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.
5 #include "chrome/browser/autocomplete/history_url_provider.h"
9 #include "base/message_loop/message_loop.h"
10 #include "base/prefs/pref_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/chrome_autocomplete_scheme_classifier.h"
15 #include "chrome/browser/autocomplete/history_quick_provider.h"
16 #include "chrome/browser/history/history_service.h"
17 #include "chrome/browser/history/history_service_factory.h"
18 #include "chrome/browser/search_engines/chrome_template_url_service_client.h"
19 #include "chrome/browser/search_engines/template_url_service_factory.h"
20 #include "chrome/common/pref_names.h"
21 #include "chrome/test/base/testing_browser_process.h"
22 #include "chrome/test/base/testing_profile.h"
23 #include "components/history/core/browser/url_database.h"
24 #include "components/metrics/proto/omnibox_event.pb.h"
25 #include "components/metrics/proto/omnibox_input_type.pb.h"
26 #include "components/omnibox/autocomplete_match.h"
27 #include "components/omnibox/autocomplete_provider.h"
28 #include "components/omnibox/autocomplete_provider_listener.h"
29 #include "components/omnibox/autocomplete_result.h"
30 #include "components/search_engines/default_search_manager.h"
31 #include "components/search_engines/search_terms_data.h"
32 #include "components/search_engines/template_url.h"
33 #include "components/search_engines/template_url_service.h"
34 #include "components/url_fixer/url_fixer.h"
35 #include "content/public/test/test_browser_thread_bundle.h"
36 #include "testing/gtest/include/gtest/gtest.h"
38 using base::ASCIIToUTF16
;
40 using base::TimeDelta
;
42 using content::TestBrowserThreadBundle
;
51 {"http://www.google.com/", "Google", 3, 3, 80},
53 // High-quality pages should get a host synthesized as a lower-quality match.
54 {"http://slashdot.org/favorite_page.html", "Favorite page", 200, 100, 80},
56 // Less popular pages should have hosts synthesized as higher-quality
58 {"http://kerneltrap.org/not_very_popular.html", "Less popular", 4, 0, 80},
60 // Unpopular pages should not appear in the results at all.
61 {"http://freshmeat.net/unpopular.html", "Unpopular", 1, 0, 80},
63 // If a host has a match, we should pick it up during host synthesis.
64 {"http://news.google.com/?ned=us&topic=n", "Google News - U.S.", 2, 2, 80},
65 {"http://news.google.com/", "Google News", 1, 1, 80},
67 // Matches that are normally not inline-autocompletable should be
68 // autocompleted if they are shorter substitutes for longer matches that would
69 // have been inline autocompleted.
70 {"http://synthesisatest.com/foo/", "Test A", 1, 1, 80},
71 {"http://synthesisbtest.com/foo/", "Test B", 1, 1, 80},
72 {"http://synthesisbtest.com/foo/bar.html", "Test B Bar", 2, 2, 80},
74 // Suggested short URLs must be "good enough" and must match user input.
75 {"http://foo.com/", "Dir", 5, 5, 80},
76 {"http://foo.com/dir/", "Dir", 2, 2, 80},
77 {"http://foo.com/dir/another/", "Dir", 5, 1, 80},
78 {"http://foo.com/dir/another/again/", "Dir", 10, 0, 80},
79 {"http://foo.com/dir/another/again/myfile.html", "File", 10, 2, 80},
81 // We throw in a lot of extra URLs here to make sure we're testing the
82 // history database's query, not just the autocomplete provider.
83 {"http://startest.com/y/a", "A", 2, 2, 80},
84 {"http://startest.com/y/b", "B", 5, 2, 80},
85 {"http://startest.com/x/c", "C", 5, 2, 80},
86 {"http://startest.com/x/d", "D", 5, 5, 80},
87 {"http://startest.com/y/e", "E", 4, 2, 80},
88 {"http://startest.com/y/f", "F", 3, 2, 80},
89 {"http://startest.com/y/g", "G", 3, 2, 80},
90 {"http://startest.com/y/h", "H", 3, 2, 80},
91 {"http://startest.com/y/i", "I", 3, 2, 80},
92 {"http://startest.com/y/j", "J", 3, 2, 80},
93 {"http://startest.com/y/k", "K", 3, 2, 80},
94 {"http://startest.com/y/l", "L", 3, 2, 80},
95 {"http://startest.com/y/m", "M", 3, 2, 80},
97 // A file: URL is useful for testing that fixup does the right thing w.r.t.
98 // the number of trailing slashes on the user's input.
99 {"file:///C:/foo.txt", "", 2, 2, 80},
101 // Results with absurdly high typed_counts so that very generic queries like
102 // "http" will give consistent results even if more data is added above.
103 {"http://bogussite.com/a", "Bogus A", 10002, 10000, 80},
104 {"http://bogussite.com/b", "Bogus B", 10001, 10000, 80},
105 {"http://bogussite.com/c", "Bogus C", 10000, 10000, 80},
107 // Domain name with number.
108 {"http://www.17173.com/", "Domain with number", 3, 3, 80},
110 // URLs to test exact-matching behavior.
111 {"http://go/", "Intranet URL", 1, 1, 80},
112 {"http://gooey/", "Intranet URL 2", 5, 5, 80},
114 // URLs for testing offset adjustment.
115 {"http://www.\xEA\xB5\x90\xEC\x9C\xA1.kr/", "Korean", 2, 2, 80},
116 {"http://spaces.com/path%20with%20spaces/foo.html", "Spaces", 2, 2, 80},
117 {"http://ms/c++%20style%20guide", "Style guide", 2, 2, 80},
119 // URLs for testing ctrl-enter behavior.
120 {"http://binky/", "Intranet binky", 2, 2, 80},
121 {"http://winky/", "Intranet winky", 2, 2, 80},
122 {"http://www.winky.com/", "Internet winky", 5, 0, 80},
124 // URLs used by EmptyVisits.
