1// Copyright (c) 2012 The Chromium Authors. All rights reserved. 2// Use of this source code is governed by a BSD-style license that can be 3// found in the LICENSE file. 4 5#include "chrome/browser/history/history_backend.h" 6 7#include <algorithm> 8#include <functional> 9#include <list> 10#include <map> 11#include <set> 12#include <vector> 13 14#include "base/basictypes.h" 15#include "base/bind.h" 16#include "base/compiler_specific.h" 17#include "base/files/file_enumerator.h" 18#include "base/memory/scoped_ptr.h" 19#include "base/memory/scoped_vector.h" 20#include "base/message_loop/message_loop.h" 21#include "base/metrics/histogram.h" 22#include "base/rand_util.h" 23#include "base/strings/string_util.h" 24#include "base/strings/utf_string_conversions.h" 25#include "base/time/time.h" 26#include "chrome/browser/autocomplete/history_url_provider.h" 27#include "chrome/browser/bookmarks/bookmark_service.h" 28#include "chrome/browser/chrome_notification_types.h" 29#include "chrome/browser/favicon/favicon_changed_details.h" 30#include "chrome/browser/history/download_row.h" 31#include "chrome/browser/history/history_db_task.h" 32#include "chrome/browser/history/history_notifications.h" 33#include "chrome/browser/history/in_memory_history_backend.h" 34#include "chrome/browser/history/page_usage_data.h" 35#include "chrome/browser/history/select_favicon_frames.h" 36#include "chrome/browser/history/top_sites.h" 37#include "chrome/browser/history/typed_url_syncable_service.h" 38#include "chrome/browser/history/visit_filter.h" 39#include "chrome/common/chrome_constants.h" 40#include "chrome/common/importer/imported_favicon_usage.h" 41#include "chrome/common/url_constants.h" 42#include "grit/chromium_strings.h" 43#include "grit/generated_resources.h" 44#include "net/base/registry_controlled_domains/registry_controlled_domain.h" 45#include "sql/error_delegate_util.h" 46#include "url/gurl.h" 47 48#if defined(OS_ANDROID) 49#include "chrome/browser/history/android/android_provider_backend.h" 50#endif 51 52using base::Time; 53using base::TimeDelta; 54using base::TimeTicks; 55 56/* The HistoryBackend consists of a number of components: 57 58 HistoryDatabase (stores past 3 months of history) 59 URLDatabase (stores a list of URLs) 60 DownloadDatabase (stores a list of downloads) 61 VisitDatabase (stores a list of visits for the URLs) 62 VisitSegmentDatabase (stores groups of URLs for the most visited view). 63 64 ArchivedDatabase (stores history older than 3 months) 65 URLDatabase (stores a list of URLs) 66 DownloadDatabase (stores a list of downloads) 67 VisitDatabase (stores a list of visits for the URLs) 68 69 (this does not store visit segments as they expire after 3 mos.) 70 71 ExpireHistoryBackend (manages moving things from HistoryDatabase to 72 the ArchivedDatabase and deleting) 73*/ 74 75namespace history { 76 77// How long we keep segment data for in days. Currently 3 months. 78// This value needs to be greater or equal to 79// MostVisitedModel::kMostVisitedScope but we don't want to introduce a direct 80// dependency between MostVisitedModel and the history backend. 81const int kSegmentDataRetention = 90; 82 83// How long we'll wait to do a commit, so that things are batched together. 84const int kCommitIntervalSeconds = 10; 85 86// The amount of time before we re-fetch the favicon. 87const int kFaviconRefetchDays = 7; 88 89// The maximum number of items we'll allow in the redirect list before 90// deleting some. 91const int kMaxRedirectCount = 32; 92 93// The number of days old a history entry can be before it is considered "old" 94// and is archived. 95const int kArchiveDaysThreshold = 90; 96 97#if defined(OS_ANDROID) 98// The maximum number of top sites to track when recording top page visit stats. 99const size_t kPageVisitStatsMaxTopSites = 50; 100#endif 101 102// Converts from PageUsageData to MostVisitedURL. |redirects| is a 103// list of redirects for this URL. Empty list means no redirects. 104MostVisitedURL MakeMostVisitedURL(const PageUsageData& page_data, 105 const RedirectList& redirects) { 106 MostVisitedURL mv; 107 mv.url = page_data.GetURL(); 108 mv.title = page_data.GetTitle(); 109 if (redirects.empty()) { 110 // Redirects must contain at least the target url. 111 mv.redirects.push_back(mv.url); 112 } else { 113 mv.redirects = redirects; 114 if (mv.redirects[mv.redirects.size() - 1] != mv.url) { 115 // The last url must be the target url. 116 mv.redirects.push_back(mv.url); 117 } 118 } 119 return mv; 120} 121 122// This task is run on a timer so that commits happen at regular intervals 123// so they are batched together. The important thing about this class is that 124// it supports canceling of the task so the reference to the backend will be 125// freed. The problem is that when history is shutting down, there is likely 126// to be one of these commits still pending and holding a reference. 127// 128// The backend can call Cancel to have this task release the reference. The 129// task will still run (if we ever get to processing the event before 130// shutdown), but it will not do anything. 131// 132// Note that this is a refcounted object and is not a task in itself. It should 133// be assigned to a RunnableMethod. 134// 135// TODO(brettw): bug 1165182: This should be replaced with a 136// base::WeakPtrFactory which will handle everything automatically (like we do 137// in ExpireHistoryBackend). 138class CommitLaterTask : public base::RefCounted<CommitLaterTask> { 139 public: 140 explicit CommitLaterTask(HistoryBackend* history_backend) 141 : history_backend_(history_backend) { 142 } 143 144 // The backend will call this function if it is being destroyed so that we 145 // release our reference. 146 void Cancel() { 147 history_backend_ = NULL; 148 } 149 150 void RunCommit() { 151 if (history_backend_.get()) 152 history_backend_->Commit(); 153 } 154 155 private: 156 friend class base::RefCounted<CommitLaterTask>; 157 158 ~CommitLaterTask() {} 159 160 scoped_refptr<HistoryBackend> history_backend_; 161}; 162 163// HistoryBackend -------------------------------------------------------------- 164 165HistoryBackend::HistoryBackend(const base::FilePath& history_dir, 166 int id, 167 Delegate* delegate, 168 BookmarkService* bookmark_service) 169 : delegate_(delegate), 170 id_(id), 171 history_dir_(history_dir), 172 scheduled_kill_db_(false), 173 expirer_(this, bookmark_service), 174 recent_redirects_(kMaxRedirectCount), 175 backend_destroy_message_loop_(NULL), 176 segment_queried_(false), 177 bookmark_service_(bookmark_service) { 178} 179 180HistoryBackend::~HistoryBackend() { 181 DCHECK(!scheduled_commit_.get()) << "Deleting without cleanup"; 182 ReleaseDBTasks(); 183 184#if defined(OS_ANDROID) 185 // Release AndroidProviderBackend before other objects. 186 android_provider_backend_.reset(); 187#endif 188 189 // First close the databases before optionally running the "destroy" task. 190 CloseAllDatabases(); 191 192 if (!backend_destroy_task_.is_null()) { 193 // Notify an interested party (typically a unit test) that we're done. 194 DCHECK(backend_destroy_message_loop_); 195 backend_destroy_message_loop_->PostTask(FROM_HERE, backend_destroy_task_); 196 } 197 198#if defined(OS_ANDROID) 199 sql::Connection::Delete(GetAndroidCacheFileName()); 200#endif 201} 202 203void HistoryBackend::Init(const std::string& languages, bool force_fail) { 204 if (!force_fail) 205 InitImpl(languages); 206 delegate_->DBLoaded(id_); 207 typed_url_syncable_service_.reset(new TypedUrlSyncableService(this)); 208 memory_pressure_listener_.reset(new base::MemoryPressureListener( 209 base::Bind(&HistoryBackend::OnMemoryPressure, base::Unretained(this)))); 210#if defined(OS_ANDROID) 211 PopulateMostVisitedURLMap(); 212#endif 213} 214 215void HistoryBackend::SetOnBackendDestroyTask(base::MessageLoop* message_loop, 216 const base::Closure& task) { 217 if (!backend_destroy_task_.is_null()) 218 DLOG(WARNING) << "Setting more than one destroy task, overriding"; 219 backend_destroy_message_loop_ = message_loop; 220 backend_destroy_task_ = task; 221} 222 223void HistoryBackend::Closing() { 224 // Any scheduled commit will have a reference to us, we must make it 225 // release that reference before we can be destroyed. 226 CancelScheduledCommit(); 227 228 // Release our reference to the delegate, this reference will be keeping the 229 // history service alive. 230 delegate_.reset(); 231} 232 233void HistoryBackend::NotifyRenderProcessHostDestruction(const void* host) { 234 tracker_.NotifyRenderProcessHostDestruction(host); 235} 236 237base::FilePath HistoryBackend::GetThumbnailFileName() const { 238 return history_dir_.Append(chrome::kThumbnailsFilename); 239} 240 241base::FilePath HistoryBackend::GetFaviconsFileName() const { 242 return history_dir_.Append(chrome::kFaviconsFilename); 243} 244 245base::FilePath HistoryBackend::GetArchivedFileName() const { 246 return history_dir_.Append(chrome::kArchivedHistoryFilename); 247} 248 249#if defined(OS_ANDROID) 250base::FilePath HistoryBackend::GetAndroidCacheFileName() const { 251 return history_dir_.Append(chrome::kAndroidCacheFilename); 252} 253#endif 254 255SegmentID HistoryBackend::GetLastSegmentID(VisitID from_visit) { 256 // Set is used to detect referrer loops. Should not happen, but can 257 // if the database is corrupt. 258 std::set<VisitID> visit_set; 259 VisitID visit_id = from_visit; 260 while (visit_id) { 261 VisitRow row; 262 if (!db_->GetRowForVisit(visit_id, &row)) 263 return 0; 264 if (row.segment_id) 265 return row.segment_id; // Found a visit in this change with a segment. 266 267 // Check the referrer of this visit, if any. 268 visit_id = row.referring_visit; 269 270 if (visit_set.find(visit_id) != visit_set.end()) { 271 NOTREACHED() << "Loop in referer chain, giving up"; 272 break; 273 } 274 visit_set.insert(visit_id); 275 } 276 return 0; 277} 278 279SegmentID HistoryBackend::UpdateSegments( 280 const GURL& url, 281 VisitID from_visit, 282 VisitID visit_id, 283 content::PageTransition transition_type, 284 const Time ts) { 285 if (!db_) 286 return 0; 287 288 // We only consider main frames. 289 if (!content::PageTransitionIsMainFrame(transition_type)) 290 return 0; 291 292 SegmentID segment_id = 0; 293 content::PageTransition t = 294 content::PageTransitionStripQualifier(transition_type); 295 296 // Are we at the beginning of a new segment? 297 // Note that navigating to an existing entry (with back/forward) reuses the 298 // same transition type. We are not adding it as a new segment in that case 299 // because if this was the target of a redirect, we might end up with 300 // 2 entries for the same final URL. Ex: User types google.net, gets 301 // redirected to google.com. A segment is created for google.net. On 302 // google.com users navigates through a link, then press back. That last 303 // navigation is for the entry google.com transition typed. We end up adding 304 // a segment for that one as well. So we end up with google.net and google.com 305 // in the segment table, showing as 2 entries in the NTP. 306 // Note also that we should still be updating the visit count for that segment 307 // which we are not doing now. It should be addressed when 308 // http://crbug.com/96860 is fixed. 309 if ((t == content::PAGE_TRANSITION_TYPED || 310 t == content::PAGE_TRANSITION_AUTO_BOOKMARK) && 311 (transition_type & content::PAGE_TRANSITION_FORWARD_BACK) == 0) { 312 // If so, create or get the segment. 313 std::string segment_name = db_->ComputeSegmentName(url); 314 URLID url_id = db_->GetRowForURL(url, NULL); 315 if (!url_id) 316 return 0; 317 318 if (!(segment_id = db_->GetSegmentNamed(segment_name))) { 319 if (!(segment_id = db_->CreateSegment(url_id, segment_name))) { 320 NOTREACHED(); 321 return 0; 322 } 323 } else { 324 // Note: if we update an existing segment, we update the url used to 325 // represent that segment in order to minimize stale most visited 326 // images. 327 db_->UpdateSegmentRepresentationURL(segment_id, url_id); 328 } 329 } else { 330 // Note: it is possible there is no segment ID set for this visit chain. 331 // This can happen if the initial navigation wasn't AUTO_BOOKMARK or 332 // TYPED. (For example GENERATED). In this case this visit doesn't count 333 // toward any segment. 334 if (!(segment_id = GetLastSegmentID(from_visit))) 335 return 0; 336 } 337 338 // Set the segment in the visit. 339 if (!db_->SetSegmentID(visit_id, segment_id)) { 340 NOTREACHED(); 341 return 0; 342 } 343 344 // Finally, increase the counter for that segment / day. 345 if (!db_->IncreaseSegmentVisitCount(segment_id, ts, 1)) { 346 NOTREACHED(); 347 return 0; 348 } 349 return segment_id; 350} 351 352void HistoryBackend::UpdateWithPageEndTime(const void* host, 353 int32 page_id, 354 const GURL& url, 355 Time end_ts) { 356 // Will be filled with the URL ID and the visit ID of the last addition. 357 VisitID visit_id = tracker_.GetLastVisit(host, page_id, url); 358 UpdateVisitDuration(visit_id, end_ts); 359} 360 361void HistoryBackend::UpdateVisitDuration(VisitID visit_id, const Time end_ts) { 362 if (!db_) 363 return; 364 365 // Get the starting visit_time for visit_id. 366 VisitRow visit_row; 367 if (db_->GetRowForVisit(visit_id, &visit_row)) { 368 // We should never have a negative duration time even when time is skewed. 369 visit_row.visit_duration = end_ts > visit_row.visit_time ? 