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\SlotRecord;
  48 use MediaWiki\User\UserIdentity;
  49 use MessageCache;
  50 use ParserCache;
  51 use ParserOptions;
  52 use ParserOutput;
  53 use RecentChangesUpdateJob;
  54 use ResourceLoaderWikiModule;
  55 use Revision;
  56 use SearchUpdate;
  57 use SiteStatsUpdate;
  58 use Title;
  59 use User;
  60 use Wikimedia\Assert\Assert;
  61 use Wikimedia\Rdbms\LBFactory;
  62 use WikiPage;
  63
  64 /**
  65  * A handle for managing updates for derived page data on edit, import, purge, etc.
  66  *
  67  * @note Avoid direct usage of DerivedPageDataUpdater.
  68  *
  69  * @todo Define interfaces for the different use cases of DerivedPageDataUpdater, particularly
  70  * providing access to post-PST content and ParserOutput to callbacks during revision creation,
  71  * which currently use WikiPage::prepareContentForEdit, and allowing updates to be triggered on
  72  * purge, import, and undeletion, which currently use WikiPage::doEditUpdates() and
  73  * Content::getSecondaryDataUpdates().
  74  *
  75  * DerivedPageDataUpdater instances are designed to be cached inside a WikiPage instance,
  76  * and re-used by callback code over the course of an update operation. It's a stepping stone
  77  * one the way to a more complete refactoring of WikiPage.
  78  *
  79  * When using a DerivedPageDataUpdater, the following life cycle must be observed:
  80  * grabCurrentRevision (optional), prepareContent (optional), prepareUpdate (required
  81  * for doUpdates). getCanonicalParserOutput, getSlots, and getSecondaryDataUpdates
  82  * require prepareContent or prepareUpdate to have been called first, to initialize the
  83  * DerivedPageDataUpdater.
  84  *
  85  * @see docs/pageupdater.txt for more information.
  86  *
  87  * MCR migration note: this replaces the relevant methods in WikiPage, and covers the use cases
  88  * of PreparedEdit.
  89  *
  90  * @internal
  91  *
  92  * @since 1.32
  93  * @ingroup Page
  94  */
  95 class DerivedPageDataUpdater implements IDBAccessObject {
  96
  97         /**
  98          * @var UserIdentity|null
  99          */
 100         private $user = null;
 101
 102         /**
 103          * @var WikiPage
 104          */
 105         private $wikiPage;
 106
 107         /**
 108          * @var ParserCache
 109          */
 110         private $parserCache;
 111
 112         /**
 113          * @var RevisionStore
 114          */
 115         private $revisionStore;
 116
 117         /**
 118          * @var Language
 119          */
 120         private $contLang;
 121
 122         /**
 123          * @var JobQueueGroup
 124          */
 125         private $jobQueueGroup;
 126
 127         /**
 128          * @var MessageCache
 129          */
 130         private $messageCache;
 131
 132         /**
 133          * @var LBFactory
 134          */
 135         private $loadbalancerFactory;
 136
 137         /**
 138          * @var string see $wgArticleCountMethod
 139          */
 140         private $articleCountMethod;
 141
 142         /**
 143          * @var boolean see $wgRCWatchCategoryMembership
 144          */
 145         private $rcWatchCategoryMembership = false;
 146
 147         /**
 148          * Stores (most of) the $options parameter of prepareUpdate().
 149          * @see prepareUpdate()
 150          */
 151         private $options = [
 152                 'changed' => true,
 153                 'created' => false,
 154                 'moved' => false,
 155                 'restored' => false,
 156                 'oldrevision' => null,
 157                 'oldcountable' => null,
 158                 'oldredirect' => null,
 159                 'triggeringUser' => null,
 160                 // causeAction/causeAgent default to 'unknown' but that's handled where it's read,
 161                 // to make the life of prepareUpdate() callers easier.
 162                 'causeAction' => null,
 163                 'causeAgent' => null,
 164         ];
 165
 166         /**
 167          * The state of the relevant row in page table before the edit.
 168          * This is determined by the first call to grabCurrentRevision, prepareContent,
 169          * or prepareUpdate (so it is only accessible in 'knows-current' or a later stage).
 170          * If pageState was not initialized when prepareUpdate() is called, prepareUpdate() will
 171          * attempt to emulate the state of the page table before the edit.
 172          *
 173          * Contains the following fields:
 174          * - oldRevision (RevisionRecord|null): the revision that was current before the change
 175          *   associated with this update. Might not be set, use getParentRevision().
 176          * - oldId (int|null): the id of the above revision. 0 if there is no such revision (the change
 177          *   was about creating a new page); null if not known (that should not happen).
 178          * - oldIsRedirect (bool|null): whether the page was a redirect before the change. Lazy-loaded,
 179          *   can be null; use wasRedirect() instead of direct access.
 180          * - oldCountable (bool|null): whether the page was countable before the change (or null
 181          *   if we don't have that information)
 182          *
 183          * @var array
 184          */
 185         private $pageState = null;
 186
 187         /**
 188          * @var RevisionSlotsUpdate|null
 189          */
 190         private $slotsUpdate = null;
 191
 192         /**
 193          * @var RevisionRecord|null
 194          */
 195         private $parentRevision = null;
 196
 197         /**
 198          * @var RevisionRecord|null
 199          */
 200         private $revision = null;
 201
 202         /**
 203          * @var RenderedRevision
 204          */
 205         private $renderedRevision = null;
 206
 207         /**
 208          * @var RevisionRenderer
 209          */
 210         private $revisionRenderer;
 211
 212         /**
 213          * A stage identifier for managing the life cycle of this instance.
 214          * Possible stages are 'new', 'knows-current', 'has-content', 'has-revision', and 'done'.
 215          *
 216          * @see docs/pageupdater.txt for documentation of the life cycle.
 217          *
 218          * @var string
 219          */
 220         private $stage = 'new';
 221
 222         /**
 223          * Transition table for managing the life cycle of DerivedPageDateUpdater instances.
 224          *
 225          * XXX: Overkill. This is a linear order, we could just count. Names are nice though,
 226          * and constants are also overkill...
 227          *
 228          * @see docs/pageupdater.txt for documentation of the life cycle.
 229          *
 230          * @var array[]
 231          */
 232         private static $transitions = [
 233                 'new' => [
 234                         'new' => true,
 235                         'knows-current' => true,
 236                         'has-content' => true,
 237                         'has-revision' => true,
 238                 ],
 239                 'knows-current' => [
 240                         'knows-current' => true,
 241                         'has-content' => true,
 242                         'has-revision' => true,
 243                 ],
 244                 'has-content' => [
 245                         'has-content' => true,
 246                         'has-revision' => true,
 247                 ],
 248                 'has-revision' => [
 249                         'has-revision' => true,
 250                         'done' => true,
 251                 ],
 252         ];
 253
 254         /**
 255          * @param WikiPage $wikiPage ,
 256          * @param RevisionStore $revisionStore
 257          * @param RevisionRenderer $revisionRenderer
 258          * @param ParserCache $parserCache
 259          * @param JobQueueGroup $jobQueueGroup
 260          * @param MessageCache $messageCache
 261          * @param Language $contLang
 262          * @param LBFactory $loadbalancerFactory
 263          */
 264         public function __construct(
 265                 WikiPage $wikiPage,
 266                 RevisionStore $revisionStore,
 267                 RevisionRenderer $revisionRenderer,
 268                 ParserCache $parserCache,
 269                 JobQueueGroup $jobQueueGroup,
 270                 MessageCache $messageCache,
 271                 Language $contLang,
 272                 LBFactory $loadbalancerFactory
 273         ) {
 274                 $this->wikiPage = $wikiPage;
 275
 276                 $this->parserCache = $parserCache;
 277                 $this->revisionStore = $revisionStore;
 278                 $this->revisionRenderer = $revisionRenderer;
 279                 $this->jobQueueGroup = $jobQueueGroup;
 280                 $this->messageCache = $messageCache;
 281                 $this->contLang = $contLang;
 282                 // XXX only needed for waiting for replicas to catch up; there should be a narrower
 283                 // interface for that.
