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_factory.h"
20 #include "chrome/browser/predictors/autocomplete_action_predictor_factory.h"
21 #include "chrome/browser/predictors/predictor_database.h"
22 #include "chrome/browser/predictors/predictor_database_factory.h"
23 #include "chrome/browser/prerender/prerender_field_trial.h"
24 #include "chrome/browser/prerender/prerender_handle.h"
25 #include "chrome/browser/prerender/prerender_manager.h"
26 #include "chrome/browser/prerender/prerender_manager_factory.h"
27 #include "chrome/browser/profiles/profile.h"
28 #include "chrome/browser/ui/omnibox/omnibox_popup_model.h"
29 #include "components/history/core/browser/history_service.h"
30 #include "components/history/core/browser/in_memory_database.h"
31 #include "components/omnibox/autocomplete_match.h"
32 #include "components/omnibox/autocomplete_result.h"
33 #include "components/omnibox/omnibox_log.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 history::HistoryService
* history_service
=
84 HistoryServiceFactory::GetForProfile(
85 profile_
, ServiceAccessType::EXPLICIT_ACCESS
);
87 history_service
->InMemoryDatabase();
90 PredictorDatabaseFactory::GetForProfile(profile_
)->autocomplete_table();
92 // Observe all main frame loads so we can wait for the first to complete
93 // before accessing DB and IO threads to build the local cache.
94 notification_registrar_
.Add(this,
95 content::NOTIFICATION_LOAD_COMPLETED_MAIN_FRAME
,
96 content::NotificationService::AllSources());
100 AutocompleteActionPredictor::~AutocompleteActionPredictor() {
101 if (main_profile_predictor_
)
102 main_profile_predictor_
->incognito_predictor_
= NULL
;
103 else if (incognito_predictor_
)
104 incognito_predictor_
->main_profile_predictor_
= NULL
;
105 if (prerender_handle_
.get())
106 prerender_handle_
->OnCancel();
109 void AutocompleteActionPredictor::RegisterTransitionalMatches(
110 const base::string16
& user_text
,
111 const AutocompleteResult
& result
) {
112 if (user_text
.length() < kMinimumUserTextLength
)
114 const base::string16
lower_user_text(base::i18n::ToLower(user_text
));
116 // Merge this in to an existing match if we already saw |user_text|
117 std::vector
<TransitionalMatch
>::iterator match_it
=
118 std::find(transitional_matches_
.begin(), transitional_matches_
.end(),
121 if (match_it
== transitional_matches_
.end()) {
122 TransitionalMatch transitional_match
;
123 transitional_match
.user_text
= lower_user_text
;
124 match_it
= transitional_matches_
.insert(transitional_matches_
.end(),
128 for (const auto& i
: result
) {
129 if (std::find(match_it
->urls
.begin(), match_it
->urls
.end(),
130 i
.destination_url
) == match_it
->urls
.end()) {
131 match_it
->urls
.push_back(i
.destination_url
);
136 void AutocompleteActionPredictor::ClearTransitionalMatches() {
137 transitional_matches_
.clear();
140 void AutocompleteActionPredictor::CancelPrerender() {
141 // If the prerender has already been abandoned, leave it to its own timeout;
142 // this normally gets called immediately after OnOmniboxOpenedUrl.
143 if (prerender_handle_
&& !prerender_handle_
->IsAbandoned()) {
144 prerender_handle_
->OnCancel();
145 prerender_handle_
.reset();
149 void AutocompleteActionPredictor::StartPrerendering(
151 content::SessionStorageNamespace
* session_storage_namespace
,
152 const gfx::Size
& size
) {
153 // Only cancel the old prerender after starting the new one, so if the URLs
154 // are the same, the underlying prerender will be reused.
155 scoped_ptr
<prerender::PrerenderHandle
> old_prerender_handle(
156 prerender_handle_
.release());
157 if (prerender::PrerenderManager
* prerender_manager
=
158 prerender::PrerenderManagerFactory::GetForProfile(profile_
)) {
159 prerender_handle_
.reset(prerender_manager
->AddPrerenderFromOmnibox(
160 url
, session_storage_namespace
, size
));
162 if (old_prerender_handle
)
163 old_prerender_handle
->OnCancel();
166 // Given a match, return a recommended action.
