/a",
488    "http://bogussite.com/b",
489    "http://bogussite.com/c",
490  };
491  ASSERT_NO_FATAL_FAILURE(RunTest(input_2, string16(), false, fixup_2,
492                                  arraysize(fixup_2)));
493  EXPECT_EQ(ASCIIToUTF16("/bogussite.com/a"),
494            matches_.front().inline_autocompletion);
495
496  // Adding a TLD to a small number like "56" should result in "www.56.com"
497  // rather than "0.0.0.56.com".
498  const std::string fixup_3[] = {"http://www.56.com/"};
499  RunTest(ASCIIToUTF16("56"), ASCIIToUTF16("com"), true, fixup_3,
500          arraysize(fixup_3));
501
502  // An input looks like a IP address like "127.0.0.1" should result in
503  // "http://127.0.0.1/".
504  const std::string fixup_4[] = {"http://127.0.0.1/"};
505  RunTest(ASCIIToUTF16("127.0.0.1"), string16(), false, fixup_4,
506          arraysize(fixup_4));
507
508  // An number "17173" should result in "http://www.17173.com/" in db.
509  const std::string fixup_5[] = {"http://www.17173.com/"};
510  RunTest(ASCIIToUTF16("17173"), string16(), false, fixup_5,
511          arraysize(fixup_5));
512}
513
514// Make sure the results for the input 'p' don't change between the first and
515// second passes.
516TEST_F(HistoryURLProviderTest, EmptyVisits) {
517  // Wait for history to create the in memory DB.
518  profile_->BlockUntilHistoryProcessesPendingRequests();
519
520  AutocompleteInput input(ASCIIToUTF16("p"), string16::npos, string16(), GURL(),
521                          false, false, true, AutocompleteInput::ALL_MATCHES);
522  autocomplete_->Start(input, false);
523  // HistoryURLProvider shouldn't be done (waiting on async results).
524  EXPECT_FALSE(autocomplete_->done());
525
526  // We should get back an entry for pandora.
527  matches_ = autocomplete_->matches();
528  ASSERT_GT(matches_.size(), 0u);
529  EXPECT_EQ(GURL("http://pandora.com/"), matches_[0].destination_url);
530  int pandora_relevance = matches_[0].relevance;
531
532  // Run the message loop. When |autocomplete_| finishes the loop is quit.
533  base::MessageLoop::current()->Run();
534  EXPECT_TRUE(autocomplete_->done());
535  matches_ = autocomplete_->matches();
536  ASSERT_GT(matches_.size(), 0u);
537  EXPECT_EQ(GURL("http://pandora.com/"), matches_[0].destination_url);
538  EXPECT_EQ(pandora_relevance, matches_[0].relevance);
539}
540
541TEST_F(HistoryURLProviderTestNoDB, NavigateWithoutDB) {
542  // Ensure that we will still produce matches for navigation when there is no
543  // database.
544  std::string navigation_1[] = {"http://test.com/"};
545  RunTest(ASCIIToUTF16("test.com"), string16(), false, navigation_1,
546          arraysize(navigation_1));
547
548  std::string navigation_2[] = {"http://slash/"};
549  RunTest(ASCIIToUTF16("slash"), string16(), false, navigation_2,
550          arraysize(navigation_2));
551
552  RunTest(ASCIIToUTF16("this is a query"), string16(), false, NULL, 0);
553}
554
555TEST_F(HistoryURLProviderTest, DontAutocompleteOnTrailingWhitespace) {
556  AutocompleteInput input(ASCIIToUTF16("slash "), string16::npos, string16(),
557                          GURL(), false, false, true,
558                          AutocompleteInput::ALL_MATCHES);
559  autocomplete_->Start(input, false);
560  if (!autocomplete_->done())
561    base::MessageLoop::current()->Run();
562
563  // None of the matches should attempt to autocomplete.
564  matches_ = autocomplete_->matches();
565  for (size_t i = 0; i < matches_.size(); ++i)
566    EXPECT_TRUE(matches_[i].inline_autocompletion.empty());
567}
568
569TEST_F(HistoryURLProviderTest, TreatEmailsAsSearches) {
570  // Visiting foo.com should not make this string be treated as a navigation.