370 end_ts - visit_row.visit_time : TimeDelta::FromMicroseconds(0); 371 db_->UpdateVisitRow(visit_row); 372 } 373} 374 375void HistoryBackend::AddPage(const HistoryAddPageArgs& request) { 376 if (!db_) 377 return; 378 379 // Will be filled with the URL ID and the visit ID of the last addition. 380 std::pair<URLID, VisitID> last_ids(0, tracker_.GetLastVisit( 381 request.id_scope, request.page_id, request.referrer)); 382 383 VisitID from_visit_id = last_ids.second; 384 385 // If a redirect chain is given, we expect the last item in that chain to be 386 // the final URL. 387 DCHECK(request.redirects.empty() || 388 request.redirects.back() == request.url); 389 390 // If the user is adding older history, we need to make sure our times 391 // are correct. 392 if (request.time < first_recorded_time_) 393 first_recorded_time_ = request.time; 394 395 content::PageTransition request_transition = request.transition; 396 content::PageTransition stripped_transition = 397 content::PageTransitionStripQualifier(request_transition); 398 bool is_keyword_generated = 399 (stripped_transition == content::PAGE_TRANSITION_KEYWORD_GENERATED); 400 401 // If the user is navigating to a not-previously-typed intranet hostname, 402 // change the transition to TYPED so that the omnibox will learn that this is 403 // a known host. 404 bool has_redirects = request.redirects.size() > 1; 405 if (content::PageTransitionIsMainFrame(request_transition) && 406 (stripped_transition != content::PAGE_TRANSITION_TYPED) && 407 !is_keyword_generated) { 408 const GURL& origin_url(has_redirects ? 409 request.redirects[0] : request.url); 410 if (origin_url.SchemeIs(content::kHttpScheme) || 411 origin_url.SchemeIs(content::kHttpsScheme) || 412 origin_url.SchemeIs(content::kFtpScheme)) { 413 std::string host(origin_url.host()); 414 size_t registry_length = 415 net::registry_controlled_domains::GetRegistryLength( 416 host, 417 net::registry_controlled_domains::EXCLUDE_UNKNOWN_REGISTRIES, 418 net::registry_controlled_domains::EXCLUDE_PRIVATE_REGISTRIES); 419 if (registry_length == 0 && !db_->IsTypedHost(host)) { 420 stripped_transition = content::PAGE_TRANSITION_TYPED; 421 request_transition = 422 content::PageTransitionFromInt( 423 stripped_transition | 424 content::PageTransitionGetQualifier(request_transition)); 425 } 426 } 427 } 428 429 if (!has_redirects) { 430 // The single entry is both a chain start and end. 431 content::PageTransition t = content::PageTransitionFromInt( 432 request_transition | 433 content::PAGE_TRANSITION_CHAIN_START | 434 content::PAGE_TRANSITION_CHAIN_END); 435 436 // No redirect case (one element means just the page itself). 437 last_ids = AddPageVisit(request.url, request.time, 438 last_ids.second, t, request.visit_source); 439 440 // Update the segment for this visit. KEYWORD_GENERATED visits should not 441 // result in changing most visited, so we don't update segments (most 442 // visited db). 443 if (!is_keyword_generated) { 444 UpdateSegments(request.url, from_visit_id, last_ids.second, t, 445 request.time); 446 447 // Update the referrer's duration. 448 UpdateVisitDuration(from_visit_id, request.time); 449 } 450 } else { 451 // Redirect case. Add the redirect chain. 452 453 content::PageTransition redirect_info = 454 content::PAGE_TRANSITION_CHAIN_START; 455 456 RedirectList redirects = request.redirects; 457 if (redirects[0].SchemeIs(chrome::kAboutScheme)) { 458 // When the redirect source + referrer is "about" we skip it. This 459 // happens when a page opens a new frame/window to about:blank and then 460 // script sets the URL to somewhere else (used to hide the referrer). It 461 // would be nice to keep all these redirects properly but we don't ever 462 // see the initial about:blank load, so we don't know where the 463 // subsequent client redirect came from. 464 // 465 // In this case, we just don't bother hooking up the source of the 466 // redirects, so we remove it. 467 redirects.erase(redirects.begin()); 468 } else if (request_transition & content::PAGE_TRANSITION_CLIENT_REDIRECT) { 469 redirect_info = content::PAGE_TRANSITION_CLIENT_REDIRECT; 470 // The first entry in the redirect chain initiated a client redirect. 471 // We don't add this to the database since the referrer is already 472 // there, so we skip over it but change the transition type of the first 473 // transition to client redirect. 474 // 475 // The referrer is invalid when restoring a session that features an 476 // https tab that redirects to a different host or to http. In this 477 // case we don't need to reconnect the new redirect with the existing 478 // chain. 479 if (request.referrer.is_valid()) { 480 DCHECK(request.referrer == redirects[0]); 481 redirects.erase(redirects.begin()); 482 483 // If the navigation entry for this visit has replaced that for the 484 // first visit, remove the CHAIN_END marker from the first visit. This 485 // can be called a lot, for example, the page cycler, and most of the 486 // time we won't have changed anything. 487 VisitRow visit_row; 488 if (request.did_replace_entry && 489 db_->GetRowForVisit(last_ids.second, &visit_row) && 490 visit_row.transition & content::PAGE_TRANSITION_CHAIN_END) { 491 visit_row.transition = content::PageTransitionFromInt( 492 visit_row.transition & ~content::PAGE_TRANSITION_CHAIN_END); 493 db_->UpdateVisitRow(visit_row); 494 } 495 } 496 } 497 498 for (size_t redirect_index = 0; redirect_index < redirects.size(); 499 redirect_index++) { 500 content::PageTransition t = 501 content::PageTransitionFromInt(stripped_transition | redirect_info); 502 503 // If this is the last transition, add a CHAIN_END marker 504 if (redirect_index == (redirects.size() - 1)) { 505 t = content::PageTransitionFromInt( 506 t | content::PAGE_TRANSITION_CHAIN_END); 507 } 508 509 // Record all redirect visits with the same timestamp. We don't display 510 // them anyway, and if we ever decide to, we can reconstruct their order 511 // from the redirect chain. 512 last_ids = AddPageVisit(redirects[redirect_index], 513 request.time, last_ids.second, 514 t, request.visit_source); 515 if (t & content::PAGE_TRANSITION_CHAIN_START) { 516 // Update the segment for this visit. 517 UpdateSegments(redirects[redirect_index], 518 from_visit_id, last_ids.second, t, request.time); 519 520 // Update the visit_details for this visit. 521 UpdateVisitDuration(from_visit_id, request.time); 522 } 523 524 // Subsequent transitions in the redirect list must all be server 525 // redirects. 526 redirect_info = content::PAGE_TRANSITION_SERVER_REDIRECT; 527 } 528 529 // Last, save this redirect chain for later so we can set titles & favicons 530 // on the redirected pages properly. 531 recent_redirects_.Put(request.url, redirects); 532 } 533 534 // TODO(brettw) bug 1140015: Add an "add page" notification so the history 535 // views can keep in sync. 536 537 // Add the last visit to the tracker so we can get outgoing transitions. 538 // TODO(evanm): Due to http://b/1194536 we lose the referrers of a subframe 539 // navigation anyway, so last_visit_id is always zero for them. But adding 540 // them here confuses main frame history, so we skip them for now. 541 if (stripped_transition != content::PAGE_TRANSITION_AUTO_SUBFRAME && 542 stripped_transition != content::PAGE_TRANSITION_MANUAL_SUBFRAME && 543 !is_keyword_generated) { 544 tracker_.AddVisit(request.id_scope, request.page_id, request.url, 545 last_ids.second); 546 } 547 548 ScheduleCommit(); 549} 550 551void HistoryBackend::InitImpl(const std::string& languages) { 552 DCHECK(!db_) << "Initializing HistoryBackend twice"; 553 // In the rare case where the db fails to initialize a dialog may get shown 554 // the blocks the caller, yet allows other messages through. For this reason 555 // we only set db_ to the created database if creation is successful. That 556 // way other methods won't do anything as db_ is still NULL. 557 558 TimeTicks beginning_time = TimeTicks::Now(); 559 560 // Compute the file names. 561 base::FilePath history_name = history_dir_.Append(chrome::kHistoryFilename); 562 base::FilePath thumbnail_name = GetFaviconsFileName(); 563 base::FilePath archived_name = GetArchivedFileName(); 564 565 // Delete the old index database files which are no longer used. 566 DeleteFTSIndexDatabases(); 567 568 // History database. 569 db_.reset(new HistoryDatabase()); 570 571 // Unretained to avoid a ref loop with db_. 572 db_->set_error_callback( 573 base::Bind(&HistoryBackend::DatabaseErrorCallback, 574 base::Unretained(this))); 575 576 sql::InitStatus status = db_->Init(history_name); 577 switch (status) { 578 case sql::INIT_OK: 579 break; 580 case sql::INIT_FAILURE: { 581 // A NULL db_ will cause all calls on this object to notice this error 582 // and to not continue. If the error callback scheduled killing the 583 // database, the task it posted has not executed yet. Try killing the 584 // database now before we close it. 585 bool kill_db = scheduled_kill_db_; 586 if (kill_db) 587 KillHistoryDatabase(); 588 UMA_HISTOGRAM_BOOLEAN("History.AttemptedToFixProfileError", kill_db); 589 delegate_->NotifyProfileError(id_, status); 590 db_.reset(); 591 return; 592 } 593 default: 594 NOTREACHED(); 595 } 596 597 // Fill the in-memory database and send it back to the history service on the 598 // main thread. 599 InMemoryHistoryBackend* mem_backend = new InMemoryHistoryBackend; 600 if (mem_backend->Init(history_name, db_.get())) 601 delegate_->SetInMemoryBackend(id_, mem_backend); // Takes ownership of 602 // pointer. 603 else 604 delete mem_backend; // Error case, run without the in-memory DB. 605 db_->BeginExclusiveMode(); // Must be after the mem backend read the data. 606 607 // Thumbnail database. 608 // TODO(shess): "thumbnail database" these days only stores 609 // favicons. Thumbnails are stored in "top sites". Consider 610 // renaming "thumbnail" references to "favicons" or something of the 611 // sort. 612 thumbnail_db_.reset(new ThumbnailDatabase()); 613 if (thumbnail_db_->Init(thumbnail_name) != sql::INIT_OK) { 614 // Unlike the main database, we don't error out when the database is too 615 // new because this error is much less severe. Generally, this shouldn't 616 // happen since the thumbnail and main database versions should be in sync. 617 // We'll just continue without thumbnails & favicons in this case or any 618 // other error. 619 LOG(WARNING) << "Could not initialize the thumbnail database."; 620 thumbnail_db_.reset(); 621 } 622 623 // Archived database. 624 if (db_->needs_version_17_migration()) { 625 // See needs_version_17_migration() decl for more. In this case, we want 626 // to delete the archived database and need to do so before we try to 627 // open the file. We can ignore any error (maybe the file doesn't exist). 628 sql::Connection::Delete(archived_name); 629 } 630 archived_db_.reset(new ArchivedDatabase()); 631 if (!archived_db_->Init(archived_name)) { 632 LOG(WARNING) << "Could not initialize the archived database."; 633 archived_db_.reset(); 634 } 635 636 // Generate the history and thumbnail database metrics only after performing 637 // any migration work. 638 if (base::RandInt(1, 100) == 50) { 639 // Only do this computation sometimes since it can be expensive. 640 db_->ComputeDatabaseMetrics(history_name); 641 if (thumbnail_db_) 642 thumbnail_db_->ComputeDatabaseMetrics(); 643 } 644 645 // Tell the expiration module about all the nice databases we made. This must 646 // happen before db_->Init() is called since the callback ForceArchiveHistory 647 // may need to expire stuff. 648 // 649 // *sigh*, this can all be cleaned up when that migration code is removed. 650 // The main DB initialization should intuitively be first (not that it 651 // actually matters) and the expirer should be set last. 652 expirer_.SetDatabases(db_.get(), archived_db_.get(), thumbnail_db_.get()); 653 654 // Open the long-running transaction. 655 db_->BeginTransaction(); 656 if (thumbnail_db_) 657 thumbnail_db_->BeginTransaction(); 658 if (archived_db_) 659 archived_db_->BeginTransaction(); 660 661 // Get the first item in our database. 662 db_->GetStartDate(&first_recorded_time_); 663 664 // Start expiring old stuff. 665 expirer_.StartArchivingOldStuff(TimeDelta::FromDays(kArchiveDaysThreshold)); 666 667#if defined(OS_ANDROID) 668 if (thumbnail_db_) { 669 android_provider_backend_.reset(new AndroidProviderBackend( 670 GetAndroidCacheFileName(), db_.get(), thumbnail_db_.get(), 671 bookmark_service_, delegate_.get())); 672 } 673#endif 674 675 HISTOGRAM_TIMES("History.InitTime", 676 TimeTicks::Now() - beginning_time); 677} 678 679void HistoryBackend::OnMemoryPressure( 680 base::MemoryPressureListener::MemoryPressureLevel memory_pressure_level) { 681 bool trim_aggressively = memory_pressure_level == 682 base::MemoryPressureListener::MEMORY_PRESSURE_CRITICAL; 683 if (db_) 684 db_->TrimMemory(trim_aggressively); 685 if (thumbnail_db_) 686 thumbnail_db_->TrimMemory(trim_aggressively); 687 if (archived_db_) 688 archived_db_->TrimMemory(trim_aggressively); 689} 690 691void HistoryBackend::CloseAllDatabases() { 692 if (db_) { 693 // Commit the long-running transaction. 694 db_->CommitTransaction(); 695 db_.reset(); 696 // Forget the first recorded time since the database is closed. 