167 AutocompleteActionPredictor::Action
168 AutocompleteActionPredictor::RecommendAction(
169 const base::string16
& user_text
,
170 const AutocompleteMatch
& match
) const {
171 bool is_in_db
= false;
172 const double confidence
= CalculateConfidence(user_text
, match
, &is_in_db
);
173 DCHECK(confidence
>= 0.0 && confidence
<= 1.0);
175 UMA_HISTOGRAM_BOOLEAN("AutocompleteActionPredictor.MatchIsInDb", is_in_db
);
178 // Multiple enties with the same URL are fine as the confidence may be
180 tracked_urls_
.push_back(std::make_pair(match
.destination_url
, confidence
));
181 UMA_HISTOGRAM_COUNTS_100("AutocompleteActionPredictor.Confidence",
185 // Map the confidence to an action.
186 Action action
= ACTION_NONE
;
187 for (int i
= 0; i
< LAST_PREDICT_ACTION
; ++i
) {
188 if (confidence
>= kConfidenceCutoff
[i
]) {
189 action
= static_cast<Action
>(i
);
194 // Downgrade prerender to preconnect if this is a search match or if omnibox
195 // prerendering is disabled. There are cases when Instant will not handle a
196 // search suggestion and in those cases it would be good to prerender the
197 // search results, however search engines have not been set up to correctly
198 // handle being prerendered and until they are we should avoid it.
199 // http://crbug.com/117495
200 if (action
== ACTION_PRERENDER
&&
201 (AutocompleteMatch::IsSearchType(match
.type
) ||
202 !prerender::IsOmniboxEnabled(profile_
))) {
203 action
= ACTION_PRECONNECT
;
209 // Return true if the suggestion type warrants a TCP/IP preconnection.
210 // i.e., it is now quite likely that the user will select the related domain.
212 bool AutocompleteActionPredictor::IsPreconnectable(
213 const AutocompleteMatch
& match
) {
214 return AutocompleteMatch::IsSearchType(match
.type
);
217 bool AutocompleteActionPredictor::IsPrerenderAbandonedForTesting() {
218 return prerender_handle_
&& prerender_handle_
->IsAbandoned();
221 void AutocompleteActionPredictor::Observe(
223 const content::NotificationSource
& source
,
224 const content::NotificationDetails
& details
) {
226 case content::NOTIFICATION_LOAD_COMPLETED_MAIN_FRAME
:
227 CreateLocalCachesFromDatabase();
228 notification_registrar_
.Remove(
230 content::NOTIFICATION_LOAD_COMPLETED_MAIN_FRAME
,
231 content::NotificationService::AllSources());
233 case chrome::NOTIFICATION_OMNIBOX_OPENED_URL
: {
234 DCHECK(initialized_
);
236 // TODO(dominich): This doesn't need to be synchronous. Investigate
237 // posting it as a task to be run later.
238 OnOmniboxOpenedUrl(*content::Details
<OmniboxLog
>(details
).ptr());
243 NOTREACHED() << "Unexpected notification observed.";
248 void AutocompleteActionPredictor::CreateLocalCachesFromDatabase() {
249 // Create local caches using the database as loaded. We will garbage collect
250 // rows from the caches and the database once the history service is
252 std::vector
<AutocompleteActionPredictorTable::Row
>* rows
=
253 new std::vector
<AutocompleteActionPredictorTable::Row
>();
254 content::BrowserThread::PostTaskAndReply(content::BrowserThread::DB
,
256 base::Bind(&AutocompleteActionPredictorTable::GetAllRows
, table_
, rows
),
257 base::Bind(&AutocompleteActionPredictor::CreateCaches
, AsWeakPtr(),
261 void AutocompleteActionPredictor::DeleteAllRows() {
266 db_id_cache_
.clear();
269 content::BrowserThread::PostTask(content::BrowserThread::DB
