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/autocomplete/autocomplete_match.h"
19 #include "chrome/browser/autocomplete/autocomplete_result.h"
20 #include "chrome/browser/chrome_notification_types.h"
21 #include "chrome/browser/history/history_notifications.h"
22 #include "chrome/browser/history/history_service.h"
23 #include "chrome/browser/history/history_service_factory.h"
24 #include "chrome/browser/history/in_memory_database.h"
25 #include "chrome/browser/omnibox/omnibox_log.h"
26 #include "chrome/browser/predictors/autocomplete_action_predictor_factory.h"
27 #include "chrome/browser/predictors/predictor_database.h"
28 #include "chrome/browser/predictors/predictor_database_factory.h"
29 #include "chrome/browser/prerender/prerender_field_trial.h"
30 #include "chrome/browser/prerender/prerender_handle.h"
31 #include "chrome/browser/prerender/prerender_manager.h"
32 #include "chrome/browser/prerender/prerender_manager_factory.h"
33 #include "chrome/browser/profiles/profile.h"
34 #include "chrome/browser/ui/omnibox/omnibox_popup_model.h"
35 #include "content/public/browser/browser_thread.h"
36 #include "content/public/browser/notification_details.h"
37 #include "content/public/browser/notification_service.h"
38 #include "content/public/browser/notification_source.h"
42 const float kConfidenceCutoff
[] = {
47 COMPILE_ASSERT(arraysize(kConfidenceCutoff
) ==
48 predictors::AutocompleteActionPredictor::LAST_PREDICT_ACTION
,
49 ConfidenceCutoff_count_mismatch
);
51 const size_t kMinimumUserTextLength
= 1;
52 const int kMinimumNumberOfHits
= 3;
56 DATABASE_ACTION_UPDATE
,
57 DATABASE_ACTION_DELETE_SOME
,
58 DATABASE_ACTION_DELETE_ALL
,
64 namespace predictors
{
66 const int AutocompleteActionPredictor::kMaximumDaysToKeepEntry
= 14;
68 AutocompleteActionPredictor::AutocompleteActionPredictor(Profile
* profile
)
70 main_profile_predictor_(NULL
),
71 incognito_predictor_(NULL
),
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_
, Profile::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 (AutocompleteResult::const_iterator
i(result
.begin()); i
!= result
.end();
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 const content::SessionStorageNamespaceMap
& session_storage_namespace_map
,
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 content::SessionStorageNamespace
* session_storage_namespace
= NULL
;
160 content::SessionStorageNamespaceMap::const_iterator it
=
161 session_storage_namespace_map
.find(std::string());
162 if (it
!= session_storage_namespace_map
.end())
163 session_storage_namespace
= it
->second
.get();
164 prerender_handle_
.reset(prerender_manager
->AddPrerenderFromOmnibox(
165 url
, session_storage_namespace
, size
));
167 if (old_prerender_handle
)
168 old_prerender_handle
->OnCancel();
171 // Given a match, return a recommended action.
172 AutocompleteActionPredictor::Action
173 AutocompleteActionPredictor::RecommendAction(
174 const base::string16
& user_text
,
175 const AutocompleteMatch
& match
) const {
176 bool is_in_db
= false;
177 const double confidence
= CalculateConfidence(user_text
, match
, &is_in_db
);
178 DCHECK(confidence
>= 0.0 && confidence
<= 1.0);
180 UMA_HISTOGRAM_BOOLEAN("AutocompleteActionPredictor.MatchIsInDb", is_in_db
);
183 // Multiple enties with the same URL are fine as the confidence may be
185 tracked_urls_
.push_back(std::make_pair(match
.destination_url
, confidence
));
186 UMA_HISTOGRAM_COUNTS_100("AutocompleteActionPredictor.Confidence",
190 // Map the confidence to an action.
191 Action action
= ACTION_NONE
;
192 for (int i
= 0; i
< LAST_PREDICT_ACTION
; ++i
) {
193 if (confidence
>= kConfidenceCutoff
[i
]) {
194 action
= static_cast<Action
>(i
);
199 // Downgrade prerender to preconnect if this is a search match or if omnibox
200 // prerendering is disabled. There are cases when Instant will not handle a
201 // search suggestion and in those cases it would be good to prerender the
202 // search results, however search engines have not been set up to correctly
203 // handle being prerendered and until they are we should avoid it.