571  // That means the result should be scored around 1200 ("what you typed")
572  // and not 1400+.
573  const std::string expected[] = {"http://user@foo.com/"};
574  ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("user@foo.com"), string16(),
575                                  false, expected, arraysize(expected)));
576  EXPECT_LE(1200, matches_[0].relevance);
577  EXPECT_LT(matches_[0].relevance, 1210);
578}
579
580TEST_F(HistoryURLProviderTest, IntranetURLsWithPaths) {
581  struct TestCase {
582    const char* input;
583    int relevance;
584  } test_cases[] = {
585    { "fooey", 0 },
586    { "fooey/", 1200 },     // 1200 for URL would still navigate by default.
587    { "fooey/a", 1200 },    // 1200 for UNKNOWN would not.
588    { "fooey/a b", 1200 },  // Also UNKNOWN.
589    { "gooey", 1410 },
590    { "gooey/", 1410 },
591    { "gooey/a", 1400 },
592    { "gooey/a b", 1400 },
593  };
594  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(test_cases); ++i) {
595    SCOPED_TRACE(test_cases[i].input);
596    if (test_cases[i].relevance == 0) {
597      RunTest(ASCIIToUTF16(test_cases[i].input), string16(), false, NULL, 0);
598    } else {
599      const std::string output[] = {
600        URLFixerUpper::FixupURL(test_cases[i].input, std::string()).spec()
601      };
602      ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16(test_cases[i].input),
603                              string16(), false, output, arraysize(output)));
604      // Actual relevance should be at least what test_cases expects and
605      // and no more than 10 more.
606      EXPECT_LE(test_cases[i].relevance, matches_[0].relevance);
607      EXPECT_LT(matches_[0].relevance, test_cases[i].relevance + 10);
608    }
609  }
610}
611
612TEST_F(HistoryURLProviderTest, IntranetURLsWithRefs) {
613  struct TestCase {
614    const char* input;
615    int relevance;
616    AutocompleteInput::Type type;
617  } test_cases[] = {
618    { "gooey", 1410, AutocompleteInput::UNKNOWN },
619    { "gooey/", 1410, AutocompleteInput::URL },
620    { "gooey#", 1200, AutocompleteInput::UNKNOWN },
621    { "gooey/#", 1200, AutocompleteInput::URL },
622    { "gooey#foo", 1200, AutocompleteInput::UNKNOWN },
623    { "gooey/#foo", 1200, AutocompleteInput::URL },
624    { "gooey# foo", 1200, AutocompleteInput::UNKNOWN },
625    { "gooey/# foo", 1200, AutocompleteInput::URL },
626  };
627  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(test_cases); ++i) {
628    SCOPED_TRACE(test_cases[i].input);
629    const std::string output[] = {
630      URLFixerUpper::FixupURL(test_cases[i].input, std::string()).spec()
631    };
632    AutocompleteInput::Type type;
633    ASSERT_NO_FATAL_FAILURE(
634        RunTest(ASCIIToUTF16(test_cases[i].input),
635                string16(), false, output, arraysize(output), &type));
636    // Actual relevance should be at least what test_cases expects and
637    // and no more than 10 more.
638    EXPECT_LE(test_cases[i].relevance, matches_[0].relevance);
639    EXPECT_LT(matches_[0].relevance, test_cases[i].relevance + 10);
640    // Input type should be what we expect.  This is important because
641    // this provider counts on SearchProvider to give queries a relevance
642    // score >1200 for UNKNOWN inputs and <1200 for URL inputs.  (That's
643    // already tested in search_provider_unittest.cc.)  For this test
644    // here to test that the user sees the correct behavior, it needs
645    // to check that the input type was identified correctly.
646    EXPECT_EQ(test_cases[i].type, type);
647  }
648}
649
650// Makes sure autocompletion happens for intranet sites that have been
651// previoulsy visited.
