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 CategoryMembershipChangeJob;
  26 use Content;
  27 use ContentHandler;
  28 use DataUpdate;
  29 use DeferrableUpdate;
  30 use DeferredUpdates;
  31 use Hooks;
  32 use IDBAccessObject;
  33 use InvalidArgumentException;
  34 use JobQueueGroup;
  35 use Language;
  36 use LinksDeletionUpdate;
  37 use LinksUpdate;
  38 use LogicException;
  39 use MediaWiki\Edit\PreparedEdit;
  40 use MediaWiki\MediaWikiServices;
  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 MWCallableUpdate;
  52 use ParserCache;
  53 use ParserOptions;
  54 use ParserOutput;
  55 use Psr\Log\LoggerAwareInterface;
  56 use Psr\Log\LoggerInterface;
  57 use Psr\Log\NullLogger;
  58 use RecentChangesUpdateJob;
  59 use ResourceLoaderWikiModule;
  60 use Revision;
  61 use SearchUpdate;
  62 use SiteStatsUpdate;
  63 use Title;
  64 use User;
  65 use Wikimedia\Assert\Assert;
  66 use Wikimedia\Rdbms\ILBFactory;
  67 use WikiPage;
  68
  69 /**
  70  * A handle for managing updates for derived page data on edit, import, purge, etc.
  71  *
  72  * @note Avoid direct usage of DerivedPageDataUpdater.
  73  *
  74  * @todo Define interfaces for the different use cases of DerivedPageDataUpdater, particularly
  75  * providing access to post-PST content and ParserOutput to callbacks during revision creation,
  76  * which currently use WikiPage::prepareContentForEdit, and allowing updates to be triggered on
  77  * purge, import, and undeletion, which currently use WikiPage::doEditUpdates() and
  78  * Content::getSecondaryDataUpdates().
  79  *
  80  * DerivedPageDataUpdater instances are designed to be cached inside a WikiPage instance,
  81  * and re-used by callback code over the course of an update operation. It's a stepping stone
  82  * one the way to a more complete refactoring of WikiPage.
  83  *
  84  * When using a DerivedPageDataUpdater, the following life cycle must be observed:
  85  * grabCurrentRevision (optional), prepareContent (optional), prepareUpdate (required
  86  * for doUpdates). getCanonicalParserOutput, getSlots, and getSecondaryDataUpdates
  87  * require prepareContent or prepareUpdate to have been called first, to initialize the
  88  * DerivedPageDataUpdater.
  89  *
  90  * @see docs/pageupdater.txt for more information.
  91  *
  92  * MCR migration note: this replaces the relevant methods in WikiPage, and covers the use cases
  93  * of PreparedEdit.
  94  *
  95  * @internal
  96  *
  97  * @since 1.32
  98  * @ingroup Page
  99  */
 100 class DerivedPageDataUpdater implements IDBAccessObject, LoggerAwareInterface {
 101
 102         /**
 103          * @var UserIdentity|null
 104          */
 105         private $user = null;
 106
 107         /**
 108          * @var WikiPage
 109          */
 110         private $wikiPage;
 111
 112         /**
 113          * @var ParserCache
 114          */
 115         private $parserCache;
 116
 117         /**
 118          * @var RevisionStore
 119          */
 120         private $revisionStore;
 121
 122         /**
 123          * @var Language
 124          */
 125         private $contLang;
 126
 127         /**
 128          * @var JobQueueGroup
 129          */
 130         private $jobQueueGroup;
 131
 132         /**
 133          * @var MessageCache
 134          */
 135         private $messageCache;
 136
 137         /**
 138          * @var ILBFactory
 139          */
 140         private $loadbalancerFactory;
 141
 142         /**
 143          * @var LoggerInterface
 144          */
 145         private $logger;
 146
 147         /**
 148          * @var string see $wgArticleCountMethod
 149          */
 150         private $articleCountMethod;
 151
 152         /**
 153          * @var boolean see $wgRCWatchCategoryMembership
 154          */
 155         private $rcWatchCategoryMembership = false;
 156
 157         /**
 158          * Stores (most of) the $options parameter of prepareUpdate().
 159          * @see prepareUpdate()
 160          */
 161         private $options = [
 162                 'changed' => true,
 163                 // newrev is true if prepareUpdate is handling the creation of a new revision,
 164                 // as opposed to a null edit or a forced update.
 165                 'newrev' => false,
 166                 'created' => false,
 167                 'moved' => false,
 168                 'restored' => false,
 169                 'oldrevision' => null,
 170                 'oldcountable' => null,
 171                 'oldredirect' => null,
 172                 'triggeringUser' => null,
 173                 // causeAction/causeAgent default to 'unknown' but that's handled where it's read,
 174                 // to make the life of prepareUpdate() callers easier.
 175                 'causeAction' => null,
 176                 'causeAgent' => null,
 177         ];
 178
 179         /**
 180          * The state of the relevant row in page table before the edit.
 181          * This is determined by the first call to grabCurrentRevision, prepareContent,
 182          * or prepareUpdate (so it is only accessible in 'knows-current' or a later stage).
 183          * If pageState was not initialized when prepareUpdate() is called, prepareUpdate() will
 184          * attempt to emulate the state of the page table before the edit.
 185          *
 186          * Contains the following fields:
 187          * - oldRevision (RevisionRecord|null): the revision that was current before the change
 188          *   associated with this update. Might not be set, use getParentRevision().
 189          * - oldId (int|null): the id of the above revision. 0 if there is no such revision (the change
 190          *   was about creating a new page); null if not known (that should not happen).
 191          * - oldIsRedirect (bool|null): whether the page was a redirect before the change. Lazy-loaded,
 192          *   can be null; use wasRedirect() instead of direct access.
 193          * - oldCountable (bool|null): whether the page was countable before the change (or null
 194          *   if we don't have that information)
 195          *
 196          * @var array
 197          */
 198         private $pageState = null;
 199
 200         /**
 201          * @var RevisionSlotsUpdate|null
 202          */
 203         private $slotsUpdate = null;
 204
 205         /**
 206          * @var RevisionRecord|null
 207          */
 208         private $parentRevision = null;
 209
 210         /**
 211          * @var RevisionRecord|null
 212          */
 213         private $revision = null;
 214
 215         /**
 216          * @var RenderedRevision
 217          */
 218         private $renderedRevision = null;
 219
 220         /**
 221          * @var RevisionRenderer
 222          */
 223         private $revisionRenderer;
 224
 225         /** @var SlotRoleRegistry */
 226         private $slotRoleRegistry;
 227
 228         /**
 229          * A stage identifier for managing the life cycle of this instance.
 230          * Possible stages are 'new', 'knows-current', 'has-content', 'has-revision', and 'done'.
 231          *
 232          * @see docs/pageupdater.txt for documentation of the life cycle.
 233          *
 234          * @var string
 235          */
 236         private $stage = 'new';
 237
 238         /**
 239          * Transition table for managing the life cycle of DerivedPageDateUpdater instances.
 240          *
 241          * XXX: Overkill. This is a linear order, we could just count. Names are nice though,
 242          * and constants are also overkill...
 243          *
 244          * @see docs/pageupdater.txt for documentation of the life cycle.