697 first_recorded_time_ = base::Time(); 698 } 699 if (thumbnail_db_) { 700 thumbnail_db_->CommitTransaction(); 701 thumbnail_db_.reset(); 702 } 703 if (archived_db_) { 704 archived_db_->CommitTransaction(); 705 archived_db_.reset(); 706 } 707} 708 709std::pair<URLID, VisitID> HistoryBackend::AddPageVisit( 710 const GURL& url, 711 Time time, 712 VisitID referring_visit, 713 content::PageTransition transition, 714 VisitSource visit_source) { 715 // Top-level frame navigations are visible, everything else is hidden 716 bool new_hidden = !content::PageTransitionIsMainFrame(transition); 717 718 // NOTE: This code must stay in sync with 719 // ExpireHistoryBackend::ExpireURLsForVisits(). 720 // TODO(pkasting): http://b/1148304 We shouldn't be marking so many URLs as 721 // typed, which would eliminate the need for this code. 722 int typed_increment = 0; 723 content::PageTransition transition_type = 724 content::PageTransitionStripQualifier(transition); 725 if ((transition_type == content::PAGE_TRANSITION_TYPED && 726 !content::PageTransitionIsRedirect(transition)) || 727 transition_type == content::PAGE_TRANSITION_KEYWORD_GENERATED) 728 typed_increment = 1; 729 730#if defined(OS_ANDROID) 731 // Only count the page visit if it came from user browsing and only count it 732 // once when cycling through a redirect chain. 733 if (visit_source == SOURCE_BROWSED && 734 (transition & content::PAGE_TRANSITION_CHAIN_END) != 0) { 735 RecordTopPageVisitStats(url); 736 } 737#endif 738 739 // See if this URL is already in the DB. 740 URLRow url_info(url); 741 URLID url_id = db_->GetRowForURL(url, &url_info); 742 if (url_id) { 743 // Update of an existing row. 744 if (content::PageTransitionStripQualifier(transition) != 745 content::PAGE_TRANSITION_RELOAD) 746 url_info.set_visit_count(url_info.visit_count() + 1); 747 if (typed_increment) 748 url_info.set_typed_count(url_info.typed_count() + typed_increment); 749 if (url_info.last_visit() < time) 750 url_info.set_last_visit(time); 751 752 // Only allow un-hiding of pages, never hiding. 753 if (!new_hidden) 754 url_info.set_hidden(false); 755 756 db_->UpdateURLRow(url_id, url_info); 757 } else { 758 // Addition of a new row. 759 url_info.set_visit_count(1); 760 url_info.set_typed_count(typed_increment); 761 url_info.set_last_visit(time); 762 url_info.set_hidden(new_hidden); 763 764 url_id = db_->AddURL(url_info); 765 if (!url_id) { 766 NOTREACHED() << "Adding URL failed."; 767 return std::make_pair(0, 0); 768 } 769 url_info.id_ = url_id; 770 } 771 772 // Add the visit with the time to the database. 773 VisitRow visit_info(url_id, time, referring_visit, transition, 0); 774 VisitID visit_id = db_->AddVisit(&visit_info, visit_source); 775 NotifyVisitObservers(visit_info); 776 777 if (visit_info.visit_time < first_recorded_time_) 778 first_recorded_time_ = visit_info.visit_time; 779 780 // Broadcast a notification of the visit. 781 if (visit_id) { 782 if (typed_url_syncable_service_.get()) 783 typed_url_syncable_service_->OnUrlVisited(transition, &url_info); 784 785 URLVisitedDetails* details = new URLVisitedDetails; 786 details->transition = transition; 787 details->row = url_info; 788 // TODO(meelapshah) Disabled due to potential PageCycler regression. 789 // Re-enable this. 790 // GetMostRecentRedirectsTo(url, &details->redirects); 791 BroadcastNotifications(chrome::NOTIFICATION_HISTORY_URL_VISITED, details); 792 } else { 793 VLOG(0) << "Failed to build visit insert statement: " 794 << "url_id = " << url_id; 795 } 796 797 return std::make_pair(url_id, visit_id); 798} 799 800void HistoryBackend::AddPagesWithDetails(const URLRows& urls, 801 VisitSource visit_source) { 802 if (!db_) 803 return; 804 805 scoped_ptr<URLsModifiedDetails> modified(new URLsModifiedDetails); 806 for (URLRows::const_iterator i = urls.begin(); i != urls.end(); ++i) { 807 DCHECK(!i->last_visit().is_null()); 808 809 // We will add to either the archived database or the main one depending on 810 // the date of the added visit. 811 URLDatabase* url_database; 812 VisitDatabase* visit_database; 813 if (IsExpiredVisitTime(i->last_visit())) { 814 if (!archived_db_) 815 return; // No archived database to save it to, just forget this. 816 url_database = archived_db_.get(); 817 visit_database = archived_db_.get(); 818 } else { 819 url_database = db_.get(); 820 visit_database = db_.get(); 821 } 822 823 URLRow existing_url; 824 URLID url_id = url_database->GetRowForURL(i->url(), &existing_url); 825 if (!url_id) { 826 // Add the page if it doesn't exist. 827 url_id = url_database->AddURL(*i); 828 if (!url_id) { 829 NOTREACHED() << "Could not add row to DB"; 830 return; 831 } 832 833 if (i->typed_count() > 0) { 834 modified->changed_urls.push_back(*i); 835 modified->changed_urls.back().set_id(url_id); // *i likely has |id_| 0. 836 } 837 } 838 839 // Sync code manages the visits itself. 840 if (visit_source != SOURCE_SYNCED) { 841 // Make up a visit to correspond to the last visit to the page. 842 VisitRow visit_info(url_id, i->last_visit(), 0, 843 content::PageTransitionFromInt( 844 content::PAGE_TRANSITION_LINK | 845 content::PAGE_TRANSITION_CHAIN_START | 846 content::PAGE_TRANSITION_CHAIN_END), 0); 847 if (!visit_database->AddVisit(&visit_info, visit_source)) { 848 NOTREACHED() << "Adding visit failed."; 849 return; 850 } 851 NotifyVisitObservers(visit_info); 852 853 if (visit_info.visit_time < first_recorded_time_) 854 first_recorded_time_ = visit_info.visit_time; 855 } 856 } 857 858 if (typed_url_syncable_service_.get()) 859 typed_url_syncable_service_->OnUrlsModified(&modified->changed_urls); 860 861 // Broadcast a notification for typed URLs that have been modified. This 862 // will be picked up by the in-memory URL database on the main thread. 863 // 864 // TODO(brettw) bug 1140015: Add an "add page" notification so the history 865 // views can keep in sync. 866 BroadcastNotifications(chrome::NOTIFICATION_HISTORY_URLS_MODIFIED, 867 modified.release()); 868 869 ScheduleCommit(); 870} 871 872bool HistoryBackend::IsExpiredVisitTime(const base::Time& time) { 873 return time < expirer_.GetCurrentArchiveTime(); 874} 875 876void HistoryBackend::SetPageTitle(const GURL& url, 877 const base::string16& title) { 878 if (!db_) 879 return; 880 881 // Search for recent redirects which should get the same title. We make a 882 // dummy list containing the exact URL visited if there are no redirects so 883 // the processing below can be the same. 884 history::RedirectList dummy_list; 885 history::RedirectList* redirects; 886 RedirectCache::iterator iter = recent_redirects_.Get(url); 887 if (iter != recent_redirects_.end()) { 888 redirects = &iter->second; 889 890 // This redirect chain should have the destination URL as the last item. 891 DCHECK(!redirects->empty()); 892 DCHECK(redirects->back() == url); 893 } else { 894 // No redirect chain stored, make up one containing the URL we want so we 895 // can use the same logic below. 896 dummy_list.push_back(url); 897 redirects = &dummy_list; 898 } 899 900 scoped_ptr<URLsModifiedDetails> details(new URLsModifiedDetails); 901 for (size_t i = 0; i < redirects->size(); i++) { 902 URLRow row; 903 URLID row_id = db_->GetRowForURL(redirects->at(i), &row); 904 if (row_id && row.title() != title) { 905 row.set_title(title); 906 db_->UpdateURLRow(row_id, row); 907 details->changed_urls.push_back(row); 908 } 909 } 910 911 // Broadcast notifications for any URLs that have changed. This will 912 // update the in-memory database and the InMemoryURLIndex. 913 if (!details->changed_urls.empty()) { 914 if (typed_url_syncable_service_.get()) 915 typed_url_syncable_service_->OnUrlsModified(&details->changed_urls); 916 BroadcastNotifications(chrome::NOTIFICATION_HISTORY_URLS_MODIFIED, 917 details.release()); 918 ScheduleCommit(); 919 } 920} 921 922void HistoryBackend::AddPageNoVisitForBookmark(const GURL& url, 923 const base::string16& title) { 924 if (!db_) 925 return; 926 927 URLRow url_info(url); 928 URLID url_id = db_->GetRowForURL(url, &url_info); 929 if (url_id) { 930 // URL is already known, nothing to do. 931 return; 932 } 933 934 if (!title.empty()) { 935 url_info.set_title(title); 936 } else { 937 url_info.set_title(UTF8ToUTF16(url.spec())); 938 } 939 940 url_info.set_last_visit(Time::Now()); 941 // Mark the page hidden. If the user types it in, it'll unhide. 942 url_info.set_hidden(true); 943 944 db_->AddURL(url_info); 945} 946 947void HistoryBackend::IterateURLs( 948 const scoped_refptr<visitedlink::VisitedLinkDelegate::URLEnumerator>& 949 iterator) { 950 if (db_) { 951 HistoryDatabase::URLEnumerator e; 952 if (db_->InitURLEnumeratorForEverything(&e)) { 953 URLRow info; 954 while (e.GetNextURL(&info)) { 955 iterator->OnURL(info.url()); 956 } 957 iterator->OnComplete(true); // Success. 958 return; 959 } 960 } 961 iterator->OnComplete(false); // Failure. 962} 963 964bool HistoryBackend::GetAllTypedURLs(URLRows* urls) { 965 if (db_) 966 return db_->GetAllTypedUrls(urls); 967 return false; 968} 969 970bool HistoryBackend::GetVisitsForURL(URLID id, VisitVector* visits) { 971 if (db_) 972 return db_->GetVisitsForURL(id, visits); 973 return false; 974} 975 976bool HistoryBackend::GetMostRecentVisitsForURL(URLID id, 977 int max_visits, 978 VisitVector* visits) { 979 if (db_) 980 return db_->GetMostRecentVisitsForURL(id, max_visits, visits); 981 return false; 982} 983 984bool HistoryBackend::UpdateURL(URLID id, const history::URLRow& url) { 985 if (db_) 986 return db_->UpdateURLRow(id, url); 987 return false; 988} 989 990bool HistoryBackend::AddVisits(const GURL& url, 991 const std::vector<VisitInfo>& visits, 992 VisitSource visit_source) { 993 if (db_) { 994 for (std::vector<VisitInfo>::const_iterator visit = visits.begin(); 995 visit != visits.end(); ++visit) { 996 if (!AddPageVisit( 997 url, visit->first, 0, visit->second, visit_source).first) { 998 return false; 999 } 1000 } 1001 ScheduleCommit(); 1002 return true; 1003 } 1004 return false; 1005} 1006 1007bool HistoryBackend::RemoveVisits(const VisitVector& visits) { 1008 if (!db_) 1009 return false; 1010 1011 expirer_.ExpireVisits(visits); 1012 ScheduleCommit(); 1013 return true; 1014} 1015 1016bool HistoryBackend::GetVisitsSource(const VisitVector& visits, 1017 VisitSourceMap* sources) { 1018 if (!db_) 1019 return false; 1020 1021 db_->GetVisitsSource(visits, sources); 1022 return true; 1023} 1024 1025bool HistoryBackend::GetURL(const GURL& url, history::URLRow* url_row) { 1026 if (db_) 1027 return db_->GetRowForURL(url, url_row) != 0; 1028 return false; 1029} 1030 1031void HistoryBackend::QueryURL(scoped_refptr<QueryURLRequest> request, 1032 const GURL& url, 1033 bool want_visits) { 1034 if (request->canceled()) 1035 return; 1036 1037 bool success = false; 1038 URLRow* row = &request->value.a; 1039 VisitVector* visits = &request->value.b; 1040 if (db_) { 1041 if (db_->GetRowForURL(url, row)) { 1042 // Have a row. 1043 success = true; 1044 1045 // Optionally query the visits. 1046 if (want_visits) 1047 db_->GetVisitsForURL(row->id(), visits); 1048 } 1049 } 1050 request->ForwardResult(request->handle(), success, row, visits); 1051} 1052 1053TypedUrlSyncableService* HistoryBackend::GetTypedUrlSyncableService() const { 1054 return typed_url_syncable_service_.get(); 1055} 1056 1057// Segment usage --------------------------------------------------------------- 1058 1059void HistoryBackend::DeleteOldSegmentData() { 1060 if (db_) 1061 db_->DeleteSegmentData(Time::Now() - 1062 TimeDelta::FromDays(kSegmentDataRetention)); 1063} 1064 1065void HistoryBackend::QuerySegmentUsage( 1066 scoped_refptr<QuerySegmentUsageRequest> request, 1067 const Time from_time, 1068 int max_result_count) { 1069 if (request->canceled()) 1070 return; 1071 1072 if (db_) { 1073 db_->QuerySegmentUsage(from_time, max_result_count, &request->value.get()); 1074 1075 // If this is the first time we query segments, invoke 1076 // DeleteOldSegmentData asynchronously. We do this to cleanup old 1077 // entries. 1078 if (!segment_queried_) { 1079 segment_queried_ = true; 1080 base::MessageLoop::current()->PostTask( 1081 FROM_HERE, 1082 base::Bind(&HistoryBackend::DeleteOldSegmentData, this)); 1083 } 1084 } 1085 request->ForwardResult(request->handle(), &request->value.get()); 1086} 1087 1088void HistoryBackend::IncreaseSegmentDuration(const GURL& url, 1089 base::Time time, 1090 base::TimeDelta delta) { 1091 if (!db_) 1092 return; 1093 1094 const std::string segment_name(VisitSegmentDatabase::ComputeSegmentName(url)); 1095 SegmentID segment_id = db_->GetSegmentNamed(segment_name); 1096 if (!segment_id) { 1097 URLID url_id = db_->GetRowForURL(url, NULL); 1098 if (!url_id) 1099 return; 1100 segment_id = db_->CreateSegment(url_id, segment_name); 1101 if (!segment_id) 1102 return; 1103 } 1104 SegmentDurationID duration_id; 1105 base::TimeDelta total_delta; 1106 if (!db_->GetSegmentDuration(segment_id, time, &duration_id, 1107 &total_delta)) { 1108 db_->CreateSegmentDuration(segment_id, time, delta); 1109 return; 1110 } 1111 total_delta += delta; 1112 db_->SetSegmentDuration(duration_id, total_delta); 1113} 1114 1115void HistoryBackend::QuerySegmentDuration( 1116 scoped_refptr<QuerySegmentUsageRequest> request, 1117 const base::Time from_time, 1118 int max_result_count) { 1119 if (request->canceled()) 1120 return; 1121 1122 if (db_) { 1123 db_->QuerySegmentDuration(from_time, max_result_count, 1124 &request->value.get()); 1125 } 1126 request->ForwardResult(request->handle(), &request->value.get()); 1127} 1128 1129// Keyword visits -------------------------------------------------------------- 1130 1131void HistoryBackend::SetKeywordSearchTermsForURL(const GURL& url, 1132 TemplateURLID keyword_id, 1133 const base::string16& term) { 1134 if (!db_) 1135 return; 1136 1137 // Get the ID for this URL. 1138 URLID url_id = db_->GetRowForURL(url, NULL); 1139 if (!url_id) { 1140 // There is a small possibility the url was deleted before the keyword 1141 // was added. Ignore the request. 