125 {"http://pandora.com/", "Pandora", 2, 2, 80},
126 // This entry is explicitly added more recently than
127 // history::kLowQualityMatchAgeLimitInDays.
128 // {"http://p/", "p", 0, 0, 80},
130 // For intranet based tests.
131 {"http://intra/one", "Intranet", 2, 2, 80},
132 {"http://intra/two", "Intranet two", 1, 1, 80},
133 {"http://intra/three", "Intranet three", 2, 2, 80},
134 {"http://moo/bar", "Intranet moo", 1, 1, 80},
135 {"http://typedhost/typedpath", "Intranet typed", 1, 1, 80},
136 {"http://typedhost/untypedpath", "Intranet untyped", 1, 0, 80},
138 {"http://x.com/one", "Internet", 2, 2, 80},
139 {"http://x.com/two", "Internet two", 1, 1, 80},
140 {"http://x.com/three", "Internet three", 2, 2, 80},
142 // For experimental HUP scoring test.
143 {"http://7.com/1a", "One", 8, 4, 4},
144 {"http://7.com/2a", "Two A", 4, 2, 8},
145 {"http://7.com/2b", "Two B", 4, 1, 8},
146 {"http://7.com/3a", "Three", 2, 1, 16},
147 {"http://7.com/4a", "Four A", 1, 1, 32},
148 {"http://7.com/4b", "Four B", 1, 1, 64},
149 {"http://7.com/5a", "Five A", 8, 0, 64}, // never typed.
152 class HistoryURLProviderTest
: public testing::Test
,
153 public AutocompleteProviderListener
{
155 struct UrlAndLegalDefault
{
157 bool allowed_to_be_default_match
;
160 HistoryURLProviderTest()
161 : sort_matches_(false) {
162 HistoryQuickProvider::set_disabled(true);
165 ~HistoryURLProviderTest() override
{
166 HistoryQuickProvider::set_disabled(false);
169 // AutocompleteProviderListener:
170 void OnProviderUpdate(bool updated_matches
) override
;
173 static KeyedService
* CreateTemplateURLService(
174 content::BrowserContext
* context
) {
175 Profile
* profile
= static_cast<Profile
*>(context
);
176 return new TemplateURLService(
177 profile
->GetPrefs(), make_scoped_ptr(new SearchTermsData
), NULL
,
178 scoped_ptr
<TemplateURLServiceClient
>(
179 new ChromeTemplateURLServiceClient(
180 HistoryServiceFactory::GetForProfile(
181 profile
, Profile::EXPLICIT_ACCESS
))),
182 NULL
, NULL
, base::Closure());
186 void SetUp() override
{
187 ASSERT_TRUE(SetUpImpl(false));
189 void TearDown() override
;
191 // Does the real setup.
192 bool SetUpImpl(bool no_db
) WARN_UNUSED_RESULT
;
194 // Fills test data into the history system.
197 // Runs an autocomplete query on |text| and checks to see that the returned
198 // results' destination URLs match those provided. Also allows checking
199 // that the input type was identified correctly.
200 void RunTest(const base::string16
& text
,
201 const std::string
& desired_tld
,
202 bool prevent_inline_autocomplete
,
203 const UrlAndLegalDefault
* expected_urls
,
205 metrics::OmniboxInputType::Type
* identified_input_type
);
207 // A version of the above without the final |type| output parameter.
208 void RunTest(const base::string16
& text
,
209 const std::string
& desired_tld
,
210 bool prevent_inline_autocomplete
,
211 const UrlAndLegalDefault
* expected_urls
,
212 size_t num_results
) {
213 metrics::OmniboxInputType::Type type
;
214 return RunTest(text
, desired_tld
, prevent_inline_autocomplete
,
215 expected_urls
, num_results
, &type
);
218 content::TestBrowserThreadBundle thread_bundle_
;
220 scoped_ptr
<TestingProfile
> profile_
;
221 HistoryService
* history_service_
;
222 scoped_refptr
<HistoryURLProvider
> autocomplete_
;
223 // Should the matches be sorted and duplicates removed?
227 class HistoryURLProviderTestNoDB
: public HistoryURLProviderTest
{
229 void SetUp() override
{
230 ASSERT_TRUE(SetUpImpl(true));
234 class HistoryURLProviderTestNoSearchProvider
: public HistoryURLProviderTest
{
236 void SetUp() override
{
237 DefaultSearchManager::SetFallbackSearchEnginesDisabledForTesting(true);
238 HistoryURLProviderTest::SetUp();
241 void TearDown() override
{
242 HistoryURLProviderTest::TearDown();
243 DefaultSearchManager::SetFallbackSearchEnginesDisabledForTesting(false);
247 void HistoryURLProviderTest::OnProviderUpdate(bool updated_matches
) {
248 if (autocomplete_
->done())
249 base::MessageLoop::current()->Quit();
252 bool HistoryURLProviderTest::SetUpImpl(bool no_db
) {
253 profile_
.reset(new TestingProfile());
254 if (!(profile_
->CreateHistoryService(true, no_db
)))
257 profile_
->BlockUntilHistoryProcessesPendingRequests();
258 profile_
->BlockUntilHistoryIndexIsRefreshed();
260 profile_
->GetPrefs()->SetString(prefs::kAcceptLanguages
, "en-US,en,ko");
261 history_service_
= HistoryServiceFactory::GetForProfile(
262 profile_
.get(), Profile::EXPLICIT_ACCESS
);
264 autocomplete_
= new HistoryURLProvider(this, profile_
.get());
265 TemplateURLServiceFactory::GetInstance()->SetTestingFactoryAndUse(
266 profile_
.get(), &HistoryURLProviderTest::CreateTemplateURLService
);
271 void HistoryURLProviderTest::TearDown() {
272 autocomplete_
= NULL
;
275 void HistoryURLProviderTest::FillData() {
276 // Most visits are a long time ago (some tests require this since we do some
277 // special logic for things visited very recently). Note that this time must
278 // be more recent than the "expire history" threshold for the data to be kept
279 // in the main database.