 245          *
 246          * @var array[]
 247          */
 248         private static $transitions = [
 249                 'new' => [
 250                         'new' => true,
 251                         'knows-current' => true,
 252                         'has-content' => true,
 253                         'has-revision' => true,
 254                 ],
 255                 'knows-current' => [
 256                         'knows-current' => true,
 257                         'has-content' => true,
 258                         'has-revision' => true,
 259                 ],
 260                 'has-content' => [
 261                         'has-content' => true,
 262                         'has-revision' => true,
 263                 ],
 264                 'has-revision' => [
 265                         'has-revision' => true,
 266                         'done' => true,
 267                 ],
 268         ];
 269
 270         /**
 271          * @param WikiPage $wikiPage ,
 272          * @param RevisionStore $revisionStore
 273          * @param RevisionRenderer $revisionRenderer
 274          * @param SlotRoleRegistry $slotRoleRegistry
 275          * @param ParserCache $parserCache
 276          * @param JobQueueGroup $jobQueueGroup
 277          * @param MessageCache $messageCache
 278          * @param Language $contLang
 279          * @param ILBFactory $loadbalancerFactory
 280          */
 281         public function __construct(
 282                 WikiPage $wikiPage,
 283                 RevisionStore $revisionStore,
 284                 RevisionRenderer $revisionRenderer,
 285                 SlotRoleRegistry $slotRoleRegistry,
 286                 ParserCache $parserCache,
 287                 JobQueueGroup $jobQueueGroup,
 288                 MessageCache $messageCache,
 289                 Language $contLang,
 290                 ILBFactory $loadbalancerFactory
 291         ) {
 292                 $this->wikiPage = $wikiPage;
 293
 294                 $this->parserCache = $parserCache;
 295                 $this->revisionStore = $revisionStore;
 296                 $this->revisionRenderer = $revisionRenderer;
 297                 $this->slotRoleRegistry = $slotRoleRegistry;
 298                 $this->jobQueueGroup = $jobQueueGroup;
 299                 $this->messageCache = $messageCache;
 300                 $this->contLang = $contLang;
 301                 // XXX only needed for waiting for replicas to catch up; there should be a narrower
 302                 // interface for that.
 303                 $this->loadbalancerFactory = $loadbalancerFactory;
 304                 $this->logger = new NullLogger();
 305         }
 306
 307         public function setLogger( LoggerInterface $logger ) {
 308                 $this->logger = $logger;
 309         }
 310
 311         /**
 312          * Transition function for managing the life cycle of this instances.
 313          *
 314          * @see docs/pageupdater.txt for documentation of the life cycle.
 315          *
 316          * @param string $newStage the new stage
 317          * @return string the previous stage
 318          *
 319          * @throws LogicException If a transition to the given stage is not possible in the current
 320          *         stage.
 321          */
 322         private function doTransition( $newStage ) {
 323                 $this->assertTransition( $newStage );
 324
 325                 $oldStage = $this->stage;
 326                 $this->stage = $newStage;
 327
 328                 return $oldStage;
 329         }
 330
 331         /**
 332          * Asserts that a transition to the given stage is possible, without performing it.
 333          *
 334          * @see docs/pageupdater.txt for documentation of the life cycle.
 335          *
 336          * @param string $newStage the new stage
 337          *
 338          * @throws LogicException If this instance is not in the expected stage
 339          */
 340         private function assertTransition( $newStage ) {
 341                 if ( empty( self::$transitions[$this->stage][$newStage] ) ) {
 342                         throw new LogicException( "Cannot transition from {$this->stage} to $newStage" );
 343                 }
 344         }
 345
 346         /**
 347          * Checks whether this DerivedPageDataUpdater can be re-used for running updates targeting
 348          * the given revision.
 349          *
 350          * @param UserIdentity|null $user The user creating the revision in question
 351          * @param RevisionRecord|null $revision New revision (after save, if already saved)
 352          * @param RevisionSlotsUpdate|null $slotsUpdate New content (before PST)
 353          * @param null|int $parentId Parent revision of the edit (use 0 for page creation)
 354          *
 355          * @return bool
 356          */
 357         public function isReusableFor(
 358                 UserIdentity $user = null,
 359                 RevisionRecord $revision = null,
 360                 RevisionSlotsUpdate $slotsUpdate = null,
 361                 $parentId = null
 362         ) {
 363                 if ( $revision
 364                         && $parentId
 365                         && $revision->getParentId() !== $parentId
 366                 ) {
 367                         throw new InvalidArgumentException( '$parentId should match the parent of $revision' );
 368                 }
 369
 370                 // NOTE: For null revisions, $user may be different from $this->revision->getUser
 371                 // and also from $revision->getUser.
 372                 // But $user should always match $this->user.
 373                 if ( $user && $this->user && $user->getName() !== $this->user->getName() ) {
 374                         return false;
 375                 }
 376
 377                 if ( $revision && $this->revision && $this->revision->getId()
 378                         && $this->revision->getId() !== $revision->getId()
 379                 ) {
 380                         return false;
 381                 }
 382
 383                 if ( $this->pageState
 384                         && $revision
 385                         && $revision->getParentId() !== null
 386                         && $this->pageState['oldId'] !== $revision->getParentId()
 387                 ) {
 388                         return false;
 389                 }
 390
 391                 if ( $this->pageState
 392                         && $parentId !== null
 393                         && $this->pageState['oldId'] !== $parentId
 394                 ) {
 395                         return false;
 396                 }
 397
 398                 // NOTE: this check is the primary reason for having the $this->slotsUpdate field!
 399                 if ( $this->slotsUpdate
 400                         && $slotsUpdate
 401                         && !$this->slotsUpdate->hasSameUpdates( $slotsUpdate )
 402                 ) {
 403                         return false;
 404                 }
 405
 406                 if ( $revision
 407                         && $this->revision
 408                         && !$this->revision->getSlots()->hasSameContent( $revision->getSlots() )
 409                 ) {
 410                         return false;
 411                 }
 412
 413                 return true;
 414         }
 415
 416         /**
 417          * @param string $articleCountMethod "any" or "link".
 418          * @see $wgArticleCountMethod
 419          */
 420         public function setArticleCountMethod( $articleCountMethod ) {
 421                 $this->articleCountMethod = $articleCountMethod;
 422         }
 423
 424         /**
 425          * @param bool $rcWatchCategoryMembership
 426          * @see $wgRCWatchCategoryMembership
 427          */
 428         public function setRcWatchCategoryMembership( $rcWatchCategoryMembership ) {
 429                 $this->rcWatchCategoryMembership = $rcWatchCategoryMembership;
 430         }
 431
 432         /**
 433          * @return Title
 434          */
 435         private function getTitle() {
 436                 // NOTE: eventually, we won't get a WikiPage passed into the constructor any more
 437                 return $this->wikiPage->getTitle();
 438         }
 439
 440         /**
 441          * @return WikiPage
 442          */
 443         private function getWikiPage() {
 444                 // NOTE: eventually, we won't get a WikiPage passed into the constructor any more
 445                 return $this->wikiPage;
 446         }
 447
 448         /**
 449          * Determines whether the page being edited already existed.
 450          * Only defined after calling grabCurrentRevision() or prepareContent() or prepareUpdate()!
 451          *
 452          * @return bool
 453          * @throws LogicException if called before grabCurrentRevision
 454          */
 455         public function pageExisted() {
 456                 $this->assertHasPageState( __METHOD__ );
 457
 458                 return $this->pageState['oldId'] > 0;
 459         }
 460
 461         /**
 462          * Returns the parent revision of the new revision wrapped by this update.
 463          * If the update is a null-edit, this will return the parent of the current (and new) revision.
 464          * This will return null if the revision wrapped by this update created the page.
 465          * Only defined after calling prepareContent() or prepareUpdate()!
 466          *
 467          * @return RevisionRecord|null the parent revision of the new revision, or null if
 468          *         the update created the page.
 469          */
 470         private function getParentRevision() {
 471                 $this->assertPrepared( __METHOD__ );
 472
 473                 if ( $this->parentRevision ) {
 474                         return $this->parentRevision;
 475                 }
 476
 477                 if ( !$this->pageState['oldId'] ) {
 478                         // If there was no current revision, there is no parent revision,
 479                         // since the page didn't exist.
