includes/Storage/DerivedPageDataUpdater.php

   1 <?php
   2 /**
   3  * A handle for managing updates for derived page data on edit, import, purge, etc.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  */
  22
  23 namespace MediaWiki\Storage;
  24
  25 use ApiStashEdit;
  26 use CategoryMembershipChangeJob;
  27 use Content;
  28 use ContentHandler;
  29 use DataUpdate;
  30 use DeferrableUpdate;
  31 use DeferredUpdates;
  32 use Hooks;
  33 use IDBAccessObject;
  34 use InvalidArgumentException;
  35 use JobQueueGroup;
  36 use Language;
  37 use LinksDeletionUpdate;
  38 use LinksUpdate;
  39 use LogicException;
  40 use MediaWiki\Edit\PreparedEdit;
  41 use MediaWiki\Revision\MutableRevisionRecord;
  42 use MediaWiki\Revision\RenderedRevision;
  43 use MediaWiki\Revision\RevisionRecord;
  44 use MediaWiki\Revision\RevisionRenderer;
  45 use MediaWiki\Revision\RevisionSlots;
  46 use MediaWiki\Revision\RevisionStore;
  47 use MediaWiki\Revision\SlotRoleRegistry;
  48 use MediaWiki\Revision\SlotRecord;
  49 use MediaWiki\User\UserIdentity;
  50 use MessageCache;
  51 use ParserCache;
  52 use ParserOptions;
  53 use ParserOutput;
  54 use RecentChangesUpdateJob;
  55 use ResourceLoaderWikiModule;
  56 use Revision;
  57 use SearchUpdate;
  58 use SiteStatsUpdate;
  59 use Title;
  60 use User;
  61 use Wikimedia\Assert\Assert;
  62 use Wikimedia\Rdbms\LBFactory;
  63 use WikiPage;
  64
  65 /**
  66  * A handle for managing updates for derived page data on edit, import, purge, etc.
  67  *
  68  * @note Avoid direct usage of DerivedPageDataUpdater.
  69  *
  70  * @todo Define interfaces for the different use cases of DerivedPageDataUpdater, particularly
  71  * providing access to post-PST content and ParserOutput to callbacks during revision creation,
  72  * which currently use WikiPage::prepareContentForEdit, and allowing updates to be triggered on
  73  * purge, import, and undeletion, which currently use WikiPage::doEditUpdates() and
  74  * Content::getSecondaryDataUpdates().
  75  *
  76  * DerivedPageDataUpdater instances are designed to be cached inside a WikiPage instance,
  77  * and re-used by callback code over the course of an update operation. It's a stepping stone
  78  * one the way to a more complete refactoring of WikiPage.
  79  *
  80  * When using a DerivedPageDataUpdater, the following life cycle must be observed:
  81  * grabCurrentRevision (optional), prepareContent (optional), prepareUpdate (required
  82  * for doUpdates). getCanonicalParserOutput, getSlots, and getSecondaryDataUpdates
  83  * require prepareContent or prepareUpdate to have been called first, to initialize the
  84  * DerivedPageDataUpdater.
  85  *
  86  * @see docs/pageupdater.txt for more information.
  87  *
  88  * MCR migration note: this replaces the relevant methods in WikiPage, and covers the use cases
  89  * of PreparedEdit.
  90  *
  91  * @internal
  92  *
  93  * @since 1.32
  94  * @ingroup Page
  95  */
  96 class DerivedPageDataUpdater implements IDBAccessObject {
  97
  98         /**
  99          * @var UserIdentity|null
 100          */
 101         private $user = null;
 102
 103         /**
 104          * @var WikiPage
 105          */
 106         private $wikiPage;
 107
 108         /**
 109          * @var ParserCache
 110          */
 111         private $parserCache;
 112
 113         /**
 114          * @var RevisionStore
 115          */
 116         private $revisionStore;
 117
 118         /**
 119          * @var Language
 120          */
 121         private $contLang;
 122
 123         /**
 124          * @var JobQueueGroup
 125          */
 126         private $jobQueueGroup;
 127
 128         /**
 129          * @var MessageCache
 130          */
 131         private $messageCache;
 132
 133         /**
 134          * @var LBFactory
 135          */
 136         private $loadbalancerFactory;
 137
 138         /**
 139          * @var string see $wgArticleCountMethod
 140          */
 141         private $articleCountMethod;
 142
 143         /**
 144          * @var boolean see $wgRCWatchCategoryMembership
 145          */
 146         private $rcWatchCategoryMembership = false;
 147
 148         /**
 149          * Stores (most of) the $options parameter of prepareUpdate().
 150          * @see prepareUpdate()
 151          */
 152         private $options = [
 153                 'changed' => true,
 154                 'created' => false,
 155                 'moved' => false,
 156                 'restored' => false,
 157                 'oldrevision' => null,
 158                 'oldcountable' => null,
 159                 'oldredirect' => null,
 160                 'triggeringUser' => null,
 161                 // causeAction/causeAgent default to 'unknown' but that's handled where it's read,
 162                 // to make the life of prepareUpdate() callers easier.
 163                 'causeAction' => null,
 164                 'causeAgent' => null,
 165         ];
 166
 167         /**
 168          * The state of the relevant row in page table before the edit.
 169          * This is determined by the first call to grabCurrentRevision, prepareContent,
 170          * or prepareUpdate (so it is only accessible in 'knows-current' or a later stage).
 171          * If pageState was not initialized when prepareUpdate() is called, prepareUpdate() will
 172          * attempt to emulate the state of the page table before the edit.
 173          *
 174          * Contains the following fields:
 175          * - oldRevision (RevisionRecord|null): the revision that was current before the change
 176          *   associated with this update. Might not be set, use getParentRevision().
 177          * - oldId (int|null): the id of the above revision. 0 if there is no such revision (the change
 178          *   was about creating a new page); null if not known (that should not happen).
 179          * - oldIsRedirect (bool|null): whether the page was a redirect before the change. Lazy-loaded,
 180          *   can be null; use wasRedirect() instead of direct access.
 181          * - oldCountable (bool|null): whether the page was countable before the change (or null
 182          *   if we don't have that information)
 183          *
 184          * @var array
 185          */
 186         private $pageState = null;
 187
 188         /**
 189          * @var RevisionSlotsUpdate|null
 190          */
 191         private $slotsUpdate = null;
 192
 193         /**
 194          * @var RevisionRecord|null
 195          */
 196         private $parentRevision = null;
 197
 198         /**
 199          * @var RevisionRecord|null
 200          */
 201         private $revision = null;
 202
 203         /**
 204          * @var RenderedRevision
 205          */
 206         private $renderedRevision = null;
 207
 208         /**
 209          * @var RevisionRenderer
 210          */
 211         private $revisionRenderer;
 212
 213         /** @var SlotRoleRegistry */
 214         private $slotRoleRegistry;
 215
 216         /**
 217          * A stage identifier for managing the life cycle of this instance.
 218          * Possible stages are 'new', 'knows-current', 'has-content', 'has-revision', and 'done'.
 219          *
 220          * @see docs/pageupdater.txt for documentation of the life cycle.
 221          *
 222          * @var string
 223          */
 224         private $stage = 'new';
 225
 226         /**
 227          * Transition table for managing the life cycle of DerivedPageDateUpdater instances.
 228          *
 229          * XXX: Overkill. This is a linear order, we could just count. Names are nice though,
 230          * and constants are also overkill...
 231          *
 232          * @see docs/pageupdater.txt for documentation of the life cycle.