 284                 $this->loadbalancerFactory = $loadbalancerFactory;
 285         }
 286
 287         /**
 288          * Transition function for managing the life cycle of this instances.
 289          *
 290          * @see docs/pageupdater.txt for documentation of the life cycle.
 291          *
 292          * @param string $newStage the new stage
 293          * @return string the previous stage
 294          *
 295          * @throws LogicException If a transition to the given stage is not possible in the current
 296          *         stage.
 297          */
 298         private function doTransition( $newStage ) {
 299                 $this->assertTransition( $newStage );
 300
 301                 $oldStage = $this->stage;
 302                 $this->stage = $newStage;
 303
 304                 return $oldStage;
 305         }
 306
 307         /**
 308          * Asserts that a transition to the given stage is possible, without performing it.
 309          *
 310          * @see docs/pageupdater.txt for documentation of the life cycle.
 311          *
 312          * @param string $newStage the new stage
 313          *
 314          * @throws LogicException If this instance is not in the expected stage
 315          */
 316         private function assertTransition( $newStage ) {
 317                 if ( empty( self::$transitions[$this->stage][$newStage] ) ) {
 318                         throw new LogicException( "Cannot transition from {$this->stage} to $newStage" );
 319                 }
 320         }
 321
 322         /**
 323          * @return bool|string
 324          */
 325         private function getWikiId() {
 326                 // TODO: get from RevisionStore
 327                 return false;
 328         }
 329
 330         /**
 331          * Checks whether this DerivedPageDataUpdater can be re-used for running updates targeting
 332          * the given revision.
 333          *
 334          * @param UserIdentity|null $user The user creating the revision in question
 335          * @param RevisionRecord|null $revision New revision (after save, if already saved)
 336          * @param RevisionSlotsUpdate|null $slotsUpdate New content (before PST)
 337          * @param null|int $parentId Parent revision of the edit (use 0 for page creation)
 338          *
 339          * @return bool
 340          */
 341         public function isReusableFor(
 342                 UserIdentity $user = null,
 343                 RevisionRecord $revision = null,
 344                 RevisionSlotsUpdate $slotsUpdate = null,
 345                 $parentId = null
 346         ) {
 347                 if ( $revision
 348                         && $parentId
 349                         && $revision->getParentId() !== $parentId
 350                 ) {
 351                         throw new InvalidArgumentException( '$parentId should match the parent of $revision' );
 352                 }
 353
 354                 if ( $revision
 355                         && $user
 356                         && $revision->getUser( RevisionRecord::RAW )->getName() !== $user->getName()
 357                 ) {
 358                         throw new InvalidArgumentException( '$user should match the author of $revision' );
 359                 }
 360
 361                 if ( $user && $this->user && $user->getName() !== $this->user->getName() ) {
 362                         return false;
 363                 }
 364
 365                 if ( $revision && $this->revision && $this->revision->getId()
 366                         && $this->revision->getId() !== $revision->getId()
 367                 ) {
 368                         return false;
 369                 }
 370
 371                 if ( $revision && !$user ) {
 372                         $user = $revision->getUser( RevisionRecord::RAW );
 373                 }
 374
 375                 if ( $this->pageState
 376                         && $revision
 377                         && $revision->getParentId() !== null
 378                         && $this->pageState['oldId'] !== $revision->getParentId()
 379                 ) {
 380                         return false;
 381                 }
 382
 383                 if ( $this->pageState
 384                         && $parentId !== null
 385                         && $this->pageState['oldId'] !== $parentId
 386                 ) {
 387                         return false;
 388                 }
 389
 390                 if ( $this->revision
 391                         && $user
 392                         && $this->revision->getUser( RevisionRecord::RAW )
 393                         && $this->revision->getUser( RevisionRecord::RAW )->getName() !== $user->getName()
 394                 ) {
 395                         return false;
 396                 }
 397
 398                 if ( $revision
 399                         && $this->user
 400                         && $this->revision->getUser( RevisionRecord::RAW )
 401                         && $revision->getUser( RevisionRecord::RAW )->getName() !== $this->user->getName()
 402                 ) {
 403                         return false;
 404                 }
 405
 406                 // NOTE: this check is the primary reason for having the $this->slotsUpdate field!
 407                 if ( $this->slotsUpdate
 408                         && $slotsUpdate
 409                         && !$this->slotsUpdate->hasSameUpdates( $slotsUpdate )
 410                 ) {
 411                         return false;
 412                 }
 413
 414                 if ( $revision
 415                         && $this->revision
 416                         && !$this->revision->getSlots()->hasSameContent( $revision->getSlots() )
 417                 ) {
 418                         return false;
 419                 }
 420
 421                 return true;
 422         }
 423
 424         /**
 425          * @param string $articleCountMethod "any" or "link".
 426          * @see $wgArticleCountMethod
 427          */
 428         public function setArticleCountMethod( $articleCountMethod ) {
 429                 $this->articleCountMethod = $articleCountMethod;
 430         }
 431
 432         /**
 433          * @param bool $rcWatchCategoryMembership
 434          * @see $wgRCWatchCategoryMembership
 435          */
 436         public function setRcWatchCategoryMembership( $rcWatchCategoryMembership ) {
 437                 $this->rcWatchCategoryMembership = $rcWatchCategoryMembership;
 438         }
 439
 440         /**
 441          * @return Title
 442          */
 443         private function getTitle() {
 444                 // NOTE: eventually, we won't get a WikiPage passed into the constructor any more
 445                 return $this->wikiPage->getTitle();
 446         }
 447
 448         /**
 449          * @return WikiPage
 450          */
 451         private function getWikiPage() {
 452                 // NOTE: eventually, we won't get a WikiPage passed into the constructor any more
 453                 return $this->wikiPage;
 454         }
 455
 456         /**
 457          * Determines whether the page being edited already existed.
 458          * Only defined after calling grabCurrentRevision() or prepareContent() or prepareUpdate()!
 459          *
 460          * @return bool
 461          * @throws LogicException if called before grabCurrentRevision
 462          */
 463         public function pageExisted() {
 464                 $this->assertHasPageState( __METHOD__ );
 465
 466                 return $this->pageState['oldId'] > 0;
 467         }
 468
 469         /**
 470          * Returns the parent revision of the new revision wrapped by this update.
 471          * If the update is a null-edit, this will return the parent of the current (and new) revision.
 472          * This will return null if the revision wrapped by this update created the page.
 473          * Only defined after calling prepareContent() or prepareUpdate()!
 474          *
 475          * @return RevisionRecord|null the parent revision of the new revision, or null if
 476          *         the update created the page.
 477          */
 478         private function getParentRevision() {
 479                 $this->assertPrepared( __METHOD__ );
 480
 481                 if ( $this->parentRevision ) {
 482                         return $this->parentRevision;
 483                 }
 484
 485                 if ( !$this->pageState['oldId'] ) {
 486                         // If there was no current revision, there is no parent revision,
 487                         // since the page didn't exist.
 488                         return null;
 489                 }
 490
 491                 $oldId = $this->revision->getParentId();
 492                 $flags = $this->useMaster() ? RevisionStore::READ_LATEST : 0;
 493                 $this->parentRevision = $oldId
 494                         ? $this->revisionStore->getRevisionById( $oldId, $flags )
 495                         : null;
 496
 497                 return $this->parentRevision;
 498         }
 499
 500         /**
 501          * Returns the revision that was the page's current revision when grabCurrentRevision()
 502          * was first called.
 503          *
 504          * During an edit, that revision will act as the logical parent of the new revision.
 505          *
 506          * Some updates are performed based on the difference between the database state at the
 507          * moment this method is first called, and the state after the edit.
 508          *
 509          * @see docs/pageupdater.txt for more information on when thie method can and should be called.
