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/predictors/autocomplete_action_predictor.h"
11 #include "base/bind.h"
12 #include "base/guid.h"
13 #include "base/i18n/case_conversion.h"
14 #include "base/metrics/histogram.h"
15 #include "base/strings/string_util.h"
16 #include "base/strings/stringprintf.h"
17 #include "base/strings/utf_string_conversions.h"
18 #include "chrome/browser/chrome_notification_types.h"
19 #include "chrome/browser/history/history_service.h"
20 #include "chrome/browser/history/history_service_factory.h"
21 #include "chrome/browser/omnibox/omnibox_log.h"
22 #include "chrome/browser/predictors/autocomplete_action_predictor_factory.h"
23 #include "chrome/browser/predictors/predictor_database.h"
24 #include "chrome/browser/predictors/predictor_database_factory.h"
25 #include "chrome/browser/prerender/prerender_field_trial.h"
26 #include "chrome/browser/prerender/prerender_handle.h"
27 #include "chrome/browser/prerender/prerender_manager.h"
28 #include "chrome/browser/prerender/prerender_manager_factory.h"
29 #include "chrome/browser/profiles/profile.h"
30 #include "chrome/browser/ui/omnibox/omnibox_popup_model.h"
31 #include "components/history/core/browser/in_memory_database.h"
32 #include "components/omnibox/autocomplete_match.h"
33 #include "components/omnibox/autocomplete_result.h"
34 #include "content/public/browser/browser_thread.h"
35 #include "content/public/browser/notification_details.h"
36 #include "content/public/browser/notification_service.h"
37 #include "content/public/browser/notification_source.h"
41 const float kConfidenceCutoff
[] = {
46 static_assert(arraysize(kConfidenceCutoff
) ==
47 predictors::AutocompleteActionPredictor::LAST_PREDICT_ACTION
,
48 "kConfidenceCutoff count should match LAST_PREDICT_ACTION");
50 const size_t kMinimumUserTextLength
= 1;
51 const int kMinimumNumberOfHits
= 3;
55 DATABASE_ACTION_UPDATE
,
56 DATABASE_ACTION_DELETE_SOME
,
57 DATABASE_ACTION_DELETE_ALL
,
63 namespace predictors
{
65 const int AutocompleteActionPredictor::kMaximumDaysToKeepEntry
= 14;
67 AutocompleteActionPredictor::AutocompleteActionPredictor(Profile
* profile
)
69 main_profile_predictor_(NULL
),
70 incognito_predictor_(NULL
),
72 history_service_observer_(this) {
73 if (profile_
->IsOffTheRecord()) {
74 main_profile_predictor_
= AutocompleteActionPredictorFactory::GetForProfile(
75 profile_
->GetOriginalProfile());
76 DCHECK(main_profile_predictor_
);
77 main_profile_predictor_
->incognito_predictor_
= this;
78 if (main_profile_predictor_
->initialized_
)
79 CopyFromMainProfile();
81 // Request the in-memory database from the history to force it to load so
82 // it's available as soon as possible.
83 HistoryService
* history_service
= HistoryServiceFactory::GetForProfile(
84 profile_
, ServiceAccessType::EXPLICIT_ACCESS
);
86 history_service
->InMemoryDatabase();
89 PredictorDatabaseFactory::GetForProfile(profile_
)->autocomplete_table();
91 // Observe all main frame loads so we can wait for the first to complete
92 // before accessing DB and IO threads to build the local cache.