 233          *
 234          * @var array[]
 235          */
 236         private static $transitions = [
 237                 'new' => [
 238                         'new' => true,
 239                         'knows-current' => true,
 240                         'has-content' => true,
 241                         'has-revision' => true,
 242                 ],
 243                 'knows-current' => [
 244                         'knows-current' => true,
 245                         'has-content' => true,
 246                         'has-revision' => true,
 247                 ],
 248                 'has-content' => [
 249                         'has-content' => true,
 250                         'has-revision' => true,
 251                 ],
 252                 'has-revision' => [
 253                         'has-revision' => true,
 254                         'done' => true,
 255                 ],
 256         ];
 257
 258         /**
 259          * @param WikiPage $wikiPage ,
 260          * @param RevisionStore $revisionStore
 261          * @param RevisionRenderer $revisionRenderer
 262          * @param SlotRoleRegistry $slotRoleRegistry
 263          * @param ParserCache $parserCache
 264          * @param JobQueueGroup $jobQueueGroup
 265          * @param MessageCache $messageCache
 266          * @param Language $contLang
 267          * @param LBFactory $loadbalancerFactory
 268          */
 269         public function __construct(
 270                 WikiPage $wikiPage,
 271                 RevisionStore $revisionStore,
 272                 RevisionRenderer $revisionRenderer,
 273                 SlotRoleRegistry $slotRoleRegistry,
 274                 ParserCache $parserCache,
 275                 JobQueueGroup $jobQueueGroup,
 276                 MessageCache $messageCache,
 277                 Language $contLang,
 278                 LBFactory $loadbalancerFactory
 279         ) {
 280                 $this->wikiPage = $wikiPage;
 281
 282                 $this->parserCache = $parserCache;
 283                 $this->revisionStore = $revisionStore;
 284                 $this->revisionRenderer = $revisionRenderer;
 285                 $this->slotRoleRegistry = $slotRoleRegistry;
 286                 $this->jobQueueGroup = $jobQueueGroup;
 287                 $this->messageCache = $messageCache;
 288                 $this->contLang = $contLang;
 289                 // XXX only needed for waiting for replicas to catch up; there should be a narrower
 290                 // interface for that.
 291                 $this->loadbalancerFactory = $loadbalancerFactory;
 292         }
 293
 294         /**
 295          * Transition function for managing the life cycle of this instances.
 296          *
 297          * @see docs/pageupdater.txt for documentation of the life cycle.
 298          *
 299          * @param string $newStage the new stage
 300          * @return string the previous stage
 301          *
 302          * @throws LogicException If a transition to the given stage is not possible in the current
 303          *         stage.
 304          */
 305         private function doTransition( $newStage ) {
 306                 $this->assertTransition( $newStage );
 307
 308                 $oldStage = $this->stage;
 309                 $this->stage = $newStage;
 310
 311                 return $oldStage;
 312         }
 313
 314         /**
 315          * Asserts that a transition to the given stage is possible, without performing it.
 316          *
 317          * @see docs/pageupdater.txt for documentation of the life cycle.
 318          *
 319          * @param string $newStage the new stage
 320          *
 321          * @throws LogicException If this instance is not in the expected stage
 322          */
 323         private function assertTransition( $newStage ) {
 324                 if ( empty( self::$transitions[$this->stage][$newStage] ) ) {
 325                         throw new LogicException( "Cannot transition from {$this->stage} to $newStage" );
 326                 }
 327         }
 328
 329         /**
 330          * @return bool|string
 331          */
 332         private function getWikiId() {
 333                 // TODO: get from RevisionStore
 334                 return false;
 335         }
 336
 337         /**
 338          * Checks whether this DerivedPageDataUpdater can be re-used for running updates targeting
 339          * the given revision.
 340          *
 341          * @param UserIdentity|null $user The user creating the revision in question
 342          * @param RevisionRecord|null $revision New revision (after save, if already saved)
 343          * @param RevisionSlotsUpdate|null $slotsUpdate New content (before PST)
 344          * @param null|int $parentId Parent revision of the edit (use 0 for page creation)
 345          *
 346          * @return bool
 347          */
 348         public function isReusableFor(
 349                 UserIdentity $user = null,
 350                 RevisionRecord $revision = null,
 351                 RevisionSlotsUpdate $slotsUpdate = null,
 352                 $parentId = null
 353         ) {
 354                 if ( $revision
 355                         && $parentId
 356                         && $revision->getParentId() !== $parentId
 357                 ) {
 358                         throw new InvalidArgumentException( '$parentId should match the parent of $revision' );
 359                 }
 360
 361                 // NOTE: For null revisions, $user may be different from $this->revision->getUser
 362                 // and also from $revision->getUser.
 363                 // But $user should always match $this->user.
 364                 if ( $user && $this->user && $user->getName() !== $this->user->getName() ) {
 365                         return false;
 366                 }
 367
 368                 if ( $revision && $this->revision && $this->revision->getId()
 369                         && $this->revision->getId() !== $revision->getId()
 370                 ) {
 371                         return false;
 372                 }
 373
 374                 if ( $this->pageState
 375                         && $revision
 376                         && $revision->getParentId() !== null
 377                         && $this->pageState['oldId'] !== $revision->getParentId()
 378                 ) {
 379                         return false;
 380                 }
 381
 382                 if ( $this->pageState
 383                         && $parentId !== null
 384                         && $this->pageState['oldId'] !== $parentId
 385                 ) {
 386                         return false;
 387                 }
 388
 389                 // NOTE: this check is the primary reason for having the $this->slotsUpdate field!
 390                 if ( $this->slotsUpdate
 391                         && $slotsUpdate
 392                         && !$this->slotsUpdate->hasSameUpdates( $slotsUpdate )
 393                 ) {
 394                         return false;
 395                 }
 396
 397                 if ( $revision
 398                         && $this->revision
 399                         && !$this->revision->getSlots()->hasSameContent( $revision->getSlots() )
 400                 ) {
 401                         return false;
 402                 }
 403
 404                 return true;
 405         }
 406
 407         /**
 408          * @param string $articleCountMethod "any" or "link".
 409          * @see $wgArticleCountMethod
 410          */
 411         public function setArticleCountMethod( $articleCountMethod ) {
 412                 $this->articleCountMethod = $articleCountMethod;
 413         }
 414
 415         /**
 416          * @param bool $rcWatchCategoryMembership
 417          * @see $wgRCWatchCategoryMembership
 418          */
 419         public function setRcWatchCategoryMembership( $rcWatchCategoryMembership ) {
 420                 $this->rcWatchCategoryMembership = $rcWatchCategoryMembership;
 421         }
 422
 423         /**
 424          * @return Title
 425          */
 426         private function getTitle() {
 427                 // NOTE: eventually, we won't get a WikiPage passed into the constructor any more
 428                 return $this->wikiPage->getTitle();
 429         }
 430
 431         /**
 432          * @return WikiPage
 433          */
 434         private function getWikiPage() {
 435                 // NOTE: eventually, we won't get a WikiPage passed into the constructor any more
 436                 return $this->wikiPage;
 437         }
 438
 439         /**
 440          * Determines whether the page being edited already existed.
 441          * Only defined after calling grabCurrentRevision() or prepareContent() or prepareUpdate()!
 442          *
 443          * @return bool
 444          * @throws LogicException if called before grabCurrentRevision
 445          */
 446         public function pageExisted() {
 447                 $this->assertHasPageState( __METHOD__ );
 448
 449                 return $this->pageState['oldId'] > 0;
 450         }
 451
 452         /**
 453          * Returns the parent revision of the new revision wrapped by this update.
 454          * If the update is a null-edit, this will return the parent of the current (and new) revision.
 455          * This will return null if the revision wrapped by this update created the page.
 456          * Only defined after calling prepareContent() or prepareUpdate()!
 457          *
 458          * @return RevisionRecord|null the parent revision of the new revision, or null if
 459          *         the update created the page.
 460          */
 461         private function getParentRevision() {
 462                 $this->assertPrepared( __METHOD__ );
 463
 464                 if ( $this->parentRevision ) {
 465                         return $this->parentRevision;
 466                 }
 467
 468                 if ( !$this->pageState['oldId'] ) {
 469                         // If there was no current revision, there is no parent revision,
 470                         // since the page didn't exist.
 471                         return null;
 472                 }
 473
 474                 $oldId = $this->revision->getParentId();
 475                 $flags = $this->useMaster() ? RevisionStore::READ_LATEST : 0;
 476                 $this->parentRevision = $oldId
 477                         ? $this->revisionStore->getRevisionById( $oldId, $flags )
 478                         : null;
 479
 480                 return $this->parentRevision;
 481         }
 482
 483         /**
 484          * Returns the revision that was the page's current revision when grabCurrentRevision()
 485          * was first called.