 480                         return null;
 481                 }
 482
 483                 $oldId = $this->revision->getParentId();
 484                 $flags = $this->useMaster() ? RevisionStore::READ_LATEST : 0;
 485                 $this->parentRevision = $oldId
 486                         ? $this->revisionStore->getRevisionById( $oldId, $flags )
 487                         : null;
 488
 489                 return $this->parentRevision;
 490         }
 491
 492         /**
 493          * Returns the revision that was the page's current revision when grabCurrentRevision()
 494          * was first called.
 495          *
 496          * During an edit, that revision will act as the logical parent of the new revision.
 497          *
 498          * Some updates are performed based on the difference between the database state at the
 499          * moment this method is first called, and the state after the edit.
 500          *
 501          * @see docs/pageupdater.txt for more information on when thie method can and should be called.
 502          *
 503          * @note After prepareUpdate() was called, grabCurrentRevision() will throw an exception
 504          * to avoid confusion, since the page's current revision is then the new revision after
 505          * the edit, which was presumably passed to prepareUpdate() as the $revision parameter.
 506          * Use getParentRevision() instead to access the revision that is the parent of the
 507          * new revision.
 508          *
 509          * @return RevisionRecord|null the page's current revision, or null if the page does not
 510          * yet exist.
 511          */
 512         public function grabCurrentRevision() {
 513                 if ( $this->pageState ) {
 514                         return $this->pageState['oldRevision'];
 515                 }
 516
 517                 $this->assertTransition( 'knows-current' );
 518
 519                 // NOTE: eventually, we won't get a WikiPage passed into the constructor any more
 520                 $wikiPage = $this->getWikiPage();
 521
 522                 // Do not call WikiPage::clear(), since the caller may already have caused page data
 523                 // to be loaded with SELECT FOR UPDATE. Just assert it's loaded now.
 524                 $wikiPage->loadPageData( self::READ_LATEST );
 525                 $rev = $wikiPage->getRevision();
 526                 $current = $rev ? $rev->getRevisionRecord() : null;
 527
 528                 $this->pageState = [
 529                         'oldRevision' => $current,
 530                         'oldId' => $rev ? $rev->getId() : 0,
 531                         'oldIsRedirect' => $wikiPage->isRedirect(), // NOTE: uses page table
 532                         'oldCountable' => $wikiPage->isCountable(), // NOTE: uses pagelinks table
 533                 ];
 534
 535                 $this->doTransition( 'knows-current' );
 536
 537                 return $this->pageState['oldRevision'];
 538         }
 539
 540         /**
 541          * Whether prepareUpdate() or prepareContent() have been called on this instance.
 542          *
 543          * @return bool
 544          */
 545         public function isContentPrepared() {
 546                 return $this->revision !== null;
 547         }
 548
 549         /**
 550          * Whether prepareUpdate() has been called on this instance.
 551          *
 552          * @note will also return null in case of a null-edit!
 553          *
 554          * @return bool
 555          */
 556         public function isUpdatePrepared() {
 557                 return $this->revision !== null && $this->revision->getId() !== null;
 558         }
 559
 560         /**
 561          * @return int
 562          */
 563         private function getPageId() {
 564                 // NOTE: eventually, we won't get a WikiPage passed into the constructor any more
 565                 return $this->wikiPage->getId();
 566         }
 567
 568         /**
 569          * Whether the content is deleted and thus not visible to the public.
 570          *
 571          * @return bool
 572          */
 573         public function isContentDeleted() {
 574                 if ( $this->revision ) {
 575                         // XXX: if that revision is the current revision, this should be skipped
 576                         return $this->revision->isDeleted( RevisionRecord::DELETED_TEXT );
 577                 } else {
 578                         // If the content has not been saved yet, it cannot have been deleted yet.
 579                         return false;
 580                 }
 581         }
 582
 583         /**
 584          * Returns the slot, modified or inherited, after PST, with no audience checks applied.
 585          *
 586          * @param string $role slot role name
 587          *
 588          * @throws PageUpdateException If the slot is neither set for update nor inherited from the
 589          *        parent revision.
 590          * @return SlotRecord
 591          */
 592         public function getRawSlot( $role ) {
 593                 return $this->getSlots()->getSlot( $role );
 594         }
 595
 596         /**
 597          * Returns the content of the given slot, with no audience checks.
 598          *
 599          * @throws PageUpdateException If the slot is neither set for update nor inherited from the
 600          *        parent revision.
 601          * @param string $role slot role name
 602          * @return Content
 603          */
 604         public function getRawContent( $role ) {
 605                 return $this->getRawSlot( $role )->getContent();
 606         }
 607
 608         /**
 609          * Returns the content model of the given slot
 610          *
 611          * @param string $role slot role name
 612          * @return string
 613          */
 614         private function getContentModel( $role ) {
 615                 return $this->getRawSlot( $role )->getModel();
 616         }
 617
 618         /**
 619          * @param string $role slot role name
 620          * @return ContentHandler
 621          */
 622         private function getContentHandler( $role ) {
 623                 // TODO: inject something like a ContentHandlerRegistry
 624                 return ContentHandler::getForModelID( $this->getContentModel( $role ) );
 625         }
 626
 627         private function useMaster() {
 628                 // TODO: can we just set a flag to true in prepareContent()?
 629                 return $this->wikiPage->wasLoadedFrom( self::READ_LATEST );
 630         }
 631
 632         /**
 633          * @return bool
 634          */
 635         public function isCountable() {
 636                 // NOTE: Keep in sync with WikiPage::isCountable.
 637
 638                 if ( !$this->getTitle()->isContentPage() ) {
 639                         return false;
 640                 }
 641
 642                 if ( $this->isContentDeleted() ) {
 643                         // This should be irrelevant: countability only applies to the current revision,
 644                         // and the current revision is never suppressed.
 645                         return false;
 646                 }
 647
 648                 if ( $this->isRedirect() ) {
 649                         return false;
 650                 }
 651
 652                 $hasLinks = null;
 653
 654                 if ( $this->articleCountMethod === 'link' ) {
 655                         // NOTE: it would be more appropriate to determine for each slot separately
 656                         // whether it has links, and use that information with that slot's
 657                         // isCountable() method. However, that would break parity with
 658                         // WikiPage::isCountable, which uses the pagelinks table to determine
 659                         // whether the current revision has links.
 660                         $hasLinks = (bool)count( $this->getCanonicalParserOutput()->getLinks() );
 661                 }
 662
 663                 foreach ( $this->getModifiedSlotRoles() as $role ) {
 664                         $roleHandler = $this->slotRoleRegistry->getRoleHandler( $role );
 665                         if ( $roleHandler->supportsArticleCount() ) {
 666                                 $content = $this->getRawContent( $role );
 667
 668                                 if ( $content->isCountable( $hasLinks ) ) {
 669                                         return true;
 670                                 }
 671                         }
 672                 }
 673
 674                 return false;
 675         }
 676
 677         /**
 678          * @return bool
 679          */
 680         public function isRedirect() {
 681                 // NOTE: main slot determines redirect status
 682                 // TODO: MCR: this should be controlled by a PageTypeHandler
 683                 $mainContent = $this->getRawContent( SlotRecord::MAIN );
 684
 685                 return $mainContent->isRedirect();
 686         }
 687
 688         /**
 689          * @param RevisionRecord $rev
 690          *
 691          * @return bool
 692          */
 693         private function revisionIsRedirect( RevisionRecord $rev ) {
 694                 // NOTE: main slot determines redirect status
 695                 $mainContent = $rev->getContent( SlotRecord::MAIN, RevisionRecord::RAW );
 696
 697                 return $mainContent->isRedirect();
 698         }
 699
 700         /**
 701          * Prepare updates based on an update which has not yet been saved.