 510          *
 511          * @note After prepareUpdate() was called, grabCurrentRevision() will throw an exception
 512          * to avoid confusion, since the page's current revision is then the new revision after
 513          * the edit, which was presumably passed to prepareUpdate() as the $revision parameter.
 514          * Use getParentRevision() instead to access the revision that is the parent of the
 515          * new revision.
 516          *
 517          * @return RevisionRecord|null the page's current revision, or null if the page does not
 518          * yet exist.
 519          */
 520         public function grabCurrentRevision() {
 521                 if ( $this->pageState ) {
 522                         return $this->pageState['oldRevision'];
 523                 }
 524
 525                 $this->assertTransition( 'knows-current' );
 526
 527                 // NOTE: eventually, we won't get a WikiPage passed into the constructor any more
 528                 $wikiPage = $this->getWikiPage();
 529
 530                 // Do not call WikiPage::clear(), since the caller may already have caused page data
 531                 // to be loaded with SELECT FOR UPDATE. Just assert it's loaded now.
 532                 $wikiPage->loadPageData( self::READ_LATEST );
 533                 $rev = $wikiPage->getRevision();
 534                 $current = $rev ? $rev->getRevisionRecord() : null;
 535
 536                 $this->pageState = [
 537                         'oldRevision' => $current,
 538                         'oldId' => $rev ? $rev->getId() : 0,
 539                         'oldIsRedirect' => $wikiPage->isRedirect(), // NOTE: uses page table
 540                         'oldCountable' => $wikiPage->isCountable(), // NOTE: uses pagelinks table
 541                 ];
 542
 543                 $this->doTransition( 'knows-current' );
 544
 545                 return $this->pageState['oldRevision'];
 546         }
 547
 548         /**
 549          * Whether prepareUpdate() or prepareContent() have been called on this instance.
 550          *
 551          * @return bool
 552          */
 553         public function isContentPrepared() {
 554                 return $this->revision !== null;
 555         }
 556
 557         /**
 558          * Whether prepareUpdate() has been called on this instance.
 559          *
 560          * @note will also return null in case of a null-edit!
 561          *
 562          * @return bool
 563          */
 564         public function isUpdatePrepared() {
 565                 return $this->revision !== null && $this->revision->getId() !== null;
 566         }
 567
 568         /**
 569          * @return int
 570          */
 571         private function getPageId() {
 572                 // NOTE: eventually, we won't get a WikiPage passed into the constructor any more
 573                 return $this->wikiPage->getId();
 574         }
 575
 576         /**
 577          * Whether the content is deleted and thus not visible to the public.
 578          *
 579          * @return bool
 580          */
 581         public function isContentDeleted() {
 582                 if ( $this->revision ) {
 583                         // XXX: if that revision is the current revision, this should be skipped
 584                         return $this->revision->isDeleted( RevisionRecord::DELETED_TEXT );
 585                 } else {
 586                         // If the content has not been saved yet, it cannot have been deleted yet.
 587                         return false;
 588                 }
 589         }
 590
 591         /**
 592          * Returns the slot, modified or inherited, after PST, with no audience checks applied.
 593          *
 594          * @param string $role slot role name
 595          *
 596          * @throws PageUpdateException If the slot is neither set for update nor inherited from the
 597          *        parent revision.
 598          * @return SlotRecord
 599          */
 600         public function getRawSlot( $role ) {
 601                 return $this->getSlots()->getSlot( $role );
 602         }
 603
 604         /**
 605          * Returns the content of the given slot, with no audience checks.
 606          *
 607          * @throws PageUpdateException If the slot is neither set for update nor inherited from the
 608          *        parent revision.
 609          * @param string $role slot role name
 610          * @return Content
 611          */
 612         public function getRawContent( $role ) {
 613                 return $this->getRawSlot( $role )->getContent();
 614         }
 615
 616         /**
 617          * Returns the content model of the given slot
 618          *
 619          * @param string $role slot role name
 620          * @return string
 621          */
 622         private function getContentModel( $role ) {
 623                 return $this->getRawSlot( $role )->getModel();
 624         }
 625
 626         /**
 627          * @param string $role slot role name
 628          * @return ContentHandler
 629          */
 630         private function getContentHandler( $role ) {
 631                 // TODO: inject something like a ContentHandlerRegistry
 632                 return ContentHandler::getForModelID( $this->getContentModel( $role ) );
 633         }
 634
 635         private function useMaster() {
 636                 // TODO: can we just set a flag to true in prepareContent()?
 637                 return $this->wikiPage->wasLoadedFrom( self::READ_LATEST );
 638         }
 639
 640         /**
 641          * @return bool
 642          */
 643         public function isCountable() {
 644                 // NOTE: Keep in sync with WikiPage::isCountable.
 645
 646                 if ( !$this->getTitle()->isContentPage() ) {
 647                         return false;
 648                 }
 649
 650                 if ( $this->isContentDeleted() ) {
 651                         // This should be irrelevant: countability only applies to the current revision,
 652                         // and the current revision is never suppressed.
 653                         return false;
 654                 }
 655
 656                 if ( $this->isRedirect() ) {
 657                         return false;
 658                 }
 659
 660                 $hasLinks = null;
 661
 662                 if ( $this->articleCountMethod === 'link' ) {
 663                         $hasLinks = (bool)count( $this->getCanonicalParserOutput()->getLinks() );
 664                 }
 665
 666                 // TODO: MCR: ask all slots if they have links [SlotHandler/PageTypeHandler]
 667                 $mainContent = $this->getRawContent( SlotRecord::MAIN );
 668                 return $mainContent->isCountable( $hasLinks );
 669         }
 670
 671         /**
 672          * @return bool
 673          */
 674         public function isRedirect() {
 675                 // NOTE: main slot determines redirect status
 676                 $mainContent = $this->getRawContent( SlotRecord::MAIN );
 677
 678                 return $mainContent->isRedirect();
 679         }
 680
 681         /**
 682          * @param RevisionRecord $rev
 683          *
 684          * @return bool
 685          */
 686         private function revisionIsRedirect( RevisionRecord $rev ) {
 687                 // NOTE: main slot determines redirect status
 688                 $mainContent = $rev->getContent( SlotRecord::MAIN, RevisionRecord::RAW );
 689
 690                 return $mainContent->isRedirect();
 691         }
 692
 693         /**
 694          * Prepare updates based on an update which has not yet been saved.
 695          *
 696          * This may be used to create derived data that is needed when creating a new revision;
 697          * particularly, this makes available the slots of the new revision via the getSlots()
 698          * method, after applying PST and slot inheritance.
 699          *
 700          * The derived data prepared for revision creation may then later be re-used by doUpdates(),
 701          * without the need to re-calculate.
 702          *
 703          * @see docs/pageupdater.txt for more information on when thie method can and should be called.
 704          *
 705          * @note Calling this method more than once with the same $slotsUpdate
 706          * has no effect. Calling this method multiple times with different content will cause
 707          * an exception.
 708          *
 709          * @note Calling this method after prepareUpdate() has been called will cause an exception.
 710          *
 711          * @param User $user The user to act as context for pre-save transformation (PST).
 712          *        Type hint should be reduced to UserIdentity at some point.
 713          * @param RevisionSlotsUpdate $slotsUpdate The new content of the slots to be updated
 714          *        by this edit, before PST.
 715          * @param bool $useStash Whether to use stashed ParserOutput
 716          */
 717         public function prepareContent(
 718                 User $user,
 719                 RevisionSlotsUpdate $slotsUpdate,
 720                 $useStash = true
 721         ) {
 722                 if ( $this->slotsUpdate ) {
 723                         if ( !$this->user ) {
 724                                 throw new LogicException(
 725                                         'Unexpected state: $this->slotsUpdate was initialized, '
 726                                         . 'but $this->user was not.'