 486          *
 487          * During an edit, that revision will act as the logical parent of the new revision.
 488          *
 489          * Some updates are performed based on the difference between the database state at the
 490          * moment this method is first called, and the state after the edit.
 491          *
 492          * @see docs/pageupdater.txt for more information on when thie method can and should be called.
 493          *
 494          * @note After prepareUpdate() was called, grabCurrentRevision() will throw an exception
 495          * to avoid confusion, since the page's current revision is then the new revision after
 496          * the edit, which was presumably passed to prepareUpdate() as the $revision parameter.
 497          * Use getParentRevision() instead to access the revision that is the parent of the
 498          * new revision.
 499          *
 500          * @return RevisionRecord|null the page's current revision, or null if the page does not
 501          * yet exist.
 502          */
 503         public function grabCurrentRevision() {
 504                 if ( $this->pageState ) {
 505                         return $this->pageState['oldRevision'];
 506                 }
 507
 508                 $this->assertTransition( 'knows-current' );
 509
 510                 // NOTE: eventually, we won't get a WikiPage passed into the constructor any more
 511                 $wikiPage = $this->getWikiPage();
 512
 513                 // Do not call WikiPage::clear(), since the caller may already have caused page data
 514                 // to be loaded with SELECT FOR UPDATE. Just assert it's loaded now.
 515                 $wikiPage->loadPageData( self::READ_LATEST );
 516                 $rev = $wikiPage->getRevision();
 517                 $current = $rev ? $rev->getRevisionRecord() : null;
 518
 519                 $this->pageState = [
 520                         'oldRevision' => $current,
 521                         'oldId' => $rev ? $rev->getId() : 0,
 522                         'oldIsRedirect' => $wikiPage->isRedirect(), // NOTE: uses page table
 523                         'oldCountable' => $wikiPage->isCountable(), // NOTE: uses pagelinks table
 524                 ];
 525
 526                 $this->doTransition( 'knows-current' );
 527
 528                 return $this->pageState['oldRevision'];
 529         }
 530
 531         /**
 532          * Whether prepareUpdate() or prepareContent() have been called on this instance.
 533          *
 534          * @return bool
 535          */
 536         public function isContentPrepared() {
 537                 return $this->revision !== null;
 538         }
 539
 540         /**
 541          * Whether prepareUpdate() has been called on this instance.
 542          *
 543          * @note will also return null in case of a null-edit!
 544          *
 545          * @return bool
 546          */
 547         public function isUpdatePrepared() {
 548                 return $this->revision !== null && $this->revision->getId() !== null;
 549         }
 550
 551         /**
 552          * @return int
 553          */
 554         private function getPageId() {
 555                 // NOTE: eventually, we won't get a WikiPage passed into the constructor any more
 556                 return $this->wikiPage->getId();
 557         }
 558
 559         /**
 560          * Whether the content is deleted and thus not visible to the public.
 561          *
 562          * @return bool
 563          */
 564         public function isContentDeleted() {
 565                 if ( $this->revision ) {
 566                         // XXX: if that revision is the current revision, this should be skipped
 567                         return $this->revision->isDeleted( RevisionRecord::DELETED_TEXT );
 568                 } else {
 569                         // If the content has not been saved yet, it cannot have been deleted yet.
 570                         return false;
 571                 }
 572         }
 573
 574         /**
 575          * Returns the slot, modified or inherited, after PST, with no audience checks applied.
 576          *
 577          * @param string $role slot role name
 578          *
 579          * @throws PageUpdateException If the slot is neither set for update nor inherited from the
 580          *        parent revision.
 581          * @return SlotRecord
 582          */
 583         public function getRawSlot( $role ) {
 584                 return $this->getSlots()->getSlot( $role );
 585         }
 586
 587         /**
 588          * Returns the content of the given slot, with no audience checks.
 589          *
 590          * @throws PageUpdateException If the slot is neither set for update nor inherited from the
 591          *        parent revision.
 592          * @param string $role slot role name
 593          * @return Content
 594          */
 595         public function getRawContent( $role ) {
 596                 return $this->getRawSlot( $role )->getContent();
 597         }
 598
 599         /**
 600          * Returns the content model of the given slot
 601          *
 602          * @param string $role slot role name
 603          * @return string
 604          */
 605         private function getContentModel( $role ) {
 606                 return $this->getRawSlot( $role )->getModel();
 607         }
 608
 609         /**
 610          * @param string $role slot role name
 611          * @return ContentHandler
 612          */
 613         private function getContentHandler( $role ) {
 614                 // TODO: inject something like a ContentHandlerRegistry
 615                 return ContentHandler::getForModelID( $this->getContentModel( $role ) );
 616         }
 617
 618         private function useMaster() {
 619                 // TODO: can we just set a flag to true in prepareContent()?
 620                 return $this->wikiPage->wasLoadedFrom( self::READ_LATEST );
 621         }
 622
 623         /**
 624          * @return bool
 625          */
 626         public function isCountable() {
 627                 // NOTE: Keep in sync with WikiPage::isCountable.
 628
 629                 if ( !$this->getTitle()->isContentPage() ) {
 630                         return false;
 631                 }
 632
 633                 if ( $this->isContentDeleted() ) {
 634                         // This should be irrelevant: countability only applies to the current revision,
 635                         // and the current revision is never suppressed.
 636                         return false;
 637                 }
 638
 639                 if ( $this->isRedirect() ) {
 640                         return false;
 641                 }
 642
 643                 $hasLinks = null;
 644
 645                 if ( $this->articleCountMethod === 'link' ) {
 646                         // NOTE: it would be more appropriate to determine for each slot separately
 647                         // whether it has links, and use that information with that slot's
 648                         // isCountable() method. However, that would break parity with
 649                         // WikiPage::isCountable, which uses the pagelinks table to determine
 650                         // whether the current revision has links.
 651                         $hasLinks = (bool)count( $this->getCanonicalParserOutput()->getLinks() );
 652                 }
 653
 654                 foreach ( $this->getModifiedSlotRoles() as $role ) {
 655                         $roleHandler = $this->slotRoleRegistry->getRoleHandler( $role );
 656                         if ( $roleHandler->supportsArticleCount() ) {
 657                                 $content = $this->getRawContent( $role );
 658
 659                                 if ( $content->isCountable( $hasLinks ) ) {
 660                                         return true;
 661                                 }
 662                         }
 663                 }
 664
 665                 return false;
 666         }
 667
 668         /**
 669          * @return bool
 670          */
 671         public function isRedirect() {
 672                 // NOTE: main slot determines redirect status
 673                 // TODO: MCR: this should be controlled by a PageTypeHandler
 674                 $mainContent = $this->getRawContent( SlotRecord::MAIN );
 675
 676                 return $mainContent->isRedirect();
 677         }
 678
 679         /**
 680          * @param RevisionRecord $rev
 681          *
 682          * @return bool
 683          */
 684         private function revisionIsRedirect( RevisionRecord $rev ) {
 685                 // NOTE: main slot determines redirect status
 686                 $mainContent = $rev->getContent( SlotRecord::MAIN, RevisionRecord::RAW );
 687
 688                 return $mainContent->isRedirect();
 689         }
 690
 691         /**
 692          * Prepare updates based on an update which has not yet been saved.
 693          *
 694          * This may be used to create derived data that is needed when creating a new revision;
 695          * particularly, this makes available the slots of the new revision via the getSlots()
 696          * method, after applying PST and slot inheritance.
 697          *
 698          * The derived data prepared for revision creation may then later be re-used by doUpdates(),
 699          * without the need to re-calculate.
 700          *
 701          * @see docs/pageupdater.txt for more information on when thie method can and should be called.
 702          *
 703          * @note Calling this method more than once with the same $slotsUpdate
 704          * has no effect. Calling this method multiple times with different content will cause
 705          * an exception.
 706          *
 707          * @note Calling this method after prepareUpdate() has been called will cause an exception.