1142 return; 1143 } 1144 1145 db_->SetKeywordSearchTermsForURL(url_id, keyword_id, term); 1146 1147 BroadcastNotifications( 1148 chrome::NOTIFICATION_HISTORY_KEYWORD_SEARCH_TERM_UPDATED, 1149 new KeywordSearchUpdatedDetails(url, keyword_id, term)); 1150 ScheduleCommit(); 1151} 1152 1153void HistoryBackend::DeleteAllSearchTermsForKeyword( 1154 TemplateURLID keyword_id) { 1155 if (!db_) 1156 return; 1157 1158 db_->DeleteAllSearchTermsForKeyword(keyword_id); 1159 // TODO(sky): bug 1168470. Need to move from archive dbs too. 1160 ScheduleCommit(); 1161} 1162 1163void HistoryBackend::GetMostRecentKeywordSearchTerms( 1164 scoped_refptr<GetMostRecentKeywordSearchTermsRequest> request, 1165 TemplateURLID keyword_id, 1166 const base::string16& prefix, 1167 int max_count) { 1168 if (request->canceled()) 1169 return; 1170 1171 if (db_) { 1172 db_->GetMostRecentKeywordSearchTerms(keyword_id, prefix, max_count, 1173 &(request->value)); 1174 } 1175 request->ForwardResult(request->handle(), &request->value); 1176} 1177 1178void HistoryBackend::DeleteKeywordSearchTermForURL(const GURL& url) { 1179 if (!db_) 1180 return; 1181 1182 URLID url_id = db_->GetRowForURL(url, NULL); 1183 if (!url_id) 1184 return; 1185 db_->DeleteKeywordSearchTermForURL(url_id); 1186 1187 BroadcastNotifications( 1188 chrome::NOTIFICATION_HISTORY_KEYWORD_SEARCH_TERM_DELETED, 1189 new KeywordSearchDeletedDetails(url)); 1190 ScheduleCommit(); 1191} 1192 1193void HistoryBackend::DeleteMatchingURLsForKeyword(TemplateURLID keyword_id, 1194 const base::string16& term) { 1195 if (!db_) 1196 return; 1197 1198 std::vector<KeywordSearchTermRow> rows; 1199 if (db_->GetKeywordSearchTermRows(term, &rows)) { 1200 std::vector<GURL> items_to_delete; 1201 URLRow row; 1202 for (std::vector<KeywordSearchTermRow>::iterator it = rows.begin(); 1203 it != rows.end(); ++it) { 1204 if (it->keyword_id == keyword_id && db_->GetURLRow(it->url_id, &row)) 1205 items_to_delete.push_back(row.url()); 1206 } 1207 DeleteURLs(items_to_delete); 1208 } 1209} 1210 1211// Downloads ------------------------------------------------------------------- 1212 1213void HistoryBackend::GetNextDownloadId(uint32* next_id) { 1214 if (db_) 1215 db_->GetNextDownloadId(next_id); 1216} 1217 1218// Get all the download entries from the database. 1219void HistoryBackend::QueryDownloads(std::vector<DownloadRow>* rows) { 1220 if (db_) 1221 db_->QueryDownloads(rows); 1222} 1223 1224// Update a particular download entry. 1225void HistoryBackend::UpdateDownload(const history::DownloadRow& data) { 1226 if (!db_) 1227 return; 1228 db_->UpdateDownload(data); 1229 ScheduleCommit(); 1230} 1231 1232void HistoryBackend::CreateDownload(const history::DownloadRow& history_info, 1233 bool* success) { 1234 if (!db_) 1235 return; 1236 *success = db_->CreateDownload(history_info); 1237 ScheduleCommit(); 1238} 1239 1240void HistoryBackend::RemoveDownloads(const std::set<uint32>& ids) { 1241 if (!db_) 1242 return; 1243 size_t downloads_count_before = db_->CountDownloads(); 1244 base::TimeTicks started_removing = base::TimeTicks::Now(); 1245 // HistoryBackend uses a long-running Transaction that is committed 1246 // periodically, so this loop doesn't actually hit the disk too hard. 1247 for (std::set<uint32>::const_iterator it = ids.begin(); 1248 it != ids.end(); ++it) { 1249 db_->RemoveDownload(*it); 1250 } 1251 ScheduleCommit(); 1252 base::TimeTicks finished_removing = base::TimeTicks::Now(); 1253 size_t downloads_count_after = db_->CountDownloads(); 1254 1255 DCHECK_LE(downloads_count_after, downloads_count_before); 1256 if (downloads_count_after > downloads_count_before) 1257 return; 1258 size_t num_downloads_deleted = downloads_count_before - downloads_count_after; 1259 UMA_HISTOGRAM_COUNTS("Download.DatabaseRemoveDownloadsCount", 1260 num_downloads_deleted); 1261 base::TimeDelta micros = (1000 * (finished_removing - started_removing)); 1262 UMA_HISTOGRAM_TIMES("Download.DatabaseRemoveDownloadsTime", micros); 1263 if (num_downloads_deleted > 0) { 1264 UMA_HISTOGRAM_TIMES("Download.DatabaseRemoveDownloadsTimePerRecord", 1265 (1000 * micros) / num_downloads_deleted); 1266 } 1267 DCHECK_GE(ids.size(), num_downloads_deleted); 1268 if (ids.size() < num_downloads_deleted) 1269 return; 1270 UMA_HISTOGRAM_COUNTS("Download.DatabaseRemoveDownloadsCountNotRemoved", 1271 ids.size() - num_downloads_deleted); 1272} 1273 1274void HistoryBackend::QueryHistory(scoped_refptr<QueryHistoryRequest> request, 1275 const base::string16& text_query, 1276 const QueryOptions& options) { 1277 if (request->canceled()) 1278 return; 1279 1280 TimeTicks beginning_time = TimeTicks::Now(); 1281 1282 if (db_) { 1283 if (text_query.empty()) { 1284 // Basic history query for the main database. 1285 QueryHistoryBasic(db_.get(), db_.get(), options, &request->value); 1286 1287 // Now query the archived database. This is a bit tricky because we don't 1288 // want to query it if the queried time range isn't going to find anything 1289 // in it. 1290 // TODO(brettw) bug 1171036: do blimpie querying for the archived database 1291 // as well. 1292 // if (archived_db_.get() && 1293 // expirer_.GetCurrentArchiveTime() - TimeDelta::FromDays(7)) { 1294 } else { 1295 // Text history query. 1296 QueryHistoryText(db_.get(), db_.get(), text_query, options, 1297 &request->value); 1298 if (archived_db_.get() && 1299 expirer_.GetCurrentArchiveTime() >= options.begin_time) { 1300 QueryHistoryText(archived_db_.get(), archived_db_.get(), text_query, 1301 options, &request->value); 1302 } 1303 } 1304 } 1305 1306 request->ForwardResult(request->handle(), &request->value); 1307 1308 UMA_HISTOGRAM_TIMES("History.QueryHistory", 1309 TimeTicks::Now() - beginning_time); 1310} 1311 1312// Basic time-based querying of history. 1313void HistoryBackend::QueryHistoryBasic(URLDatabase* url_db, 1314 VisitDatabase* visit_db, 1315 const QueryOptions& options, 1316 QueryResults* result) { 1317 // First get all visits. 1318 VisitVector visits; 1319 bool has_more_results = visit_db->GetVisibleVisitsInRange(options, &visits); 1320 DCHECK(static_cast<int>(visits.size()) <= options.EffectiveMaxCount()); 1321 1322 // Now add them and the URL rows to the results. 1323 URLResult url_result; 1324 for (size_t i = 0; i < visits.size(); i++) { 1325 const VisitRow visit = visits[i]; 1326 1327 // Add a result row for this visit, get the URL info from the DB. 1328 if (!url_db->GetURLRow(visit.url_id, &url_result)) { 1329 VLOG(0) << "Failed to get id " << visit.url_id 1330 << " from history.urls."; 1331 continue; // DB out of sync and URL doesn't exist, try to recover. 1332 } 1333 1334 if (!url_result.url().is_valid()) { 1335 VLOG(0) << "Got invalid URL from history.urls with id " 1336 << visit.url_id << ": " 1337 << url_result.url().possibly_invalid_spec(); 1338 continue; // Don't report invalid URLs in case of corruption. 1339 } 1340 1341 // The archived database may be out of sync with respect to starring, 1342 // titles, last visit date, etc. Therefore, we query the main DB if the 1343 // current URL database is not the main one. 1344 if (url_db == db_.get()) { 1345 // Currently querying the archived DB, update with the main database to 1346 // catch any interesting stuff. This will update it if it exists in the 1347 // main DB, and do nothing otherwise. 1348 db_->GetRowForURL(url_result.url(), &url_result); 1349 } 1350 1351 url_result.set_visit_time(visit.visit_time); 1352 1353 // Set whether the visit was blocked for a managed user by looking at the 1354 // transition type. 1355 url_result.set_blocked_visit( 1356 (visit.transition & content::PAGE_TRANSITION_BLOCKED) != 0); 1357 1358 // We don't set any of the query-specific parts of the URLResult, since 1359 // snippets and stuff don't apply to basic querying. 1360 result->AppendURLBySwapping(&url_result); 1361 } 1362 1363 if (!has_more_results && options.begin_time <= first_recorded_time_) 1364 result->set_reached_beginning(true); 1365} 1366 1367// Text-based querying of history. 1368void HistoryBackend::QueryHistoryText(URLDatabase* url_db, 1369 VisitDatabase* visit_db, 1370 const base::string16& text_query, 1371 const QueryOptions& options, 1372 QueryResults* result) { 1373 URLRows text_matches; 1374 url_db->GetTextMatches(text_query, &text_matches); 1375 1376 std::vector<URLResult> matching_visits; 1377 VisitVector visits; // Declare outside loop to prevent re-construction. 1378 for (size_t i = 0; i < text_matches.size(); i++) { 1379 const URLRow& text_match = text_matches[i]; 1380 // Get all visits for given URL match. 1381 visit_db->GetVisibleVisitsForURL(text_match.id(), options, &visits); 1382 for (size_t j = 0; j < visits.size(); j++) { 1383 URLResult url_result(text_match); 1384 url_result.set_visit_time(visits[j].visit_time); 1385 matching_visits.push_back(url_result); 1386 } 1387 } 1388 1389 std::sort(matching_visits.begin(), matching_visits.end(), 1390 URLResult::CompareVisitTime); 1391 1392 size_t max_results = options.max_count == 0 ? 1393 std::numeric_limits<size_t>::max() : static_cast<int>(options.max_count); 1394 for (std::vector<URLResult>::iterator it = matching_visits.begin(); 1395 it != matching_visits.end() && result->size() < max_results; ++it) { 1396 result->AppendURLBySwapping(&(*it)); 1397 } 1398 1399 if (matching_visits.size() == result->size() && 1400 options.begin_time <= first_recorded_time_) 1401 result->set_reached_beginning(true); 1402} 1403 1404// Frontend to GetMostRecentRedirectsFrom from the history thread. 1405void HistoryBackend::QueryRedirectsFrom( 1406 scoped_refptr<QueryRedirectsRequest> request, 1407 const GURL& url) { 1408 if (request->canceled()) 1409 return; 1410 bool success = GetMostRecentRedirectsFrom(url, &request->value); 1411 request->ForwardResult(request->handle(), url, success, &request->value); 1412} 1413 1414void HistoryBackend::QueryRedirectsTo( 1415 scoped_refptr<QueryRedirectsRequest> request, 1416 const GURL& url) { 1417 if (request->canceled()) 1418 return; 1419 bool success = GetMostRecentRedirectsTo(url, &request->value); 1420 request->ForwardResult(request->handle(), url, success, &request->value); 1421} 1422 1423void HistoryBackend::GetVisibleVisitCountToHost( 1424 scoped_refptr<GetVisibleVisitCountToHostRequest> request, 1425 const GURL& url) { 1426 if (request->canceled()) 1427 return; 1428 int count = 0; 1429 Time first_visit; 1430 const bool success = db_.get() && 1431 db_->GetVisibleVisitCountToHost(url, &count, &first_visit); 1432 request->ForwardResult(request->handle(), success, count, first_visit); 1433} 1434 1435void HistoryBackend::QueryTopURLsAndRedirects( 1436 scoped_refptr<QueryTopURLsAndRedirectsRequest> request, 1437 int result_count) { 1438 if (request->canceled()) 1439 return; 1440 1441 if (!db_) { 1442 request->ForwardResult(request->handle(), false, NULL, NULL); 1443 return; 1444 } 1445 1446 std::vector<GURL>* top_urls = &request->value.a; 1447 history::RedirectMap* redirects = &request->value.b; 1448 1449 ScopedVector<PageUsageData> data; 1450 db_->QuerySegmentUsage(base::Time::Now() - base::TimeDelta::FromDays(90), 1451 result_count, &data.get()); 1452 1453 for (size_t i = 0; i < data.size(); ++i) { 1454 top_urls->push_back(data[i]->GetURL()); 1455 RefCountedVector<GURL>* list = new RefCountedVector<GURL>; 1456 GetMostRecentRedirectsFrom(top_urls->back(), &list->data); 1457 (*redirects)[top_urls->back()] = list; 1458 } 1459 1460 request->ForwardResult(request->handle(), true, top_urls, redirects); 1461} 1462 1463// Will replace QueryTopURLsAndRedirectsRequest. 1464void HistoryBackend::QueryMostVisitedURLs( 1465 scoped_refptr<QueryMostVisitedURLsRequest> request, 1466 int result_count, 1467 int days_back) { 1468 if (request->canceled()) 1469 return; 1470 1471 if (!db_) { 1472 // No History Database - return an empty list. 1473 request->ForwardResult(request->handle(), MostVisitedURLList()); 1474 return; 1475 } 1476 1477 MostVisitedURLList* result = &request->value; 1478 QueryMostVisitedURLsImpl(result_count, days_back, result); 1479 request->ForwardResult(request->handle(), *result); 1480} 1481 1482void HistoryBackend::QueryFilteredURLs( 1483 scoped_refptr<QueryFilteredURLsRequest> request, 1484 int result_count, 1485 const history::VisitFilter& filter, 1486 bool extended_info) { 1487 if (request->canceled()) 1488 return; 1489 1490 base::Time request_start = base::Time::Now(); 1491 1492 if (!db_) { 1493 // No History Database - return an empty list. 1494 request->ForwardResult(request->handle(), FilteredURLList()); 1495 return; 1496 } 1497 1498 VisitVector visits; 1499 db_->GetDirectVisitsDuringTimes(filter, 0, &visits); 1500 1501 std::map<URLID, double> score_map; 1502 for (size_t i = 0; i < visits.size(); ++i) { 1503 score_map[visits[i].url_id] += filter.GetVisitScore(visits[i]); 1504 } 1505 1506 // TODO(georgey): experiment with visit_segment database granularity (it is 1507 // currently 24 hours) to use it directly instead of using visits database, 1508 // which is considerably slower. 