281 // TODO(brettw) It would be nice if we could test this behavior, in which
282 // case the time would be specifed in the test_db structure.
283 const Time now
= Time::Now();
285 for (size_t i
= 0; i
< arraysize(test_db
); ++i
) {
286 const TestURLInfo
& cur
= test_db
[i
];
287 const GURL
current_url(cur
.url
);
288 history_service_
->AddPageWithDetails(
289 current_url
, base::UTF8ToUTF16(cur
.title
), cur
.visit_count
,
290 cur
.typed_count
, now
- TimeDelta::FromDays(cur
.age_in_days
), false,
291 history::SOURCE_BROWSED
);
294 history_service_
->AddPageWithDetails(
295 GURL("http://p/"), base::UTF8ToUTF16("p"), 0, 0,
297 TimeDelta::FromDays(history::kLowQualityMatchAgeLimitInDays
- 1),
298 false, history::SOURCE_BROWSED
);
301 void HistoryURLProviderTest::RunTest(
302 const base::string16
& text
,
303 const std::string
& desired_tld
,
304 bool prevent_inline_autocomplete
,
305 const UrlAndLegalDefault
* expected_urls
,
307 metrics::OmniboxInputType::Type
* identified_input_type
) {
308 AutocompleteInput
input(text
, base::string16::npos
, desired_tld
, GURL(),
309 metrics::OmniboxEventProto::INVALID_SPEC
,
310 prevent_inline_autocomplete
, false, true, true,
311 ChromeAutocompleteSchemeClassifier(profile_
.get()));
312 *identified_input_type
= input
.type();
313 autocomplete_
->Start(input
, false);
314 if (!autocomplete_
->done())
315 base::MessageLoop::current()->Run();
317 matches_
= autocomplete_
->matches();
319 TemplateURLService
* service
=
320 TemplateURLServiceFactory::GetForProfile(profile_
.get());
321 for (ACMatches::iterator i
= matches_
.begin(); i
!= matches_
.end(); ++i
)
322 i
->ComputeStrippedDestinationURL(service
);
323 AutocompleteResult::DedupMatchesByDestination(
324 input
.current_page_classification(), false, &matches_
);
325 std::sort(matches_
.begin(), matches_
.end(),
326 &AutocompleteMatch::MoreRelevant
);
328 ASSERT_EQ(num_results
, matches_
.size()) << "Input text: " << text
329 << "\nTLD: \"" << desired_tld
<< "\"";
330 for (size_t i
= 0; i
< num_results
; ++i
) {
331 EXPECT_EQ(expected_urls
[i
].url
, matches_
[i
].destination_url
.spec());
332 EXPECT_EQ(expected_urls
[i
].allowed_to_be_default_match
,
333 matches_
[i
].allowed_to_be_default_match
);
337 TEST_F(HistoryURLProviderTest
, PromoteShorterURLs
) {
338 // Test that hosts get synthesized below popular pages.
339 const UrlAndLegalDefault expected_nonsynth
[] = {
340 { "http://slashdot.org/favorite_page.html", false },
341 { "http://slashdot.org/", false }
343 RunTest(ASCIIToUTF16("slash"), std::string(), true, expected_nonsynth
,
344 arraysize(expected_nonsynth
));
346 // Test that hosts get synthesized above less popular pages.
347 const UrlAndLegalDefault expected_synth
[] = {
348 { "http://kerneltrap.org/", false },
349 { "http://kerneltrap.org/not_very_popular.html", false }
351 RunTest(ASCIIToUTF16("kernel"), std::string(), true, expected_synth
,
352 arraysize(expected_synth
));
354 // Test that unpopular pages are ignored completely.
355 RunTest(ASCIIToUTF16("fresh"), std::string(), true, NULL
, 0);
357 // Test that if we create or promote shorter suggestions that would not
358 // normally be inline autocompletable, we make them inline autocompletable if
359 // the original suggestion (that we replaced as "top") was inline
361 const UrlAndLegalDefault expected_synthesisa
[] = {
362 { "http://synthesisatest.com/", true },
363 { "http://synthesisatest.com/foo/", true }
365 RunTest(ASCIIToUTF16("synthesisa"), std::string(), false, expected_synthesisa
,
366 arraysize(expected_synthesisa
));
367 EXPECT_LT(matches_
.front().relevance
, 1200);
368 const UrlAndLegalDefault expected_synthesisb
[] = {
369 { "http://synthesisbtest.com/foo/", true },
370 { "http://synthesisbtest.com/foo/bar.html", true }
372 RunTest(ASCIIToUTF16("synthesisb"), std::string(), false, expected_synthesisb
,
373 arraysize(expected_synthesisb
));
374 EXPECT_GE(matches_
.front().relevance
, 1410);
376 // Test that if we have a synthesized host that matches a suggestion, they
377 // get combined into one.
378 const UrlAndLegalDefault expected_combine
[] = {
379 { "http://news.google.com/", false },
380 { "http://news.google.com/?ned=us&topic=n", false },
382 ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("news"), std::string(), true,
384 arraysize(expected_combine
)));
385 // The title should also have gotten set properly on the host for the
386 // synthesized one, since it was also in the results.
387 EXPECT_EQ(ASCIIToUTF16("Google News"), matches_
.front().description
);
389 // Test that short URL matching works correctly as the user types more
391 // The entry for foo.com is the best of all five foo.com* entries.