 708          *
 709          * @param User $user The user to act as context for pre-save transformation (PST).
 710          *        Type hint should be reduced to UserIdentity at some point.
 711          * @param RevisionSlotsUpdate $slotsUpdate The new content of the slots to be updated
 712          *        by this edit, before PST.
 713          * @param bool $useStash Whether to use stashed ParserOutput
 714          */
 715         public function prepareContent(
 716                 User $user,
 717                 RevisionSlotsUpdate $slotsUpdate,
 718                 $useStash = true
 719         ) {
 720                 if ( $this->slotsUpdate ) {
 721                         if ( !$this->user ) {
 722                                 throw new LogicException(
 723                                         'Unexpected state: $this->slotsUpdate was initialized, '
 724                                         . 'but $this->user was not.'
 725                                 );
 726                         }
 727
 728                         if ( $this->user->getName() !== $user->getName() ) {
 729                                 throw new LogicException( 'Can\'t call prepareContent() again for different user! '
 730                                         . 'Expected ' . $this->user->getName() . ', got ' . $user->getName()
 731                                 );
 732                         }
 733
 734                         if ( !$this->slotsUpdate->hasSameUpdates( $slotsUpdate ) ) {
 735                                 throw new LogicException(
 736                                         'Can\'t call prepareContent() again with different slot content!'
 737                                 );
 738                         }
 739
 740                         return; // prepareContent() already done, nothing to do
 741                 }
 742
 743                 $this->assertTransition( 'has-content' );
 744
 745                 $wikiPage = $this->getWikiPage(); // TODO: use only for legacy hooks!
 746                 $title = $this->getTitle();
 747
 748                 $parentRevision = $this->grabCurrentRevision();
 749
 750                 $this->slotsOutput = [];
 751                 $this->canonicalParserOutput = null;
 752
 753                 // The edit may have already been prepared via api.php?action=stashedit
 754                 $stashedEdit = false;
 755
 756                 // TODO: MCR: allow output for all slots to be stashed.
 757                 if ( $useStash && $slotsUpdate->isModifiedSlot( SlotRecord::MAIN ) ) {
 758                         $mainContent = $slotsUpdate->getModifiedSlot( SlotRecord::MAIN )->getContent();
 759                         $legacyUser = User::newFromIdentity( $user );
 760                         $stashedEdit = ApiStashEdit::checkCache( $title, $mainContent, $legacyUser );
 761                 }
 762
 763                 $userPopts = ParserOptions::newFromUserAndLang( $user, $this->contLang );
 764                 Hooks::run( 'ArticlePrepareTextForEdit', [ $wikiPage, $userPopts ] );
 765
 766                 $this->user = $user;
 767                 $this->slotsUpdate = $slotsUpdate;
 768
 769                 if ( $parentRevision ) {
 770                         $this->revision = MutableRevisionRecord::newFromParentRevision( $parentRevision );
 771                 } else {
 772                         $this->revision = new MutableRevisionRecord( $title );
 773                 }
 774
 775                 // NOTE: user and timestamp must be set, so they can be used for
 776                 // {{subst:REVISIONUSER}} and {{subst:REVISIONTIMESTAMP}} in PST!
 777                 $this->revision->setTimestamp( wfTimestampNow() );
 778                 $this->revision->setUser( $user );
 779
 780                 // Set up ParserOptions to operate on the new revision
 781                 $oldCallback = $userPopts->getCurrentRevisionCallback();
 782                 $userPopts->setCurrentRevisionCallback(
 783                         function ( Title $parserTitle, $parser = false ) use ( $title, $oldCallback ) {
 784                                 if ( $parserTitle->equals( $title ) ) {
 785                                         $legacyRevision = new Revision( $this->revision );
 786                                         return $legacyRevision;
 787                                 } else {
 788                                         return call_user_func( $oldCallback, $parserTitle, $parser );
 789                                 }
 790                         }
 791                 );
 792
 793                 $pstContentSlots = $this->revision->getSlots();
 794
 795                 foreach ( $slotsUpdate->getModifiedRoles() as $role ) {
 796                         $slot = $slotsUpdate->getModifiedSlot( $role );
 797
 798                         if ( $slot->isInherited() ) {
 799                                 // No PST for inherited slots! Note that "modified" slots may still be inherited
 800                                 // from an earlier version, e.g. for rollbacks.
 801                                 $pstSlot = $slot;
 802                         } elseif ( $role === SlotRecord::MAIN && $stashedEdit ) {
 803                                 // TODO: MCR: allow PST content for all slots to be stashed.
 804                                 $pstSlot = SlotRecord::newUnsaved( $role, $stashedEdit->pstContent );
 805                         } else {
 806                                 $content = $slot->getContent();
 807                                 $pstContent = $content->preSaveTransform( $title, $this->user, $userPopts );
 808                                 $pstSlot = SlotRecord::newUnsaved( $role, $pstContent );
 809                         }
 810
 811                         $pstContentSlots->setSlot( $pstSlot );
 812                 }
 813
 814                 foreach ( $slotsUpdate->getRemovedRoles() as $role ) {
 815                         $pstContentSlots->removeSlot( $role );
 816                 }
 817
 818                 $this->options['created'] = ( $parentRevision === null );
 819                 $this->options['changed'] = ( $parentRevision === null
 820                         || !$pstContentSlots->hasSameContent( $parentRevision->getSlots() ) );
 821
 822                 $this->doTransition( 'has-content' );
 823
 824                 if ( !$this->options['changed'] ) {
 825                         // null-edit!
 826
 827                         // TODO: move this into MutableRevisionRecord
 828                         // TODO: This needs to behave differently for a forced dummy edit!
 829                         $this->revision->setId( $parentRevision->getId() );
 830                         $this->revision->setTimestamp( $parentRevision->getTimestamp() );
 831                         $this->revision->setPageId( $parentRevision->getPageId() );
 832                         $this->revision->setParentId( $parentRevision->getParentId() );
 833                         $this->revision->setUser( $parentRevision->getUser( RevisionRecord::RAW ) );
 834                         $this->revision->setComment( $parentRevision->getComment( RevisionRecord::RAW ) );
 835                         $this->revision->setMinorEdit( $parentRevision->isMinor() );
 836                         $this->revision->setVisibility( $parentRevision->getVisibility() );
 837
 838                         // prepareUpdate() is redundant for null-edits
 839                         $this->doTransition( 'has-revision' );
 840                 } else {
 841                         $this->parentRevision = $parentRevision;
 842                 }
 843
 844                 $renderHints = [ 'use-master' => $this->useMaster(), 'audience' => RevisionRecord::RAW ];
 845
 846                 if ( $stashedEdit ) {
 847                         /** @var ParserOutput $output */
 848                         $output = $stashedEdit->output;
 849
 850                         // TODO: this should happen when stashing the ParserOutput, not now!
 851                         $output->setCacheTime( $stashedEdit->timestamp );
 852
 853                         $renderHints['known-revision-output'] = $output;
 854                 }
 855
 856                 // NOTE: we want a canonical rendering, so don't pass $this->user or ParserOptions
 857                 // NOTE: the revision is either new or current, so we can bypass audience checks.
 858                 $this->renderedRevision = $this->revisionRenderer->getRenderedRevision(
 859                         $this->revision,
 860                         null,
 861                         null,
 862                         $renderHints
 863                 );
 864         }
 865
 866         /**
 867          * Returns the update's target revision - that is, the revision that will be the current
 868          * revision after the update.
 869          *
 870          * @note Callers must treat the returned RevisionRecord's content as immutable, even
 871          * if it is a MutableRevisionRecord instance. Other aspects of a MutableRevisionRecord
 872          * returned from here, such as the user or the comment, may be changed, but may not
 873          * be reflected in ParserOutput until after prepareUpdate() has been called.
 874          *
 875          * @todo This is currently used by PageUpdater::makeNewRevision() to construct an unsaved
 876          * MutableRevisionRecord instance. Introduce something like an UnsavedRevisionFactory service
 877          * for that purpose instead!