1509 ScopedVector<PageUsageData> data; 1510 data.reserve(score_map.size()); 1511 for (std::map<URLID, double>::iterator it = score_map.begin(); 1512 it != score_map.end(); ++it) { 1513 PageUsageData* pud = new PageUsageData(it->first); 1514 pud->SetScore(it->second); 1515 data.push_back(pud); 1516 } 1517 1518 // Limit to the top |result_count| results. 1519 std::sort(data.begin(), data.end(), PageUsageData::Predicate); 1520 if (result_count && implicit_cast<int>(data.size()) > result_count) 1521 data.resize(result_count); 1522 1523 for (size_t i = 0; i < data.size(); ++i) { 1524 URLRow info; 1525 if (db_->GetURLRow(data[i]->GetID(), &info)) { 1526 data[i]->SetURL(info.url()); 1527 data[i]->SetTitle(info.title()); 1528 } 1529 } 1530 1531 FilteredURLList& result = request->value; 1532 for (size_t i = 0; i < data.size(); ++i) { 1533 PageUsageData* current_data = data[i]; 1534 FilteredURL url(*current_data); 1535 1536 if (extended_info) { 1537 VisitVector visits; 1538 db_->GetVisitsForURL(current_data->GetID(), &visits); 1539 if (visits.size() > 0) { 1540 url.extended_info.total_visits = visits.size(); 1541 for (size_t i = 0; i < visits.size(); ++i) { 1542 url.extended_info.duration_opened += 1543 visits[i].visit_duration.InSeconds(); 1544 if (visits[i].visit_time > url.extended_info.last_visit_time) { 1545 url.extended_info.last_visit_time = visits[i].visit_time; 1546 } 1547 } 1548 // TODO(macourteau): implement the url.extended_info.visits stat. 1549 } 1550 } 1551 result.push_back(url); 1552 } 1553 1554 int delta_time = std::max(1, std::min(999, 1555 static_cast<int>((base::Time::Now() - request_start).InMilliseconds()))); 1556 STATIC_HISTOGRAM_POINTER_BLOCK( 1557 "NewTabPage.SuggestedSitesLoadTime", 1558 Add(delta_time), 1559 base::LinearHistogram::FactoryGet("NewTabPage.SuggestedSitesLoadTime", 1560 1, 1000, 100, base::Histogram::kUmaTargetedHistogramFlag)); 1561 1562 request->ForwardResult(request->handle(), result); 1563} 1564 1565void HistoryBackend::QueryMostVisitedURLsImpl(int result_count, 1566 int days_back, 1567 MostVisitedURLList* result) { 1568 if (!db_) 1569 return; 1570 1571 ScopedVector<PageUsageData> data; 1572 db_->QuerySegmentUsage(base::Time::Now() - 1573 base::TimeDelta::FromDays(days_back), 1574 result_count, &data.get()); 1575 1576 for (size_t i = 0; i < data.size(); ++i) { 1577 PageUsageData* current_data = data[i]; 1578 RedirectList redirects; 1579 GetMostRecentRedirectsFrom(current_data->GetURL(), &redirects); 1580 MostVisitedURL url = MakeMostVisitedURL(*current_data, redirects); 1581 result->push_back(url); 1582 } 1583} 1584 1585void HistoryBackend::GetRedirectsFromSpecificVisit( 1586 VisitID cur_visit, history::RedirectList* redirects) { 1587 // Follow any redirects from the given visit and add them to the list. 1588 // It *should* be impossible to get a circular chain here, but we check 1589 // just in case to avoid infinite loops. 1590 GURL cur_url; 1591 std::set<VisitID> visit_set; 1592 visit_set.insert(cur_visit); 1593 while (db_->GetRedirectFromVisit(cur_visit, &cur_visit, &cur_url)) { 1594 if (visit_set.find(cur_visit) != visit_set.end()) { 1595 NOTREACHED() << "Loop in visit chain, giving up"; 1596 return; 1597 } 1598 visit_set.insert(cur_visit); 1599 redirects->push_back(cur_url); 1600 } 1601} 1602 1603void HistoryBackend::GetRedirectsToSpecificVisit( 1604 VisitID cur_visit, 1605 history::RedirectList* redirects) { 1606 // Follow redirects going to cur_visit. These are added to |redirects| in 1607 // the order they are found. If a redirect chain looks like A -> B -> C and 1608 // |cur_visit| = C, redirects will be {B, A} in that order. 1609 if (!db_) 1610 return; 1611 1612 GURL cur_url; 1613 std::set<VisitID> visit_set; 1614 visit_set.insert(cur_visit); 1615 while (db_->GetRedirectToVisit(cur_visit, &cur_visit, &cur_url)) { 1616 if (visit_set.find(cur_visit) != visit_set.end()) { 1617 NOTREACHED() << "Loop in visit chain, giving up"; 1618 return; 1619 } 1620 visit_set.insert(cur_visit); 1621 redirects->push_back(cur_url); 1622 } 1623} 1624 1625bool HistoryBackend::GetMostRecentRedirectsFrom( 1626 const GURL& from_url, 1627 history::RedirectList* redirects) { 1628 redirects->clear(); 1629 if (!db_) 1630 return false; 1631 1632 URLID from_url_id = db_->GetRowForURL(from_url, NULL); 1633 VisitID cur_visit = db_->GetMostRecentVisitForURL(from_url_id, NULL); 1634 if (!cur_visit) 1635 return false; // No visits for URL. 1636 1637 GetRedirectsFromSpecificVisit(cur_visit, redirects); 1638 return true; 1639} 1640 1641bool HistoryBackend::GetMostRecentRedirectsTo( 1642 const GURL& to_url, 1643 history::RedirectList* redirects) { 1644 redirects->clear(); 1645 if (!db_) 1646 return false; 1647 1648 URLID to_url_id = db_->GetRowForURL(to_url, NULL); 1649 VisitID cur_visit = db_->GetMostRecentVisitForURL(to_url_id, NULL); 1650 if (!cur_visit) 1651 return false; // No visits for URL. 1652 1653 GetRedirectsToSpecificVisit(cur_visit, redirects); 1654 return true; 1655} 1656 1657void HistoryBackend::ScheduleAutocomplete(HistoryURLProvider* provider, 1658 HistoryURLProviderParams* params) { 1659 // ExecuteWithDB should handle the NULL database case. 1660 provider->ExecuteWithDB(this, db_.get(), params); 1661} 1662 1663void HistoryBackend::DeleteFTSIndexDatabases() { 1664 // Find files on disk matching the text databases file pattern so we can 1665 // quickly test for and delete them. 1666 base::FilePath::StringType filepattern = 1667 FILE_PATH_LITERAL("History Index *"); 1668 base::FileEnumerator enumerator( 1669 history_dir_, false, base::FileEnumerator::FILES, filepattern); 1670 int num_databases_deleted = 0; 1671 base::FilePath current_file; 1672 while (!(current_file = enumerator.Next()).empty()) { 1673 if (sql::Connection::Delete(current_file)) 1674 num_databases_deleted++; 1675 } 1676 UMA_HISTOGRAM_COUNTS("History.DeleteFTSIndexDatabases", 1677 num_databases_deleted); 1678} 1679 1680void HistoryBackend::GetFavicons( 1681 const std::vector<GURL>& icon_urls, 1682 int icon_types, 1683 int desired_size_in_dip, 1684 const std::vector<ui::ScaleFactor>& desired_scale_factors, 1685 std::vector<chrome::FaviconBitmapResult>* bitmap_results) { 1686 UpdateFaviconMappingsAndFetchImpl(NULL, icon_urls, icon_types, 1687 desired_size_in_dip, desired_scale_factors, 1688 bitmap_results); 1689} 1690 1691void HistoryBackend::GetLargestFaviconForURL( 1692 const GURL& page_url, 1693 const std::vector<int>& icon_types, 1694 int minimum_size_in_pixels, 1695 chrome::FaviconBitmapResult* favicon_bitmap_result) { 1696 DCHECK(favicon_bitmap_result); 1697 1698 if (!db_ || !thumbnail_db_) 1699 return; 1700 1701 TimeTicks beginning_time = TimeTicks::Now(); 1702 1703 std::vector<IconMapping> icon_mappings; 1704 if (!thumbnail_db_->GetIconMappingsForPageURL(page_url, &icon_mappings) || 1705 icon_mappings.empty()) 1706 return; 1707 1708 int required_icon_types = 0; 1709 for (std::vector<int>::const_iterator i = icon_types.begin(); 1710 i != icon_types.end(); ++i) { 1711 required_icon_types |= *i; 1712 } 1713 1714 // Find the largest bitmap for each IconType placing in 1715 // |largest_favicon_bitmaps|. 1716 std::map<chrome::IconType, FaviconBitmap> largest_favicon_bitmaps; 1717 for (std::vector<IconMapping>::const_iterator i = icon_mappings.begin(); 1718 i != icon_mappings.end(); ++i) { 1719 if (!(i->icon_type & required_icon_types)) 1720 continue; 1721 std::vector<FaviconBitmapIDSize> bitmap_id_sizes; 1722 thumbnail_db_->GetFaviconBitmapIDSizes(i->icon_id, &bitmap_id_sizes); 1723 FaviconBitmap& largest = largest_favicon_bitmaps[i->icon_type]; 1724 for (std::vector<FaviconBitmapIDSize>::const_iterator j = 1725 bitmap_id_sizes.begin(); j != bitmap_id_sizes.end(); ++j) { 1726 if (largest.bitmap_id == 0 || 1727 (largest.pixel_size.width() < j->pixel_size.width() && 1728 largest.pixel_size.height() < j->pixel_size.height())) { 1729 largest.icon_id = i->icon_id; 1730 largest.bitmap_id = j->bitmap_id; 1731 largest.pixel_size = j->pixel_size; 1732 } 1733 } 1734 } 1735 if (largest_favicon_bitmaps.empty()) 1736 return; 1737 1738 // Find an icon which is larger than minimum_size_in_pixels in the order of 1739 // icon_types. 1740 FaviconBitmap largest_icon; 1741 for (std::vector<int>::const_iterator t = icon_types.begin(); 1742 t != icon_types.end(); ++t) { 1743 for (std::map<chrome::IconType, FaviconBitmap>::const_iterator f = 1744 largest_favicon_bitmaps.begin(); f != largest_favicon_bitmaps.end(); 1745 ++f) { 1746 if (f->first & *t && 1747 (largest_icon.bitmap_id == 0 || 1748 (largest_icon.pixel_size.height() < f->second.pixel_size.height() && 1749 largest_icon.pixel_size.width() < f->second.pixel_size.width()))) { 1750 largest_icon = f->second; 1751 } 1752 } 1753 if (largest_icon.pixel_size.width() > minimum_size_in_pixels && 1754 largest_icon.pixel_size.height() > minimum_size_in_pixels) 1755 break; 1756 } 1757 1758 GURL icon_url; 1759 chrome::IconType icon_type; 1760 if (!thumbnail_db_->GetFaviconHeader(largest_icon.icon_id, &icon_url, 1761 &icon_type)) { 1762 return; 1763 } 1764 1765 base::Time last_updated; 1766 chrome::FaviconBitmapResult bitmap_result; 1767 bitmap_result.icon_url = icon_url; 1768 bitmap_result.icon_type = icon_type; 1769 if (!thumbnail_db_->GetFaviconBitmap(largest_icon.bitmap_id, 1770 &last_updated, 1771 &bitmap_result.bitmap_data, 1772 &bitmap_result.pixel_size)) { 1773 return; 1774 } 1775 1776 bitmap_result.expired = (Time::Now() - last_updated) > 1777 TimeDelta::FromDays(kFaviconRefetchDays); 1778 if (bitmap_result.is_valid()) 1779 *favicon_bitmap_result = bitmap_result; 1780 1781 HISTOGRAM_TIMES("History.GetLargestFaviconForURL", 1782 TimeTicks::Now() - beginning_time); 1783} 1784 1785void HistoryBackend::GetFaviconsForURL( 1786 const GURL& page_url, 1787 int icon_types, 1788 int desired_size_in_dip, 1789 const std::vector<ui::ScaleFactor>& desired_scale_factors, 1790 std::vector<chrome::FaviconBitmapResult>* bitmap_results) { 1791 DCHECK(bitmap_results); 1792 GetFaviconsFromDB(page_url, icon_types, desired_size_in_dip, 1793 desired_scale_factors, bitmap_results); 1794} 1795 1796void HistoryBackend::GetFaviconForID( 1797 chrome::FaviconID favicon_id, 1798 int desired_size_in_dip, 1799 ui::ScaleFactor desired_scale_factor, 1800 std::vector<chrome::FaviconBitmapResult>* bitmap_results) { 1801 std::vector<chrome::FaviconID> favicon_ids; 1802 favicon_ids.push_back(favicon_id); 1803 std::vector<ui::ScaleFactor> desired_scale_factors; 1804 desired_scale_factors.push_back(desired_scale_factor); 1805 1806 // Get results from DB. 1807 GetFaviconBitmapResultsForBestMatch(favicon_ids, 1808 desired_size_in_dip, 1809 desired_scale_factors, 1810 bitmap_results); 1811} 1812 1813void HistoryBackend::UpdateFaviconMappingsAndFetch( 1814 const GURL& page_url, 1815 const std::vector<GURL>& icon_urls, 1816 int icon_types, 1817 int desired_size_in_dip, 1818 const std::vector<ui::ScaleFactor>& desired_scale_factors, 1819 std::vector<chrome::FaviconBitmapResult>* bitmap_results) { 1820 UpdateFaviconMappingsAndFetchImpl(&page_url, icon_urls, icon_types, 1821 desired_size_in_dip, desired_scale_factors, 1822 bitmap_results); 1823} 1824 1825void HistoryBackend::MergeFavicon( 1826 const GURL& page_url, 1827 const GURL& icon_url, 1828 chrome::IconType icon_type, 1829 scoped_refptr<base::RefCountedMemory> bitmap_data, 1830 const gfx::Size& pixel_size) { 1831 if (!thumbnail_db_ || !db_) 1832 return; 1833 1834 chrome::FaviconID favicon_id = 1835 thumbnail_db_->GetFaviconIDForFaviconURL(icon_url, icon_type, NULL); 1836 1837 if (!favicon_id) { 1838 // There is no favicon at |icon_url|, create it. 1839 favicon_id = thumbnail_db_->AddFavicon(icon_url, icon_type); 1840 } 1841 1842 std::vector<FaviconBitmapIDSize> bitmap_id_sizes; 1843 thumbnail_db_->GetFaviconBitmapIDSizes(favicon_id, &bitmap_id_sizes); 1844 1845 // If there is already a favicon bitmap of |pixel_size| at |icon_url|, 1846 // replace it. 1847 bool bitmap_identical = false; 1848 bool replaced_bitmap = false; 1849 for (size_t i = 0; i < bitmap_id_sizes.size(); ++i) { 1850 if (bitmap_id_sizes[i].pixel_size == pixel_size) { 1851 if (IsFaviconBitmapDataEqual(bitmap_id_sizes[i].bitmap_id, bitmap_data)) { 1852 thumbnail_db_->SetFaviconBitmapLastUpdateTime( 1853 bitmap_id_sizes[i].bitmap_id, base::Time::Now()); 1854 bitmap_identical = true; 1855 } else { 1856 thumbnail_db_->SetFaviconBitmap(bitmap_id_sizes[i].bitmap_id, 1857 bitmap_data, base::Time::Now()); 1858 replaced_bitmap = true; 1859 } 1860 break; 1861 } 1862 } 1863 1864 // Create a vector of the pixel sizes of the favicon bitmaps currently at 1865 // |icon_url|. 1866 std::vector<gfx::Size> favicon_sizes; 1867 for (size_t i = 0; i < bitmap_id_sizes.size(); ++i) 1868 favicon_sizes.push_back(bitmap_id_sizes[i].pixel_size); 1869 1870 if (!replaced_bitmap && !bitmap_identical) { 1871 // Set the preexisting favicon bitmaps as expired as the preexisting favicon 1872 // bitmaps are not consistent with the merged in data. 1873 thumbnail_db_->SetFaviconOutOfDate(favicon_id); 1874 1875 // Delete an arbitrary favicon bitmap to avoid going over the limit of 1876 // |kMaxFaviconBitmapsPerIconURL|. 1877 if (bitmap_id_sizes.size() >= kMaxFaviconBitmapsPerIconURL) { 1878 thumbnail_db_->DeleteFaviconBitmap(bitmap_id_sizes[0].bitmap_id); 1879 favicon_sizes.erase(favicon_sizes.begin()); 1880 } 1881 thumbnail_db_->AddFaviconBitmap(favicon_id, bitmap_data, base::Time::Now(), 1882 pixel_size); 1883 favicon_sizes.push_back(pixel_size); 1884 } 1885 1886 // A site may have changed the favicons that it uses for |page_url|. 