 702          *
 703          * This may be used to create derived data that is needed when creating a new revision;
 704          * particularly, this makes available the slots of the new revision via the getSlots()
 705          * method, after applying PST and slot inheritance.
 706          *
 707          * The derived data prepared for revision creation may then later be re-used by doUpdates(),
 708          * without the need to re-calculate.
 709          *
 710          * @see docs/pageupdater.txt for more information on when thie method can and should be called.
 711          *
 712          * @note Calling this method more than once with the same $slotsUpdate
 713          * has no effect. Calling this method multiple times with different content will cause
 714          * an exception.
 715          *
 716          * @note Calling this method after prepareUpdate() has been called will cause an exception.
 717          *
 718          * @param User $user The user to act as context for pre-save transformation (PST).
 719          *        Type hint should be reduced to UserIdentity at some point.
 720          * @param RevisionSlotsUpdate $slotsUpdate The new content of the slots to be updated
 721          *        by this edit, before PST.
 722          * @param bool $useStash Whether to use stashed ParserOutput
 723          */
 724         public function prepareContent(
 725                 User $user,
 726                 RevisionSlotsUpdate $slotsUpdate,
 727                 $useStash = true
 728         ) {
 729                 if ( $this->slotsUpdate ) {
 730                         if ( !$this->user ) {
 731                                 throw new LogicException(
 732                                         'Unexpected state: $this->slotsUpdate was initialized, '
 733                                         . 'but $this->user was not.'
 734                                 );
 735                         }
 736
 737                         if ( $this->user->getName() !== $user->getName() ) {
 738                                 throw new LogicException( 'Can\'t call prepareContent() again for different user! '
 739                                         . 'Expected ' . $this->user->getName() . ', got ' . $user->getName()
 740                                 );
 741                         }
 742
 743                         if ( !$this->slotsUpdate->hasSameUpdates( $slotsUpdate ) ) {
 744                                 throw new LogicException(
 745                                         'Can\'t call prepareContent() again with different slot content!'
 746                                 );
 747                         }
 748
 749                         return; // prepareContent() already done, nothing to do
 750                 }
 751
 752                 $this->assertTransition( 'has-content' );
 753
 754                 $wikiPage = $this->getWikiPage(); // TODO: use only for legacy hooks!
 755                 $title = $this->getTitle();
 756
 757                 $parentRevision = $this->grabCurrentRevision();
 758
 759                 // The edit may have already been prepared via api.php?action=stashedit
 760                 $stashedEdit = false;
 761
 762                 // TODO: MCR: allow output for all slots to be stashed.
 763                 if ( $useStash && $slotsUpdate->isModifiedSlot( SlotRecord::MAIN ) ) {
 764                         $editStash = MediaWikiServices::getInstance()->getPageEditStash();
 765                         $stashedEdit = $editStash->checkCache(
 766                                 $title,
 767                                 $slotsUpdate->getModifiedSlot( SlotRecord::MAIN )->getContent(),
 768                                 User::newFromIdentity( $user )
 769                         );
 770                 }
 771
 772                 $userPopts = ParserOptions::newFromUserAndLang( $user, $this->contLang );
 773                 Hooks::run( 'ArticlePrepareTextForEdit', [ $wikiPage, $userPopts ] );
 774
 775                 $this->user = $user;
 776                 $this->slotsUpdate = $slotsUpdate;
 777
 778                 if ( $parentRevision ) {
 779                         $this->revision = MutableRevisionRecord::newFromParentRevision( $parentRevision );
 780                 } else {
 781                         $this->revision = new MutableRevisionRecord( $title );
 782                 }
 783
 784                 // NOTE: user and timestamp must be set, so they can be used for
 785                 // {{subst:REVISIONUSER}} and {{subst:REVISIONTIMESTAMP}} in PST!
 786                 $this->revision->setTimestamp( wfTimestampNow() );
 787                 $this->revision->setUser( $user );
 788
 789                 // Set up ParserOptions to operate on the new revision
 790                 $oldCallback = $userPopts->getCurrentRevisionCallback();
 791                 $userPopts->setCurrentRevisionCallback(
 792                         function ( Title $parserTitle, $parser = false ) use ( $title, $oldCallback ) {
 793                                 if ( $parserTitle->equals( $title ) ) {
 794                                         $legacyRevision = new Revision( $this->revision );
 795                                         return $legacyRevision;
 796                                 } else {
 797                                         return call_user_func( $oldCallback, $parserTitle, $parser );
 798                                 }
 799                         }
 800                 );
 801
 802                 $pstContentSlots = $this->revision->getSlots();
 803
 804                 foreach ( $slotsUpdate->getModifiedRoles() as $role ) {
 805                         $slot = $slotsUpdate->getModifiedSlot( $role );
 806
 807                         if ( $slot->isInherited() ) {
 808                                 // No PST for inherited slots! Note that "modified" slots may still be inherited
 809                                 // from an earlier version, e.g. for rollbacks.
 810                                 $pstSlot = $slot;
 811                         } elseif ( $role === SlotRecord::MAIN && $stashedEdit ) {
 812                                 // TODO: MCR: allow PST content for all slots to be stashed.
 813                                 $pstSlot = SlotRecord::newUnsaved( $role, $stashedEdit->pstContent );
 814                         } else {
 815                                 $content = $slot->getContent();
 816                                 $pstContent = $content->preSaveTransform( $title, $this->user, $userPopts );
 817                                 $pstSlot = SlotRecord::newUnsaved( $role, $pstContent );
 818                         }
 819
 820                         $pstContentSlots->setSlot( $pstSlot );
 821                 }
 822
 823                 foreach ( $slotsUpdate->getRemovedRoles() as $role ) {
 824                         $pstContentSlots->removeSlot( $role );
 825                 }
 826
 827                 $this->options['created'] = ( $parentRevision === null );
 828                 $this->options['changed'] = ( $parentRevision === null
 829                         || !$pstContentSlots->hasSameContent( $parentRevision->getSlots() ) );
 830
 831                 $this->doTransition( 'has-content' );
 832
 833                 if ( !$this->options['changed'] ) {
 834                         // null-edit!
 835
 836                         // TODO: move this into MutableRevisionRecord
 837                         // TODO: This needs to behave differently for a forced dummy edit!
 838                         $this->revision->setId( $parentRevision->getId() );
 839                         $this->revision->setTimestamp( $parentRevision->getTimestamp() );
 840                         $this->revision->setPageId( $parentRevision->getPageId() );
 841                         $this->revision->setParentId( $parentRevision->getParentId() );
 842                         $this->revision->setUser( $parentRevision->getUser( RevisionRecord::RAW ) );
 843                         $this->revision->setComment( $parentRevision->getComment( RevisionRecord::RAW ) );
 844                         $this->revision->setMinorEdit( $parentRevision->isMinor() );
 845                         $this->revision->setVisibility( $parentRevision->getVisibility() );
 846
 847                         // prepareUpdate() is redundant for null-edits
 848                         $this->doTransition( 'has-revision' );
 849                 } else {
 850                         $this->parentRevision = $parentRevision;
 851                 }
 852
 853                 $renderHints = [ 'use-master' => $this->useMaster(), 'audience' => RevisionRecord::RAW ];
 854
 855                 if ( $stashedEdit ) {
 856                         /** @var ParserOutput $output */
 857                         $output = $stashedEdit->output;
 858                         // TODO: this should happen when stashing the ParserOutput, not now!