 878          *
 879          * @return RevisionRecord
 880          */
 881         public function getRevision() {
 882                 $this->assertPrepared( __METHOD__ );
 883                 return $this->revision;
 884         }
 885
 886         /**
 887          * @return RenderedRevision
 888          */
 889         public function getRenderedRevision() {
 890                 $this->assertPrepared( __METHOD__ );
 891
 892                 return $this->renderedRevision;
 893         }
 894
 895         private function assertHasPageState( $method ) {
 896                 if ( !$this->pageState ) {
 897                         throw new LogicException(
 898                                 'Must call grabCurrentRevision() or prepareContent() '
 899                                 . 'or prepareUpdate() before calling ' . $method
 900                         );
 901                 }
 902         }
 903
 904         private function assertPrepared( $method ) {
 905                 if ( !$this->revision ) {
 906                         throw new LogicException(
 907                                 'Must call prepareContent() or prepareUpdate() before calling ' . $method
 908                         );
 909                 }
 910         }
 911
 912         private function assertHasRevision( $method ) {
 913                 if ( !$this->revision->getId() ) {
 914                         throw new LogicException(
 915                                 'Must call prepareUpdate() before calling ' . $method
 916                         );
 917                 }
 918         }
 919
 920         /**
 921          * Whether the edit creates the page.
 922          *
 923          * @return bool
 924          */
 925         public function isCreation() {
 926                 $this->assertPrepared( __METHOD__ );
 927                 return $this->options['created'];
 928         }
 929
 930         /**
 931          * Whether the edit created, or should create, a new revision (that is, it's not a null-edit).
 932          *
 933          * @warning at present, "null-revisions" that do not change content but do have a revision
 934          * record would return false after prepareContent(), but true after prepareUpdate()!
 935          * This should probably be fixed.
 936          *
 937          * @return bool
 938          */
 939         public function isChange() {
 940                 $this->assertPrepared( __METHOD__ );
 941                 return $this->options['changed'];
 942         }
 943
 944         /**
 945          * Whether the page was a redirect before the edit.
 946          *
 947          * @return bool
 948          */
 949         public function wasRedirect() {
 950                 $this->assertHasPageState( __METHOD__ );
 951
 952                 if ( $this->pageState['oldIsRedirect'] === null ) {
 953                         /** @var RevisionRecord $rev */
 954                         $rev = $this->pageState['oldRevision'];
 955                         if ( $rev ) {
 956                                 $this->pageState['oldIsRedirect'] = $this->revisionIsRedirect( $rev );
 957                         } else {
 958                                 $this->pageState['oldIsRedirect'] = false;
 959                         }
 960                 }
 961
 962                 return $this->pageState['oldIsRedirect'];
 963         }
 964
 965         /**
 966          * Returns the slots of the target revision, after PST.
 967          *
 968          * @note Callers must treat the returned RevisionSlots instance as immutable, even
 969          * if it is a MutableRevisionSlots instance.
 970          *
 971          * @return RevisionSlots
 972          */
 973         public function getSlots() {
 974                 $this->assertPrepared( __METHOD__ );
 975                 return $this->revision->getSlots();
 976         }
 977
 978         /**
 979          * Returns the RevisionSlotsUpdate for this updater.
 980          *
 981          * @return RevisionSlotsUpdate
 982          */
 983         private function getRevisionSlotsUpdate() {
 984                 $this->assertPrepared( __METHOD__ );
 985
 986                 if ( !$this->slotsUpdate ) {
 987                         $old = $this->getParentRevision();
 988                         $this->slotsUpdate = RevisionSlotsUpdate::newFromRevisionSlots(
 989                                 $this->revision->getSlots(),
 990                                 $old ? $old->getSlots() : null
 991                         );
 992                 }
 993                 return $this->slotsUpdate;
 994         }
 995
 996         /**
 997          * Returns the role names of the slots touched by the new revision,
 998          * including removed roles.
 999          *
1000          * @return string[]
1001          */
1002         public function getTouchedSlotRoles() {
1003                 return $this->getRevisionSlotsUpdate()->getTouchedRoles();
1004         }
1005
1006         /**
1007          * Returns the role names of the slots modified by the new revision,
1008          * not including removed roles.
1009          *
1010          * @return string[]
1011          */
1012         public function getModifiedSlotRoles() {
1013                 return $this->getRevisionSlotsUpdate()->getModifiedRoles();
1014         }
1015
1016         /**
1017          * Returns the role names of the slots removed by the new revision.
1018          *
1019          * @return string[]
1020          */
1021         public function getRemovedSlotRoles() {
1022                 return $this->getRevisionSlotsUpdate()->getRemovedRoles();
1023         }
1024
1025         /**
1026          * Prepare derived data updates targeting the given Revision.
1027          *
1028          * Calling this method requires the given revision to be present in the database.
1029          * This may be right after a new revision has been created, or when re-generating
1030          * derived data e.g. in ApiPurge, RefreshLinksJob, and the refreshLinks
1031          * script.
1032          *
1033          * @see docs/pageupdater.txt for more information on when thie method can and should be called.
1034          *
1035          * @note Calling this method more than once with the same revision has no effect.
1036          * $options are only used for the first call. Calling this method multiple times with
1037          * different revisions will cause an exception.
1038          *
1039          * @note If grabCurrentRevision() (or prepareContent()) has been called before
1040          * calling this method, $revision->getParentRevision() has to refer to the revision that
1041          * was the current revision at the time grabCurrentRevision() was called.
1042          *
1043          * @param RevisionRecord $revision
1044          * @param array $options Array of options, following indexes are used:
1045          * - changed: bool, whether the revision changed the content (default true)
1046          * - created: bool, whether the revision created the page (default false)
1047          * - moved: bool, whether the page was moved (default false)
1048          * - restored: bool, whether the page was undeleted (default false)
1049          * - oldrevision: Revision object for the pre-update revision (default null)
1050          * - triggeringUser: The user triggering the update (UserIdentity, defaults to the
1051          *   user who created the revision)
1052          * - oldredirect: bool, null, or string 'no-change' (default null):
1053          *    - bool: whether the page was counted as a redirect before that
1054          *      revision, only used in changed is true and created is false
1055          *    - null or 'no-change': don't update the redirect status.
1056          * - oldcountable: bool, null, or string 'no-change' (default null):
1057          *    - bool: whether the page was counted as an article before that
1058          *      revision, only used in changed is true and created is false
1059          *    - null: if created is false, don't update the article count; if created
1060          *      is true, do update the article count
1061          *    - 'no-change': don't update the article count, ever
1062          *    When set to null, pageState['oldCountable'] will be used instead if available.
1063          *  - causeAction: an arbitrary string identifying the reason for the update.
1064          *    See DataUpdate::getCauseAction(). (default 'unknown')
1065          *  - causeAgent: name of the user who caused the update. See DataUpdate::getCauseAgent().
1066          *    (string, default 'unknown')
1067          */
1068         public function prepareUpdate( RevisionRecord $revision, array $options = [] ) {
1069                 Assert::parameter(
1070                         !isset( $options['oldrevision'] )
1071                         || $options['oldrevision'] instanceof Revision
1072                         || $options['oldrevision'] instanceof RevisionRecord,
1073                         '$options["oldrevision"]',
1074                         'must be a RevisionRecord (or Revision)'
1075                 );
1076                 Assert::parameter(
1077                         !isset( $options['triggeringUser'] )
1078                         || $options['triggeringUser'] instanceof UserIdentity,
1079                         '$options["triggeringUser"]',
1080                         'must be a UserIdentity'
1081                 );
1082
1083                 if ( !$revision->getId() ) {
1084                         throw new InvalidArgumentException(
1085                                 'Revision must have an ID set for it to be used with prepareUpdate()!'
1086                         );
1087                 }
1088
1089                 if ( $this->revision && $this->revision->getId() ) {
1090                         if ( $this->revision->getId() === $revision->getId() ) {
1091                                 return; // nothing to do!