, FROM_HERE
,
270 base::Bind(&AutocompleteActionPredictorTable::DeleteAllRows
,
274 UMA_HISTOGRAM_ENUMERATION("AutocompleteActionPredictor.DatabaseAction",
275 DATABASE_ACTION_DELETE_ALL
, DATABASE_ACTION_COUNT
);
278 void AutocompleteActionPredictor::DeleteRowsWithURLs(
279 const history::URLRows
& rows
) {
283 std::vector
<AutocompleteActionPredictorTable::Row::Id
> id_list
;
285 for (DBCacheMap::iterator it
= db_cache_
.begin(); it
!= db_cache_
.end();) {
286 if (std::find_if(rows
.begin(), rows
.end(),
287 history::URLRow::URLRowHasURL(it
->first
.url
)) != rows
.end()) {
288 const DBIdCacheMap::iterator id_it
= db_id_cache_
.find(it
->first
);
289 DCHECK(id_it
!= db_id_cache_
.end());
290 id_list
.push_back(id_it
->second
);
291 db_id_cache_
.erase(id_it
);
292 db_cache_
.erase(it
++);
299 content::BrowserThread::PostTask(content::BrowserThread::DB
, FROM_HERE
,
300 base::Bind(&AutocompleteActionPredictorTable::DeleteRows
, table_
,
304 UMA_HISTOGRAM_ENUMERATION("AutocompleteActionPredictor.DatabaseAction",
305 DATABASE_ACTION_DELETE_SOME
, DATABASE_ACTION_COUNT
);
308 void AutocompleteActionPredictor::OnOmniboxOpenedUrl(const OmniboxLog
& log
) {
309 if (log
.text
.length() < kMinimumUserTextLength
)
312 // Do not attempt to learn from omnibox interactions where the omnibox
313 // dropdown is closed. In these cases the user text (|log.text|) that we
314 // learn from is either empty or effectively identical to the destination
315 // string. In either case, it can't teach us much. Also do not attempt
316 // to learn from paste-and-go actions even if the popup is open because
317 // the paste-and-go destination has no relation to whatever text the user
319 if (!log
.is_popup_open
|| log
.is_paste_and_go
)
322 // Abandon the current prerender. If it is to be used, it will be used very
323 // soon, so use the lower timeout.
324 if (prerender_handle_
) {
325 prerender_handle_
->OnNavigateAway();
326 // Don't release |prerender_handle_| so it is canceled if it survives to the
327 // next StartPrerendering call.
330 UMA_HISTOGRAM_BOOLEAN(
331 base::StringPrintf("Prerender.OmniboxNavigationsCouldPrerender%s",
332 prerender::PrerenderManager::GetModeString()).c_str(),
333 prerender::IsOmniboxEnabled(profile_
));
335 const AutocompleteMatch
& match
= log
.result
.match_at(log
.selected_index
);
336 const GURL
& opened_url
= match
.destination_url
;
337 const base::string16
lower_user_text(base::i18n::ToLower(log
.text
));
339 // Traverse transitional matches for those that have a user_text that is a
340 // prefix of |lower_user_text|.
341 std::vector
<AutocompleteActionPredictorTable::Row
> rows_to_add
;
342 std::vector
<AutocompleteActionPredictorTable::Row
> rows_to_update
;
344 for (std::vector
<TransitionalMatch
>::const_iterator it
=
345 transitional_matches_
.begin(); it
!= transitional_matches_
.end();
347 if (!base::StartsWith(lower_user_text
, it
->user_text
, true))
350 // Add entries to the database for those matches.
351 for (std::vector
<GURL
>::const_iterator url_it
= it
->urls
.begin();
352 url_it
!= it
->urls
.end(); ++url_it
) {
353 DCHECK(it
->user_text
.length() >= kMinimumUserTextLength
);
354 const DBCacheKey key
= { it
->user_text
, *url_it
};
355 const bool is_hit
= (*url_it
== opened_url
);
357 AutocompleteActionPredictorTable::Row row