 859                         $output->setCacheTime( $stashedEdit->timestamp );
 860
 861                         $renderHints['known-revision-output'] = $output;
 862
 863                         $this->logger->debug( __METHOD__ . ': using stashed edit output...' );
 864                 }
 865
 866                 // NOTE: we want a canonical rendering, so don't pass $this->user or ParserOptions
 867                 // NOTE: the revision is either new or current, so we can bypass audience checks.
 868                 $this->renderedRevision = $this->revisionRenderer->getRenderedRevision(
 869                         $this->revision,
 870                         null,
 871                         null,
 872                         $renderHints
 873                 );
 874         }
 875
 876         /**
 877          * Returns the update's target revision - that is, the revision that will be the current
 878          * revision after the update.
 879          *
 880          * @note Callers must treat the returned RevisionRecord's content as immutable, even
 881          * if it is a MutableRevisionRecord instance. Other aspects of a MutableRevisionRecord
 882          * returned from here, such as the user or the comment, may be changed, but may not
 883          * be reflected in ParserOutput until after prepareUpdate() has been called.
 884          *
 885          * @todo This is currently used by PageUpdater::makeNewRevision() to construct an unsaved
 886          * MutableRevisionRecord instance. Introduce something like an UnsavedRevisionFactory service
 887          * for that purpose instead!
 888          *
 889          * @return RevisionRecord
 890          */
 891         public function getRevision() {
 892                 $this->assertPrepared( __METHOD__ );
 893                 return $this->revision;
 894         }
 895
 896         /**
 897          * @return RenderedRevision
 898          */
 899         public function getRenderedRevision() {
 900                 $this->assertPrepared( __METHOD__ );
 901
 902                 return $this->renderedRevision;
 903         }
 904
 905         private function assertHasPageState( $method ) {
 906                 if ( !$this->pageState ) {
 907                         throw new LogicException(
 908                                 'Must call grabCurrentRevision() or prepareContent() '
 909                                 . 'or prepareUpdate() before calling ' . $method
 910                         );
 911                 }
 912         }
 913
 914         private function assertPrepared( $method ) {
 915                 if ( !$this->revision ) {
 916                         throw new LogicException(
 917                                 'Must call prepareContent() or prepareUpdate() before calling ' . $method
 918                         );
 919                 }
 920         }
 921
 922         private function assertHasRevision( $method ) {
 923                 if ( !$this->revision->getId() ) {
 924                         throw new LogicException(
 925                                 'Must call prepareUpdate() before calling ' . $method
 926                         );
 927                 }
 928         }
 929
 930         /**
 931          * Whether the edit creates the page.
 932          *
 933          * @return bool
 934          */
 935         public function isCreation() {
 936                 $this->assertPrepared( __METHOD__ );
 937                 return $this->options['created'];
 938         }
 939
 940         /**
 941          * Whether the edit created, or should create, a new revision (that is, it's not a null-edit).
 942          *
 943          * @warning at present, "null-revisions" that do not change content but do have a revision
 944          * record would return false after prepareContent(), but true after prepareUpdate()!
 945          * This should probably be fixed.
 946          *
 947          * @return bool
 948          */
 949         public function isChange() {
 950                 $this->assertPrepared( __METHOD__ );
 951                 return $this->options['changed'];
 952         }
 953
 954         /**
 955          * Whether the page was a redirect before the edit.
 956          *
 957          * @return bool
 958          */
 959         public function wasRedirect() {
 960                 $this->assertHasPageState( __METHOD__ );
 961
 962                 if ( $this->pageState['oldIsRedirect'] === null ) {
 963                         /** @var RevisionRecord $rev */
 964                         $rev = $this->pageState['oldRevision'];
 965                         if ( $rev ) {
 966                                 $this->pageState['oldIsRedirect'] = $this->revisionIsRedirect( $rev );
 967                         } else {
 968                                 $this->pageState['oldIsRedirect'] = false;
 969                         }
 970                 }
 971
 972                 return $this->pageState['oldIsRedirect'];
 973         }
 974
 975         /**
 976          * Returns the slots of the target revision, after PST.
 977          *
 978          * @note Callers must treat the returned RevisionSlots instance as immutable, even
 979          * if it is a MutableRevisionSlots instance.
 980          *
 981          * @return RevisionSlots
 982          */
 983         public function getSlots() {
 984                 $this->assertPrepared( __METHOD__ );
 985                 return $this->revision->getSlots();
 986         }
 987
 988         /**
 989          * Returns the RevisionSlotsUpdate for this updater.
 990          *
 991          * @return RevisionSlotsUpdate
 992          */
 993         private function getRevisionSlotsUpdate() {
 994                 $this->assertPrepared( __METHOD__ );
 995
 996                 if ( !$this->slotsUpdate ) {
 997                         $old = $this->getParentRevision();
 998                         $this->slotsUpdate = RevisionSlotsUpdate::newFromRevisionSlots(
 999                                 $this->revision->getSlots(),
1000                                 $old ? $old->getSlots() : null
1001                         );
1002                 }
1003                 return $this->slotsUpdate;
1004         }
1005
1006         /**
1007          * Returns the role names of the slots touched by the new revision,
1008          * including removed roles.
1009          *
1010          * @return string[]
1011          */
1012         public function getTouchedSlotRoles() {
1013                 return $this->getRevisionSlotsUpdate()->getTouchedRoles();
1014         }
1015
1016         /**
1017          * Returns the role names of the slots modified by the new revision,
1018          * not including removed roles.
1019          *
1020          * @return string[]
1021          */
1022         public function getModifiedSlotRoles() {
1023                 return $this->getRevisionSlotsUpdate()->getModifiedRoles();
1024         }
1025
1026         /**
1027          * Returns the role names of the slots removed by the new revision.
1028          *
1029          * @return string[]
1030          */
1031         public function getRemovedSlotRoles() {
1032                 return $this->getRevisionSlotsUpdate()->getRemovedRoles();
1033         }
1034
1035         /**
1036          * Prepare derived data updates targeting the given Revision.
1037          *
1038          * Calling this method requires the given revision to be present in the database.
1039          * This may be right after a new revision has been created, or when re-generating
1040          * derived data e.g. in ApiPurge, RefreshLinksJob, and the refreshLinks
1041          * script.
1042          *
1043          * @see docs/pageupdater.txt for more information on when thie method can and should be called.
1044          *
1045          * @note Calling this method more than once with the same revision has no effect.
1046          * $options are only used for the first call. Calling this method multiple times with
1047          * different revisions will cause an exception.
1048          *
1049          * @note If grabCurrentRevision() (or prepareContent()) has been called before
1050          * calling this method, $revision->getParentRevision() has to refer to the revision that
1051          * was the current revision at the time grabCurrentRevision() was called.
1052          *
1053          * @param RevisionRecord $revision
1054          * @param array $options Array of options, following indexes are used:
1055          * - changed: bool, whether the revision changed the content (default true)
1056          * - created: bool, whether the revision created the page (default false)
1057          * - moved: bool, whether the page was moved (default false)
1058          * - restored: bool, whether the page was undeleted (default false)
1059          * - oldrevision: Revision object for the pre-update revision (default null)
1060          * - triggeringUser: The user triggering the update (UserIdentity, defaults to the
1061          *   user who created the revision)
1062          * - oldredirect: bool, null, or string 'no-change' (default null):
1063          *    - bool: whether the page was counted as a redirect before that
1064          *      revision, only used in changed is true and created is false
1065          *    - null or 'no-change': don't update the redirect status.
1066          * - oldcountable: bool, null, or string 'no-change' (default null):
1067          *    - bool: whether the page was counted as an article before that
1068          *      revision, only used in changed is true and created is false
1069          *    - null: if created is false, don't update the article count; if created
1070          *      is true, do update the article count
1071          *    - 'no-change': don't update the article count, ever
1072          *    When set to null, pageState['oldCountable'] will be used instead if available.