93 notification_registrar_
.Add(this,
94 content::NOTIFICATION_LOAD_COMPLETED_MAIN_FRAME
,
95 content::NotificationService::AllSources());
99 AutocompleteActionPredictor::~AutocompleteActionPredictor() {
100 if (main_profile_predictor_
)
101 main_profile_predictor_
->incognito_predictor_
= NULL
;
102 else if (incognito_predictor_
)
103 incognito_predictor_
->main_profile_predictor_
= NULL
;
104 if (prerender_handle_
.get())
105 prerender_handle_
->OnCancel();
108 void AutocompleteActionPredictor::RegisterTransitionalMatches(
109 const base::string16
& user_text
,
110 const AutocompleteResult
& result
) {
111 if (user_text
.length() < kMinimumUserTextLength
)
113 const base::string16
lower_user_text(base::i18n::ToLower(user_text
));
115 // Merge this in to an existing match if we already saw |user_text|
116 std::vector
<TransitionalMatch
>::iterator match_it
=
117 std::find(transitional_matches_
.begin(), transitional_matches_
.end(),
120 if (match_it
== transitional_matches_
.end()) {
121 TransitionalMatch transitional_match
;
122 transitional_match
.user_text
= lower_user_text
;
123 match_it
= transitional_matches_
.insert(transitional_matches_
.end(),
127 for (const auto& i
: result
) {
128 if (std::find(match_it
->urls
.begin(), match_it
->urls
.end(),
129 i
.destination_url
) == match_it
->urls
.end()) {
130 match_it
->urls
.push_back(i
.destination_url
);
135 void AutocompleteActionPredictor::ClearTransitionalMatches() {
136 transitional_matches_
.clear();
139 void AutocompleteActionPredictor::CancelPrerender() {
140 // If the prerender has already been abandoned, leave it to its own timeout;
141 // this normally gets called immediately after OnOmniboxOpenedUrl.
142 if (prerender_handle_
&& !prerender_handle_
->IsAbandoned()) {
143 prerender_handle_
->OnCancel();
144 prerender_handle_
.reset();
148 void AutocompleteActionPredictor::StartPrerendering(
150 content::SessionStorageNamespace
* session_storage_namespace
,
151 const gfx::Size
& size
) {
152 // Only cancel the old prerender after starting the new one, so if the URLs
153 // are the same, the underlying prerender will be reused.
154 scoped_ptr
<prerender::PrerenderHandle
> old_prerender_handle(
155 prerender_handle_
.release());
156 if (prerender::PrerenderManager
* prerender_manager
=
157 prerender::PrerenderManagerFactory::GetForProfile(profile_
)) {
158 prerender_handle_
.reset(prerender_manager
->AddPrerenderFromOmnibox(
159 url
, session_storage_namespace
, size
));
161 if (old_prerender_handle
)
162 old_prerender_handle
->OnCancel();
165 // Given a match, return a recommended action.
166 AutocompleteActionPredictor::Action
167 AutocompleteActionPredictor::RecommendAction(
168 const base::string16
& user_text
,
169 const AutocompleteMatch
& match
) const {
170 bool is_in_db
= false;
171 const double confidence
= CalculateConfidence(user_text
, match
, &is_in_db
);
172 DCHECK(confidence
>= 0.0 && confidence
<= 1.0);
174 UMA_HISTOGRAM_BOOLEAN("AutocompleteActionPredictor.MatchIsInDb", is_in_db
);
177 // Multiple enties with the same URL are fine as the confidence may be
179 tracked_urls_
.push_back(std::make_pair(match
.destination_url
, confidence
));
180 UMA_HISTOGRAM_COUNTS_100("AutocompleteActionPredictor.Confidence",
184 // Map the confidence to an action.
185 Action action
= ACTION_NONE
;
186 for (int i
= 0; i
< LAST_PREDICT_ACTION
; ++i
) {
187 if (confidence
>= kConfidenceCutoff
[i
]) {
188 action
= static_cast<Action
>(i
);
193 // Downgrade prerender to preconnect if this is a search match or if omnibox
194 // prerendering is disabled. There are cases when Instant will not handle a
195 // search suggestion and in those cases it would be good to prerender the
196 // search results, however search engines have not been set up to correctly
197 // handle being prerendered and until they are we should avoid it.
198 // http://crbug.com/117495
199 if (action
== ACTION_PRERENDER
&&
200 (AutocompleteMatch::IsSearchType(match
.type
) ||
201 !prerender::IsOmniboxEnabled(profile_
))) {
202 action
= ACTION_PRECONNECT
;
208 // Return true if the suggestion type warrants a TCP/IP preconnection.