;
358 row
.user_text
= key
.user_text
;
361 DBCacheMap::iterator it
= db_cache_
.find(key
);
362 if (it
== db_cache_
.end()) {
363 row
.id
= base::GenerateGUID();
364 row
.number_of_hits
= is_hit
? 1 : 0;
365 row
.number_of_misses
= is_hit
? 0 : 1;
367 rows_to_add
.push_back(row
);
369 DCHECK(db_id_cache_
.find(key
) != db_id_cache_
.end());
370 row
.id
= db_id_cache_
.find(key
)->second
;
371 row
.number_of_hits
= it
->second
.number_of_hits
+ (is_hit
? 1 : 0);
372 row
.number_of_misses
= it
->second
.number_of_misses
+ (is_hit
? 0 : 1);
374 rows_to_update
.push_back(row
);
378 if (rows_to_add
.size() > 0 || rows_to_update
.size() > 0)
379 AddAndUpdateRows(rows_to_add
, rows_to_update
);
381 ClearTransitionalMatches();
383 // Check against tracked urls and log accuracy for the confidence we
385 for (std::vector
<std::pair
<GURL
, double> >::const_iterator it
=
386 tracked_urls_
.begin(); it
!= tracked_urls_
.end();
388 if (opened_url
== it
->first
) {
389 UMA_HISTOGRAM_COUNTS_100("AutocompleteActionPredictor.AccurateCount",
393 tracked_urls_
.clear();
396 void AutocompleteActionPredictor::AddAndUpdateRows(
397 const AutocompleteActionPredictorTable::Rows
& rows_to_add
,
398 const AutocompleteActionPredictorTable::Rows
& rows_to_update
) {
402 for (AutocompleteActionPredictorTable::Rows::const_iterator it
=
403 rows_to_add
.begin(); it
!= rows_to_add
.end(); ++it
) {
404 const DBCacheKey key
= { it
->user_text
, it
->url
};
405 DBCacheValue value
= { it
->number_of_hits
, it
->number_of_misses
};
407 DCHECK(db_cache_
.find(key
) == db_cache_
.end());
409 db_cache_
[key
] = value
;
410 db_id_cache_
[key
] = it
->id
;
411 UMA_HISTOGRAM_ENUMERATION("AutocompleteActionPredictor.DatabaseAction",
412 DATABASE_ACTION_ADD
, DATABASE_ACTION_COUNT
);
414 for (AutocompleteActionPredictorTable::Rows::const_iterator it
=
415 rows_to_update
.begin(); it
!= rows_to_update
.end(); ++it
) {
416 const DBCacheKey key
= { it
->user_text
, it
->url
};
418 DBCacheMap::iterator db_it
= db_cache_
.find(key
);
419 DCHECK(db_it
!= db_cache_
.end());
420 DCHECK(db_id_cache_
.find(key
) != db_id_cache_
.end());
422 db_it
->second
.number_of_hits
= it
->number_of_hits
;
423 db_it
->second
.number_of_misses
= it
->number_of_misses
;
424 UMA_HISTOGRAM_ENUMERATION("AutocompleteActionPredictor.DatabaseAction",
425 DATABASE_ACTION_UPDATE
, DATABASE_ACTION_COUNT
);
429 content::BrowserThread::PostTask(content::BrowserThread::DB
, FROM_HERE
,
430 base::Bind(&AutocompleteActionPredictorTable::AddAndUpdateRows
,
431 table_
, rows_to_add
, rows_to_update
));
435 void AutocompleteActionPredictor::CreateCaches(
436 std::vector
<AutocompleteActionPredictorTable::Row
>* rows
) {
437 CHECK(content::BrowserThread::CurrentlyOn(content::BrowserThread::UI
));
438 DCHECK(!profile_
->IsOffTheRecord());
439 DCHECK(!initialized_
);
440 DCHECK(db_cache_
.empty());
441 DCHECK(db_id_cache_
.empty());
443 for (std::vector
<AutocompleteActionPredictorTable::Row
>::const_iterator it
=
444 rows
->begin(); it
!= rows
->end(); ++it
) {
445 const DBCacheKey key
= { it
->user_text
, it
->url
};
446 const DBCacheValue value
= { it
->number_of_hits
, it
->number_of_misses
};
447 db_cache_
[key
] = value
;
448 db_id_cache_
[key
] = it
->id
;
451 // If the history service is ready, delete any old or invalid entries.