392 const UrlAndLegalDefault short_1
[] = {
393 { "http://foo.com/", false },
394 { "http://foo.com/dir/another/again/myfile.html", false },
395 { "http://foo.com/dir/", false }
397 RunTest(ASCIIToUTF16("foo"), std::string(), true, short_1
,
400 // When the user types the whole host, make sure we don't get two results for
402 const UrlAndLegalDefault short_2
[] = {
403 { "http://foo.com/", true },
404 { "http://foo.com/dir/another/again/myfile.html", false },
405 { "http://foo.com/dir/", false },
406 { "http://foo.com/dir/another/", false }
408 RunTest(ASCIIToUTF16("foo.com"), std::string(), true, short_2
,
410 RunTest(ASCIIToUTF16("foo.com/"), std::string(), true, short_2
,
413 // The filename is the second best of the foo.com* entries, but there is a
414 // shorter URL that's "good enough". The host doesn't match the user input
415 // and so should not appear.
416 const UrlAndLegalDefault short_3
[] = {
417 { "http://foo.com/d", true },
418 { "http://foo.com/dir/another/", false },
419 { "http://foo.com/dir/another/again/myfile.html", false },
420 { "http://foo.com/dir/", false }
422 RunTest(ASCIIToUTF16("foo.com/d"), std::string(), true, short_3
,
425 // We shouldn't promote shorter URLs than the best if they're not good
427 const UrlAndLegalDefault short_4
[] = {
428 { "http://foo.com/dir/another/a", true },
429 { "http://foo.com/dir/another/again/myfile.html", false },
430 { "http://foo.com/dir/another/again/", false }
432 RunTest(ASCIIToUTF16("foo.com/dir/another/a"), std::string(), true, short_4
,
435 // Exact matches should always be best no matter how much more another match
437 const UrlAndLegalDefault short_5a
[] = {
438 { "http://gooey/", true },
439 { "http://www.google.com/", true },
440 { "http://go/", true }
442 const UrlAndLegalDefault short_5b
[] = {
443 { "http://go/", true },
444 { "http://gooey/", true },
445 { "http://www.google.com/", true }
447 RunTest(ASCIIToUTF16("g"), std::string(), false, short_5a
,
448 arraysize(short_5a
));
449 RunTest(ASCIIToUTF16("go"), std::string(), false, short_5b
,
450 arraysize(short_5b
));
453 TEST_F(HistoryURLProviderTest
, CullRedirects
) {
454 // URLs we will be using, plus the visit counts they will initially get
455 // (the redirect set below will also increment the visit counts). We want
456 // the results to be in A,B,C order. Note also that our visit counts are
457 // all high enough so that domain synthesizing won't get triggered.
462 {"http://redirects/A", 30},
463 {"http://redirects/B", 20},
464 {"http://redirects/C", 10}
466 for (size_t i
= 0; i
< arraysize(test_cases
); ++i
) {
467 history_service_
->AddPageWithDetails(GURL(test_cases
[i
].url
),
468 ASCIIToUTF16("Title"), test_cases
[i
].count
, test_cases
[i
].count
,
469 Time::Now(), false, history::SOURCE_BROWSED
);
472 // Create a B->C->A redirect chain, but set the visit counts such that they
473 // will appear in A,B,C order in the results. The autocomplete query will
474 // search for the most recent visit when looking for redirects, so this will
475 // be found even though the previous visits had no redirects.
476 history::RedirectList redirects_to_a
;
477 redirects_to_a
.push_back(GURL(test_cases
[1].url
));
478 redirects_to_a
.push_back(GURL(test_cases
[2].url
));
479 redirects_to_a
.push_back(GURL(test_cases
[0].url
));
480 history_service_
->AddPage(GURL(test_cases
[0].url
), base::Time::Now(),
481 NULL
, 0, GURL(), redirects_to_a
, ui::PAGE_TRANSITION_TYPED
,
482 history::SOURCE_BROWSED
, true);
484 // Because all the results are part of a redirect chain with other results,
485 // all but the first one (A) should be culled. We should get the default
486 // "what you typed" result, plus this one.
487 const base::string16
typing(ASCIIToUTF16("http://redirects/"));
488 const UrlAndLegalDefault expected_results
[] = {
489 { base::UTF16ToUTF8(typing
), true },
490 { test_cases
[0].url
, false }
492 RunTest(typing
, std::string(), true, expected_results
,
493 arraysize(expected_results
));
496 TEST_F(HistoryURLProviderTestNoSearchProvider
, WhatYouTypedNoSearchProvider
) {
497 // When no search provider is available, make sure we provide WYT matches
498 // for text that could be a URL.
500 const UrlAndLegalDefault results_1
[] = {
501 { "http://wytmatch/", true }
503 RunTest(ASCIIToUTF16("wytmatch"), std::string(), false, results_1
,
504 arraysize(results_1
));
506 RunTest(ASCIIToUTF16("wytmatch foo bar"), std::string(), false, NULL
, 0);
507 RunTest(ASCIIToUTF16("wytmatch+foo+bar"), std::string(), false, NULL
, 0);
509 const UrlAndLegalDefault results_2
[] = {
510 { "http://wytmatch+foo+bar.com/", true }
512 RunTest(ASCIIToUTF16("wytmatch+foo+bar.com"), std::string(), false,
513 results_2
, arraysize(results_2
));
516 TEST_F(HistoryURLProviderTest
, WhatYouTyped
) {
517 // Make sure we suggest a What You Typed match at the right times.