 727                                 );
 728                         }
 729
 730                         if ( $this->user->getName() !== $user->getName() ) {
 731                                 throw new LogicException( 'Can\'t call prepareContent() again for different user! '
 732                                         . 'Expected ' . $this->user->getName() . ', got ' . $user->getName()
 733                                 );
 734                         }
 735
 736                         if ( !$this->slotsUpdate->hasSameUpdates( $slotsUpdate ) ) {
 737                                 throw new LogicException(
 738                                         'Can\'t call prepareContent() again with different slot content!'
 739                                 );
 740                         }
 741
 742                         return; // prepareContent() already done, nothing to do
 743                 }
 744
 745                 $this->assertTransition( 'has-content' );
 746
 747                 $wikiPage = $this->getWikiPage(); // TODO: use only for legacy hooks!
 748                 $title = $this->getTitle();
 749
 750                 $parentRevision = $this->grabCurrentRevision();
 751
 752                 $this->slotsOutput = [];
 753                 $this->canonicalParserOutput = null;
 754
 755                 // The edit may have already been prepared via api.php?action=stashedit
 756                 $stashedEdit = false;
 757
 758                 // TODO: MCR: allow output for all slots to be stashed.
 759                 if ( $useStash && $slotsUpdate->isModifiedSlot( SlotRecord::MAIN ) ) {
 760                         $mainContent = $slotsUpdate->getModifiedSlot( SlotRecord::MAIN )->getContent();
 761                         $legacyUser = User::newFromIdentity( $user );
 762                         $stashedEdit = ApiStashEdit::checkCache( $title, $mainContent, $legacyUser );
 763                 }
 764
 765                 $userPopts = ParserOptions::newFromUserAndLang( $user, $this->contLang );
 766                 Hooks::run( 'ArticlePrepareTextForEdit', [ $wikiPage, $userPopts ] );
 767
 768                 $this->user = $user;
 769                 $this->slotsUpdate = $slotsUpdate;
 770
 771                 if ( $parentRevision ) {
 772                         $this->revision = MutableRevisionRecord::newFromParentRevision( $parentRevision );
 773                 } else {
 774                         $this->revision = new MutableRevisionRecord( $title );
 775                 }
 776
 777                 // NOTE: user and timestamp must be set, so they can be used for
 778                 // {{subst:REVISIONUSER}} and {{subst:REVISIONTIMESTAMP}} in PST!
 779                 $this->revision->setTimestamp( wfTimestampNow() );
 780                 $this->revision->setUser( $user );
 781
 782                 // Set up ParserOptions to operate on the new revision
 783                 $oldCallback = $userPopts->getCurrentRevisionCallback();
 784                 $userPopts->setCurrentRevisionCallback(
 785                         function ( Title $parserTitle, $parser = false ) use ( $title, $oldCallback ) {
 786                                 if ( $parserTitle->equals( $title ) ) {
 787                                         $legacyRevision = new Revision( $this->revision );
 788                                         return $legacyRevision;
 789                                 } else {
 790                                         return call_user_func( $oldCallback, $parserTitle, $parser );
 791                                 }
 792                         }
 793                 );
 794
 795                 $pstContentSlots = $this->revision->getSlots();
 796
 797                 foreach ( $slotsUpdate->getModifiedRoles() as $role ) {
 798                         $slot = $slotsUpdate->getModifiedSlot( $role );
 799
 800                         if ( $slot->isInherited() ) {
 801                                 // No PST for inherited slots! Note that "modified" slots may still be inherited
 802                                 // from an earlier version, e.g. for rollbacks.
 803                                 $pstSlot = $slot;
 804                         } elseif ( $role === SlotRecord::MAIN && $stashedEdit ) {
 805                                 // TODO: MCR: allow PST content for all slots to be stashed.
 806                                 $pstSlot = SlotRecord::newUnsaved( $role, $stashedEdit->pstContent );
 807                         } else {
 808                                 $content = $slot->getContent();
 809                                 $pstContent = $content->preSaveTransform( $title, $this->user, $userPopts );
 810                                 $pstSlot = SlotRecord::newUnsaved( $role, $pstContent );
 811                         }
 812
 813                         $pstContentSlots->setSlot( $pstSlot );
 814                 }
 815
 816                 foreach ( $slotsUpdate->getRemovedRoles() as $role ) {
 817                         $pstContentSlots->removeSlot( $role );
 818                 }
 819
 820                 $this->options['created'] = ( $parentRevision === null );
 821                 $this->options['changed'] = ( $parentRevision === null
 822                         || !$pstContentSlots->hasSameContent( $parentRevision->getSlots() ) );
 823
 824                 $this->doTransition( 'has-content' );
 825
 826                 if ( !$this->options['changed'] ) {
 827                         // null-edit!
 828
 829                         // TODO: move this into MutableRevisionRecord
 830                         // TODO: This needs to behave differently for a forced dummy edit!
 831                         $this->revision->setId( $parentRevision->getId() );
 832                         $this->revision->setTimestamp( $parentRevision->getTimestamp() );
 833                         $this->revision->setPageId( $parentRevision->getPageId() );
 834                         $this->revision->setParentId( $parentRevision->getParentId() );
 835                         $this->revision->setUser( $parentRevision->getUser( RevisionRecord::RAW ) );
 836                         $this->revision->setComment( $parentRevision->getComment( RevisionRecord::RAW ) );
 837                         $this->revision->setMinorEdit( $parentRevision->isMinor() );
 838                         $this->revision->setVisibility( $parentRevision->getVisibility() );
 839
 840                         // prepareUpdate() is redundant for null-edits
 841                         $this->doTransition( 'has-revision' );
 842                 } else {
 843                         $this->parentRevision = $parentRevision;
 844                 }
 845
 846                 $renderHints = [ 'use-master' => $this->useMaster(), 'audience' => RevisionRecord::RAW ];
 847
 848                 if ( $stashedEdit ) {
 849                         /** @var ParserOutput $output */
 850                         $output = $stashedEdit->output;
 851
 852                         // TODO: this should happen when stashing the ParserOutput, not now!
 853                         $output->setCacheTime( $stashedEdit->timestamp );
 854
 855                         $renderHints['known-revision-output'] = $output;
 856                 }
 857
 858                 // NOTE: we want a canonical rendering, so don't pass $this->user or ParserOptions
 859                 // NOTE: the revision is either new or current, so we can bypass audience checks.
 860                 $this->renderedRevision = $this->revisionRenderer->getRenderedRevision(
 861                         $this->revision,
 862                         null,
 863                         null,
 864                         $renderHints
 865                 );
 866         }
 867
 868         /**
 869          * Returns the update's target revision - that is, the revision that will be the current
 870          * revision after the update.
 871          *
 872          * @note Callers must treat the returned RevisionRecord's content as immutable, even
 873          * if it is a MutableRevisionRecord instance. Other aspects of a MutableRevisionRecord
 874          * returned from here, such as the user or the comment, may be changed, but may not
 875          * be reflected in ParserOutput until after prepareUpdate() has been called.
 876          *
 877          * @todo This is currently used by PageUpdater::makeNewRevision() to construct an unsaved
 878          * MutableRevisionRecord instance. Introduce something like an UnsavedRevisionFactory service
 879          * for that purpose instead!
 880          *
 881          * @return RevisionRecord
 882          */
 883         public function getRevision() {
 884                 $this->assertPrepared( __METHOD__ );
 885                 return $this->revision;
 886         }
 887
 888         /**
 889          * @return RenderedRevision
 890          */
 891         public function getRenderedRevision() {
 892                 $this->assertPrepared( __METHOD__ );
 893
 894                 return $this->renderedRevision;
 895         }
 896
 897         private function assertHasPageState( $method ) {
 898                 if ( !$this->pageState ) {
 899                         throw new LogicException(
 900                                 'Must call grabCurrentRevision() or prepareContent() '
 901                                 . 'or prepareUpdate() before calling ' . $method
 902                         );
 903                 }
 904         }
 905
 906         private function assertPrepared( $method ) {
 907                 if ( !$this->revision ) {
 908                         throw new LogicException(
 909                                 'Must call prepareContent() or prepareUpdate() before calling ' . $method
 910                         );
 911                 }
 912         }
 913
 914         private function assertHasRevision( $method ) {
 915                 if ( !$this->revision->getId() ) {
 916                         throw new LogicException(
 917                                 'Must call prepareUpdate() before calling ' . $method
 918                         );
 919                 }
 920         }
 921
 922         /**
 923          * Whether the edit creates the page.