1073          *  - causeAction: an arbitrary string identifying the reason for the update.
1074          *    See DataUpdate::getCauseAction(). (default 'unknown')
1075          *  - causeAgent: name of the user who caused the update. See DataUpdate::getCauseAgent().
1076          *    (string, default 'unknown')
1077          */
1078         public function prepareUpdate( RevisionRecord $revision, array $options = [] ) {
1079                 Assert::parameter(
1080                         !isset( $options['oldrevision'] )
1081                         || $options['oldrevision'] instanceof Revision
1082                         || $options['oldrevision'] instanceof RevisionRecord,
1083                         '$options["oldrevision"]',
1084                         'must be a RevisionRecord (or Revision)'
1085                 );
1086                 Assert::parameter(
1087                         !isset( $options['triggeringUser'] )
1088                         || $options['triggeringUser'] instanceof UserIdentity,
1089                         '$options["triggeringUser"]',
1090                         'must be a UserIdentity'
1091                 );
1092
1093                 if ( !$revision->getId() ) {
1094                         throw new InvalidArgumentException(
1095                                 'Revision must have an ID set for it to be used with prepareUpdate()!'
1096                         );
1097                 }
1098
1099                 if ( $this->revision && $this->revision->getId() ) {
1100                         if ( $this->revision->getId() === $revision->getId() ) {
1101                                 return; // nothing to do!
1102                         } else {
1103                                 throw new LogicException(
1104                                         'Trying to re-use DerivedPageDataUpdater with revision '
1105                                         . $revision->getId()
1106                                         . ', but it\'s already bound to revision '
1107                                         . $this->revision->getId()
1108                                 );
1109                         }
1110                 }
1111
1112                 if ( $this->revision
1113                         && !$this->revision->getSlots()->hasSameContent( $revision->getSlots() )
1114                 ) {
1115                         throw new LogicException(
1116                                 'The Revision provided has mismatching content!'
1117                         );
1118                 }
1119
1120                 // Override fields defined in $this->options with values from $options.
1121                 $this->options = array_intersect_key( $options, $this->options ) + $this->options;
1122
1123                 if ( $this->revision ) {
1124                         $oldId = $this->pageState['oldId'] ?? 0;
1125                         $this->options['newrev'] = ( $revision->getId() !== $oldId );
1126                 } elseif ( isset( $this->options['oldrevision'] ) ) {
1127                         /** @var Revision|RevisionRecord $oldRev */
1128                         $oldRev = $this->options['oldrevision'];
1129                         $oldId = $oldRev->getId();
1130                         $this->options['newrev'] = ( $revision->getId() !== $oldId );
1131                 } else {
1132                         $oldId = $revision->getParentId();
1133                 }
1134
1135                 if ( $oldId !== null ) {
1136                         // XXX: what if $options['changed'] disagrees?
1137                         // MovePage creates a dummy revision with changed = false!
1138                         // We may want to explicitly distinguish between "no new revision" (null-edit)
1139                         // and "new revision without new content" (dummy revision).
1140
1141                         if ( $oldId === $revision->getParentId() ) {
1142                                 // NOTE: this may still be a NullRevision!
1143                                 // New revision!
1144                                 $this->options['changed'] = true;
1145                         } elseif ( $oldId === $revision->getId() ) {
1146                                 // Null-edit!
1147                                 $this->options['changed'] = false;
1148                         } else {
1149                                 // This indicates that calling code has given us the wrong Revision object
1150                                 throw new LogicException(
1151                                         'The Revision mismatches old revision ID: '
1152                                         . 'Old ID is ' . $oldId
1153                                         . ', parent ID is ' . $revision->getParentId()
1154                                         . ', revision ID is ' . $revision->getId()
1155                                 );
1156                         }
1157                 }
1158
1159                 // If prepareContent() was used to generate the PST content (which is indicated by
1160                 // $this->slotsUpdate being set), and this is not a null-edit, then the given
1161                 // revision must have the acting user as the revision author. Otherwise, user
1162                 // signatures generated by PST would mismatch the user in the revision record.
1163                 if ( $this->user !== null && $this->options['changed'] && $this->slotsUpdate ) {
1164                         $user = $revision->getUser();
1165                         if ( !$this->user->equals( $user ) ) {
1166                                 throw new LogicException(
1167                                         'The Revision provided has a mismatching actor: expected '
1168                                         . $this->user->getName()
1169                                         . ', got '
1170                                         . $user->getName()
1171                                 );
1172                         }
1173                 }
1174
1175                 // If $this->pageState was not yet initialized by grabCurrentRevision or prepareContent,
1176                 // emulate the state of the page table before the edit, as good as we can.
1177                 if ( !$this->pageState ) {
1178                         $this->pageState = [
1179                                 'oldIsRedirect' => isset( $this->options['oldredirect'] )
1180                                         && is_bool( $this->options['oldredirect'] )
1181                                                 ? $this->options['oldredirect']
1182                                                 : null,
1183                                 'oldCountable' => isset( $this->options['oldcountable'] )
1184                                         && is_bool( $this->options['oldcountable'] )
1185                                                 ? $this->options['oldcountable']
1186                                                 : null,
1187                         ];
1188
1189                         if ( $this->options['changed'] ) {
1190                                 // The edit created a new revision
1191                                 $this->pageState['oldId'] = $revision->getParentId();
1192
1193                                 if ( isset( $this->options['oldrevision'] ) ) {
1194                                         $rev = $this->options['oldrevision'];
1195                                         $this->pageState['oldRevision'] = $rev instanceof Revision
1196                                                 ? $rev->getRevisionRecord()
1197                                                 : $rev;
1198                                 }
1199                         } else {
1200                                 // This is a null-edit, so the old revision IS the new revision!
1201                                 $this->pageState['oldId'] = $revision->getId();
1202                                 $this->pageState['oldRevision'] = $revision;
1203                         }
1204                 }
1205
1206                 // "created" is forced here
1207                 $this->options['created'] = ( $this->pageState['oldId'] === 0 );
1208
1209                 $this->revision = $revision;
1210
1211                 $this->doTransition( 'has-revision' );
1212
1213                 // NOTE: in case we have a User object, don't override with a UserIdentity.
1214                 // We already checked that $revision->getUser() mathces $this->user;
1215                 if ( !$this->user ) {
1216                         $this->user = $revision->getUser( RevisionRecord::RAW );
1217                 }
1218
1219                 // Prune any output that depends on the revision ID.
1220                 if ( $this->renderedRevision ) {
1221                         $this->renderedRevision->updateRevision( $revision );
1222                 } else {
1223
1224                         // NOTE: we want a canonical rendering, so don't pass $this->user or ParserOptions
1225                         // NOTE: the revision is either new or current, so we can bypass audience checks.
1226                         $this->renderedRevision = $this->revisionRenderer->getRenderedRevision(
1227                                 $this->revision,
1228                                 null,
1229                                 null,
1230                                 [ 'use-master' => $this->useMaster(), 'audience' => RevisionRecord::RAW ]
1231                         );
1232
1233                         // XXX: Since we presumably are dealing with the current revision,
1234                         // we could try to get the ParserOutput from the parser cache.
1235                 }
1236
1237                 // TODO: optionally get ParserOutput from the ParserCache here.
1238                 // Move the logic used by RefreshLinksJob here!
1239         }
1240
1241         /**
1242          * @deprecated This only exists for B/C, use the getters on DerivedPageDataUpdater directly!