518 RunTest(ASCIIToUTF16("wytmatch"), std::string(), false, NULL
, 0);
519 RunTest(ASCIIToUTF16("wytmatch foo bar"), std::string(), false, NULL
, 0);
520 RunTest(ASCIIToUTF16("wytmatch+foo+bar"), std::string(), false, NULL
, 0);
521 RunTest(ASCIIToUTF16("wytmatch+foo+bar.com"), std::string(), false, NULL
, 0);
523 const UrlAndLegalDefault results_1
[] = {
524 { "http://www.wytmatch.com/", true }
526 RunTest(ASCIIToUTF16("wytmatch"), "com", false, results_1
,
527 arraysize(results_1
));
529 const UrlAndLegalDefault results_2
[] = {
530 { "http://wytmatch%20foo%20bar/", true }
532 RunTest(ASCIIToUTF16("http://wytmatch foo bar"), std::string(), false,
533 results_2
, arraysize(results_2
));
535 const UrlAndLegalDefault results_3
[] = {
536 { "https://wytmatch%20foo%20bar/", true }
538 RunTest(ASCIIToUTF16("https://wytmatch foo bar"), std::string(), false,
539 results_3
, arraysize(results_3
));
542 TEST_F(HistoryURLProviderTest
, Fixup
) {
543 // Test for various past crashes we've had.
544 RunTest(ASCIIToUTF16("\\"), std::string(), false, NULL
, 0);
545 RunTest(ASCIIToUTF16("#"), std::string(), false, NULL
, 0);
546 RunTest(ASCIIToUTF16("%20"), std::string(), false, NULL
, 0);
547 const UrlAndLegalDefault fixup_crash
[] = {
548 { "http://%EF%BD%A5@s/", true }
550 RunTest(base::WideToUTF16(L
"\uff65@s"), std::string(), false, fixup_crash
,
551 arraysize(fixup_crash
));
552 RunTest(base::WideToUTF16(L
"\u2015\u2015@ \uff7c"), std::string(), false,
555 // Fixing up "file:" should result in an inline autocomplete offset of just
556 // after "file:", not just after "file://".
557 const base::string16
input_1(ASCIIToUTF16("file:"));
558 const UrlAndLegalDefault fixup_1
[] = {
559 { "file:///C:/foo.txt", true }
561 ASSERT_NO_FATAL_FAILURE(RunTest(input_1
, std::string(), false, fixup_1
,
562 arraysize(fixup_1
)));
563 EXPECT_EQ(ASCIIToUTF16("///C:/foo.txt"),
564 matches_
.front().inline_autocompletion
);
566 // Fixing up "http:/" should result in an inline autocomplete offset of just
567 // after "http:/", not just after "http:".
568 const base::string16
input_2(ASCIIToUTF16("http:/"));
569 const UrlAndLegalDefault fixup_2
[] = {
570 { "http://bogussite.com/a", true },
571 { "http://bogussite.com/b", true },
572 { "http://bogussite.com/c", true }
574 ASSERT_NO_FATAL_FAILURE(RunTest(input_2
, std::string(), false, fixup_2
,
575 arraysize(fixup_2
)));
576 EXPECT_EQ(ASCIIToUTF16("/bogussite.com/a"),
577 matches_
.front().inline_autocompletion
);
579 // Adding a TLD to a small number like "56" should result in "www.56.com"
580 // rather than "0.0.0.56.com".
581 const UrlAndLegalDefault fixup_3
[] = {
582 { "http://www.56.com/", true }
584 RunTest(ASCIIToUTF16("56"), "com", true, fixup_3
, arraysize(fixup_3
));
586 // An input looks like a IP address like "127.0.0.1" should result in
587 // "http://127.0.0.1/".
588 const UrlAndLegalDefault fixup_4
[] = {
589 { "http://127.0.0.1/", true }
591 RunTest(ASCIIToUTF16("127.0.0.1"), std::string(), false, fixup_4
,
594 // An number "17173" should result in "http://www.17173.com/" in db.
595 const UrlAndLegalDefault fixup_5
[] = {
596 { "http://www.17173.com/", true }
598 RunTest(ASCIIToUTF16("17173"), std::string(), false, fixup_5
,
602 // Make sure the results for the input 'p' don't change between the first and
604 TEST_F(HistoryURLProviderTest
, EmptyVisits
) {
605 // Wait for history to create the in memory DB.
606 profile_
->BlockUntilHistoryProcessesPendingRequests();
608 AutocompleteInput
input(ASCIIToUTF16("p"), base::string16::npos
,
609 std::string(), GURL(),
610 metrics::OmniboxEventProto::INVALID_SPEC
, false,
612 ChromeAutocompleteSchemeClassifier(profile_
.get()));
613 autocomplete_
->Start(input
, false);
614 // HistoryURLProvider shouldn't be done (waiting on async results).
615 EXPECT_FALSE(autocomplete_
->done());
617 // We should get back an entry for pandora.
618 matches_
= autocomplete_
->matches();
619 ASSERT_GT(matches_
.size(), 0u);
620 EXPECT_EQ(GURL("http://pandora.com/"), matches_
[0].destination_url
);
621 int pandora_relevance
= matches_
[0].relevance
;
623 // Run the message loop. When |autocomplete_| finishes the loop is quit.
624 base::MessageLoop::current()->Run();
625 EXPECT_TRUE(autocomplete_
->done());
626 matches_
= autocomplete_
->matches();
627 ASSERT_GT(matches_
.size(), 0u);
628 EXPECT_EQ(GURL("http://pandora.com/"), matches_
[0].destination_url
);
629 EXPECT_EQ(pandora_relevance
, matches_
[0].relevance
);
632 TEST_F(HistoryURLProviderTestNoDB
, NavigateWithoutDB
) {
633 // Ensure that we will still produce matches for navigation when there is no
635 UrlAndLegalDefault navigation_1
[] = {
636 { "http://test.com/", true }
638 RunTest(ASCIIToUTF16("test.com"), std::string(), false, navigation_1
,
639 arraysize(navigation_1
));
641 UrlAndLegalDefault navigation_2
[] = {
642 { "http://slash/", true }
644 RunTest(ASCIIToUTF16("slash"), std::string(), false, navigation_2
,
645 arraysize(navigation_2
));
647 RunTest(ASCIIToUTF16("this is a query"), std::string(), false, NULL
, 0);
650 TEST_F(HistoryURLProviderTest
, DontAutocompleteOnTrailingWhitespace
) {
651 AutocompleteInput
input(ASCIIToUTF16("slash "), base::string16::npos
,
652 std::string(), GURL(),
653 metrics::OmniboxEventProto::INVALID_SPEC
, false,
655 ChromeAutocompleteSchemeClassifier(profile_
.get()));
656 autocomplete_
->Start(input
, false);
657 if (!autocomplete_
->done())
658 base::MessageLoop::current()->Run();
660 // None of the matches should attempt to autocomplete.