 924          *
 925          * @return bool
 926          */
 927         public function isCreation() {
 928                 $this->assertPrepared( __METHOD__ );
 929                 return $this->options['created'];
 930         }
 931
 932         /**
 933          * Whether the edit created, or should create, a new revision (that is, it's not a null-edit).
 934          *
 935          * @warning at present, "null-revisions" that do not change content but do have a revision
 936          * record would return false after prepareContent(), but true after prepareUpdate()!
 937          * This should probably be fixed.
 938          *
 939          * @return bool
 940          */
 941         public function isChange() {
 942                 $this->assertPrepared( __METHOD__ );
 943                 return $this->options['changed'];
 944         }
 945
 946         /**
 947          * Whether the page was a redirect before the edit.
 948          *
 949          * @return bool
 950          */
 951         public function wasRedirect() {
 952                 $this->assertHasPageState( __METHOD__ );
 953
 954                 if ( $this->pageState['oldIsRedirect'] === null ) {
 955                         /** @var RevisionRecord $rev */
 956                         $rev = $this->pageState['oldRevision'];
 957                         if ( $rev ) {
 958                                 $this->pageState['oldIsRedirect'] = $this->revisionIsRedirect( $rev );
 959                         } else {
 960                                 $this->pageState['oldIsRedirect'] = false;
 961                         }
 962                 }
 963
 964                 return $this->pageState['oldIsRedirect'];
 965         }
 966
 967         /**
 968          * Returns the slots of the target revision, after PST.
 969          *
 970          * @note Callers must treat the returned RevisionSlots instance as immutable, even
 971          * if it is a MutableRevisionSlots instance.
 972          *
 973          * @return RevisionSlots
 974          */
 975         public function getSlots() {
 976                 $this->assertPrepared( __METHOD__ );
 977                 return $this->revision->getSlots();
 978         }
 979
 980         /**
 981          * Returns the RevisionSlotsUpdate for this updater.
 982          *
 983          * @return RevisionSlotsUpdate
 984          */
 985         private function getRevisionSlotsUpdate() {
 986                 $this->assertPrepared( __METHOD__ );
 987
 988                 if ( !$this->slotsUpdate ) {
 989                         $old = $this->getParentRevision();
 990                         $this->slotsUpdate = RevisionSlotsUpdate::newFromRevisionSlots(
 991                                 $this->revision->getSlots(),
 992                                 $old ? $old->getSlots() : null
 993                         );
 994                 }
 995                 return $this->slotsUpdate;
 996         }
 997
 998         /**
 999          * Returns the role names of the slots touched by the new revision,
1000          * including removed roles.
1001          *
1002          * @return string[]
1003          */
1004         public function getTouchedSlotRoles() {
1005                 return $this->getRevisionSlotsUpdate()->getTouchedRoles();
1006         }
1007
1008         /**
1009          * Returns the role names of the slots modified by the new revision,
1010          * not including removed roles.
1011          *
1012          * @return string[]
1013          */
1014         public function getModifiedSlotRoles() {
1015                 return $this->getRevisionSlotsUpdate()->getModifiedRoles();
1016         }
1017
1018         /**
1019          * Returns the role names of the slots removed by the new revision.
1020          *
1021          * @return string[]
1022          */
1023         public function getRemovedSlotRoles() {
1024                 return $this->getRevisionSlotsUpdate()->getRemovedRoles();
1025         }
1026
1027         /**
1028          * Prepare derived data updates targeting the given Revision.
1029          *
1030          * Calling this method requires the given revision to be present in the database.
1031          * This may be right after a new revision has been created, or when re-generating
1032          * derived data e.g. in ApiPurge, RefreshLinksJob, and the refreshLinks
1033          * script.
1034          *
1035          * @see docs/pageupdater.txt for more information on when thie method can and should be called.
1036          *
1037          * @note Calling this method more than once with the same revision has no effect.
1038          * $options are only used for the first call. Calling this method multiple times with
1039          * different revisions will cause an exception.
1040          *
1041          * @note If grabCurrentRevision() (or prepareContent()) has been called before
1042          * calling this method, $revision->getParentRevision() has to refer to the revision that
1043          * was the current revision at the time grabCurrentRevision() was called.
1044          *
1045          * @param RevisionRecord $revision
1046          * @param array $options Array of options, following indexes are used:
1047          * - changed: bool, whether the revision changed the content (default true)
1048          * - created: bool, whether the revision created the page (default false)
1049          * - moved: bool, whether the page was moved (default false)
1050          * - restored: bool, whether the page was undeleted (default false)
1051          * - oldrevision: Revision object for the pre-update revision (default null)
1052          * - triggeringUser: The user triggering the update (UserIdentity, defaults to the
1053          *   user who created the revision)
1054          * - oldredirect: bool, null, or string 'no-change' (default null):
1055          *    - bool: whether the page was counted as a redirect before that
1056          *      revision, only used in changed is true and created is false
1057          *    - null or 'no-change': don't update the redirect status.
1058          * - oldcountable: bool, null, or string 'no-change' (default null):
1059          *    - bool: whether the page was counted as an article before that
1060          *      revision, only used in changed is true and created is false
1061          *    - null: if created is false, don't update the article count; if created
1062          *      is true, do update the article count
1063          *    - 'no-change': don't update the article count, ever
1064          *    When set to null, pageState['oldCountable'] will be used instead if available.
1065          *  - causeAction: an arbitrary string identifying the reason for the update.
1066          *    See DataUpdate::getCauseAction(). (default 'unknown')
1067          *  - causeAgent: name of the user who caused the update. See DataUpdate::getCauseAgent().
1068          *    (string, default 'unknown')
1069          */
1070         public function prepareUpdate( RevisionRecord $revision, array $options = [] ) {
1071                 Assert::parameter(
1072                         !isset( $options['oldrevision'] )
1073                         || $options['oldrevision'] instanceof Revision
1074                         || $options['oldrevision'] instanceof RevisionRecord,
1075                         '$options["oldrevision"]',
1076                         'must be a RevisionRecord (or Revision)'
1077                 );
1078                 Assert::parameter(
1079                         !isset( $options['triggeringUser'] )
1080                         || $options['triggeringUser'] instanceof UserIdentity,
1081                         '$options["triggeringUser"]',
1082                         'must be a UserIdentity'
1083                 );
1084
1085                 if ( !$revision->getId() ) {
1086                         throw new InvalidArgumentException(
1087                                 'Revision must have an ID set for it to be used with prepareUpdate()!'
1088                         );
1089                 }
1090
1091                 if ( $this->revision && $this->revision->getId() ) {
1092                         if ( $this->revision->getId() === $revision->getId() ) {
1093                                 return; // nothing to do!
1094                         } else {
1095                                 throw new LogicException(
1096                                         'Trying to re-use DerivedPageDataUpdater with revision '
1097                                         . $revision->getId()
1098                                         . ', but it\'s already bound to revision '
1099                                         . $this->revision->getId()
1100                                 );
1101                         }
1102                 }
1103
1104                 if ( $this->revision
1105                         && !$this->revision->getSlots()->hasSameContent( $revision->getSlots() )
1106                 ) {
1107                         throw new LogicException(
1108                                 'The Revision provided has mismatching content!'
1109                         );
1110                 }
1111
1112                 // Override fields defined in $this->options with values from $options.