209 // i.e., it is now quite likely that the user will select the related domain.
211 bool AutocompleteActionPredictor::IsPreconnectable(
212 const AutocompleteMatch
& match
) {
213 return AutocompleteMatch::IsSearchType(match
.type
);
216 bool AutocompleteActionPredictor::IsPrerenderAbandonedForTesting() {
217 return prerender_handle_
&& prerender_handle_
->IsAbandoned();
220 void AutocompleteActionPredictor::Observe(
222 const content::NotificationSource
& source
,
223 const content::NotificationDetails
& details
) {
225 case content::NOTIFICATION_LOAD_COMPLETED_MAIN_FRAME
:
226 CreateLocalCachesFromDatabase();
227 notification_registrar_
.Remove(
229 content::NOTIFICATION_LOAD_COMPLETED_MAIN_FRAME
,
230 content::NotificationService::AllSources());
232 case chrome::NOTIFICATION_OMNIBOX_OPENED_URL
: {
233 DCHECK(initialized_
);
235 // TODO(dominich): This doesn't need to be synchronous. Investigate
236 // posting it as a task to be run later.
237 OnOmniboxOpenedUrl(*content::Details
<OmniboxLog
>(details
).ptr());
242 NOTREACHED() << "Unexpected notification observed.";
247 void AutocompleteActionPredictor::CreateLocalCachesFromDatabase() {
248 // Create local caches using the database as loaded. We will garbage collect
249 // rows from the caches and the database once the history service is
251 std::vector
<AutocompleteActionPredictorTable::Row
>* rows
=
252 new std::vector
<AutocompleteActionPredictorTable::Row
>();
253 content::BrowserThread::PostTaskAndReply(content::BrowserThread::DB
,
255 base::Bind(&AutocompleteActionPredictorTable::GetAllRows
, table_
, rows
),
256 base::Bind(&AutocompleteActionPredictor::CreateCaches
, AsWeakPtr(),
260 void AutocompleteActionPredictor::DeleteAllRows() {
265 db_id_cache_
.clear();
268 content::BrowserThread::PostTask(content::BrowserThread::DB
, FROM_HERE
,
269 base::Bind(&AutocompleteActionPredictorTable::DeleteAllRows
,
273 UMA_HISTOGRAM_ENUMERATION("AutocompleteActionPredictor.DatabaseAction",
274 DATABASE_ACTION_DELETE_ALL
, DATABASE_ACTION_COUNT
);
277 void AutocompleteActionPredictor::DeleteRowsWithURLs(
278 const history::URLRows
& rows
) {
282 std::vector
<AutocompleteActionPredictorTable::Row::Id
> id_list
;
284 for (DBCacheMap::iterator it
= db_cache_
.begin(); it
!= db_cache_
.end();) {
285 if (std::find_if(rows
.begin(), rows
.end(),
286 history::URLRow::URLRowHasURL(it
->first
.url
)) != rows
.end()) {
287 const DBIdCacheMap::iterator id_it
= db_id_cache_
.find(it
->first
);
288 DCHECK(id_it
!= db_id_cache_
.end());
289 id_list
.push_back(id_it
->second
);
290 db_id_cache_
.erase(id_it
);
291 db_cache_
.erase(it
++);
298 content::BrowserThread::PostTask(content::BrowserThread::DB
, FROM_HERE
,
299 base::Bind(&AutocompleteActionPredictorTable::DeleteRows
, table_
,
303 UMA_HISTOGRAM_ENUMERATION("AutocompleteActionPredictor.DatabaseAction",
304 DATABASE_ACTION_DELETE_SOME
, DATABASE_ACTION_COUNT
);
307 void AutocompleteActionPredictor::OnOmniboxOpenedUrl(const OmniboxLog
& log
) {
308 if (log
.text
.length() < kMinimumUserTextLength
)
311 // Do not attempt to learn from omnibox interactions where the omnibox
312 // dropdown is closed. In these cases the user text (|log.text|) that we
313 // learn from is either empty or effectively identical to the destination
314 // string. In either case, it can't teach us much. Also do not attempt
315 // to learn from paste-and-go actions even if the popup is open because
316 // the paste-and-go destination has no relation to whatever text the user
318 if (!log
.is_popup_open
|| log
.is_paste_and_go
)
321 // Abandon the current prerender. If it is to be used, it will be used very
322 // soon, so use the lower timeout.