661 matches_
= autocomplete_
->matches();
662 for (size_t i
= 0; i
< matches_
.size(); ++i
) {
663 EXPECT_TRUE(matches_
[i
].inline_autocompletion
.empty());
664 EXPECT_FALSE(matches_
[i
].allowed_to_be_default_match
);
668 TEST_F(HistoryURLProviderTest
, TreatEmailsAsSearches
) {
669 // Visiting foo.com should not make this string be treated as a navigation.
670 // That means the result should be scored around 1200 ("what you typed")
672 const UrlAndLegalDefault expected
[] = {
673 { "http://user@foo.com/", true }
675 ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("user@foo.com"), std::string(),
676 false, expected
, arraysize(expected
)));
677 EXPECT_LE(1200, matches_
[0].relevance
);
678 EXPECT_LT(matches_
[0].relevance
, 1210);
681 TEST_F(HistoryURLProviderTest
, IntranetURLsWithPaths
) {
687 { "fooey/", 1200 }, // 1200 for URL would still navigate by default.
688 { "fooey/a", 1200 }, // 1200 for UNKNOWN would not.
689 { "fooey/a b", 1200 }, // Also UNKNOWN.
693 { "gooey/a b", 1400 },
695 for (size_t i
= 0; i
< arraysize(test_cases
); ++i
) {
696 SCOPED_TRACE(test_cases
[i
].input
);
697 if (test_cases
[i
].relevance
== 0) {
698 RunTest(ASCIIToUTF16(test_cases
[i
].input
), std::string(), false, NULL
, 0);
700 const UrlAndLegalDefault output
[] = {
701 {url_fixer::FixupURL(test_cases
[i
].input
, std::string()).spec(),
703 ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16(test_cases
[i
].input
),
704 std::string(), false, output
, arraysize(output
)));
705 // Actual relevance should be at least what test_cases expects and
706 // and no more than 10 more.
707 EXPECT_LE(test_cases
[i
].relevance
, matches_
[0].relevance
);
708 EXPECT_LT(matches_
[0].relevance
, test_cases
[i
].relevance
+ 10);
713 // Makes sure autocompletion happens for intranet sites that have been
714 // previoulsy visited.
715 TEST_F(HistoryURLProviderTest
, IntranetURLCompletion
) {
716 sort_matches_
= true;
718 const UrlAndLegalDefault expected1
[] = {
719 { "http://intra/three", true },
720 { "http://intra/two", true }
722 ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("intra/t"), std::string(), false,
723 expected1
, arraysize(expected1
)));
724 EXPECT_LE(1410, matches_
[0].relevance
);
725 EXPECT_LT(matches_
[0].relevance
, 1420);
726 EXPECT_EQ(matches_
[0].relevance
- 1, matches_
[1].relevance
);
728 const UrlAndLegalDefault expected2
[] = {
729 { "http://moo/b", true },
730 { "http://moo/bar", true }
732 ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("moo/b"), std::string(), false,
733 expected2
, arraysize(expected2
)));
734 // The url what you typed match should be around 1400, otherwise the
735 // search what you typed match is going to be first.
736 EXPECT_LE(1400, matches_
[0].relevance
);
737 EXPECT_LT(matches_
[0].relevance
, 1410);
739 const UrlAndLegalDefault expected3
[] = {
740 { "http://intra/one", true },
741 { "http://intra/three", true },
742 { "http://intra/two", true }
744 RunTest(ASCIIToUTF16("intra"), std::string(), false, expected3
,
745 arraysize(expected3
));
747 const UrlAndLegalDefault expected4
[] = {
748 { "http://intra/one", true },
749 { "http://intra/three", true },
750 { "http://intra/two", true }
752 RunTest(ASCIIToUTF16("intra/"), std::string(), false, expected4
,
753 arraysize(expected4
));
755 const UrlAndLegalDefault expected5
[] = {
756 { "http://intra/one", true }
758 ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("intra/o"), std::string(), false,
759 expected5
, arraysize(expected5
)));
760 EXPECT_LE(1410, matches_
[0].relevance
);
761 EXPECT_LT(matches_
[0].relevance
, 1420);
763 const UrlAndLegalDefault expected6
[] = {
764 { "http://intra/x", true }
766 ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("intra/x"), std::string(), false,
767 expected6
, arraysize(expected6
)));
768 EXPECT_LE(1400, matches_
[0].relevance
);
769 EXPECT_LT(matches_
[0].relevance
, 1410);
771 const UrlAndLegalDefault expected7
[] = {
772 { "http://typedhost/untypedpath", true }
774 ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16("typedhost/untypedpath"),
775 std::string(), false, expected7
,
776 arraysize(expected7
)));
777 EXPECT_LE(1400, matches_
[0].relevance
);
778 EXPECT_LT(matches_
[0].relevance
, 1410);
781 TEST_F(HistoryURLProviderTest
, CrashDueToFixup
) {
782 // This test passes if we don't crash. The results don't matter.
783 const char* const test_cases
[] = {
788 for (size_t i
= 0; i
< arraysize(test_cases
); ++i
) {
789 AutocompleteInput
input(ASCIIToUTF16(test_cases
[i
]), base::string16::npos
,
790 std::string(), GURL(),
791 metrics::OmniboxEventProto::INVALID_SPEC
,
792 false, false, true, true,
793 ChromeAutocompleteSchemeClassifier(profile_
.get()));
794 autocomplete_
->Start(input
, false);
795 if (!autocomplete_
->done())
796 base::MessageLoop::current()->Run();
800 TEST_F(HistoryURLProviderTest
, CullSearchResults
) {
801 // Set up a default search engine.