1887 // Example Scenario: 1888 // page_url = news.google.com 1889 // Initial State: www.google.com/favicon.ico 16x16, 32x32 1890 // MergeFavicon(news.google.com, news.google.com/news_specific.ico, ..., 1891 // ..., 16x16) 1892 // 1893 // Difficulties: 1894 // 1. Sync requires that a call to GetFaviconsForURL() returns the 1895 // |bitmap_data| passed into MergeFavicon(). 1896 // - It is invalid for the 16x16 bitmap for www.google.com/favicon.ico to 1897 // stay mapped to news.google.com because it would be unclear which 16x16 1898 // bitmap should be returned via GetFaviconsForURL(). 1899 // 1900 // 2. www.google.com/favicon.ico may be mapped to more than just 1901 // news.google.com (eg www.google.com). 1902 // - The 16x16 bitmap cannot be deleted from www.google.com/favicon.ico 1903 // 1904 // To resolve these problems, we copy all of the favicon bitmaps previously 1905 // mapped to news.google.com (|page_url|) and add them to the favicon at 1906 // news.google.com/news_specific.ico (|icon_url|). The favicon sizes for 1907 // |icon_url| are set to default to indicate that |icon_url| has incomplete 1908 // / incorrect data. 1909 // Difficulty 1: All but news.google.com/news_specific.ico are unmapped from 1910 // news.google.com 1911 // Difficulty 2: The favicon bitmaps for www.google.com/favicon.ico are not 1912 // modified. 1913 1914 std::vector<IconMapping> icon_mappings; 1915 thumbnail_db_->GetIconMappingsForPageURL(page_url, icon_type, &icon_mappings); 1916 1917 // Copy the favicon bitmaps mapped to |page_url| to the favicon at |icon_url| 1918 // till the limit of |kMaxFaviconBitmapsPerIconURL| is reached. 1919 for (size_t i = 0; i < icon_mappings.size(); ++i) { 1920 if (favicon_sizes.size() >= kMaxFaviconBitmapsPerIconURL) 1921 break; 1922 1923 if (icon_mappings[i].icon_url == icon_url) 1924 continue; 1925 1926 std::vector<FaviconBitmap> bitmaps_to_copy; 1927 thumbnail_db_->GetFaviconBitmaps(icon_mappings[i].icon_id, 1928 &bitmaps_to_copy); 1929 for (size_t j = 0; j < bitmaps_to_copy.size(); ++j) { 1930 // Do not add a favicon bitmap at a pixel size for which there is already 1931 // a favicon bitmap mapped to |icon_url|. The one there is more correct 1932 // and having multiple equally sized favicon bitmaps for |page_url| is 1933 // ambiguous in terms of GetFaviconsForURL(). 1934 std::vector<gfx::Size>::iterator it = std::find(favicon_sizes.begin(), 1935 favicon_sizes.end(), bitmaps_to_copy[j].pixel_size); 1936 if (it != favicon_sizes.end()) 1937 continue; 1938 1939 // Add the favicon bitmap as expired as it is not consistent with the 1940 // merged in data. 1941 thumbnail_db_->AddFaviconBitmap(favicon_id, 1942 bitmaps_to_copy[j].bitmap_data, base::Time(), 1943 bitmaps_to_copy[j].pixel_size); 1944 favicon_sizes.push_back(bitmaps_to_copy[j].pixel_size); 1945 1946 if (favicon_sizes.size() >= kMaxFaviconBitmapsPerIconURL) 1947 break; 1948 } 1949 } 1950 1951 // Update the favicon mappings such that only |icon_url| is mapped to 1952 // |page_url|. 1953 bool mapping_changed = false; 1954 if (icon_mappings.size() != 1 || icon_mappings[0].icon_url != icon_url) { 1955 std::vector<chrome::FaviconID> favicon_ids; 1956 favicon_ids.push_back(favicon_id); 1957 SetFaviconMappingsForPageAndRedirects(page_url, icon_type, favicon_ids); 1958 mapping_changed = true; 1959 } 1960 1961 if (mapping_changed || !bitmap_identical) 1962 SendFaviconChangedNotificationForPageAndRedirects(page_url); 1963 ScheduleCommit(); 1964} 1965 1966void HistoryBackend::SetFavicons( 1967 const GURL& page_url, 1968 chrome::IconType icon_type, 1969 const std::vector<chrome::FaviconBitmapData>& favicon_bitmap_data) { 1970 if (!thumbnail_db_ || !db_) 1971 return; 1972 1973 DCHECK(ValidateSetFaviconsParams(favicon_bitmap_data)); 1974 1975 // Build map of FaviconBitmapData for each icon url. 1976 typedef std::map<GURL, std::vector<chrome::FaviconBitmapData> > 1977 BitmapDataByIconURL; 1978 BitmapDataByIconURL grouped_by_icon_url; 1979 for (size_t i = 0; i < favicon_bitmap_data.size(); ++i) { 1980 const GURL& icon_url = favicon_bitmap_data[i].icon_url; 1981 grouped_by_icon_url[icon_url].push_back(favicon_bitmap_data[i]); 1982 } 1983 1984 // Track whether the method modifies or creates any favicon bitmaps, favicons 1985 // or icon mappings. 1986 bool data_modified = false; 1987 1988 std::vector<chrome::FaviconID> icon_ids; 1989 for (BitmapDataByIconURL::const_iterator it = grouped_by_icon_url.begin(); 1990 it != grouped_by_icon_url.end(); ++it) { 1991 const GURL& icon_url = it->first; 1992 chrome::FaviconID icon_id = 1993 thumbnail_db_->GetFaviconIDForFaviconURL(icon_url, icon_type, NULL); 1994 1995 if (!icon_id) { 1996 // TODO(pkotwicz): Remove the favicon sizes attribute from 1997 // ThumbnailDatabase::AddFavicon(). 1998 icon_id = thumbnail_db_->AddFavicon(icon_url, icon_type); 1999 data_modified = true; 2000 } 2001 icon_ids.push_back(icon_id); 2002 2003 if (!data_modified) 2004 SetFaviconBitmaps(icon_id, it->second, &data_modified); 2005 else 2006 SetFaviconBitmaps(icon_id, it->second, NULL); 2007 } 2008 2009 data_modified |= 2010 SetFaviconMappingsForPageAndRedirects(page_url, icon_type, icon_ids); 2011 2012 if (data_modified) { 2013 // Send notification to the UI as an icon mapping, favicon, or favicon 2014 // bitmap was changed by this function. 2015 SendFaviconChangedNotificationForPageAndRedirects(page_url); 2016 } 2017 ScheduleCommit(); 2018} 2019 2020void HistoryBackend::SetFaviconsOutOfDateForPage(const GURL& page_url) { 2021 std::vector<IconMapping> icon_mappings; 2022 2023 if (!thumbnail_db_ || 2024 !thumbnail_db_->GetIconMappingsForPageURL(page_url, 2025 &icon_mappings)) 2026 return; 2027 2028 for (std::vector<IconMapping>::iterator m = icon_mappings.begin(); 2029 m != icon_mappings.end(); ++m) { 2030 thumbnail_db_->SetFaviconOutOfDate(m->icon_id); 2031 } 2032 ScheduleCommit(); 2033} 2034 2035void HistoryBackend::CloneFavicons(const GURL& old_page_url, 2036 const GURL& new_page_url) { 2037 if (!thumbnail_db_) 2038 return; 2039 2040 // Prevent cross-domain cloning. 2041 if (old_page_url.GetOrigin() != new_page_url.GetOrigin()) 2042 return; 2043 2044 thumbnail_db_->CloneIconMappings(old_page_url, new_page_url); 2045 ScheduleCommit(); 2046} 2047 2048void HistoryBackend::SetImportedFavicons( 2049 const std::vector<ImportedFaviconUsage>& favicon_usage) { 2050 if (!db_ || !thumbnail_db_) 2051 return; 2052 2053 Time now = Time::Now(); 2054 2055 // Track all URLs that had their favicons set or updated. 2056 std::set<GURL> favicons_changed; 2057 2058 for (size_t i = 0; i < favicon_usage.size(); i++) { 2059 chrome::FaviconID favicon_id = thumbnail_db_->GetFaviconIDForFaviconURL( 2060 favicon_usage[i].favicon_url, chrome::FAVICON, NULL); 2061 if (!favicon_id) { 2062 // This favicon doesn't exist yet, so we create it using the given data. 2063 // TODO(pkotwicz): Pass in real pixel size. 2064 favicon_id = thumbnail_db_->AddFavicon( 2065 favicon_usage[i].favicon_url, 2066 chrome::FAVICON, 2067 new base::RefCountedBytes(favicon_usage[i].png_data), 2068 now, 2069 gfx::Size()); 2070 } 2071 2072 // Save the mapping from all the URLs to the favicon. 2073 BookmarkService* bookmark_service = GetBookmarkService(); 2074 for (std::set<GURL>::const_iterator url = favicon_usage[i].urls.begin(); 2075 url != favicon_usage[i].urls.end(); ++url) { 2076 URLRow url_row; 2077 if (!db_->GetRowForURL(*url, &url_row)) { 2078 // If the URL is present as a bookmark, add the url in history to 2079 // save the favicon mapping. This will match with what history db does 2080 // for regular bookmarked URLs with favicons - when history db is 2081 // cleaned, we keep an entry in the db with 0 visits as long as that 2082 // url is bookmarked. 2083 if (bookmark_service && bookmark_service_->IsBookmarked(*url)) { 2084 URLRow url_info(*url); 2085 url_info.set_visit_count(0); 2086 url_info.set_typed_count(0); 2087 url_info.set_last_visit(base::Time()); 2088 url_info.set_hidden(false); 2089 db_->AddURL(url_info); 2090 thumbnail_db_->AddIconMapping(*url, favicon_id); 2091 favicons_changed.insert(*url); 2092 } 2093 } else { 2094 if (!thumbnail_db_->GetIconMappingsForPageURL( 2095 *url, chrome::FAVICON, NULL)) { 2096 // URL is present in history, update the favicon *only* if it is not 2097 // set already. 2098 thumbnail_db_->AddIconMapping(*url, favicon_id); 2099 favicons_changed.insert(*url); 2100 } 2101 } 2102 } 2103 } 2104 2105 if (!favicons_changed.empty()) { 2106 // Send the notification about the changed favicon URLs. 2107 FaviconChangedDetails* changed_details = new FaviconChangedDetails; 2108 changed_details->urls.swap(favicons_changed); 2109 BroadcastNotifications(chrome::NOTIFICATION_FAVICON_CHANGED, 2110 changed_details); 2111 } 2112} 2113 2114void HistoryBackend::UpdateFaviconMappingsAndFetchImpl( 2115 const GURL* page_url, 2116 const std::vector<GURL>& icon_urls, 2117 int icon_types, 2118 int desired_size_in_dip, 2119 const std::vector<ui::ScaleFactor>& desired_scale_factors, 2120 std::vector<chrome::FaviconBitmapResult>* bitmap_results) { 2121 // If |page_url| is specified, |icon_types| must be either a single icon 2122 // type or icon types which are equivalent. 2123 DCHECK(!page_url || 2124 icon_types == chrome::FAVICON || 2125 icon_types == chrome::TOUCH_ICON || 2126 icon_types == chrome::TOUCH_PRECOMPOSED_ICON || 2127 icon_types == (chrome::TOUCH_ICON | chrome::TOUCH_PRECOMPOSED_ICON)); 2128 bitmap_results->clear(); 2129 2130 if (!thumbnail_db_) { 2131 return; 2132 } 2133 2134 std::vector<chrome::FaviconID> favicon_ids; 2135 2136 // The icon type for which the mappings will the updated and data will be 2137 // returned. 2138 chrome::IconType selected_icon_type = chrome::INVALID_ICON; 2139 2140 for (size_t i = 0; i < icon_urls.size(); ++i) { 2141 const GURL& icon_url = icon_urls[i]; 2142 chrome::IconType icon_type_out; 2143 const chrome::FaviconID favicon_id = 2144 thumbnail_db_->GetFaviconIDForFaviconURL( 2145 icon_url, icon_types, &icon_type_out); 2146 2147 if (favicon_id) { 2148 // Return and update icon mappings only for the largest icon type. As 2149 // |icon_urls| is not sorted in terms of icon type, clear |favicon_ids| 2150 // if an |icon_url| with a larger icon type is found. 2151 if (icon_type_out > selected_icon_type) { 2152 selected_icon_type = icon_type_out; 2153 favicon_ids.clear(); 2154 } 2155 if (icon_type_out == selected_icon_type) 2156 favicon_ids.push_back(favicon_id); 2157 } 2158 } 2159 2160 if (page_url && !favicon_ids.empty()) { 2161 bool mappings_updated = 2162 SetFaviconMappingsForPageAndRedirects(*page_url, selected_icon_type, 2163 favicon_ids); 2164 if (mappings_updated) { 2165 SendFaviconChangedNotificationForPageAndRedirects(*page_url); 2166 ScheduleCommit(); 2167 } 2168 } 2169 2170 GetFaviconBitmapResultsForBestMatch(favicon_ids, desired_size_in_dip, 2171 desired_scale_factors, bitmap_results); 2172} 2173 2174void HistoryBackend::SetFaviconBitmaps( 2175 chrome::FaviconID icon_id, 2176 const std::vector<chrome::FaviconBitmapData>& favicon_bitmap_data, 2177 bool* favicon_bitmaps_changed) { 2178 if (favicon_bitmaps_changed) 2179 *favicon_bitmaps_changed = false; 2180 2181 std::vector<FaviconBitmapIDSize> bitmap_id_sizes; 2182 thumbnail_db_->GetFaviconBitmapIDSizes(icon_id, &bitmap_id_sizes); 2183 2184 std::vector<chrome::FaviconBitmapData> to_add = favicon_bitmap_data; 2185 2186 for (size_t i = 0; i < bitmap_id_sizes.size(); ++i) { 2187 const gfx::Size& pixel_size = bitmap_id_sizes[i].pixel_size; 2188 std::vector<chrome::FaviconBitmapData>::iterator match_it = to_add.end(); 2189 for (std::vector<chrome::FaviconBitmapData>::iterator it = to_add.begin(); 2190 it != to_add.end(); ++it) { 2191 if (it->pixel_size == pixel_size) { 2192 match_it = it; 2193 break; 2194 } 2195 } 2196 2197 FaviconBitmapID bitmap_id = bitmap_id_sizes[i].bitmap_id; 2198 if (match_it == to_add.end()) { 2199 thumbnail_db_->DeleteFaviconBitmap(bitmap_id); 2200 2201 if (favicon_bitmaps_changed) 2202 *favicon_bitmaps_changed = true; 2203 } else { 2204 if (favicon_bitmaps_changed && 2205 !*favicon_bitmaps_changed && 2206 IsFaviconBitmapDataEqual(bitmap_id, match_it->bitmap_data)) { 2207 thumbnail_db_->SetFaviconBitmapLastUpdateTime( 2208 bitmap_id, base::Time::Now()); 2209 } else { 2210 thumbnail_db_->SetFaviconBitmap(bitmap_id, match_it->bitmap_data, 2211 base::Time::Now()); 2212 2213 if (favicon_bitmaps_changed) 2214 *favicon_bitmaps_changed = true; 2215 } 2216 to_add.erase(match_it); 2217 } 2218 } 2219 2220 for (size_t i = 0; i < to_add.size(); ++i) { 2221 thumbnail_db_->AddFaviconBitmap(icon_id, to_add[i].bitmap_data, 2222 base::Time::Now(), to_add[i].pixel_size); 2223 2224 if (favicon_bitmaps_changed) 2225 *favicon_bitmaps_changed = true; 2226 } 2227} 2228 2229bool HistoryBackend::ValidateSetFaviconsParams( 2230 const std::vector<chrome::FaviconBitmapData>& favicon_bitmap_data) const { 2231 typedef std::map<GURL, size_t> BitmapsPerIconURL; 2232 BitmapsPerIconURL num_bitmaps_per_icon_url; 2233 for (size_t i = 0; i < favicon_bitmap_data.size(); ++i) { 2234 if (!favicon_bitmap_data[i].bitmap_data.get()) 2235 return false; 2236 2237 const GURL& icon_url = favicon_bitmap_data[i].icon_url; 2238 if (!num_bitmaps_per_icon_url.count(icon_url)) 2239 num_bitmaps_per_icon_url[icon_url] = 1u; 2240 else 2241 ++num_bitmaps_per_icon_url[icon_url]; 2242 } 2243 2244 if (num_bitmaps_per_icon_url.size() > kMaxFaviconsPerPage) 2245 return false; 2246 2247 for (BitmapsPerIconURL::const_iterator it = num_bitmaps_per_icon_url.begin(); 2248 it != num_bitmaps_per_icon_url.end(); ++it) { 2249 if (it->second > kMaxFaviconBitmapsPerIconURL) 2250 return false; 2251 } 2252 return true; 2253} 2254 2255bool HistoryBackend::IsFaviconBitmapDataEqual( 2256 FaviconBitmapID bitmap_id, 2257 const scoped_refptr<base::RefCountedMemory>& new_bitmap_data) { 2258 if (!new_bitmap_data.get()) 2259 return false; 2260 2261 scoped_refptr<base::RefCountedMemory> original_bitmap_data; 2262 thumbnail_db_->GetFaviconBitmap(bitmap_id, 2263 NULL, 2264 &original_bitmap_data, 2265 NULL); 2266 return new_bitmap_data->Equals(original_bitmap_data); 2267} 2268 2269bool HistoryBackend::GetFaviconsFromDB( 2270 const GURL& page_url, 2271 int icon_types, 2272 int desired_size_in_dip, 2273 const std::vector<ui::ScaleFactor>& desired_scale_factors, 2274 std::vector<chrome::FaviconBitmapResult>* favicon_bitmap_results) { 2275 DCHECK(favicon_bitmap_results); 2276 favicon_bitmap_results->clear(); 2277 2278 if (!db_ || !thumbnail_db_) 2279 return false; 2280 2281 // Time the query. 