1113                 $this->options = array_intersect_key( $options, $this->options ) + $this->options;
1114
1115                 if ( isset( $this->pageState['oldId'] ) ) {
1116                         $oldId = $this->pageState['oldId'];
1117                 } elseif ( isset( $this->options['oldrevision'] ) ) {
1118                         /** @var Revision|RevisionRecord $oldRev */
1119                         $oldRev = $this->options['oldrevision'];
1120                         $oldId = $oldRev->getId();
1121                 } else {
1122                         $oldId = $revision->getParentId();
1123                 }
1124
1125                 if ( $oldId !== null ) {
1126                         // XXX: what if $options['changed'] disagrees?
1127                         // MovePage creates a dummy revision with changed = false!
1128                         // We may want to explicitly distinguish between "no new revision" (null-edit)
1129                         // and "new revision without new content" (dummy revision).
1130
1131                         if ( $oldId === $revision->getParentId() ) {
1132                                 // NOTE: this may still be a NullRevision!
1133                                 // New revision!
1134                                 $this->options['changed'] = true;
1135                         } elseif ( $oldId === $revision->getId() ) {
1136                                 // Null-edit!
1137                                 $this->options['changed'] = false;
1138                         } else {
1139                                 // This indicates that calling code has given us the wrong Revision object
1140                                 throw new LogicException(
1141                                         'The Revision mismatches old revision ID: '
1142                                         . 'Old ID is ' . $oldId
1143                                         . ', parent ID is ' . $revision->getParentId()
1144                                         . ', revision ID is ' . $revision->getId()
1145                                 );
1146                         }
1147                 }
1148
1149                 // If prepareContent() was used to generate the PST content (which is indicated by
1150                 // $this->slotsUpdate being set), and this is not a null-edit, then the given
1151                 // revision must have the acting user as the revision author. Otherwise, user
1152                 // signatures generated by PST would mismatch the user in the revision record.
1153                 if ( $this->user !== null && $this->options['changed'] && $this->slotsUpdate ) {
1154                         $user = $revision->getUser();
1155                         if ( !$this->user->equals( $user ) ) {
1156                                 throw new LogicException(
1157                                         'The Revision provided has a mismatching actor: expected '
1158                                         . $this->user->getName()
1159                                         . ', got '
1160                                         . $user->getName()
1161                                 );
1162                         }
1163                 }
1164
1165                 // If $this->pageState was not yet initialized by grabCurrentRevision or prepareContent,
1166                 // emulate the state of the page table before the edit, as good as we can.
1167                 if ( !$this->pageState ) {
1168                         $this->pageState = [
1169                                 'oldIsRedirect' => isset( $this->options['oldredirect'] )
1170                                         && is_bool( $this->options['oldredirect'] )
1171                                                 ? $this->options['oldredirect']
1172                                                 : null,
1173                                 'oldCountable' => isset( $this->options['oldcountable'] )
1174                                         && is_bool( $this->options['oldcountable'] )
1175                                                 ? $this->options['oldcountable']
1176                                                 : null,
1177                         ];
1178
1179                         if ( $this->options['changed'] ) {
1180                                 // The edit created a new revision
1181                                 $this->pageState['oldId'] = $revision->getParentId();
1182
1183                                 if ( isset( $this->options['oldrevision'] ) ) {
1184                                         $rev = $this->options['oldrevision'];
1185                                         $this->pageState['oldRevision'] = $rev instanceof Revision
1186                                                 ? $rev->getRevisionRecord()
1187                                                 : $rev;
1188                                 }
1189                         } else {
1190                                 // This is a null-edit, so the old revision IS the new revision!
1191                                 $this->pageState['oldId'] = $revision->getId();
1192                                 $this->pageState['oldRevision'] = $revision;
1193                         }
1194                 }
1195
1196                 // "created" is forced here
1197                 $this->options['created'] = ( $this->pageState['oldId'] === 0 );
1198
1199                 $this->revision = $revision;
1200
1201                 $this->doTransition( 'has-revision' );
1202
1203                 // NOTE: in case we have a User object, don't override with a UserIdentity.
1204                 // We already checked that $revision->getUser() mathces $this->user;
1205                 if ( !$this->user ) {
1206                         $this->user = $revision->getUser( RevisionRecord::RAW );
1207                 }
1208
1209                 // Prune any output that depends on the revision ID.
1210                 if ( $this->renderedRevision ) {
1211                         $this->renderedRevision->updateRevision( $revision );
1212                 } else {
1213
1214                         // NOTE: we want a canonical rendering, so don't pass $this->user or ParserOptions
1215                         // NOTE: the revision is either new or current, so we can bypass audience checks.
1216                         $this->renderedRevision = $this->revisionRenderer->getRenderedRevision(
1217                                 $this->revision,
1218                                 null,
1219                                 null,
1220                                 [ 'use-master' => $this->useMaster(), 'audience' => RevisionRecord::RAW ]
1221                         );
1222
1223                         // XXX: Since we presumably are dealing with the current revision,
1224                         // we could try to get the ParserOutput from the parser cache.
1225                 }
1226
1227                 // TODO: optionally get ParserOutput from the ParserCache here.
1228                 // Move the logic used by RefreshLinksJob here!
1229         }
1230
1231         /**
1232          * @deprecated This only exists for B/C, use the getters on DerivedPageDataUpdater directly!
1233          * @return PreparedEdit
1234          */
1235         public function getPreparedEdit() {
1236                 $this->assertPrepared( __METHOD__ );
1237
1238                 $slotsUpdate = $this->getRevisionSlotsUpdate();
1239                 $preparedEdit = new PreparedEdit();
1240
1241                 $preparedEdit->popts = $this->getCanonicalParserOptions();
1242                 $preparedEdit->output = $this->getCanonicalParserOutput();
1243                 $preparedEdit->pstContent = $this->revision->getContent( SlotRecord::MAIN );
1244                 $preparedEdit->newContent =
1245                         $slotsUpdate->isModifiedSlot( SlotRecord::MAIN )
1246                         ? $slotsUpdate->getModifiedSlot( SlotRecord::MAIN )->getContent()
1247                         : $this->revision->getContent( SlotRecord::MAIN ); // XXX: can we just remove this?
1248                 $preparedEdit->oldContent = null; // unused. // XXX: could get this from the parent revision
1249                 $preparedEdit->revid = $this->revision ? $this->revision->getId() : null;
1250                 $preparedEdit->timestamp = $preparedEdit->output->getCacheTime();
1251                 $preparedEdit->format = $preparedEdit->pstContent->getDefaultFormat();
1252
1253                 return $preparedEdit;
1254         }
1255
1256         /**
1257          * @param string $role
1258          * @param bool $generateHtml
1259          * @return ParserOutput
1260          */
1261         public function getSlotParserOutput( $role, $generateHtml = true ) {
1262                 return $this->getRenderedRevision()->getSlotParserOutput(
1263                         $role,
1264                         [ 'generate-html' => $generateHtml ]
1265                 );
1266         }
1267
1268         /**
1269          * @return ParserOutput
1270          */
1271         public function getCanonicalParserOutput() {
1272                 return $this->getRenderedRevision()->getRevisionParserOutput();
1273         }
1274
1275         /**
1276          * @return ParserOptions
1277          */
1278         public function getCanonicalParserOptions() {
1279                 return $this->getRenderedRevision()->getOptions();
1280         }
1281
1282         /**
1283          * @param bool $recursive
1284          *
1285          * @return DeferrableUpdate[]
1286          */
1287         public function getSecondaryDataUpdates( $recursive = false ) {
1288                 if ( $this->isContentDeleted() ) {
1289                         // This shouldn't happen, since the current content is always public,
1290                         // and DataUpates are only needed for current content.
1291                         return [];
1292                 }
1293
1294                 $output = $this->getCanonicalParserOutput();
1295
1296                 // Construct a LinksUpdate for the combined canonical output.