1243          * @return PreparedEdit
1244          */
1245         public function getPreparedEdit() {
1246                 $this->assertPrepared( __METHOD__ );
1247
1248                 $slotsUpdate = $this->getRevisionSlotsUpdate();
1249                 $preparedEdit = new PreparedEdit();
1250
1251                 $preparedEdit->popts = $this->getCanonicalParserOptions();
1252                 $preparedEdit->parserOutputCallback = [ $this, 'getCanonicalParserOutput' ];
1253                 $preparedEdit->pstContent = $this->revision->getContent( SlotRecord::MAIN );
1254                 $preparedEdit->newContent =
1255                         $slotsUpdate->isModifiedSlot( SlotRecord::MAIN )
1256                         ? $slotsUpdate->getModifiedSlot( SlotRecord::MAIN )->getContent()
1257                         : $this->revision->getContent( SlotRecord::MAIN ); // XXX: can we just remove this?
1258                 $preparedEdit->oldContent = null; // unused. // XXX: could get this from the parent revision
1259                 $preparedEdit->revid = $this->revision ? $this->revision->getId() : null;
1260                 $preparedEdit->timestamp = $preparedEdit->output->getCacheTime();
1261                 $preparedEdit->format = $preparedEdit->pstContent->getDefaultFormat();
1262
1263                 return $preparedEdit;
1264         }
1265
1266         /**
1267          * @param string $role
1268          * @param bool $generateHtml
1269          * @return ParserOutput
1270          */
1271         public function getSlotParserOutput( $role, $generateHtml = true ) {
1272                 return $this->getRenderedRevision()->getSlotParserOutput(
1273                         $role,
1274                         [ 'generate-html' => $generateHtml ]
1275                 );
1276         }
1277
1278         /**
1279          * @return ParserOutput
1280          */
1281         public function getCanonicalParserOutput() {
1282                 return $this->getRenderedRevision()->getRevisionParserOutput();
1283         }
1284
1285         /**
1286          * @return ParserOptions
1287          */
1288         public function getCanonicalParserOptions() {
1289                 return $this->getRenderedRevision()->getOptions();
1290         }
1291
1292         /**
1293          * @param bool $recursive
1294          *
1295          * @return DeferrableUpdate[]
1296          */
1297         public function getSecondaryDataUpdates( $recursive = false ) {
1298                 if ( $this->isContentDeleted() ) {
1299                         // This shouldn't happen, since the current content is always public,
1300                         // and DataUpates are only needed for current content.
1301                         return [];
1302                 }
1303
1304                 $output = $this->getCanonicalParserOutput();
1305
1306                 // Construct a LinksUpdate for the combined canonical output.
1307                 $linksUpdate = new LinksUpdate(
1308                         $this->getTitle(),
1309                         $output,
1310                         $recursive
1311                 );
1312
1313                 $allUpdates = [ $linksUpdate ];
1314
1315                 // NOTE: Run updates for all slots, not just the modified slots! Otherwise,
1316                 // info for an inherited slot may end up being removed. This is also needed
1317                 // to ensure that purges are effective.
1318                 $renderedRevision = $this->getRenderedRevision();
1319                 foreach ( $this->getSlots()->getSlotRoles() as $role ) {
1320                         $slot = $this->getRawSlot( $role );
1321                         $content = $slot->getContent();
1322                         $handler = $content->getContentHandler();
1323
1324                         $updates = $handler->getSecondaryDataUpdates(
1325                                 $this->getTitle(),
1326                                 $content,
1327                                 $role,
1328                                 $renderedRevision
1329                         );
1330                         $allUpdates = array_merge( $allUpdates, $updates );
1331
1332                         // TODO: remove B/C hack in 1.32!
1333                         // NOTE: we assume that the combined output contains all relevant meta-data for
1334                         // all slots!
1335                         $legacyUpdates = $content->getSecondaryDataUpdates(
1336                                 $this->getTitle(),
1337                                 null,
1338                                 $recursive,
1339                                 $output
1340                         );
1341
1342                         // HACK: filter out redundant and incomplete LinksUpdates
1343                         $legacyUpdates = array_filter( $legacyUpdates, function ( $update ) {
1344                                 return !( $update instanceof LinksUpdate );
1345                         } );
1346
1347                         $allUpdates = array_merge( $allUpdates, $legacyUpdates );
1348                 }
1349
1350                 // XXX: if a slot was removed by an earlier edit, but deletion updates failed to run at
1351                 // that time, we don't know for which slots to run deletion updates when purging a page.
1352                 // We'd have to examine the entire history of the page to determine that. Perhaps there
1353                 // could be a "try extra hard" mode for that case that would run a DB query to find all
1354                 // roles/models ever used on the page. On the other hand, removing slots should be quite
1355                 // rare, so perhaps this isn't worth the trouble.
1356
1357                 // TODO: consolidate with similar logic in WikiPage::getDeletionUpdates()
1358                 $wikiPage = $this->getWikiPage();
1359                 $parentRevision = $this->getParentRevision();
1360                 foreach ( $this->getRemovedSlotRoles() as $role ) {
1361                         // HACK: we should get the content model of the removed slot from a SlotRoleHandler!
1362                         // For now, find the slot in the parent revision - if the slot was removed, it should
1363                         // always exist in the parent revision.
1364                         $parentSlot = $parentRevision->getSlot( $role, RevisionRecord::RAW );
1365                         $content = $parentSlot->getContent();
1366                         $handler = $content->getContentHandler();
1367
1368                         $updates = $handler->getDeletionUpdates(
1369                                 $this->getTitle(),
1370                                 $role
1371                         );
1372                         $allUpdates = array_merge( $allUpdates, $updates );
1373
1374                         // TODO: remove B/C hack in 1.32!
1375                         $legacyUpdates = $content->getDeletionUpdates( $wikiPage );
1376
1377                         // HACK: filter out redundant and incomplete LinksDeletionUpdate
1378                         $legacyUpdates = array_filter( $legacyUpdates, function ( $update ) {
1379                                 return !( $update instanceof LinksDeletionUpdate );
1380                         } );
1381
1382                         $allUpdates = array_merge( $allUpdates, $legacyUpdates );
1383                 }
1384
1385                 // TODO: hard deprecate SecondaryDataUpdates in favor of RevisionDataUpdates in 1.33!
1386                 Hooks::run(
1387                         'RevisionDataUpdates',
1388                         [ $this->getTitle(), $renderedRevision, &$allUpdates ]
1389                 );
1390
1391                 return $allUpdates;
1392         }
1393
1394         /**
1395          * Do standard updates after page edit, purge, or import.
1396          * Update links tables, site stats, search index, title cache, message cache, etc.
1397          * Purges pages that depend on this page when appropriate.
1398          * With a 10% chance, triggers pruning the recent changes table.
1399          *
1400          * @note prepareUpdate() must be called before calling this method!
1401          *
1402          * MCR migration note: this replaces WikiPage::doEditUpdates.
1403          */
1404         public function doUpdates() {
1405                 $this->assertTransition( 'done' );
1406
1407                 // TODO: move logic into a PageEventEmitter service
1408
1409                 $wikiPage = $this->getWikiPage(); // TODO: use only for legacy hooks!
1410
1411                 $legacyUser = User::newFromIdentity( $this->user );
1412                 $legacyRevision = new Revision( $this->revision );
1413
1414                 $userParserOptions = ParserOptions::newFromUser( $legacyUser );
1415                 // Decide whether to save the final canonical parser ouput based on the fact that
1416                 // users are typically redirected to viewing pages right after they edit those pages.
1417                 // Due to vary-revision-id, getting/saving that output here might require a reparse.
1418                 if ( $userParserOptions->matchesForCacheKey( $this->getCanonicalParserOptions() ) ) {
1419                         // Whether getting the final output requires a reparse or not, the user will
1420                         // need canonical output anyway, since that is what their parser options use.