1092                         } else {
1093                                 throw new LogicException(
1094                                         'Trying to re-use DerivedPageDataUpdater with revision '
1095                                         . $revision->getId()
1096                                         . ', but it\'s already bound to revision '
1097                                         . $this->revision->getId()
1098                                 );
1099                         }
1100                 }
1101
1102                 if ( $this->revision
1103                         && !$this->revision->getSlots()->hasSameContent( $revision->getSlots() )
1104                 ) {
1105                         throw new LogicException(
1106                                 'The Revision provided has mismatching content!'
1107                         );
1108                 }
1109
1110                 // Override fields defined in $this->options with values from $options.
1111                 $this->options = array_intersect_key( $options, $this->options ) + $this->options;
1112
1113                 if ( isset( $this->pageState['oldId'] ) ) {
1114                         $oldId = $this->pageState['oldId'];
1115                 } elseif ( isset( $this->options['oldrevision'] ) ) {
1116                         /** @var Revision|RevisionRecord $oldRev */
1117                         $oldRev = $this->options['oldrevision'];
1118                         $oldId = $oldRev->getId();
1119                 } else {
1120                         $oldId = $revision->getParentId();
1121                 }
1122
1123                 if ( $oldId !== null ) {
1124                         // XXX: what if $options['changed'] disagrees?
1125                         // MovePage creates a dummy revision with changed = false!
1126                         // We may want to explicitly distinguish between "no new revision" (null-edit)
1127                         // and "new revision without new content" (dummy revision).
1128
1129                         if ( $oldId === $revision->getParentId() ) {
1130                                 // NOTE: this may still be a NullRevision!
1131                                 // New revision!
1132                                 $this->options['changed'] = true;
1133                         } elseif ( $oldId === $revision->getId() ) {
1134                                 // Null-edit!
1135                                 $this->options['changed'] = false;
1136                         } else {
1137                                 // This indicates that calling code has given us the wrong Revision object
1138                                 throw new LogicException(
1139                                         'The Revision mismatches old revision ID: '
1140                                         . 'Old ID is ' . $oldId
1141                                         . ', parent ID is ' . $revision->getParentId()
1142                                         . ', revision ID is ' . $revision->getId()
1143                                 );
1144                         }
1145                 }
1146
1147                 // If prepareContent() was used to generate the PST content (which is indicated by
1148                 // $this->slotsUpdate being set), and this is not a null-edit, then the given
1149                 // revision must have the acting user as the revision author. Otherwise, user
1150                 // signatures generated by PST would mismatch the user in the revision record.
1151                 if ( $this->user !== null && $this->options['changed'] && $this->slotsUpdate ) {
1152                         $user = $revision->getUser();
1153                         if ( !$this->user->equals( $user ) ) {
1154                                 throw new LogicException(
1155                                         'The Revision provided has a mismatching actor: expected '
1156                                         . $this->user->getName()
1157                                         . ', got '
1158                                         . $user->getName()
1159                                 );
1160                         }
1161                 }
1162
1163                 // If $this->pageState was not yet initialized by grabCurrentRevision or prepareContent,
1164                 // emulate the state of the page table before the edit, as good as we can.
1165                 if ( !$this->pageState ) {
1166                         $this->pageState = [
1167                                 'oldIsRedirect' => isset( $this->options['oldredirect'] )
1168                                         && is_bool( $this->options['oldredirect'] )
1169                                                 ? $this->options['oldredirect']
1170                                                 : null,
1171                                 'oldCountable' => isset( $this->options['oldcountable'] )
1172                                         && is_bool( $this->options['oldcountable'] )
1173                                                 ? $this->options['oldcountable']
1174                                                 : null,
1175                         ];
1176
1177                         if ( $this->options['changed'] ) {
1178                                 // The edit created a new revision
1179                                 $this->pageState['oldId'] = $revision->getParentId();
1180
1181                                 if ( isset( $this->options['oldrevision'] ) ) {
1182                                         $rev = $this->options['oldrevision'];
1183                                         $this->pageState['oldRevision'] = $rev instanceof Revision
1184                                                 ? $rev->getRevisionRecord()
1185                                                 : $rev;
1186                                 }
1187                         } else {
1188                                 // This is a null-edit, so the old revision IS the new revision!
1189                                 $this->pageState['oldId'] = $revision->getId();
1190                                 $this->pageState['oldRevision'] = $revision;
1191                         }
1192                 }
1193
1194                 // "created" is forced here
1195                 $this->options['created'] = ( $this->pageState['oldId'] === 0 );
1196
1197                 $this->revision = $revision;
1198
1199                 $this->doTransition( 'has-revision' );
1200
1201                 // NOTE: in case we have a User object, don't override with a UserIdentity.
1202                 // We already checked that $revision->getUser() mathces $this->user;
1203                 if ( !$this->user ) {
1204                         $this->user = $revision->getUser( RevisionRecord::RAW );
1205                 }
1206
1207                 // Prune any output that depends on the revision ID.
1208                 if ( $this->renderedRevision ) {
1209                         $this->renderedRevision->updateRevision( $revision );
1210                 } else {
1211
1212                         // NOTE: we want a canonical rendering, so don't pass $this->user or ParserOptions
1213                         // NOTE: the revision is either new or current, so we can bypass audience checks.
1214                         $this->renderedRevision = $this->revisionRenderer->getRenderedRevision(
1215                                 $this->revision,
1216                                 null,
1217                                 null,
1218                                 [ 'use-master' => $this->useMaster(), 'audience' => RevisionRecord::RAW ]
1219                         );
1220
1221                         // XXX: Since we presumably are dealing with the current revision,
1222                         // we could try to get the ParserOutput from the parser cache.
1223                 }
1224
1225                 // TODO: optionally get ParserOutput from the ParserCache here.
1226                 // Move the logic used by RefreshLinksJob here!
1227         }
1228
1229         /**
1230          * @deprecated This only exists for B/C, use the getters on DerivedPageDataUpdater directly!
1231          * @return PreparedEdit
1232          */
1233         public function getPreparedEdit() {
1234                 $this->assertPrepared( __METHOD__ );
1235
1236                 $slotsUpdate = $this->getRevisionSlotsUpdate();
1237                 $preparedEdit = new PreparedEdit();
1238
1239                 $preparedEdit->popts = $this->getCanonicalParserOptions();
1240                 $preparedEdit->output = $this->getCanonicalParserOutput();
1241                 $preparedEdit->pstContent = $this->revision->getContent( SlotRecord::MAIN );
1242                 $preparedEdit->newContent =
1243                         $slotsUpdate->isModifiedSlot( SlotRecord::MAIN )
1244                         ? $slotsUpdate->getModifiedSlot( SlotRecord::MAIN )->getContent()
1245                         : $this->revision->getContent( SlotRecord::MAIN ); // XXX: can we just remove this?
1246                 $preparedEdit->oldContent = null; // unused. // XXX: could get this from the parent revision
1247                 $preparedEdit->revid = $this->revision ? $this->revision->getId() : null;
1248                 $preparedEdit->timestamp = $preparedEdit->output->getCacheTime();
1249                 $preparedEdit->format = $preparedEdit->pstContent->getDefaultFormat();
1250
1251                 return $preparedEdit;
1252         }
1253
1254         /**
1255          * @param string $role
1256          * @param bool $generateHtml
1257          * @return ParserOutput
1258          */
1259         public function getSlotParserOutput( $role, $generateHtml = true ) {
1260                 return $this->getRenderedRevision()->getSlotParserOutput(
1261                         $role,
1262                         [ 'generate-html' => $generateHtml ]
1263                 );
1264         }
1265
1266         /**
1267          * @return ParserOutput
1268          */
1269         public function getCanonicalParserOutput() {
1270                 return $this->getRenderedRevision()->getRevisionParserOutput();
1271         }
1272
1273         /**
1274          * @return ParserOptions
1275          */
1276         public function getCanonicalParserOptions() {
1277                 return $this->getRenderedRevision()->getOptions();
1278         }
1279
1280         /**
1281          * @param bool $recursive
1282          *
1283          * @return DeferrableUpdate[]
1284          */
1285         public function getSecondaryDataUpdates( $recursive = false ) {
1286                 if ( $this->isContentDeleted() ) {
1287                         // This shouldn't happen, since the current content is always public,
1288                         // and DataUpates are only needed for current content.