204 // http://crbug.com/117495
205 if (action
== ACTION_PRERENDER
&&
206 (AutocompleteMatch::IsSearchType(match
.type
) ||
207 !prerender::IsOmniboxEnabled(profile_
))) {
208 action
= ACTION_PRECONNECT
;
214 // Return true if the suggestion type warrants a TCP/IP preconnection.
215 // i.e., it is now quite likely that the user will select the related domain.
217 bool AutocompleteActionPredictor::IsPreconnectable(
218 const AutocompleteMatch
& match
) {
219 return AutocompleteMatch::IsSearchType(match
.type
);
222 bool AutocompleteActionPredictor::IsPrerenderAbandonedForTesting() {
223 return prerender_handle_
&& prerender_handle_
->IsAbandoned();
226 void AutocompleteActionPredictor::Observe(
228 const content::NotificationSource
& source
,
229 const content::NotificationDetails
& details
) {
231 case content::NOTIFICATION_LOAD_COMPLETED_MAIN_FRAME
:
232 CreateLocalCachesFromDatabase();
233 notification_registrar_
.Remove(
235 content::NOTIFICATION_LOAD_COMPLETED_MAIN_FRAME
,
236 content::NotificationService::AllSources());
239 case chrome::NOTIFICATION_HISTORY_URLS_DELETED
: {
240 DCHECK(initialized_
);
241 const content::Details
<const history::URLsDeletedDetails
>
242 urls_deleted_details
=
243 content::Details
<const history::URLsDeletedDetails
>(details
);
244 if (urls_deleted_details
->all_history
)
247 DeleteRowsWithURLs(urls_deleted_details
->rows
);
251 case chrome::NOTIFICATION_OMNIBOX_OPENED_URL
: {
252 DCHECK(initialized_
);
254 // TODO(dominich): This doesn't need to be synchronous. Investigate
255 // posting it as a task to be run later.
256 OnOmniboxOpenedUrl(*content::Details
<OmniboxLog
>(details
).ptr());
260 case chrome::NOTIFICATION_HISTORY_LOADED
: {
261 TryDeleteOldEntries(content::Details
<HistoryService
>(details
).ptr());
263 notification_registrar_
.Remove(this,
264 chrome::NOTIFICATION_HISTORY_LOADED
,
265 content::Source
<Profile
>(profile_
));
270 NOTREACHED() << "Unexpected notification observed.";
275 void AutocompleteActionPredictor::CreateLocalCachesFromDatabase() {
276 // Create local caches using the database as loaded. We will garbage collect
277 // rows from the caches and the database once the history service is
279 std::vector
<AutocompleteActionPredictorTable::Row
>* rows
=
280 new std::vector
<AutocompleteActionPredictorTable::Row
>();
281 content::BrowserThread::PostTaskAndReply(content::BrowserThread::DB
,
283 base::Bind(&AutocompleteActionPredictorTable::GetAllRows
, table_
, rows
),
284 base::Bind(&AutocompleteActionPredictor::CreateCaches
, AsWeakPtr(),
288 void AutocompleteActionPredictor::DeleteAllRows() {
293 db_id_cache_
.clear();
296 content::BrowserThread::PostTask(content::BrowserThread::DB
, FROM_HERE
,
297 base::Bind(&AutocompleteActionPredictorTable::DeleteAllRows
,
301 UMA_HISTOGRAM_ENUMERATION("AutocompleteActionPredictor.DatabaseAction",
302 DATABASE_ACTION_DELETE_ALL
, DATABASE_ACTION_COUNT
);
305 void AutocompleteActionPredictor::DeleteRowsWithURLs(
306 const history::URLRows
& rows
) {
310 std::vector
<AutocompleteActionPredictorTable::Row::Id
> id_list
;
312 for (DBCacheMap::iterator it