1297                 $linksUpdate = new LinksUpdate(
1298                         $this->getTitle(),
1299                         $output,
1300                         $recursive
1301                 );
1302
1303                 $allUpdates = [ $linksUpdate ];
1304
1305                 // NOTE: Run updates for all slots, not just the modified slots! Otherwise,
1306                 // info for an inherited slot may end up being removed. This is also needed
1307                 // to ensure that purges are effective.
1308                 $renderedRevision = $this->getRenderedRevision();
1309                 foreach ( $this->getSlots()->getSlotRoles() as $role ) {
1310                         $slot = $this->getRawSlot( $role );
1311                         $content = $slot->getContent();
1312                         $handler = $content->getContentHandler();
1313
1314                         $updates = $handler->getSecondaryDataUpdates(
1315                                 $this->getTitle(),
1316                                 $content,
1317                                 $role,
1318                                 $renderedRevision
1319                         );
1320                         $allUpdates = array_merge( $allUpdates, $updates );
1321
1322                         // TODO: remove B/C hack in 1.32!
1323                         // NOTE: we assume that the combined output contains all relevant meta-data for
1324                         // all slots!
1325                         $legacyUpdates = $content->getSecondaryDataUpdates(
1326                                 $this->getTitle(),
1327                                 null,
1328                                 $recursive,
1329                                 $output
1330                         );
1331
1332                         // HACK: filter out redundant and incomplete LinksUpdates
1333                         $legacyUpdates = array_filter( $legacyUpdates, function ( $update ) {
1334                                 return !( $update instanceof LinksUpdate );
1335                         } );
1336
1337                         $allUpdates = array_merge( $allUpdates, $legacyUpdates );
1338                 }
1339
1340                 // XXX: if a slot was removed by an earlier edit, but deletion updates failed to run at
1341                 // that time, we don't know for which slots to run deletion updates when purging a page.
1342                 // We'd have to examine the entire history of the page to determine that. Perhaps there
1343                 // could be a "try extra hard" mode for that case that would run a DB query to find all
1344                 // roles/models ever used on the page. On the other hand, removing slots should be quite
1345                 // rare, so perhaps this isn't worth the trouble.
1346
1347                 // TODO: consolidate with similar logic in WikiPage::getDeletionUpdates()
1348                 $wikiPage = $this->getWikiPage();
1349                 $parentRevision = $this->getParentRevision();
1350                 foreach ( $this->getRemovedSlotRoles() as $role ) {
1351                         // HACK: we should get the content model of the removed slot from a SlotRoleHandler!
1352                         // For now, find the slot in the parent revision - if the slot was removed, it should
1353                         // always exist in the parent revision.
1354                         $parentSlot = $parentRevision->getSlot( $role, RevisionRecord::RAW );
1355                         $content = $parentSlot->getContent();
1356                         $handler = $content->getContentHandler();
1357
1358                         $updates = $handler->getDeletionUpdates(
1359                                 $this->getTitle(),
1360                                 $role
1361                         );
1362                         $allUpdates = array_merge( $allUpdates, $updates );
1363
1364                         // TODO: remove B/C hack in 1.32!
1365                         $legacyUpdates = $content->getDeletionUpdates( $wikiPage );
1366
1367                         // HACK: filter out redundant and incomplete LinksDeletionUpdate
1368                         $legacyUpdates = array_filter( $legacyUpdates, function ( $update ) {
1369                                 return !( $update instanceof LinksDeletionUpdate );
1370                         } );
1371
1372                         $allUpdates = array_merge( $allUpdates, $legacyUpdates );
1373                 }
1374
1375                 // TODO: hard deprecate SecondaryDataUpdates in favor of RevisionDataUpdates in 1.33!
1376                 Hooks::run(
1377                         'RevisionDataUpdates',
1378                         [ $this->getTitle(), $renderedRevision, &$allUpdates ]
1379                 );
1380
1381                 return $allUpdates;
1382         }
1383
1384         /**
1385          * Do standard updates after page edit, purge, or import.
1386          * Update links tables, site stats, search index, title cache, message cache, etc.
1387          * Purges pages that depend on this page when appropriate.
1388          * With a 10% chance, triggers pruning the recent changes table.
1389          *
1390          * @note prepareUpdate() must be called before calling this method!
1391          *
1392          * MCR migration note: this replaces WikiPage::doEditUpdates.
1393          */
1394         public function doUpdates() {
1395                 $this->assertTransition( 'done' );
1396
1397                 // TODO: move logic into a PageEventEmitter service
1398
1399                 $wikiPage = $this->getWikiPage(); // TODO: use only for legacy hooks!
1400
1401                 $legacyUser = User::newFromIdentity( $this->user );
1402                 $legacyRevision = new Revision( $this->revision );
1403
1404                 $this->doParserCacheUpdate();
1405
1406                 $this->doSecondaryDataUpdates( [
1407                         // T52785 do not update any other pages on a null edit
1408                         'recursive' => $this->options['changed'],
1409                         'defer' => DeferredUpdates::POSTSEND,
1410                 ] );
1411
1412                 // TODO: MCR: check if *any* changed slot supports categories!
1413                 if ( $this->rcWatchCategoryMembership
1414                         && $this->getContentHandler( SlotRecord::MAIN )->supportsCategories() === true
1415                         && ( $this->options['changed'] || $this->options['created'] )
1416                         && !$this->options['restored']
1417                 ) {
1418                         // Note: jobs are pushed after deferred updates, so the job should be able to see
1419                         // the recent change entry (also done via deferred updates) and carry over any
1420                         // bot/deletion/IP flags, ect.
1421                         $this->jobQueueGroup->lazyPush(
1422                                 new CategoryMembershipChangeJob(
1423                                         $this->getTitle(),
1424                                         [
1425                                                 'pageId' => $this->getPageId(),
1426                                                 'revTimestamp' => $this->revision->getTimestamp(),
1427                                         ]
1428                                 )
1429                         );
1430                 }
1431
1432                 // TODO: replace legacy hook! Use a listener on PageEventEmitter instead!
1433                 $editInfo = $this->getPreparedEdit();
1434                 Hooks::run( 'ArticleEditUpdates', [ &$wikiPage, &$editInfo, $this->options['changed'] ] );
1435
1436                 // TODO: replace legacy hook! Use a listener on PageEventEmitter instead!
1437                 if ( Hooks::run( 'ArticleEditUpdatesDeleteFromRecentchanges', [ &$wikiPage ] ) ) {
1438                         // Flush old entries from the `recentchanges` table
1439                         if ( mt_rand( 0, 9 ) == 0 ) {
1440                                 $this->jobQueueGroup->lazyPush( RecentChangesUpdateJob::newPurgeJob() );
1441                         }
1442                 }
1443
1444                 $id = $this->getPageId();
1445                 $title = $this->getTitle();
1446                 $dbKey = $title->getPrefixedDBkey();
1447                 $shortTitle = $title->getDBkey();
1448
1449                 if ( !$title->exists() ) {
1450                         wfDebug( __METHOD__ . ": Page doesn't exist any more, bailing out\n" );
1451
1452                         $this->doTransition( 'done' );
1453                         return;
1454                 }
1455
1456                 if ( $this->options['oldcountable'] === 'no-change' ||
1457                         ( !$this->options['changed'] && !$this->options['moved'] )
1458                 ) {
1459                         $good = 0;
1460                 } elseif ( $this->options['created'] ) {
1461                         $good = (int)$this->isCountable();
1462                 } elseif ( $this->options['oldcountable'] !== null ) {
1463                         $good = (int)$this->isCountable()
1464                                 - (int)$this->options['oldcountable'];
1465                 } elseif ( isset( $this->pageState['oldCountable'] ) ) {
1466                         $good = (int)$this->isCountable()
1467                                 - (int)$this->pageState['oldCountable'];
1468                 } else {
1469                         $good = 0;
1470                 }
1471                 $edits = $this->options['changed'] ? 1 : 0;
1472                 $pages = $this->options['created'] ? 1 : 0;
1473
1474                 DeferredUpdates::addUpdate( SiteStatsUpdate::factory(
1475                         [ 'edits' => $edits, 'articles' => $good, 'pages' => $pages ]
1476                 ) );
1477
1478                 // TODO: make search infrastructure aware of slots!