323 if (prerender_handle_
) {
324 prerender_handle_
->OnNavigateAway();
325 // Don't release |prerender_handle_| so it is canceled if it survives to the
326 // next StartPrerendering call.
329 UMA_HISTOGRAM_BOOLEAN(
330 base::StringPrintf("Prerender.OmniboxNavigationsCouldPrerender%s",
331 prerender::PrerenderManager::GetModeString()).c_str(),
332 prerender::IsOmniboxEnabled(profile_
));
334 const AutocompleteMatch
& match
= log
.result
.match_at(log
.selected_index
);
335 const GURL
& opened_url
= match
.destination_url
;
336 const base::string16
lower_user_text(base::i18n::ToLower(log
.text
));
338 // Traverse transitional matches for those that have a user_text that is a
339 // prefix of |lower_user_text|.
340 std::vector
<AutocompleteActionPredictorTable::Row
> rows_to_add
;
341 std::vector
<AutocompleteActionPredictorTable::Row
> rows_to_update
;
343 for (std::vector
<TransitionalMatch
>::const_iterator it
=
344 transitional_matches_
.begin(); it
!= transitional_matches_
.end();
346 if (!StartsWith(lower_user_text
, it
->user_text
, true))
349 // Add entries to the database for those matches.
350 for (std::vector
<GURL
>::const_iterator url_it
= it
->urls
.begin();
351 url_it
!= it
->urls
.end(); ++url_it
) {
352 DCHECK(it
->user_text
.length() >= kMinimumUserTextLength
);
353 const DBCacheKey key
= { it
->user_text
, *url_it
};
354 const bool is_hit
= (*url_it
== opened_url
);
356 AutocompleteActionPredictorTable::Row row
;
357 row
.user_text
= key
.user_text
;
360 DBCacheMap::iterator it
= db_cache_
.find(key
);
361 if (it
== db_cache_
.end()) {
362 row
.id
= base::GenerateGUID();
363 row
.number_of_hits
= is_hit
? 1 : 0;
364 row
.number_of_misses
= is_hit
? 0 : 1;
366 rows_to_add
.push_back(row
);
368 DCHECK(db_id_cache_
.find(key
) != db_id_cache_
.end());
369 row
.id
= db_id_cache_
.find(key
)->second
;
370 row
.number_of_hits
= it
->second
.number_of_hits
+ (is_hit
? 1 : 0);
371 row
.number_of_misses
= it
->second
.number_of_misses
+ (is_hit
? 0 : 1);
373 rows_to_update
.push_back(row
);
377 if (rows_to_add
.size() > 0 || rows_to_update
.size() > 0)
378 AddAndUpdateRows(rows_to_add
, rows_to_update
);
380 ClearTransitionalMatches();
382 // Check against tracked urls and log accuracy for the confidence we
384 for (std::vector
<std::pair
<GURL
, double> >::const_iterator it
=
385 tracked_urls_
.begin(); it
!= tracked_urls_
.end();
387 if (opened_url
== it
->first
) {
388 UMA_HISTOGRAM_COUNTS_100("AutocompleteActionPredictor.AccurateCount",
392 tracked_urls_
.clear();
395 void AutocompleteActionPredictor::AddAndUpdateRows(
396 const AutocompleteActionPredictorTable::Rows
& rows_to_add
,
397 const AutocompleteActionPredictorTable::Rows
& rows_to_update
) {
401 for (AutocompleteActionPredictorTable::Rows::const_iterator it
=
402 rows_to_add
.begin(); it
!= rows_to_add
.end(); ++it
) {
403 const DBCacheKey key
= { it
->user_text
, it
->url
};
404 DBCacheValue value
= { it
->number_of_hits
, it
->number_of_misses
};
406 DCHECK(db_cache_
.find(key
) == db_cache_
.end());
408 db_cache_
[key
] = value
;
409 db_id_cache_
[key
] = it
->id
;
410 UMA_HISTOGRAM_ENUMERATION("AutocompleteActionPredictor.DatabaseAction",