652TEST_F(HistoryURLProviderTest, IntranetURLCompletion) {
653  sort_matches_ = true;
654
655  const std::string expected1[] = {
656    "http://intra/three",
657    "http://intra/two",
658  };
659  ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("intra/t"), string16(), false,
660                                  expected1, arraysize(expected1)));
661  EXPECT_LE(1410, matches_[0].relevance);
662  EXPECT_LT(matches_[0].relevance, 1420);
663  EXPECT_EQ(matches_[0].relevance - 1, matches_[1].relevance);
664
665  const std::string expected2[] = {
666    "http://moo/b",
667    "http://moo/bar",
668  };
669  ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("moo/b"), string16(), false,
670                                  expected2, arraysize(expected2)));
671  // The url what you typed match should be around 1400, otherwise the
672  // search what you typed match is going to be first.
673  EXPECT_LE(1400, matches_[0].relevance);
674  EXPECT_LT(matches_[0].relevance, 1410);
675
676  const std::string expected3[] = {
677    "http://intra/one",
678    "http://intra/three",
679    "http://intra/two",
680  };
681  RunTest(ASCIIToUTF16("intra"), string16(), false, expected3,
682          arraysize(expected3));
683
684  const std::string expected4[] = {
685    "http://intra/one",
686    "http://intra/three",
687    "http://intra/two",
688  };
689  RunTest(ASCIIToUTF16("intra/"), string16(), false, expected4,
690          arraysize(expected4));
691
692  const std::string expected5[] = {
693    "http://intra/one",
694  };
695  ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("intra/o"), string16(), false,
696                                  expected5, arraysize(expected5)));
697  EXPECT_LE(1410, matches_[0].relevance);
698  EXPECT_LT(matches_[0].relevance, 1420);
699
700  const std::string expected6[] = {
701    "http://intra/x",
702  };
703  ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("intra/x"), string16(), false,
704                                  expected6, arraysize(expected6)));
705  EXPECT_LE(1400, matches_[0].relevance);
706  EXPECT_LT(matches_[0].relevance, 1410);
707
708  const std::string expected7[] = {
709    "http://typedhost/untypedpath",
710  };
711  ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("typedhost/untypedpath"),
712      string16(), false, expected7, arraysize(expected7)));
713  EXPECT_LE(1400, matches_[0].relevance);
714  EXPECT_LT(matches_[0].relevance, 1410);
715}
716
717TEST_F(HistoryURLProviderTest, CrashDueToFixup) {
718  // This test passes if we don't crash.  The results don't matter.
719  const char* const test_cases[] = {
720    "//c",
721    "\\@st"
722  };
723  for (size_t i = 0; i < arraysize(test_cases); ++i) {
724    AutocompleteInput input(ASCIIToUTF16(test_cases[i]), string16::npos,
725                            string16(), GURL(), false, false, true,
726                            AutocompleteInput::ALL_MATCHES);
727    autocomplete_->Start(input, false);
728    if (!autocomplete_->done())
729      base::MessageLoop::current()->Run();
730  }
731}
732
733TEST_F(HistoryURLProviderTest, CullSearchResults) {
734  // Set up a default search engine.
735  TemplateURLData data;
736  data.SetKeyword(ASCIIToUTF16("TestEngine"));
737  data.SetURL("http://testsearch.com/?q={searchTerms}");
738  TemplateURLService* template_url_service =
739      TemplateURLServiceFactory::GetForProfile(profile_.get());
740  TemplateURL* template_url = new TemplateURL(profile_.get(), data);
741  template_url_service->Add(template_url);
742  template_url_service->SetDefaultSearchProvider(template_url);
743  template_url_service->Load();
744
745  // URLs we will be using, plus the visit counts they will initially get
746  // (the redirect set below will also increment the visit counts). We want
747  // the results to be in A,B,C order. Note also that our visit counts are
748  // all high enough so that domain synthesizing won't get triggered.
749  struct TestCase {
750    const char* url;
751    int count;
752  } test_cases[] = {
753    {"https://testsearch.com/", 30},
754    {"https://testsearch.com/?q=foobar", 20},
755    {"http://foobar.com/", 10}
756  };
757  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(test_cases); ++i) {
758    history_service_->AddPageWithDetails(GURL(test_cases[i].url),
759        UTF8ToUTF16("Title"), test_cases[i].count, test_cases[i].count,
760        Time::Now(), false, history::SOURCE_BROWSED);
761  }
762
763  // We should not see search URLs when typing a previously used query.
764  const std::string expected_when_searching_query[] = {
765    test_cases[2].url
766  };
767  RunTest(ASCIIToUTF16("foobar"), string16(), true,
768      expected_when_searching_query, arraysize(expected_when_searching_query));
769
770  // We should not see search URLs when typing the search engine name.