= db_cache_
.begin(); it
!= db_cache_
.end();) {
313 if (std::find_if(rows
.begin(), rows
.end(),
314 history::URLRow::URLRowHasURL(it
->first
.url
)) != rows
.end()) {
315 const DBIdCacheMap::iterator id_it
= db_id_cache_
.find(it
->first
);
316 DCHECK(id_it
!= db_id_cache_
.end());
317 id_list
.push_back(id_it
->second
);
318 db_id_cache_
.erase(id_it
);
319 db_cache_
.erase(it
++);
326 content::BrowserThread::PostTask(content::BrowserThread::DB
, FROM_HERE
,
327 base::Bind(&AutocompleteActionPredictorTable::DeleteRows
, table_
,
331 UMA_HISTOGRAM_ENUMERATION("AutocompleteActionPredictor.DatabaseAction",
332 DATABASE_ACTION_DELETE_SOME
, DATABASE_ACTION_COUNT
);
335 void AutocompleteActionPredictor::OnOmniboxOpenedUrl(const OmniboxLog
& log
) {
336 if (log
.text
.length() < kMinimumUserTextLength
)
339 // Do not attempt to learn from omnibox interactions where the omnibox
340 // dropdown is closed. In these cases the user text (|log.text|) that we
341 // learn from is either empty or effectively identical to the destination
342 // string. In either case, it can't teach us much. Also do not attempt
343 // to learn from paste-and-go actions even if the popup is open because
344 // the paste-and-go destination has no relation to whatever text the user
346 if (!log
.is_popup_open
|| log
.is_paste_and_go
)
349 // Abandon the current prerender. If it is to be used, it will be used very
350 // soon, so use the lower timeout.
351 if (prerender_handle_
) {
352 prerender_handle_
->OnNavigateAway();
353 // Don't release |prerender_handle_| so it is canceled if it survives to the
354 // next StartPrerendering call.
357 UMA_HISTOGRAM_BOOLEAN(
358 base::StringPrintf("Prerender.OmniboxNavigationsCouldPrerender%s",
359 prerender::PrerenderManager::GetModeString()).c_str(),
360 prerender::IsOmniboxEnabled(profile_
));
362 const AutocompleteMatch
& match
= log
.result
.match_at(log
.selected_index
);
363 const GURL
& opened_url
= match
.destination_url
;
364 const base::string16
lower_user_text(base::i18n::ToLower(log
.text
));
366 // Traverse transitional matches for those that have a user_text that is a
367 // prefix of |lower_user_text|.
368 std::vector
<AutocompleteActionPredictorTable::Row
> rows_to_add
;
369 std::vector
<AutocompleteActionPredictorTable::Row
> rows_to_update
;
371 for (std::vector
<TransitionalMatch
>::const_iterator it
=
372 transitional_matches_
.begin(); it
!= transitional_matches_
.end();
374 if (!StartsWith(lower_user_text
, it
->user_text
, true))
377 // Add entries to the database for those matches.
378 for (std::vector
<GURL
>::const_iterator url_it
= it
->urls
.begin();
379 url_it
!= it
->urls
.end(); ++url_it
) {
380 DCHECK(it
->user_text
.length() >= kMinimumUserTextLength
);
381 const DBCacheKey key
= { it
->user_text
, *url_it
};
382 const bool is_hit
= (*url_it
== opened_url
);
384 AutocompleteActionPredictorTable::Row row
;
385 row
.user_text
= key
.user_text
;
388 DBCacheMap::iterator it
= db_cache_
.find(key
);
389 if (it
== db_cache_
.end()) {
390 row
.id
= base::GenerateGUID();
391 row
.number_of_hits
= is_hit
? 1 : 0;