1289                         return [];
1290                 }
1291
1292                 $output = $this->getCanonicalParserOutput();
1293
1294                 // Construct a LinksUpdate for the combined canonical output.
1295                 $linksUpdate = new LinksUpdate(
1296                         $this->getTitle(),
1297                         $output,
1298                         $recursive
1299                 );
1300
1301                 $allUpdates = [ $linksUpdate ];
1302
1303                 // NOTE: Run updates for all slots, not just the modified slots! Otherwise,
1304                 // info for an inherited slot may end up being removed. This is also needed
1305                 // to ensure that purges are effective.
1306                 $renderedRevision = $this->getRenderedRevision();
1307                 foreach ( $this->getSlots()->getSlotRoles() as $role ) {
1308                         $slot = $this->getRawSlot( $role );
1309                         $content = $slot->getContent();
1310                         $handler = $content->getContentHandler();
1311
1312                         $updates = $handler->getSecondaryDataUpdates(
1313                                 $this->getTitle(),
1314                                 $content,
1315                                 $role,
1316                                 $renderedRevision
1317                         );
1318                         $allUpdates = array_merge( $allUpdates, $updates );
1319
1320                         // TODO: remove B/C hack in 1.32!
1321                         // NOTE: we assume that the combined output contains all relevant meta-data for
1322                         // all slots!
1323                         $legacyUpdates = $content->getSecondaryDataUpdates(
1324                                 $this->getTitle(),
1325                                 null,
1326                                 $recursive,
1327                                 $output
1328                         );
1329
1330                         // HACK: filter out redundant and incomplete LinksUpdates
1331                         $legacyUpdates = array_filter( $legacyUpdates, function ( $update ) {
1332                                 return !( $update instanceof LinksUpdate );
1333                         } );
1334
1335                         $allUpdates = array_merge( $allUpdates, $legacyUpdates );
1336                 }
1337
1338                 // XXX: if a slot was removed by an earlier edit, but deletion updates failed to run at
1339                 // that time, we don't know for which slots to run deletion updates when purging a page.
1340                 // We'd have to examine the entire history of the page to determine that. Perhaps there
1341                 // could be a "try extra hard" mode for that case that would run a DB query to find all
1342                 // roles/models ever used on the page. On the other hand, removing slots should be quite
1343                 // rare, so perhaps this isn't worth the trouble.
1344
1345                 // TODO: consolidate with similar logic in WikiPage::getDeletionUpdates()
1346                 $wikiPage = $this->getWikiPage();
1347                 $parentRevision = $this->getParentRevision();
1348                 foreach ( $this->getRemovedSlotRoles() as $role ) {
1349                         // HACK: we should get the content model of the removed slot from a SlotRoleHandler!
1350                         // For now, find the slot in the parent revision - if the slot was removed, it should
1351                         // always exist in the parent revision.
1352                         $parentSlot = $parentRevision->getSlot( $role, RevisionRecord::RAW );
1353                         $content = $parentSlot->getContent();
1354                         $handler = $content->getContentHandler();
1355
1356                         $updates = $handler->getDeletionUpdates(
1357                                 $this->getTitle(),
1358                                 $role
1359                         );
1360                         $allUpdates = array_merge( $allUpdates, $updates );
1361
1362                         // TODO: remove B/C hack in 1.32!
1363                         $legacyUpdates = $content->getDeletionUpdates( $wikiPage );
1364
1365                         // HACK: filter out redundant and incomplete LinksDeletionUpdate
1366                         $legacyUpdates = array_filter( $legacyUpdates, function ( $update ) {
1367                                 return !( $update instanceof LinksDeletionUpdate );
1368                         } );
1369
1370                         $allUpdates = array_merge( $allUpdates, $legacyUpdates );
1371                 }
1372
1373                 // TODO: hard deprecate SecondaryDataUpdates in favor of RevisionDataUpdates in 1.33!
1374                 Hooks::run(
1375                         'RevisionDataUpdates',
1376                         [ $this->getTitle(), $renderedRevision, &$allUpdates ]
1377                 );
1378
1379                 return $allUpdates;
1380         }
1381
1382         /**
1383          * Do standard updates after page edit, purge, or import.
1384          * Update links tables, site stats, search index, title cache, message cache, etc.
1385          * Purges pages that depend on this page when appropriate.
1386          * With a 10% chance, triggers pruning the recent changes table.
1387          *
1388          * @note prepareUpdate() must be called before calling this method!
1389          *
1390          * MCR migration note: this replaces WikiPage::doEditUpdates.
1391          */
1392         public function doUpdates() {
1393                 $this->assertTransition( 'done' );
1394
1395                 // TODO: move logic into a PageEventEmitter service
1396
1397                 $wikiPage = $this->getWikiPage(); // TODO: use only for legacy hooks!
1398
1399                 $legacyUser = User::newFromIdentity( $this->user );
1400                 $legacyRevision = new Revision( $this->revision );
1401
1402                 $this->doParserCacheUpdate();
1403
1404                 $this->doSecondaryDataUpdates( [
1405                         // T52785 do not update any other pages on a null edit
1406                         'recursive' => $this->options['changed'],
1407                         'defer' => DeferredUpdates::POSTSEND,
1408                 ] );
1409
1410                 // TODO: MCR: check if *any* changed slot supports categories!
1411                 if ( $this->rcWatchCategoryMembership
1412                         && $this->getContentHandler( SlotRecord::MAIN )->supportsCategories() === true
1413                         && ( $this->options['changed'] || $this->options['created'] )
1414                         && !$this->options['restored']
1415                 ) {
1416                         // Note: jobs are pushed after deferred updates, so the job should be able to see
1417                         // the recent change entry (also done via deferred updates) and carry over any
1418                         // bot/deletion/IP flags, ect.
1419                         $this->jobQueueGroup->lazyPush(
1420                                 CategoryMembershipChangeJob::newSpec(
1421                                         $this->getTitle(),
1422                                         $this->revision->getTimestamp()
1423                                 )
1424                         );
1425                 }
1426
1427                 // TODO: replace legacy hook! Use a listener on PageEventEmitter instead!
1428                 $editInfo = $this->getPreparedEdit();
1429                 Hooks::run( 'ArticleEditUpdates', [ &$wikiPage, &$editInfo, $this->options['changed'] ] );
1430
1431                 // TODO: replace legacy hook! Use a listener on PageEventEmitter instead!
1432                 if ( Hooks::run( 'ArticleEditUpdatesDeleteFromRecentchanges', [ &$wikiPage ] ) ) {
1433                         // Flush old entries from the `recentchanges` table
1434                         if ( mt_rand( 0, 9 ) == 0 ) {
1435                                 $this->jobQueueGroup->lazyPush( RecentChangesUpdateJob::newPurgeJob() );
1436                         }
1437                 }
1438
1439                 $id = $this->getPageId();
1440                 $title = $this->getTitle();
1441                 $dbKey = $title->getPrefixedDBkey();
1442                 $shortTitle = $title->getDBkey();
1443
1444                 if ( !$title->exists() ) {
1445                         wfDebug( __METHOD__ . ": Page doesn't exist any more, bailing out\n" );
1446
1447                         $this->doTransition( 'done' );
1448                         return;
1449                 }
1450
1451                 if ( $this->options['oldcountable'] === 'no-change' ||
1452                         ( !$this->options['changed'] && !$this->options['moved'] )
1453                 ) {
1454                         $good = 0;
1455                 } elseif ( $this->options['created'] ) {
1456                         $good = (int)$this->isCountable();
1457                 } elseif ( $this->options['oldcountable'] !== null ) {
1458                         $good = (int)$this->isCountable()
1459                                 - (int)$this->options['oldcountable'];
1460                 } elseif ( isset( $this->pageState['oldCountable'] ) ) {
1461                         $good = (int)$this->isCountable()
1462                                 - (int)$this->pageState['oldCountable'];
1463                 } else {
1464                         $good = 0;
1465                 }
1466                 $edits = $this->options['changed'] ? 1 : 0;
1467                 $pages = $this->options['created'] ? 1 : 0;
1468
1469                 DeferredUpdates::addUpdate( SiteStatsUpdate::factory(
1470                         [ 'edits' => $edits, 'articles' => $good, 'pages' => $pages ]
1471                 ) );
1472
1473                 // TODO: make search infrastructure aware of slots!