1479                 $mainSlot = $this->revision->getSlot( SlotRecord::MAIN );
1480                 if ( !$mainSlot->isInherited() && !$this->isContentDeleted() ) {
1481                         DeferredUpdates::addUpdate( new SearchUpdate( $id, $dbKey, $mainSlot->getContent() ) );
1482                 }
1483
1484                 // If this is another user's talk page, update newtalk.
1485                 // Don't do this if $options['changed'] = false (null-edits) nor if
1486                 // it's a minor edit and the user making the edit doesn't generate notifications for those.
1487                 if ( $this->options['changed']
1488                         && $title->getNamespace() == NS_USER_TALK
1489                         && $shortTitle != $legacyUser->getTitleKey()
1490                         && !( $this->revision->isMinor() && $legacyUser->isAllowed( 'nominornewtalk' ) )
1491                 ) {
1492                         $recipient = User::newFromName( $shortTitle, false );
1493                         if ( !$recipient ) {
1494                                 wfDebug( __METHOD__ . ": invalid username\n" );
1495                         } else {
1496                                 // Allow extensions to prevent user notification
1497                                 // when a new message is added to their talk page
1498                                 // TODO: replace legacy hook!  Use a listener on PageEventEmitter instead!
1499                                 if ( Hooks::run( 'ArticleEditUpdateNewTalk', [ &$wikiPage, $recipient ] ) ) {
1500                                         if ( User::isIP( $shortTitle ) ) {
1501                                                 // An anonymous user
1502                                                 $recipient->setNewtalk( true, $legacyRevision );
1503                                         } elseif ( $recipient->isLoggedIn() ) {
1504                                                 $recipient->setNewtalk( true, $legacyRevision );
1505                                         } else {
1506                                                 wfDebug( __METHOD__ . ": don't need to notify a nonexistent user\n" );
1507                                         }
1508                                 }
1509                         }
1510                 }
1511
1512                 if ( $title->getNamespace() == NS_MEDIAWIKI
1513                         && $this->getRevisionSlotsUpdate()->isModifiedSlot( SlotRecord::MAIN )
1514                 ) {
1515                         $mainContent = $this->isContentDeleted() ? null : $this->getRawContent( SlotRecord::MAIN );
1516
1517                         $this->messageCache->updateMessageOverride( $title, $mainContent );
1518                 }
1519
1520                 // TODO: move onArticleCreate and onArticle into a PageEventEmitter service
1521                 if ( $this->options['created'] ) {
1522                         WikiPage::onArticleCreate( $title );
1523                 } elseif ( $this->options['changed'] ) { // T52785
1524                         WikiPage::onArticleEdit( $title, $legacyRevision, $this->getTouchedSlotRoles() );
1525                 }
1526
1527                 $oldRevision = $this->getParentRevision();
1528                 $oldLegacyRevision = $oldRevision ? new Revision( $oldRevision ) : null;
1529
1530                 // TODO: In the wiring, register a listener for this on the new PageEventEmitter
1531                 ResourceLoaderWikiModule::invalidateModuleCache(
1532                         $title, $oldLegacyRevision, $legacyRevision, $this->getWikiId() ?: wfWikiID()
1533                 );
1534
1535                 $this->doTransition( 'done' );
1536         }
1537
1538         /**
1539          * Do secondary data updates (such as updating link tables).
1540          *
1541          * MCR note: this method is temporarily exposed via WikiPage::doSecondaryDataUpdates.
1542          *
1543          * @param array $options
1544          *   - recursive: make the update recursive, i.e. also update pages which transclude the
1545          *     current page or otherwise depend on it (default: false)
1546          *   - defer: one of the DeferredUpdates constants, or false to run immediately after waiting
1547          *     for replication of the changes from the SecondaryDataUpdates hooks (default: false)
1548          *   - transactionTicket: a transaction ticket from LBFactory::getEmptyTransactionTicket(),
1549          *     only when defer is false (default: null)
1550          * @since 1.32
1551          */
1552         public function doSecondaryDataUpdates( array $options = [] ) {
1553                 $this->assertHasRevision( __METHOD__ );
1554                 $options += [
1555                         'recursive' => false,
1556                         'defer' => false,
1557                         'transactionTicket' => null,
1558                 ];
1559                 $deferValues = [ false, DeferredUpdates::PRESEND, DeferredUpdates::POSTSEND ];
1560                 if ( !in_array( $options['defer'], $deferValues, true ) ) {
1561                         throw new InvalidArgumentException( 'invalid value for defer: ' . $options['defer'] );
1562                 }
1563                 Assert::parameterType( 'integer|null', $options['transactionTicket'],
1564                         '$options[\'transactionTicket\']' );
1565
1566                 $updates = $this->getSecondaryDataUpdates( $options['recursive'] );
1567
1568                 $triggeringUser = $this->options['triggeringUser'] ?? $this->user;
1569                 if ( !$triggeringUser instanceof User ) {
1570                         $triggeringUser = User::newFromIdentity( $triggeringUser );
1571                 }
1572                 $causeAction = $this->options['causeAction'] ?? 'unknown';
1573                 $causeAgent = $this->options['causeAgent'] ?? 'unknown';
1574                 $legacyRevision = new Revision( $this->revision );
1575
1576                 if ( $options['defer'] === false && $options['transactionTicket'] !== null ) {
1577                         // For legacy hook handlers doing updates via LinksUpdateConstructed, make sure
1578                         // any pending writes they made get flushed before the doUpdate() calls below.
1579                         // This avoids snapshot-clearing errors in LinksUpdate::acquirePageLock().
1580                         $this->loadbalancerFactory->commitAndWaitForReplication(
1581                                 __METHOD__, $options['transactionTicket']
1582                         );
1583                 }
1584
1585                 foreach ( $updates as $update ) {
1586                         if ( $update instanceof DataUpdate ) {
1587                                 $update->setCause( $causeAction, $causeAgent );
1588                         }
1589                         if ( $update instanceof LinksUpdate ) {
1590                                 $update->setRevision( $legacyRevision );
1591                                 $update->setTriggeringUser( $triggeringUser );
1592                         }
1593                         if ( $options['defer'] === false ) {
1594                                 if ( $options['transactionTicket'] !== null ) {
1595                                         $update->setTransactionTicket( $options['transactionTicket'] );
1596                                 }
1597                                 $update->doUpdate();
1598                         } else {
1599                                 DeferredUpdates::addUpdate( $update, $options['defer'] );
1600                         }
1601                 }
1602         }
1603
1604         public function doParserCacheUpdate() {
1605                 $this->assertHasRevision( __METHOD__ );
1606
1607                 $wikiPage = $this->getWikiPage(); // TODO: ParserCache should accept a RevisionRecord instead
1608
1609                 // NOTE: this may trigger the first parsing of the new content after an edit (when not
1610                 // using pre-generated stashed output).
1611                 // XXX: we may want to use the PoolCounter here. This would perhaps allow the initial parse
1612                 // to be performed post-send. The client could already follow a HTTP redirect to the
1613                 // page view, but would then have to wait for a response until rendering is complete.
1614                 $output = $this->getCanonicalParserOutput();
1615
1616                 // Save it to the parser cache. Use the revision timestamp in the case of a
1617                 // freshly saved edit, as that matches page_touched and a mismatch would trigger an
1618                 // unnecessary reparse.
1619                 $timestamp = $this->options['changed'] ? $this->revision->getTimestamp()
1620                         : $output->getTimestamp();
1621                 $this->parserCache->save(
1622                         $output, $wikiPage, $this->getCanonicalParserOptions(),
1623                         $timestamp, $this->revision->getId()
1624                 );
1625         }
1626
1627 }