392 row
.number_of_misses
= is_hit
? 0 : 1;
394 rows_to_add
.push_back(row
);
396 DCHECK(db_id_cache_
.find(key
) != db_id_cache_
.end());
397 row
.id
= db_id_cache_
.find(key
)->second
;
398 row
.number_of_hits
= it
->second
.number_of_hits
+ (is_hit
? 1 : 0);
399 row
.number_of_misses
= it
->second
.number_of_misses
+ (is_hit
? 0 : 1);
401 rows_to_update
.push_back(row
);
405 if (rows_to_add
.size() > 0 || rows_to_update
.size() > 0)
406 AddAndUpdateRows(rows_to_add
, rows_to_update
);
408 ClearTransitionalMatches();
410 // Check against tracked urls and log accuracy for the confidence we
412 for (std::vector
<std::pair
<GURL
, double> >::const_iterator it
=
413 tracked_urls_
.begin(); it
!= tracked_urls_
.end();
415 if (opened_url
== it
->first
) {
416 UMA_HISTOGRAM_COUNTS_100("AutocompleteActionPredictor.AccurateCount",
420 tracked_urls_
.clear();
423 void AutocompleteActionPredictor::AddAndUpdateRows(
424 const AutocompleteActionPredictorTable::Rows
& rows_to_add
,
425 const AutocompleteActionPredictorTable::Rows
& rows_to_update
) {
429 for (AutocompleteActionPredictorTable::Rows::const_iterator it
=
430 rows_to_add
.begin(); it
!= rows_to_add
.end(); ++it
) {
431 const DBCacheKey key
= { it
->user_text
, it
->url
};
432 DBCacheValue value
= { it
->number_of_hits
, it
->number_of_misses
};
434 DCHECK(db_cache_
.find(key
) == db_cache_
.end());
436 db_cache_
[key
] = value
;
437 db_id_cache_
[key
] = it
->id
;
438 UMA_HISTOGRAM_ENUMERATION("AutocompleteActionPredictor.DatabaseAction",
439 DATABASE_ACTION_ADD
, DATABASE_ACTION_COUNT
);
441 for (AutocompleteActionPredictorTable::Rows::const_iterator it
=
442 rows_to_update
.begin(); it
!= rows_to_update
.end(); ++it
) {
443 const DBCacheKey key
= { it
->user_text
, it
->url
};
445 DBCacheMap::iterator db_it
= db_cache_
.find(key
);
446 DCHECK(db_it
!= db_cache_
.end());
447 DCHECK(db_id_cache_
.find(key
) != db_id_cache_
.end());
449 db_it
->second
.number_of_hits
= it
->number_of_hits
;
450 db_it
->second
.number_of_misses
= it
->number_of_misses
;
451 UMA_HISTOGRAM_ENUMERATION("AutocompleteActionPredictor.DatabaseAction",
452 DATABASE_ACTION_UPDATE
, DATABASE_ACTION_COUNT
);
456 content::BrowserThread::PostTask(content::BrowserThread::DB
, FROM_HERE
,
457 base::Bind(&AutocompleteActionPredictorTable::AddAndUpdateRows
,
458 table_
, rows_to_add
, rows_to_update
));
462 void AutocompleteActionPredictor::CreateCaches(
463 std::vector
<AutocompleteActionPredictorTable::Row
>* rows
) {
464 CHECK(content::BrowserThread::CurrentlyOn(content::BrowserThread::UI
));
465 DCHECK(!profile_
->IsOffTheRecord());
466 DCHECK(!initialized_
);
467 DCHECK(db_cache_
.empty());
468 DCHECK(db_id_cache_
.empty());
470 for (std::vector
<AutocompleteActionPredictorTable::Row
>::const_iterator it
=
471 rows
->begin(); it
!= rows
->end(); ++it
) {
472 const DBCacheKey key
= { it
->user_text
, it
->url
};
473 const DBCacheValue value
= { it
->number_of_hits
, it
->number_of_misses
};
474 db_cache_
[key
] = value
;
475 db_id_cache_
[key
] = it
->id
;
478 // If the history service is ready, delete any old or invalid entries.