802 TemplateURLData data
;
803 data
.SetKeyword(ASCIIToUTF16("TestEngine"));
804 data
.SetURL("http://testsearch.com/?q={searchTerms}");
805 TemplateURLService
* template_url_service
=
806 TemplateURLServiceFactory::GetForProfile(profile_
.get());
807 TemplateURL
* template_url
= new TemplateURL(data
);
808 template_url_service
->Add(template_url
);
809 template_url_service
->SetUserSelectedDefaultSearchProvider(template_url
);
810 template_url_service
->Load();
812 // URLs we will be using, plus the visit counts they will initially get
813 // (the redirect set below will also increment the visit counts). We want
814 // the results to be in A,B,C order. Note also that our visit counts are
815 // all high enough so that domain synthesizing won't get triggered.
820 {"https://testsearch.com/", 30},
821 {"https://testsearch.com/?q=foobar", 20},
822 {"http://foobar.com/", 10}
824 for (size_t i
= 0; i
< arraysize(test_cases
); ++i
) {
825 history_service_
->AddPageWithDetails(GURL(test_cases
[i
].url
),
826 base::UTF8ToUTF16("Title"), test_cases
[i
].count
, test_cases
[i
].count
,
827 Time::Now(), false, history::SOURCE_BROWSED
);
830 // We should not see search URLs when typing a previously used query.
831 const UrlAndLegalDefault expected_when_searching_query
[] = {
832 { test_cases
[2].url
, false }
834 RunTest(ASCIIToUTF16("foobar"), std::string(), true,
835 expected_when_searching_query
, arraysize(expected_when_searching_query
));
837 // We should not see search URLs when typing the search engine name.
838 const UrlAndLegalDefault expected_when_searching_site
[] = {
839 { test_cases
[0].url
, false }
841 RunTest(ASCIIToUTF16("testsearch"), std::string(), true,
842 expected_when_searching_site
, arraysize(expected_when_searching_site
));
845 TEST_F(HistoryURLProviderTest
, SuggestExactInput
) {
846 const size_t npos
= std::string::npos
;
852 const char* contents
;
853 // Offsets of the ACMatchClassifications, terminated by npos.
855 // The index of the ACMatchClassification that should have the MATCH bit
856 // set, npos if no ACMatchClassification should have the MATCH bit set.
857 size_t match_classification_index
;
859 { "http://www.somesite.com", false,
860 "http://www.somesite.com", {0, npos
, npos
}, 0 },
861 { "www.somesite.com", true,
862 "www.somesite.com", {0, npos
, npos
}, 0 },
863 { "www.somesite.com", false,
864 "http://www.somesite.com", {0, 7, npos
}, 1 },
865 { "somesite.com", true,
866 "somesite.com", {0, npos
, npos
}, 0 },
867 { "somesite.com", false,
868 "http://somesite.com", {0, 7, npos
}, 1 },
870 "w", {0, npos
, npos
}, 0 },
872 "http://w", {0, 7, npos
}, 1 },
874 "w.com", {0, npos
, npos
}, 0 },
876 "http://w.com", {0, 7, npos
}, 1 },
878 "www.w.com", {0, npos
, npos
}, 0 },
879 { "www.w.com", false,
880 "http://www.w.com", {0, 7, npos
}, 1 },
881 { "view-source:w", true,
882 "view-source:w", {0, npos
, npos
}, 0 },
883 { "view-source:www.w.com/", true,
884 "view-source:www.w.com", {0, npos
, npos
}, npos
},
885 { "view-source:www.w.com/", false,
886 "view-source:http://www.w.com", {0, npos
, npos
}, npos
},
887 { "view-source:http://www.w.com/", false,
888 "view-source:http://www.w.com", {0, npos
, npos
}, 0 },
889 { " view-source:", true,
890 "view-source:", {0, npos
, npos
}, 0 },
891 { "http:////////w.com", false,
892 "http://w.com", {0, npos
, npos
}, npos
},
893 { " http:////////www.w.com", false,
894 "http://www.w.com", {0, npos
, npos
}, npos
},
895 { "http:a///www.w.com", false,
896 "http://a///www.w.com", {0, npos
, npos
}, npos
},
897 { "mailto://a@b.com", true,
898 "mailto://a@b.com", {0, npos
, npos
}, 0 },
899 { "mailto://a@b.com", false,
900 "mailto://a@b.com", {0, npos
, npos
}, 0 },
902 for (size_t i
= 0; i
< arraysize(test_cases
); ++i
) {
903 SCOPED_TRACE(testing::Message() << "Index " << i
<< " input: "
904 << test_cases
[i
].input
<< ", trim_http: "
905 << test_cases
[i
].trim_http
);
907 AutocompleteInput
input(ASCIIToUTF16(test_cases
[i
].input
),
908 base::string16::npos
, std::string(),
910 metrics::OmniboxEventProto::INVALID_SPEC
, false,
912 ChromeAutocompleteSchemeClassifier(profile_
.get()));
913 AutocompleteMatch
match(autocomplete_
->SuggestExactInput(
914 input
.text(), input
.canonicalized_url(), test_cases
[i
].trim_http
));
915 EXPECT_EQ(ASCIIToUTF16(test_cases
[i
].contents
), match
.contents
);
916 for (size_t match_index
= 0; match_index
< match
.contents_class
.size();
918 EXPECT_EQ(test_cases
[i
].offsets
[match_index
],
919 match
.contents_class
[match_index
].offset
);
920 EXPECT_EQ(ACMatchClassification::URL
|
921 (match_index
== test_cases
[i
].match_classification_index
?