411 DATABASE_ACTION_ADD
, DATABASE_ACTION_COUNT
);
413 for (AutocompleteActionPredictorTable::Rows::const_iterator it
=
414 rows_to_update
.begin(); it
!= rows_to_update
.end(); ++it
) {
415 const DBCacheKey key
= { it
->user_text
, it
->url
};
417 DBCacheMap::iterator db_it
= db_cache_
.find(key
);
418 DCHECK(db_it
!= db_cache_
.end());
419 DCHECK(db_id_cache_
.find(key
) != db_id_cache_
.end());
421 db_it
->second
.number_of_hits
= it
->number_of_hits
;
422 db_it
->second
.number_of_misses
= it
->number_of_misses
;
423 UMA_HISTOGRAM_ENUMERATION("AutocompleteActionPredictor.DatabaseAction",
424 DATABASE_ACTION_UPDATE
, DATABASE_ACTION_COUNT
);
428 content::BrowserThread::PostTask(content::BrowserThread::DB
, FROM_HERE
,
429 base::Bind(&AutocompleteActionPredictorTable::AddAndUpdateRows
,
430 table_
, rows_to_add
, rows_to_update
));
434 void AutocompleteActionPredictor::CreateCaches(
435 std::vector
<AutocompleteActionPredictorTable::Row
>* rows
) {
436 CHECK(content::BrowserThread::CurrentlyOn(content::BrowserThread::UI
));
437 DCHECK(!profile_
->IsOffTheRecord());
438 DCHECK(!initialized_
);
439 DCHECK(db_cache_
.empty());
440 DCHECK(db_id_cache_
.empty());
442 for (std::vector
<AutocompleteActionPredictorTable::Row
>::const_iterator it
=
443 rows
->begin(); it
!= rows
->end(); ++it
) {
444 const DBCacheKey key
= { it
->user_text
, it
->url
};
445 const DBCacheValue value
= { it
->number_of_hits
, it
->number_of_misses
};
446 db_cache_
[key
] = value
;
447 db_id_cache_
[key
] = it
->id
;
450 // If the history service is ready, delete any old or invalid entries.
451 HistoryService
* history_service
= HistoryServiceFactory::GetForProfile(
452 profile_
, ServiceAccessType::EXPLICIT_ACCESS
);
453 if (!TryDeleteOldEntries(history_service
)) {
454 // Wait for the notification that the history service is ready and the URL
457 history_service_observer_
.Add(history_service
);
461 bool AutocompleteActionPredictor::TryDeleteOldEntries(HistoryService
* service
) {
462 CHECK(content::BrowserThread::CurrentlyOn(content::BrowserThread::UI
));
463 DCHECK(!profile_
->IsOffTheRecord());
464 DCHECK(!initialized_
);
469 history::URLDatabase
* url_db
= service
->InMemoryDatabase();
473 DeleteOldEntries(url_db
);
477 void AutocompleteActionPredictor::DeleteOldEntries(
478 history::URLDatabase
* url_db
) {
479 CHECK(content::BrowserThread::CurrentlyOn(content::BrowserThread::UI
));
480 DCHECK(!profile_
->IsOffTheRecord());
481 DCHECK(!initialized_
);
482 DCHECK(table_
.get());
484 std::vector
<AutocompleteActionPredictorTable::Row::Id
> ids_to_delete
;
485 DeleteOldIdsFromCaches(url_db
, &ids_to_delete
);
487 content::BrowserThread::PostTask(content::BrowserThread::DB
, FROM_HERE
,
488 base::Bind(&AutocompleteActionPredictorTable::DeleteRows
, table_
,
491 FinishInitialization();
492 if (incognito_predictor_
)
493 incognito_predictor_
->CopyFromMainProfile();
496 void AutocompleteActionPredictor::DeleteOldIdsFromCaches(
497 history::URLDatabase
* url_db
,
498 std::vector
<AutocompleteActionPredictorTable::Row::Id
>* id_list
) {
499 CHECK(content::BrowserThread::CurrentlyOn(content::BrowserThread::UI