1474                 $mainSlot = $this->revision->getSlot( SlotRecord::MAIN );
1475                 if ( !$mainSlot->isInherited() && !$this->isContentDeleted() ) {
1476                         DeferredUpdates::addUpdate( new SearchUpdate( $id, $dbKey, $mainSlot->getContent() ) );
1477                 }
1478
1479                 // If this is another user's talk page, update newtalk.
1480                 // Don't do this if $options['changed'] = false (null-edits) nor if
1481                 // it's a minor edit and the user making the edit doesn't generate notifications for those.
1482                 if ( $this->options['changed']
1483                         && $title->getNamespace() == NS_USER_TALK
1484                         && $shortTitle != $legacyUser->getTitleKey()
1485                         && !( $this->revision->isMinor() && $legacyUser->isAllowed( 'nominornewtalk' ) )
1486                 ) {
1487                         $recipient = User::newFromName( $shortTitle, false );
1488                         if ( !$recipient ) {
1489                                 wfDebug( __METHOD__ . ": invalid username\n" );
1490                         } else {
1491                                 // Allow extensions to prevent user notification
1492                                 // when a new message is added to their talk page
1493                                 // TODO: replace legacy hook!  Use a listener on PageEventEmitter instead!
1494                                 if ( Hooks::run( 'ArticleEditUpdateNewTalk', [ &$wikiPage, $recipient ] ) ) {
1495                                         if ( User::isIP( $shortTitle ) ) {
1496                                                 // An anonymous user
1497                                                 $recipient->setNewtalk( true, $legacyRevision );
1498                                         } elseif ( $recipient->isLoggedIn() ) {
1499                                                 $recipient->setNewtalk( true, $legacyRevision );
1500                                         } else {
1501                                                 wfDebug( __METHOD__ . ": don't need to notify a nonexistent user\n" );
1502                                         }
1503                                 }
1504                         }
1505                 }
1506
1507                 if ( $title->getNamespace() == NS_MEDIAWIKI
1508                         && $this->getRevisionSlotsUpdate()->isModifiedSlot( SlotRecord::MAIN )
1509                 ) {
1510                         $mainContent = $this->isContentDeleted() ? null : $this->getRawContent( SlotRecord::MAIN );
1511
1512                         $this->messageCache->updateMessageOverride( $title, $mainContent );
1513                 }
1514
1515                 // TODO: move onArticleCreate and onArticle into a PageEventEmitter service
1516                 if ( $this->options['created'] ) {
1517                         WikiPage::onArticleCreate( $title );
1518                 } elseif ( $this->options['changed'] ) { // T52785
1519                         WikiPage::onArticleEdit( $title, $legacyRevision, $this->getTouchedSlotRoles() );
1520                 }
1521
1522                 $oldRevision = $this->getParentRevision();
1523                 $oldLegacyRevision = $oldRevision ? new Revision( $oldRevision ) : null;
1524
1525                 // TODO: In the wiring, register a listener for this on the new PageEventEmitter
1526                 ResourceLoaderWikiModule::invalidateModuleCache(
1527                         $title, $oldLegacyRevision, $legacyRevision, $this->getWikiId() ?: wfWikiID()
1528                 );
1529
1530                 $this->doTransition( 'done' );
1531         }
1532
1533         /**
1534          * Do secondary data updates (such as updating link tables).
1535          *
1536          * MCR note: this method is temporarily exposed via WikiPage::doSecondaryDataUpdates.
1537          *
1538          * @param array $options
1539          *   - recursive: make the update recursive, i.e. also update pages which transclude the
1540          *     current page or otherwise depend on it (default: false)
1541          *   - defer: one of the DeferredUpdates constants, or false to run immediately after waiting
1542          *     for replication of the changes from the SecondaryDataUpdates hooks (default: false)
1543          *   - transactionTicket: a transaction ticket from LBFactory::getEmptyTransactionTicket(),
1544          *     only when defer is false (default: null)
1545          * @since 1.32
1546          */
1547         public function doSecondaryDataUpdates( array $options = [] ) {
1548                 $this->assertHasRevision( __METHOD__ );
1549                 $options += [
1550                         'recursive' => false,
1551                         'defer' => false,
1552                         'transactionTicket' => null,
1553                 ];
1554                 $deferValues = [ false, DeferredUpdates::PRESEND, DeferredUpdates::POSTSEND ];
1555                 if ( !in_array( $options['defer'], $deferValues, true ) ) {
1556                         throw new InvalidArgumentException( 'invalid value for defer: ' . $options['defer'] );
1557                 }
1558                 Assert::parameterType( 'integer|null', $options['transactionTicket'],
1559                         '$options[\'transactionTicket\']' );
1560
1561                 $updates = $this->getSecondaryDataUpdates( $options['recursive'] );
1562
1563                 $triggeringUser = $this->options['triggeringUser'] ?? $this->user;
1564                 if ( !$triggeringUser instanceof User ) {
1565                         $triggeringUser = User::newFromIdentity( $triggeringUser );
1566                 }
1567                 $causeAction = $this->options['causeAction'] ?? 'unknown';
1568                 $causeAgent = $this->options['causeAgent'] ?? 'unknown';
1569                 $legacyRevision = new Revision( $this->revision );
1570
1571                 if ( $options['defer'] === false && $options['transactionTicket'] !== null ) {
1572                         // For legacy hook handlers doing updates via LinksUpdateConstructed, make sure
1573                         // any pending writes they made get flushed before the doUpdate() calls below.
1574                         // This avoids snapshot-clearing errors in LinksUpdate::acquirePageLock().
1575                         $this->loadbalancerFactory->commitAndWaitForReplication(
1576                                 __METHOD__, $options['transactionTicket']
1577                         );
1578                 }
1579
1580                 foreach ( $updates as $update ) {
1581                         if ( $update instanceof DataUpdate ) {
1582                                 $update->setCause( $causeAction, $causeAgent );
1583                         }
1584                         if ( $update instanceof LinksUpdate ) {
1585                                 $update->setRevision( $legacyRevision );
1586                                 $update->setTriggeringUser( $triggeringUser );
1587                         }
1588                         if ( $options['defer'] === false ) {
1589                                 if ( $options['transactionTicket'] !== null ) {
1590                                         $update->setTransactionTicket( $options['transactionTicket'] );
1591                                 }
1592                                 $update->doUpdate();
1593                         } else {
1594                                 DeferredUpdates::addUpdate( $update, $options['defer'] );
1595                         }
1596                 }
1597         }
1598
1599         public function doParserCacheUpdate() {
1600                 $this->assertHasRevision( __METHOD__ );
1601
1602                 $wikiPage = $this->getWikiPage(); // TODO: ParserCache should accept a RevisionRecord instead
1603
1604                 // NOTE: this may trigger the first parsing of the new content after an edit (when not
1605                 // using pre-generated stashed output).
1606                 // XXX: we may want to use the PoolCounter here. This would perhaps allow the initial parse
1607                 // to be performed post-send. The client could already follow a HTTP redirect to the
1608                 // page view, but would then have to wait for a response until rendering is complete.
1609                 $output = $this->getCanonicalParserOutput();
1610
1611                 // Save it to the parser cache. Use the revision timestamp in the case of a
1612                 // freshly saved edit, as that matches page_touched and a mismatch would trigger an
1613                 // unnecessary reparse.
1614                 $timestamp = $this->options['changed'] ? $this->revision->getTimestamp()
1615                         : $output->getTimestamp();
1616                 $this->parserCache->save(
1617                         $output, $wikiPage, $this->getCanonicalParserOptions(),
1618                         $timestamp, $this->revision->getId()
1619                 );
1620         }
1621
1622 }