922 ACMatchClassification::MATCH
: 0),
923 match
.contents_class
[match_index
].style
);
925 EXPECT_EQ(npos
, test_cases
[i
].offsets
[match
.contents_class
.size()]);
929 TEST_F(HistoryURLProviderTest
, HUPScoringExperiment
) {
930 HUPScoringParams max_2000_no_time_decay
;
931 max_2000_no_time_decay
.typed_count_buckets
.buckets().push_back(
932 std::make_pair(0.0, 2000));
933 HUPScoringParams max_1250_no_time_decay
;
934 max_1250_no_time_decay
.typed_count_buckets
.buckets().push_back(
935 std::make_pair(0.0, 1250));
936 HUPScoringParams max_1000_no_time_decay
;
937 max_1000_no_time_decay
.typed_count_buckets
.buckets().push_back(
938 std::make_pair(0.0, 1000));
940 HUPScoringParams max_1100_with_time_decay_and_max_cap
;
941 max_1100_with_time_decay_and_max_cap
.typed_count_buckets
.
942 set_relevance_cap(1400);
943 max_1100_with_time_decay_and_max_cap
.typed_count_buckets
.
944 set_half_life_days(16);
945 max_1100_with_time_decay_and_max_cap
.typed_count_buckets
.buckets().push_back(
946 std::make_pair(0.5, 1100));
947 max_1100_with_time_decay_and_max_cap
.typed_count_buckets
.buckets().push_back(
948 std::make_pair(0.24, 200));
949 max_1100_with_time_decay_and_max_cap
.typed_count_buckets
.buckets().push_back(
950 std::make_pair(0.0, 100));
952 HUPScoringParams max_1100_visit_typed_decays
;
953 max_1100_visit_typed_decays
.typed_count_buckets
.set_half_life_days(16);
954 max_1100_visit_typed_decays
.typed_count_buckets
.buckets().push_back(
955 std::make_pair(0.5, 1100));
956 max_1100_visit_typed_decays
.typed_count_buckets
.buckets().push_back(
957 std::make_pair(0.0, 100));
958 max_1100_visit_typed_decays
.visited_count_buckets
.set_half_life_days(16);
959 max_1100_visit_typed_decays
.visited_count_buckets
.buckets().push_back(
960 std::make_pair(0.5, 550));
961 max_1100_visit_typed_decays
.visited_count_buckets
.buckets().push_back(
962 std::make_pair(0.0, 50));
964 const int kMaxMatches
= 3;
967 HUPScoringParams scoring_params
;
968 struct ExpectedMatch
{
970 int control_relevance
;
971 int experiment_relevance
;
973 ExpectedMatch matches
[kMaxMatches
];
975 // Max score 2000 -> no demotion.
976 { "7.com/1", max_2000_no_time_decay
,
977 {{"7.com/1a", 1413, 1413}, {NULL
, 0, 0}, {NULL
, 0, 0}} },
979 // Limit score to 1250/1000 and make sure that the top match is unchanged.
980 { "7.com/1", max_1250_no_time_decay
,
981 {{"7.com/1a", 1413, 1413}, {NULL
, 0, 0}, {NULL
, 0, 0}} },
982 { "7.com/2", max_1250_no_time_decay
,
983 {{"7.com/2a", 1413, 1413}, {"7.com/2b", 1412, 1250}, {NULL
, 0, 0}} },
984 { "7.com/4", max_1000_no_time_decay
,
985 {{"7.com/4", 1203, 1203}, {"7.com/4a", 1202, 1000},
986 {"7.com/4b", 1201, 999}} },
988 // Max relevance cap is 1400 and half-life is 16 days.
989 { "7.com/1", max_1100_with_time_decay_and_max_cap
,
990 {{"7.com/1a", 1413, 1413}, {NULL
, 0, 0}, {NULL
, 0, 0}} },
991 { "7.com/4", max_1100_with_time_decay_and_max_cap
,
992 {{"7.com/4", 1203, 1203}, {"7.com/4a", 1202, 200},
993 {"7.com/4b", 1201, 100}} },
995 // Max relevance cap is 1400 and half-life is 16 days for both visit/typed.
996 { "7.com/5", max_1100_visit_typed_decays
,
997 {{"7.com/5", 1203, 1203}, {"7.com/5a", 1202, 50}, {NULL
, 0, 0}} },
999 for (size_t i
= 0; i
< arraysize(test_cases
); ++i
) {
1000 SCOPED_TRACE(test_cases
[i
].input
);
1001 UrlAndLegalDefault output
[kMaxMatches
];
1003 for (max_matches
= 0; max_matches
< kMaxMatches
; ++max_matches
) {
1004 if (test_cases
[i
].matches
[max_matches
].url
== NULL
)
1006 output
[max_matches
].url
=
1007 url_fixer::FixupURL(test_cases
[i
].matches
[max_matches
].url
,
1008 std::string()).spec();
1009 output
[max_matches
].allowed_to_be_default_match
= true;
1011 autocomplete_
->scoring_params_
= test_cases
[i
].scoring_params
;
1013 // Test the control (scoring disabled).
1014 autocomplete_
->scoring_params_
.experimental_scoring_enabled
= false;
1015 ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16(test_cases
[i
].input
),
1016 std::string(), false, output
, max_matches
));
1017 for (int j
= 0; j
< max_matches
; ++j
) {
1018 EXPECT_EQ(test_cases
[i
].matches
[j
].control_relevance
,
1019 matches_
[j
].relevance
);
1022 // Test the experiment (scoring enabled).
1023 autocomplete_
->scoring_params_
.experimental_scoring_enabled
= true;
1024 ASSERT_NO_FATAL_FAILURE(RunTest(ASCIIToUTF16(test_cases
[i
].input
),
1025 std::string(), false, output
, max_matches
));
1026 for (int j
= 0; j
< max_matches
; ++j
) {
1027 EXPECT_EQ(test_cases
[i
].matches
[j
].experiment_relevance
,
1028 matches_
[j
].relevance
);