2282 TimeTicks beginning_time = TimeTicks::Now(); 2283 2284 // Get FaviconIDs for |page_url| and one of |icon_types|. 2285 std::vector<IconMapping> icon_mappings; 2286 thumbnail_db_->GetIconMappingsForPageURL(page_url, icon_types, 2287 &icon_mappings); 2288 std::vector<chrome::FaviconID> favicon_ids; 2289 for (size_t i = 0; i < icon_mappings.size(); ++i) 2290 favicon_ids.push_back(icon_mappings[i].icon_id); 2291 2292 // Populate |favicon_bitmap_results| and |icon_url_sizes|. 2293 bool success = GetFaviconBitmapResultsForBestMatch(favicon_ids, 2294 desired_size_in_dip, desired_scale_factors, favicon_bitmap_results); 2295 UMA_HISTOGRAM_TIMES("History.GetFavIconFromDB", // historical name 2296 TimeTicks::Now() - beginning_time); 2297 return success && !favicon_bitmap_results->empty(); 2298} 2299 2300bool HistoryBackend::GetFaviconBitmapResultsForBestMatch( 2301 const std::vector<chrome::FaviconID>& candidate_favicon_ids, 2302 int desired_size_in_dip, 2303 const std::vector<ui::ScaleFactor>& desired_scale_factors, 2304 std::vector<chrome::FaviconBitmapResult>* favicon_bitmap_results) { 2305 favicon_bitmap_results->clear(); 2306 2307 if (candidate_favicon_ids.empty()) 2308 return true; 2309 2310 // Find the FaviconID and the FaviconBitmapIDs which best match 2311 // |desired_size_in_dip| and |desired_scale_factors|. 2312 // TODO(pkotwicz): Select bitmap results from multiple favicons once 2313 // content::FaviconStatus supports multiple icon URLs. 2314 chrome::FaviconID best_favicon_id = 0; 2315 std::vector<FaviconBitmapID> best_bitmap_ids; 2316 float highest_score = kSelectFaviconFramesInvalidScore; 2317 for (size_t i = 0; i < candidate_favicon_ids.size(); ++i) { 2318 std::vector<FaviconBitmapIDSize> bitmap_id_sizes; 2319 thumbnail_db_->GetFaviconBitmapIDSizes(candidate_favicon_ids[i], 2320 &bitmap_id_sizes); 2321 2322 // Build vector of gfx::Size from |bitmap_id_sizes|. 2323 std::vector<gfx::Size> sizes; 2324 for (size_t j = 0; j < bitmap_id_sizes.size(); ++j) 2325 sizes.push_back(bitmap_id_sizes[j].pixel_size); 2326 2327 std::vector<size_t> candidate_bitmap_indices; 2328 float score = 0; 2329 SelectFaviconFrameIndices(sizes, 2330 desired_scale_factors, 2331 desired_size_in_dip, 2332 &candidate_bitmap_indices, 2333 &score); 2334 if (score > highest_score) { 2335 highest_score = score; 2336 best_favicon_id = candidate_favicon_ids[i], 2337 best_bitmap_ids.clear(); 2338 for (size_t j = 0; j < candidate_bitmap_indices.size(); ++j) { 2339 size_t candidate_index = candidate_bitmap_indices[j]; 2340 best_bitmap_ids.push_back( 2341 bitmap_id_sizes[candidate_index].bitmap_id); 2342 } 2343 } 2344 } 2345 2346 // Construct FaviconBitmapResults from |best_favicon_id| and 2347 // |best_bitmap_ids|. 2348 GURL icon_url; 2349 chrome::IconType icon_type; 2350 if (!thumbnail_db_->GetFaviconHeader(best_favicon_id, &icon_url, 2351 &icon_type)) { 2352 return false; 2353 } 2354 2355 for (size_t i = 0; i < best_bitmap_ids.size(); ++i) { 2356 base::Time last_updated; 2357 chrome::FaviconBitmapResult bitmap_result; 2358 bitmap_result.icon_url = icon_url; 2359 bitmap_result.icon_type = icon_type; 2360 if (!thumbnail_db_->GetFaviconBitmap(best_bitmap_ids[i], 2361 &last_updated, 2362 &bitmap_result.bitmap_data, 2363 &bitmap_result.pixel_size)) { 2364 return false; 2365 } 2366 2367 bitmap_result.expired = (Time::Now() - last_updated) > 2368 TimeDelta::FromDays(kFaviconRefetchDays); 2369 if (bitmap_result.is_valid()) 2370 favicon_bitmap_results->push_back(bitmap_result); 2371 } 2372 return true; 2373} 2374 2375bool HistoryBackend::SetFaviconMappingsForPageAndRedirects( 2376 const GURL& page_url, 2377 chrome::IconType icon_type, 2378 const std::vector<chrome::FaviconID>& icon_ids) { 2379 if (!thumbnail_db_) 2380 return false; 2381 2382 // Find all the pages whose favicons we should set, we want to set it for 2383 // all the pages in the redirect chain if it redirected. 2384 history::RedirectList redirects; 2385 GetCachedRecentRedirects(page_url, &redirects); 2386 2387 bool mappings_changed = false; 2388 2389 // Save page <-> favicon associations. 2390 for (history::RedirectList::const_iterator i(redirects.begin()); 2391 i != redirects.end(); ++i) { 2392 mappings_changed |= SetFaviconMappingsForPage(*i, icon_type, icon_ids); 2393 } 2394 return mappings_changed; 2395} 2396 2397bool HistoryBackend::SetFaviconMappingsForPage( 2398 const GURL& page_url, 2399 chrome::IconType icon_type, 2400 const std::vector<chrome::FaviconID>& icon_ids) { 2401 DCHECK_LE(icon_ids.size(), kMaxFaviconsPerPage); 2402 bool mappings_changed = false; 2403 2404 // Two icon types are considered 'equivalent' if one of the icon types is 2405 // TOUCH_ICON and the other is TOUCH_PRECOMPOSED_ICON. 2406 // 2407 // Sets the icon mappings from |page_url| for |icon_type| to the favicons 2408 // with |icon_ids|. Mappings for |page_url| to favicons of type |icon_type| 2409 // whose FaviconID is not in |icon_ids| are removed. All icon mappings for 2410 // |page_url| to favicons of a type equivalent to |icon_type| are removed. 2411 // Remove any favicons which are orphaned as a result of the removal of the 2412 // icon mappings. 2413 2414 std::vector<chrome::FaviconID> unmapped_icon_ids = icon_ids; 2415 2416 std::vector<IconMapping> icon_mappings; 2417 thumbnail_db_->GetIconMappingsForPageURL(page_url, &icon_mappings); 2418 2419 for (std::vector<IconMapping>::iterator m = icon_mappings.begin(); 2420 m != icon_mappings.end(); ++m) { 2421 std::vector<chrome::FaviconID>::iterator icon_id_it = std::find( 2422 unmapped_icon_ids.begin(), unmapped_icon_ids.end(), m->icon_id); 2423 2424 // If the icon mapping already exists, avoid removing it and adding it back. 2425 if (icon_id_it != unmapped_icon_ids.end()) { 2426 unmapped_icon_ids.erase(icon_id_it); 2427 continue; 2428 } 2429 2430 if ((icon_type == chrome::TOUCH_ICON && 2431 m->icon_type == chrome::TOUCH_PRECOMPOSED_ICON) || 2432 (icon_type == chrome::TOUCH_PRECOMPOSED_ICON && 2433 m->icon_type == chrome::TOUCH_ICON) || (icon_type == m->icon_type)) { 2434 thumbnail_db_->DeleteIconMapping(m->mapping_id); 2435 2436 // Removing the icon mapping may have orphaned the associated favicon so 2437 // we must recheck it. This is not super fast, but this case will get 2438 // triggered rarely, since normally a page will always map to the same 2439 // favicon IDs. It will mostly happen for favicons we import. 2440 if (!thumbnail_db_->HasMappingFor(m->icon_id)) 2441 thumbnail_db_->DeleteFavicon(m->icon_id); 2442 mappings_changed = true; 2443 } 2444 } 2445 2446 for (size_t i = 0; i < unmapped_icon_ids.size(); ++i) { 2447 thumbnail_db_->AddIconMapping(page_url, unmapped_icon_ids[i]); 2448 mappings_changed = true; 2449 } 2450 return mappings_changed; 2451} 2452 2453void HistoryBackend::GetCachedRecentRedirects( 2454 const GURL& page_url, 2455 history::RedirectList* redirect_list) { 2456 RedirectCache::iterator iter = recent_redirects_.Get(page_url); 2457 if (iter != recent_redirects_.end()) { 2458 *redirect_list = iter->second; 2459 2460 // The redirect chain should have the destination URL as the last item. 2461 DCHECK(!redirect_list->empty()); 2462 DCHECK(redirect_list->back() == page_url); 2463 } else { 2464 // No known redirects, construct mock redirect chain containing |page_url|. 2465 redirect_list->push_back(page_url); 2466 } 2467} 2468 2469void HistoryBackend::SendFaviconChangedNotificationForPageAndRedirects( 2470 const GURL& page_url) { 2471 history::RedirectList redirect_list; 2472 GetCachedRecentRedirects(page_url, &redirect_list); 2473 2474 FaviconChangedDetails* changed_details = new FaviconChangedDetails; 2475 for (size_t i = 0; i < redirect_list.size(); ++i) 2476 changed_details->urls.insert(redirect_list[i]); 2477 2478 BroadcastNotifications(chrome::NOTIFICATION_FAVICON_CHANGED, 2479 changed_details); 2480} 2481 2482void HistoryBackend::Commit() { 2483 if (!db_) 2484 return; 2485 2486 // Note that a commit may not actually have been scheduled if a caller 2487 // explicitly calls this instead of using ScheduleCommit. Likewise, we 2488 // may reset the flag written by a pending commit. But this is OK! It 2489 // will merely cause extra commits (which is kind of the idea). We 2490 // could optimize more for this case (we may get two extra commits in 2491 // some cases) but it hasn't been important yet. 2492 CancelScheduledCommit(); 2493 2494 db_->CommitTransaction(); 2495 DCHECK(db_->transaction_nesting() == 0) << "Somebody left a transaction open"; 2496 db_->BeginTransaction(); 2497 2498 if (thumbnail_db_) { 2499 thumbnail_db_->CommitTransaction(); 2500 DCHECK(thumbnail_db_->transaction_nesting() == 0) << 2501 "Somebody left a transaction open"; 2502 thumbnail_db_->BeginTransaction(); 2503 } 2504 2505 if (archived_db_) { 2506 archived_db_->CommitTransaction(); 2507 archived_db_->BeginTransaction(); 2508 } 2509} 2510 2511void HistoryBackend::ScheduleCommit() { 2512 if (scheduled_commit_.get()) 2513 return; 2514 scheduled_commit_ = new CommitLaterTask(this); 2515 base::MessageLoop::current()->PostDelayedTask( 2516 FROM_HERE, 2517 base::Bind(&CommitLaterTask::RunCommit, scheduled_commit_.get()), 2518 base::TimeDelta::FromSeconds(kCommitIntervalSeconds)); 2519} 2520 2521void HistoryBackend::CancelScheduledCommit() { 2522 if (scheduled_commit_.get()) { 2523 scheduled_commit_->Cancel(); 2524 scheduled_commit_ = NULL; 2525 } 2526} 2527 2528void HistoryBackend::ProcessDBTaskImpl() { 2529 if (!db_) { 2530 // db went away, release all the refs. 2531 ReleaseDBTasks(); 2532 return; 2533 } 2534 2535 // Remove any canceled tasks. 2536 while (!db_task_requests_.empty() && db_task_requests_.front()->canceled()) { 2537 db_task_requests_.front()->Release(); 2538 db_task_requests_.pop_front(); 2539 } 2540 if (db_task_requests_.empty()) 2541 return; 2542 2543 // Run the first task. 2544 HistoryDBTaskRequest* request = db_task_requests_.front(); 2545 db_task_requests_.pop_front(); 2546 if (request->value->RunOnDBThread(this, db_.get())) { 2547 // The task is done. Notify the callback. 2548 request->ForwardResult(); 2549 // We AddRef'd the request before adding, need to release it now. 2550 request->Release(); 2551 } else { 2552 // Tasks wants to run some more. Schedule it at the end of current tasks. 2553 db_task_requests_.push_back(request); 2554 // And process it after an invoke later. 2555 base::MessageLoop::current()->PostTask( 2556 FROM_HERE, base::Bind(&HistoryBackend::ProcessDBTaskImpl, this)); 2557 } 2558} 2559 2560void HistoryBackend::ReleaseDBTasks() { 2561 for (std::list<HistoryDBTaskRequest*>::iterator i = 2562 db_task_requests_.begin(); i != db_task_requests_.end(); ++i) { 2563 (*i)->Release(); 2564 } 2565 db_task_requests_.clear(); 2566} 2567 2568//////////////////////////////////////////////////////////////////////////////// 2569// 2570// Generic operations 2571// 2572//////////////////////////////////////////////////////////////////////////////// 2573 2574void HistoryBackend::DeleteURLs(const std::vector<GURL>& urls) { 2575 expirer_.DeleteURLs(urls); 2576 2577 db_->GetStartDate(&first_recorded_time_); 2578 // Force a commit, if the user is deleting something for privacy reasons, we 2579 // want to get it on disk ASAP. 2580 Commit(); 2581} 2582 2583void HistoryBackend::DeleteURL(const GURL& url) { 2584 expirer_.DeleteURL(url); 2585 2586 db_->GetStartDate(&first_recorded_time_); 2587 // Force a commit, if the user is deleting something for privacy reasons, we 2588 // want to get it on disk ASAP. 2589 Commit(); 2590} 2591 2592void HistoryBackend::ExpireHistoryBetween( 2593 const std::set<GURL>& restrict_urls, 2594 Time begin_time, 2595 Time end_time) { 2596 if (!db_) 2597 return; 2598 2599 if (begin_time.is_null() && (end_time.is_null() || end_time.is_max()) && 2600 restrict_urls.empty()) { 2601 // Special case deleting all history so it can be faster and to reduce the 2602 // possibility of an information leak. 2603 DeleteAllHistory(); 2604 } else { 2605 // Clearing parts of history, have the expirer do the depend 2606 expirer_.ExpireHistoryBetween(restrict_urls, begin_time, end_time); 2607 2608 // Force a commit, if the user is deleting something for privacy reasons, 2609 // we want to get it on disk ASAP. 2610 Commit(); 2611 } 2612 2613 if (begin_time <= first_recorded_time_) 2614 db_->GetStartDate(&first_recorded_time_); 2615} 2616 2617void HistoryBackend::ExpireHistoryForTimes( 2618 const std::set<base::Time>& times, 2619 base::Time begin_time, base::Time end_time) { 2620 if (times.empty() || !db_) 2621 return; 2622 2623 DCHECK(*times.begin() >= begin_time) 2624 << "Min time is before begin time: " 2625 << times.begin()->ToJsTime() << " v.s. " << begin_time.ToJsTime(); 2626 DCHECK(*times.rbegin() < end_time) 2627 << "Max time is after end time: " 2628 << times.rbegin()->ToJsTime() << " v.s. " << end_time.ToJsTime(); 2629 2630 history::QueryOptions options; 2631 options.begin_time = begin_time; 2632 options.end_time = end_time; 2633 options.duplicate_policy = QueryOptions::KEEP_ALL_DUPLICATES; 2634 QueryResults results; 2635 QueryHistoryBasic(db_.get(), db_.get(), options, &results); 2636 2637 // 1st pass: find URLs that are visited at one of |times|. 