));
500 DCHECK(!profile_
->IsOffTheRecord());
501 DCHECK(!initialized_
);
506 for (DBCacheMap::iterator it
= db_cache_
.begin(); it
!= db_cache_
.end();) {
507 history::URLRow url_row
;
509 if ((url_db
->GetRowForURL(it
->first
.url
, &url_row
) == 0) ||
510 ((base::Time::Now() - url_row
.last_visit()).InDays() >
511 kMaximumDaysToKeepEntry
)) {
512 const DBIdCacheMap::iterator id_it
= db_id_cache_
.find(it
->first
);
513 DCHECK(id_it
!= db_id_cache_
.end());
514 id_list
->push_back(id_it
->second
);
515 db_id_cache_
.erase(id_it
);
516 db_cache_
.erase(it
++);
523 void AutocompleteActionPredictor::CopyFromMainProfile() {
524 CHECK(content::BrowserThread::CurrentlyOn(content::BrowserThread::UI
));
525 DCHECK(profile_
->IsOffTheRecord());
526 DCHECK(!initialized_
);
527 DCHECK(main_profile_predictor_
);
528 DCHECK(main_profile_predictor_
->initialized_
);
530 db_cache_
= main_profile_predictor_
->db_cache_
;
531 db_id_cache_
= main_profile_predictor_
->db_id_cache_
;
532 FinishInitialization();
535 void AutocompleteActionPredictor::FinishInitialization() {
536 CHECK(content::BrowserThread::CurrentlyOn(content::BrowserThread::UI
));
537 DCHECK(!initialized_
);
539 // Incognito and normal profiles should listen only to omnibox notifications
540 // from their own profile, but both should listen to history deletions from
541 // the main profile, since opening the history page in either case actually
542 // opens the non-incognito history (and lets users delete from there).
543 notification_registrar_
.Add(this, chrome::NOTIFICATION_OMNIBOX_OPENED_URL
,
544 content::Source
<Profile
>(profile_
));
548 double AutocompleteActionPredictor::CalculateConfidence(
549 const base::string16
& user_text
,
550 const AutocompleteMatch
& match
,
551 bool* is_in_db
) const {
552 const DBCacheKey key
= { user_text
, match
.destination_url
};
555 if (user_text
.length() < kMinimumUserTextLength
)
558 const DBCacheMap::const_iterator iter
= db_cache_
.find(key
);
559 if (iter
== db_cache_
.end())
563 return CalculateConfidenceForDbEntry(iter
);
566 double AutocompleteActionPredictor::CalculateConfidenceForDbEntry(
567 DBCacheMap::const_iterator iter
) const {
568 const DBCacheValue
& value
= iter
->second
;
569 if (value
.number_of_hits
< kMinimumNumberOfHits
)
572 const double number_of_hits
= static_cast<double>(value
.number_of_hits
);
573 return number_of_hits
/ (number_of_hits
+ value
.number_of_misses
);
576 void AutocompleteActionPredictor::Shutdown() {
577 history_service_observer_
.RemoveAll();
580 void AutocompleteActionPredictor::OnURLsDeleted(
581 HistoryService
* history_service
,
584 const history::URLRows
& deleted_rows
,
585 const std::set
<GURL
>& favicon_urls
) {
592 DeleteRowsWithURLs(deleted_rows
);
595 void AutocompleteActionPredictor::OnHistoryServiceLoaded(
596 HistoryService
* history_service
) {
597 TryDeleteOldEntries(history_service
);
598 history_service_observer_
.Remove(history_service
);
601 AutocompleteActionPredictor::TransitionalMatch::TransitionalMatch() {
604 AutocompleteActionPredictor::TransitionalMatch::~TransitionalMatch() {
607 } // namespace predictors