479 HistoryService
* history_service
=
480 HistoryServiceFactory::GetForProfile(profile_
, Profile::EXPLICIT_ACCESS
);
481 if (!TryDeleteOldEntries(history_service
)) {
482 // Wait for the notification that the history service is ready and the URL
484 notification_registrar_
.Add(this, chrome::NOTIFICATION_HISTORY_LOADED
,
485 content::Source
<Profile
>(profile_
));
489 bool AutocompleteActionPredictor::TryDeleteOldEntries(HistoryService
* service
) {
490 CHECK(content::BrowserThread::CurrentlyOn(content::BrowserThread::UI
));
491 DCHECK(!profile_
->IsOffTheRecord());
492 DCHECK(!initialized_
);
497 history::URLDatabase
* url_db
= service
->InMemoryDatabase();
501 DeleteOldEntries(url_db
);
505 void AutocompleteActionPredictor::DeleteOldEntries(
506 history::URLDatabase
* url_db
) {
507 CHECK(content::BrowserThread::CurrentlyOn(content::BrowserThread::UI
));
508 DCHECK(!profile_
->IsOffTheRecord());
509 DCHECK(!initialized_
);
510 DCHECK(table_
.get());
512 std::vector
<AutocompleteActionPredictorTable::Row::Id
> ids_to_delete
;
513 DeleteOldIdsFromCaches(url_db
, &ids_to_delete
);
515 content::BrowserThread::PostTask(content::BrowserThread::DB
, FROM_HERE
,
516 base::Bind(&AutocompleteActionPredictorTable::DeleteRows
, table_
,
519 FinishInitialization();
520 if (incognito_predictor_
)
521 incognito_predictor_
->CopyFromMainProfile();
524 void AutocompleteActionPredictor::DeleteOldIdsFromCaches(
525 history::URLDatabase
* url_db
,
526 std::vector
<AutocompleteActionPredictorTable::Row::Id
>* id_list
) {
527 CHECK(content::BrowserThread::CurrentlyOn(content::BrowserThread::UI
));
528 DCHECK(!profile_
->IsOffTheRecord());
529 DCHECK(!initialized_
);
534 for (DBCacheMap::iterator it
= db_cache_
.begin(); it
!= db_cache_
.end();) {
535 history::URLRow url_row
;
537 if ((url_db
->GetRowForURL(it
->first
.url
, &url_row
) == 0) ||
538 ((base::Time::Now() - url_row
.last_visit()).InDays() >
539 kMaximumDaysToKeepEntry
)) {
540 const DBIdCacheMap::iterator id_it
= db_id_cache_
.find(it
->first
);
541 DCHECK(id_it
!= db_id_cache_
.end());
542 id_list
->push_back(id_it
->second
);
543 db_id_cache_
.erase(id_it
);
544 db_cache_
.erase(it
++);
551 void AutocompleteActionPredictor::CopyFromMainProfile() {
552 CHECK(content::BrowserThread::CurrentlyOn(content::BrowserThread::UI
));
553 DCHECK(profile_
->IsOffTheRecord());
554 DCHECK(!initialized_
);
555 DCHECK(main_profile_predictor_
);
556 DCHECK(main_profile_predictor_
->initialized_
);
558 db_cache_
= main_profile_predictor_
->db_cache_
;
559 db_id_cache_
= main_profile_predictor_
->db_id_cache_
;
560 FinishInitialization();
563 void AutocompleteActionPredictor::FinishInitialization() {
564 CHECK(content::BrowserThread::CurrentlyOn(content::BrowserThread::UI
));
565 DCHECK(!initialized_
);
567 // Incognito and normal profiles should listen only to omnibox notifications
568 // from their own profile, but both should listen to history deletions from
569 // the main profile, since opening the history page in either case actually
570 // opens the non-incognito history (and lets users delete from there).
571 notification_registrar_
.Add(this, chrome::NOTIFICATION_OMNIBOX_OPENED_URL
,
572 content::Source
<Profile
>(profile_
));
573 notification_registrar_
.Add(this, chrome::NOTIFICATION_HISTORY_URLS_DELETED
,
574 content::Source
<Profile
>(profile_
->GetOriginalProfile()));
578 double AutocompleteActionPredictor::CalculateConfidence(
579 const base::string16
& user_text
,
580 const AutocompleteMatch
& match
,
581 bool* is_in_db
) const {
582 const DBCacheKey key
= { user_text
, match
.destination_url
};
585 if (user_text
.length() < kMinimumUserTextLength
)
588 const DBCacheMap::const_iterator iter
= db_cache_
.find(key
);
589 if (iter
== db_cache_
.end())
593 return CalculateConfidenceForDbEntry(iter
);
596 double AutocompleteActionPredictor::CalculateConfidenceForDbEntry(
597 DBCacheMap::const_iterator iter
) const {
598 const DBCacheValue
& value
= iter
->second
;
599 if (value
.number_of_hits
< kMinimumNumberOfHits
)
602 const double number_of_hits
= static_cast<double>(value
.number_of_hits
);
603 return number_of_hits
/ (number_of_hits
+ value
.number_of_misses
);
606 AutocompleteActionPredictor::TransitionalMatch::TransitionalMatch() {
609 AutocompleteActionPredictor::TransitionalMatch::~TransitionalMatch() {
612 } // namespace predictors