452 history::HistoryService
* history_service
=
453 HistoryServiceFactory::GetForProfile(profile_
,
454 ServiceAccessType::EXPLICIT_ACCESS
);
455 if (!TryDeleteOldEntries(history_service
)) {
456 // Wait for the notification that the history service is ready and the URL
459 history_service_observer_
.Add(history_service
);
463 bool AutocompleteActionPredictor::TryDeleteOldEntries(
464 history::HistoryService
* service
) {
465 CHECK(content::BrowserThread::CurrentlyOn(content::BrowserThread::UI
));
466 DCHECK(!profile_
->IsOffTheRecord());
467 DCHECK(!initialized_
);
472 history::URLDatabase
* url_db
= service
->InMemoryDatabase();
476 DeleteOldEntries(url_db
);
480 void AutocompleteActionPredictor::DeleteOldEntries(
481 history::URLDatabase
* url_db
) {
482 CHECK(content::BrowserThread::CurrentlyOn(content::BrowserThread::UI
));
483 DCHECK(!profile_
->IsOffTheRecord());
484 DCHECK(!initialized_
);
485 DCHECK(table_
.get());
487 std::vector
<AutocompleteActionPredictorTable::Row::Id
> ids_to_delete
;
488 DeleteOldIdsFromCaches(url_db
, &ids_to_delete
);
490 content::BrowserThread::PostTask(content::BrowserThread::DB
, FROM_HERE
,
491 base::Bind(&AutocompleteActionPredictorTable::DeleteRows
, table_
,
494 FinishInitialization();
495 if (incognito_predictor_
)
496 incognito_predictor_
->CopyFromMainProfile();
499 void AutocompleteActionPredictor::DeleteOldIdsFromCaches(
500 history::URLDatabase
* url_db
,
501 std::vector
<AutocompleteActionPredictorTable::Row::Id
>* id_list
) {
502 CHECK(content::BrowserThread::CurrentlyOn(content::BrowserThread::UI
));
503 DCHECK(!profile_
->IsOffTheRecord());
504 DCHECK(!initialized_
);
509 for (DBCacheMap::iterator it
= db_cache_
.begin(); it
!= db_cache_
.end();) {
510 history::URLRow url_row
;
512 if ((url_db
->GetRowForURL(it
->first
.url
, &url_row
) == 0) ||
513 ((base::Time::Now() - url_row
.last_visit()).InDays() >
514 kMaximumDaysToKeepEntry
)) {
515 const DBIdCacheMap::iterator id_it
= db_id_cache_
.find(it
->first
);
516 DCHECK(id_it
!= db_id_cache_
.end());
517 id_list
->push_back(id_it
->second
);
518 db_id_cache_
.erase(id_it
);
519 db_cache_
.erase(it
++);
526 void AutocompleteActionPredictor::CopyFromMainProfile() {
527 CHECK(content::BrowserThread::CurrentlyOn(content::BrowserThread::UI
));
528 DCHECK(profile_
->IsOffTheRecord());
529 DCHECK(!initialized_
);
530 DCHECK(main_profile_predictor_
);
531 DCHECK(main_profile_predictor_
->initialized_
);
533 db_cache_
= main_profile_predictor_
->db_cache_
;
534 db_id_cache_
= main_profile_predictor_
->db_id_cache_
;
535 FinishInitialization();
538 void AutocompleteActionPredictor::FinishInitialization() {
539 CHECK(content::BrowserThread::CurrentlyOn(content::BrowserThread::UI
));
540 DCHECK(!initialized_
);
542 // Incognito and normal profiles should listen only to omnibox notifications
543 // from their own profile, but both should listen to history deletions from
544 // the main profile, since opening the history page in either case actually
545 // opens the non-incognito history (and lets users delete from there).
546 notification_registrar_
.Add(this, chrome::NOTIFICATION_OMNIBOX_OPENED_URL
,
547 content::Source
<Profile
>(profile_
));
551 double AutocompleteActionPredictor::CalculateConfidence(
552 const base::string16
& user_text
,
553 const AutocompleteMatch
& match
,
554 bool* is_in_db
) const {
555 const DBCacheKey key
= { user_text
, match
.destination_url
};
558 if (user_text
.length() < kMinimumUserTextLength
)
561 const DBCacheMap::const_iterator iter
= db_cache_
.find(key
);
562 if (iter
== db_cache_
.end())
566 return CalculateConfidenceForDbEntry(iter
);
569 double AutocompleteActionPredictor::CalculateConfidenceForDbEntry(
570 DBCacheMap::const_iterator iter
) const {
571 const DBCacheValue
& value
= iter
->second
;
572 if (value
.number_of_hits
< kMinimumNumberOfHits
)
575 const double number_of_hits
= static_cast<double>(value
.number_of_hits
);
576 return number_of_hits
/ (number_of_hits
+ value
.number_of_misses
);
579 void AutocompleteActionPredictor::Shutdown() {
580 history_service_observer_
.RemoveAll();
583 void AutocompleteActionPredictor::OnURLsDeleted(
584 history::HistoryService
* history_service
,
587 const history::URLRows
& deleted_rows
,
588 const std::set
<GURL
>& favicon_urls
) {
595 DeleteRowsWithURLs(deleted_rows
);
598 void AutocompleteActionPredictor::OnHistoryServiceLoaded(
599 history::HistoryService
* history_service
) {
600 TryDeleteOldEntries(history_service
);
601 history_service_observer_
.Remove(history_service
);
604 AutocompleteActionPredictor::TransitionalMatch::TransitionalMatch() {
607 AutocompleteActionPredictor::TransitionalMatch::~TransitionalMatch() {
610 } // namespace predictors