2638 std::set<GURL> urls; 2639 for (size_t i = 0; i < results.size(); ++i) { 2640 if (times.count(results[i].visit_time()) > 0) 2641 urls.insert(results[i].url()); 2642 } 2643 if (urls.empty()) 2644 return; 2645 2646 // 2nd pass: collect all visit times of those URLs. 2647 std::vector<base::Time> times_to_expire; 2648 for (size_t i = 0; i < results.size(); ++i) { 2649 if (urls.count(results[i].url())) 2650 times_to_expire.push_back(results[i].visit_time()); 2651 } 2652 2653 // Put the times in reverse chronological order and remove 2654 // duplicates (for expirer_.ExpireHistoryForTimes()). 2655 std::sort(times_to_expire.begin(), times_to_expire.end(), 2656 std::greater<base::Time>()); 2657 times_to_expire.erase( 2658 std::unique(times_to_expire.begin(), times_to_expire.end()), 2659 times_to_expire.end()); 2660 2661 // Expires by times and commit. 2662 DCHECK(!times_to_expire.empty()); 2663 expirer_.ExpireHistoryForTimes(times_to_expire); 2664 Commit(); 2665 2666 DCHECK(times_to_expire.back() >= first_recorded_time_); 2667 // Update |first_recorded_time_| if we expired it. 2668 if (times_to_expire.back() == first_recorded_time_) 2669 db_->GetStartDate(&first_recorded_time_); 2670} 2671 2672void HistoryBackend::ExpireHistory( 2673 const std::vector<history::ExpireHistoryArgs>& expire_list) { 2674 if (db_) { 2675 bool update_first_recorded_time = false; 2676 2677 for (std::vector<history::ExpireHistoryArgs>::const_iterator it = 2678 expire_list.begin(); it != expire_list.end(); ++it) { 2679 expirer_.ExpireHistoryBetween(it->urls, it->begin_time, it->end_time); 2680 2681 if (it->begin_time < first_recorded_time_) 2682 update_first_recorded_time = true; 2683 } 2684 Commit(); 2685 2686 // Update |first_recorded_time_| if any deletion might have affected it. 2687 if (update_first_recorded_time) 2688 db_->GetStartDate(&first_recorded_time_); 2689 } 2690} 2691 2692void HistoryBackend::URLsNoLongerBookmarked(const std::set<GURL>& urls) { 2693 if (!db_) 2694 return; 2695 2696 for (std::set<GURL>::const_iterator i = urls.begin(); i != urls.end(); ++i) { 2697 URLRow url_row; 2698 if (!db_->GetRowForURL(*i, &url_row)) 2699 continue; // The URL isn't in the db; nothing to do. 2700 2701 VisitVector visits; 2702 db_->GetVisitsForURL(url_row.id(), &visits); 2703 2704 if (visits.empty()) 2705 expirer_.DeleteURL(*i); // There are no more visits; nuke the URL. 2706 } 2707} 2708 2709void HistoryBackend::DatabaseErrorCallback(int error, sql::Statement* stmt) { 2710 if (!scheduled_kill_db_ && sql::IsErrorCatastrophic(error)) { 2711 scheduled_kill_db_ = true; 2712 // Don't just do the close/delete here, as we are being called by |db| and 2713 // that seems dangerous. 2714 // TODO(shess): Consider changing KillHistoryDatabase() to use 2715 // RazeAndClose(). Then it can be cleared immediately. 2716 base::MessageLoop::current()->PostTask( 2717 FROM_HERE, 2718 base::Bind(&HistoryBackend::KillHistoryDatabase, this)); 2719 } 2720} 2721 2722void HistoryBackend::KillHistoryDatabase() { 2723 scheduled_kill_db_ = false; 2724 if (!db_) 2725 return; 2726 2727 // Rollback transaction because Raze() cannot be called from within a 2728 // transaction. 2729 db_->RollbackTransaction(); 2730 bool success = db_->Raze(); 2731 UMA_HISTOGRAM_BOOLEAN("History.KillHistoryDatabaseResult", success); 2732 2733#if defined(OS_ANDROID) 2734 // Release AndroidProviderBackend before other objects. 2735 android_provider_backend_.reset(); 2736#endif 2737 2738 // The expirer keeps tabs on the active databases. Tell it about the 2739 // databases which will be closed. 2740 expirer_.SetDatabases(NULL, NULL, NULL); 2741 2742 // Reopen a new transaction for |db_| for the sake of CloseAllDatabases(). 2743 db_->BeginTransaction(); 2744 CloseAllDatabases(); 2745} 2746 2747void HistoryBackend::ProcessDBTask( 2748 scoped_refptr<HistoryDBTaskRequest> request) { 2749 DCHECK(request.get()); 2750 if (request->canceled()) 2751 return; 2752 2753 bool task_scheduled = !db_task_requests_.empty(); 2754 // Make sure we up the refcount of the request. ProcessDBTaskImpl will 2755 // release when done with the task. 2756 request->AddRef(); 2757 db_task_requests_.push_back(request.get()); 2758 if (!task_scheduled) { 2759 // No other tasks are scheduled. Process request now. 2760 ProcessDBTaskImpl(); 2761 } 2762} 2763 2764void HistoryBackend::BroadcastNotifications( 2765 int type, 2766 HistoryDetails* details_deleted) { 2767 // |delegate_| may be NULL if |this| is in the process of closing (closed by 2768 // HistoryService -> HistoryBackend::Closing(). 2769 if (delegate_) 2770 delegate_->BroadcastNotifications(type, details_deleted); 2771 else 2772 delete details_deleted; 2773} 2774 2775void HistoryBackend::NotifySyncURLsDeleted(bool all_history, 2776 bool archived, 2777 URLRows* rows) { 2778 if (typed_url_syncable_service_.get()) 2779 typed_url_syncable_service_->OnUrlsDeleted(all_history, archived, rows); 2780} 2781 2782// Deleting -------------------------------------------------------------------- 2783 2784void HistoryBackend::DeleteAllHistory() { 2785 // Our approach to deleting all history is: 2786 // 1. Copy the bookmarks and their dependencies to new tables with temporary 2787 // names. 2788 // 2. Delete the original tables. Since tables can not share pages, we know 2789 // that any data we don't want to keep is now in an unused page. 2790 // 3. Renaming the temporary tables to match the original. 2791 // 4. Vacuuming the database to delete the unused pages. 2792 // 2793 // Since we are likely to have very few bookmarks and their dependencies 2794 // compared to all history, this is also much faster than just deleting from 2795 // the original tables directly. 2796 2797 // Get the bookmarked URLs. 2798 std::vector<BookmarkService::URLAndTitle> starred_urls; 2799 BookmarkService* bookmark_service = GetBookmarkService(); 2800 if (bookmark_service) 2801 bookmark_service_->GetBookmarks(&starred_urls); 2802 2803 URLRows kept_urls; 2804 for (size_t i = 0; i < starred_urls.size(); i++) { 2805 URLRow row; 2806 if (!db_->GetRowForURL(starred_urls[i].url, &row)) 2807 continue; 2808 2809 // Clear the last visit time so when we write these rows they are "clean." 2810 row.set_last_visit(Time()); 2811 row.set_visit_count(0); 2812 row.set_typed_count(0); 2813 kept_urls.push_back(row); 2814 } 2815 2816 // Clear thumbnail and favicon history. The favicons for the given URLs will 2817 // be kept. 2818 if (!ClearAllThumbnailHistory(kept_urls)) { 2819 LOG(ERROR) << "Thumbnail history could not be cleared"; 2820 // We continue in this error case. If the user wants to delete their 2821 // history, we should delete as much as we can. 2822 } 2823 2824 // ClearAllMainHistory will change the IDs of the URLs in kept_urls. 2825 // Therefore, we clear the list afterwards to make sure nobody uses this 2826 // invalid data. 2827 if (!ClearAllMainHistory(kept_urls)) 2828 LOG(ERROR) << "Main history could not be cleared"; 2829 kept_urls.clear(); 2830 2831 // Delete archived history. 2832 if (archived_db_) { 2833 // Close the database and delete the file. 2834 archived_db_.reset(); 2835 base::FilePath archived_file_name = GetArchivedFileName(); 2836 sql::Connection::Delete(archived_file_name); 2837 2838 // Now re-initialize the database (which may fail). 2839 archived_db_.reset(new ArchivedDatabase()); 2840 if (!archived_db_->Init(archived_file_name)) { 2841 LOG(WARNING) << "Could not initialize the archived database."; 2842 archived_db_.reset(); 2843 } else { 2844 // Open our long-running transaction on this database. 2845 archived_db_->BeginTransaction(); 2846 } 2847 } 2848 2849 db_->GetStartDate(&first_recorded_time_); 2850 2851 // Send out the notification that history is cleared. The in-memory database 2852 // will pick this up and clear itself. 2853 URLsDeletedDetails* details = new URLsDeletedDetails; 2854 details->all_history = true; 2855 NotifySyncURLsDeleted(true, false, NULL); 2856 BroadcastNotifications(chrome::NOTIFICATION_HISTORY_URLS_DELETED, details); 2857} 2858 2859bool HistoryBackend::ClearAllThumbnailHistory(const URLRows& kept_urls) { 2860 if (!thumbnail_db_) { 2861 // When we have no reference to the thumbnail database, maybe there was an 2862 // error opening it. In this case, we just try to blow it away to try to 2863 // fix the error if it exists. This may fail, in which case either the 2864 // file doesn't exist or there's no more we can do. 2865 sql::Connection::Delete(GetFaviconsFileName()); 2866 2867 // Older version of the database. 2868 sql::Connection::Delete(GetThumbnailFileName()); 2869 return true; 2870 } 2871 2872 // Urls to retain mappings for. 2873 std::vector<GURL> urls_to_keep; 2874 for (URLRows::const_iterator i = kept_urls.begin(); 2875 i != kept_urls.end(); ++i) { 2876 urls_to_keep.push_back(i->url()); 2877 } 2878 2879 // Isolate from any long-running transaction. 2880 thumbnail_db_->CommitTransaction(); 2881 thumbnail_db_->BeginTransaction(); 2882 2883 // TODO(shess): If this fails, perhaps the database should be razed 2884 // or deleted. 2885 if (!thumbnail_db_->RetainDataForPageUrls(urls_to_keep)) { 2886 thumbnail_db_->RollbackTransaction(); 2887 thumbnail_db_->BeginTransaction(); 2888 return false; 2889 } 2890 2891#if defined(OS_ANDROID) 2892 // TODO (michaelbai): Add the unit test once AndroidProviderBackend is 2893 // avaliable in HistoryBackend. 2894 db_->ClearAndroidURLRows(); 2895#endif 2896 2897 // Vacuum to remove all the pages associated with the dropped tables. There 2898 // must be no transaction open on the table when we do this. We assume that 2899 // our long-running transaction is open, so we complete it and start it again. 2900 DCHECK(thumbnail_db_->transaction_nesting() == 1); 2901 thumbnail_db_->CommitTransaction(); 2902 thumbnail_db_->Vacuum(); 2903 thumbnail_db_->BeginTransaction(); 2904 return true; 2905} 2906 2907bool HistoryBackend::ClearAllMainHistory(const URLRows& kept_urls) { 2908 // Create the duplicate URL table. We will copy the kept URLs into this. 2909 if (!db_->CreateTemporaryURLTable()) 2910 return false; 2911 2912 // Insert the URLs into the temporary table. 2913 for (URLRows::const_iterator i = kept_urls.begin(); i != kept_urls.end(); 2914 ++i) { 2915 db_->AddTemporaryURL(*i); 2916 } 2917 2918 // Replace the original URL table with the temporary one. 2919 if (!db_->CommitTemporaryURLTable()) 2920 return false; 2921 2922 // Delete the old tables and recreate them empty. 2923 db_->RecreateAllTablesButURL(); 2924 2925 // Vacuum to reclaim the space from the dropped tables. This must be done 2926 // when there is no transaction open, and we assume that our long-running 2927 // transaction is currently open. 2928 db_->CommitTransaction(); 2929 db_->Vacuum(); 2930 db_->BeginTransaction(); 2931 db_->GetStartDate(&first_recorded_time_); 2932 2933 return true; 2934} 2935 2936BookmarkService* HistoryBackend::GetBookmarkService() { 2937 if (bookmark_service_) 2938 bookmark_service_->BlockTillLoaded(); 2939 return bookmark_service_; 2940} 2941 2942void HistoryBackend::NotifyVisitObservers(const VisitRow& visit) { 2943 BriefVisitInfo info; 2944 info.url_id = visit.url_id; 2945 info.time = visit.visit_time; 2946 info.transition = visit.transition; 2947 // If we don't have a delegate yet during setup or shutdown, we will drop 2948 // these notifications. 2949 if (delegate_) 2950 delegate_->NotifyVisitDBObserversOnAddVisit(info); 2951} 2952 2953#if defined(OS_ANDROID) 2954void HistoryBackend::PopulateMostVisitedURLMap() { 2955 MostVisitedURLList most_visited_urls; 2956 QueryMostVisitedURLsImpl(kPageVisitStatsMaxTopSites, kSegmentDataRetention, 2957 &most_visited_urls); 2958 2959 DCHECK_LE(most_visited_urls.size(), kPageVisitStatsMaxTopSites); 2960 for (size_t i = 0; i < most_visited_urls.size(); ++i) { 2961 most_visited_urls_map_[most_visited_urls[i].url] = i; 2962 for (size_t j = 0; j < most_visited_urls[i].redirects.size(); ++j) 2963 most_visited_urls_map_[most_visited_urls[i].redirects[j]] = i; 2964 } 2965} 2966 2967void HistoryBackend::RecordTopPageVisitStats(const GURL& url) { 2968 int rank = kPageVisitStatsMaxTopSites; 2969 std::map<GURL, int>::const_iterator it = most_visited_urls_map_.find(url); 2970 if (it != most_visited_urls_map_.end()) 2971 rank = (*it).second; 2972 UMA_HISTOGRAM_ENUMERATION("History.TopSitesVisitsByRank", 2973 rank, kPageVisitStatsMaxTopSites + 1); 2974} 2975#endif 2976 2977} // namespace history 2978