771  const std::string expected_when_searching_site[] = {
772    test_cases[0].url
773  };
774  RunTest(ASCIIToUTF16("testsearch"), string16(), true,
775      expected_when_searching_site, arraysize(expected_when_searching_site));
776}
777
778TEST_F(HistoryURLProviderTest, SuggestExactInput) {
779  const size_t npos = std::string::npos;
780  struct TestCase {
781    // Inputs:
782    const char* input;
783    bool trim_http;
784    // Expected Outputs:
785    const char* contents;
786    // Offsets of the ACMatchClassifications, terminated by npos.
787    size_t offsets[3];
788    // The index of the ACMatchClassification that should have the MATCH bit
789    // set, npos if no ACMatchClassification should have the MATCH bit set.
790    size_t match_classification_index;
791  } test_cases[] = {
792    { "http://www.somesite.com", false,
793      "http://www.somesite.com", {0, npos, npos}, 0 },
794    { "www.somesite.com", true,
795      "www.somesite.com", {0, npos, npos}, 0 },
796    { "www.somesite.com", false,
797      "http://www.somesite.com", {0, 7, npos}, 1 },
798    { "somesite.com", true,
799      "somesite.com", {0, npos, npos}, 0 },
800    { "somesite.com", false,
801      "http://somesite.com", {0, 7, npos}, 1 },
802    { "w", true,
803      "w", {0, npos, npos}, 0 },
804    { "w", false,
805      "http://w", {0, 7, npos}, 1 },
806    { "w.com", true,
807      "w.com", {0, npos, npos}, 0 },
808    { "w.com", false,
809      "http://w.com", {0, 7, npos}, 1 },
810    { "www.w.com", true,
811      "www.w.com", {0, npos, npos}, 0 },
812    { "www.w.com", false,
813      "http://www.w.com", {0, 7, npos}, 1 },
814    { "view-source:www.w.com/", true,
815      "view-source:www.w.com", {0, npos, npos}, npos },
816    { "view-source:www.w.com/", false,
817      "view-source:http://www.w.com", {0, npos, npos}, npos },
818    { "view-source:http://www.w.com/", false,
819      "view-source:http://www.w.com", {0, npos, npos}, 0 },
820    { "   view-source:", true,
821      "view-source:", {0, npos, npos}, 0 },
822    { "http:////////w.com", false,
823      "http://w.com", {0, npos, npos}, npos },
824    { "    http:////////www.w.com", false,
825      "http://www.w.com", {0, npos, npos}, npos },
826    { "http:a///www.w.com", false,
827      "http://a///www.w.com", {0, npos, npos}, npos },
828    { "mailto://a@b.com", true,
829      "mailto://a@b.com", {0, npos, npos}, 0 },
830    { "mailto://a@b.com", false,
831      "mailto://a@b.com", {0, npos, npos}, 0 },
832  };
833  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(test_cases); ++i) {
834    SCOPED_TRACE(testing::Message() << "Index " << i << " input: "
835                                    << test_cases[i].input << ", trim_http: "
836                                    << test_cases[i].trim_http);
837
838    AutocompleteInput input(ASCIIToUTF16(test_cases[i].input), string16::npos,
839                            string16(), GURL("about:blank"),
840                            false, false, true, AutocompleteInput::ALL_MATCHES);
841    AutocompleteMatch match = HistoryURLProvider::SuggestExactInput(
842        autocomplete_.get(), input, test_cases[i].trim_http);
843    EXPECT_EQ(ASCIIToUTF16(test_cases[i].contents), match.contents);
844    for (size_t match_index = 0; match_index < match.contents_class.size();
845         ++match_index) {
846      EXPECT_EQ(test_cases[i].offsets[match_index],
847                match.contents_class[match_index].offset);
848      EXPECT_EQ(ACMatchClassification::URL |
849                (match_index == test_cases[i].match_classification_index ?
850                 ACMatchClassification::MATCH : 0),
851                match.contents_class[match_index].style);
852    }
853    EXPECT_EQ(npos, test_cases[i].offsets[match.contents_class.size()]);
854  }
855}
856