1421                         // A reparse now at least has the benefit of various warm process caches.
1422                         $this->doParserCacheUpdate();
1423                 } else {
1424                         // If the user does not have canonical parse options, then don't risk another parse
1425                         // to make output they cannot use on the page refresh that typically occurs after
1426                         // editing. Doing the parser output save post-send will still benefit *other* users.
1427                         DeferredUpdates::addCallableUpdate( function () {
1428                                 $this->doParserCacheUpdate();
1429                         } );
1430                 }
1431
1432                 // Defer the getCannonicalParserOutput() call triggered by getSecondaryDataUpdates()
1433                 // by wrapping the code that schedules the secondary updates in a callback itself
1434                 $wrapperUpdate = new MWCallableUpdate(
1435                         function () {
1436                                 $this->doSecondaryDataUpdates( [
1437                                         // T52785 do not update any other pages on a null edit
1438                                         'recursive' => $this->options['changed']
1439                                 ] );
1440                         },
1441                         __METHOD__
1442                 );
1443                 $wrapperUpdate->setTransactionRoundRequirement( $wrapperUpdate::TRX_ROUND_ABSENT );
1444                 DeferredUpdates::addUpdate( $wrapperUpdate );
1445
1446                 // TODO: MCR: check if *any* changed slot supports categories!
1447                 if ( $this->rcWatchCategoryMembership
1448                         && $this->getContentHandler( SlotRecord::MAIN )->supportsCategories() === true
1449                         && ( $this->options['changed'] || $this->options['created'] )
1450                         && !$this->options['restored']
1451                 ) {
1452                         // Note: jobs are pushed after deferred updates, so the job should be able to see
1453                         // the recent change entry (also done via deferred updates) and carry over any
1454                         // bot/deletion/IP flags, ect.
1455                         $this->jobQueueGroup->lazyPush(
1456                                 CategoryMembershipChangeJob::newSpec(
1457                                         $this->getTitle(),
1458                                         $this->revision->getTimestamp()
1459                                 )
1460                         );
1461                 }
1462
1463                 // TODO: replace legacy hook! Use a listener on PageEventEmitter instead!
1464                 // @note: Extensions should *avoid* calling getCannonicalParserOutput() when using
1465                 // this hook whenever possible in order to avoid unnecessary additional parses.
1466                 $editInfo = $this->getPreparedEdit();
1467                 Hooks::run( 'ArticleEditUpdates',
1468                         [ &$wikiPage, &$editInfo, $this->options['changed'] ] );
1469
1470                 // TODO: replace legacy hook! Use a listener on PageEventEmitter instead!
1471                 if ( Hooks::run( 'ArticleEditUpdatesDeleteFromRecentchanges', [ &$wikiPage ] ) ) {
1472                         // Flush old entries from the `recentchanges` table
1473                         if ( mt_rand( 0, 9 ) == 0 ) {
1474                                 $this->jobQueueGroup->lazyPush( RecentChangesUpdateJob::newPurgeJob() );
1475                         }
1476                 }
1477
1478                 $id = $this->getPageId();
1479                 $title = $this->getTitle();
1480                 $dbKey = $title->getPrefixedDBkey();
1481                 $shortTitle = $title->getDBkey();
1482
1483                 if ( !$title->exists() ) {
1484                         wfDebug( __METHOD__ . ": Page doesn't exist any more, bailing out\n" );
1485
1486                         $this->doTransition( 'done' );
1487                         return;
1488                 }
1489
1490                 DeferredUpdates::addCallableUpdate( function () {
1491                         if (
1492                                 $this->options['oldcountable'] === 'no-change' ||
1493                                 ( !$this->options['changed'] && !$this->options['moved'] )
1494                         ) {
1495                                 $good = 0;
1496                         } elseif ( $this->options['created'] ) {
1497                                 $good = (int)$this->isCountable();
1498                         } elseif ( $this->options['oldcountable'] !== null ) {
1499                                 $good = (int)$this->isCountable()
1500                                         - (int)$this->options['oldcountable'];
1501                         } elseif ( isset( $this->pageState['oldCountable'] ) ) {
1502                                 $good = (int)$this->isCountable()
1503                                         - (int)$this->pageState['oldCountable'];
1504                         } else {
1505                                 $good = 0;
1506                         }
1507                         $edits = $this->options['changed'] ? 1 : 0;
1508                         $pages = $this->options['created'] ? 1 : 0;
1509
1510                         DeferredUpdates::addUpdate( SiteStatsUpdate::factory(
1511                                 [ 'edits' => $edits, 'articles' => $good, 'pages' => $pages ]
1512                         ) );
1513                 } );
1514
1515                 // TODO: make search infrastructure aware of slots!
1516                 $mainSlot = $this->revision->getSlot( SlotRecord::MAIN );
1517                 if ( !$mainSlot->isInherited() && !$this->isContentDeleted() ) {
1518                         DeferredUpdates::addUpdate( new SearchUpdate( $id, $dbKey, $mainSlot->getContent() ) );
1519                 }
1520
1521                 // If this is another user's talk page, update newtalk.
1522                 // Don't do this if $options['changed'] = false (null-edits) nor if
1523                 // it's a minor edit and the user making the edit doesn't generate notifications for those.
1524                 if ( $this->options['changed']
1525                         && $title->getNamespace() == NS_USER_TALK
1526                         && $shortTitle != $legacyUser->getTitleKey()
1527                         && !( $this->revision->isMinor() && $legacyUser->isAllowed( 'nominornewtalk' ) )
1528                 ) {
1529                         $recipient = User::newFromName( $shortTitle, false );
1530                         if ( !$recipient ) {
1531                                 wfDebug( __METHOD__ . ": invalid username\n" );
1532                         } else {
1533                                 // Allow extensions to prevent user notification
1534                                 // when a new message is added to their talk page
1535                                 // TODO: replace legacy hook!  Use a listener on PageEventEmitter instead!
1536                                 if ( Hooks::run( 'ArticleEditUpdateNewTalk', [ &$wikiPage, $recipient ] ) ) {
1537                                         if ( User::isIP( $shortTitle ) ) {
1538                                                 // An anonymous user
1539                                                 $recipient->setNewtalk( true, $legacyRevision );
1540                                         } elseif ( $recipient->isLoggedIn() ) {
1541                                                 $recipient->setNewtalk( true, $legacyRevision );
1542                                         } else {
1543                                                 wfDebug( __METHOD__ . ": don't need to notify a nonexistent user\n" );
1544                                         }
1545                                 }
1546                         }
1547                 }
1548
1549                 if ( $title->getNamespace() == NS_MEDIAWIKI
1550                         && $this->getRevisionSlotsUpdate()->isModifiedSlot( SlotRecord::MAIN )
1551                 ) {
1552                         $mainContent = $this->isContentDeleted() ? null : $this->getRawContent( SlotRecord::MAIN );
1553
1554                         $this->messageCache->updateMessageOverride( $title, $mainContent );
1555                 }
1556
1557                 // TODO: move onArticleCreate and onArticle into a PageEventEmitter service
1558                 if ( $this->options['created'] ) {
1559                         WikiPage::onArticleCreate( $title );
1560                 } elseif ( $this->options['changed'] ) { // T52785
1561                         WikiPage::onArticleEdit( $title, $legacyRevision, $this->getTouchedSlotRoles() );
1562                 }
1563
1564                 $oldRevision = $this->getParentRevision();
1565                 $oldLegacyRevision = $oldRevision ? new Revision( $oldRevision ) : null;
1566
1567                 // TODO: In the wiring, register a listener for this on the new PageEventEmitter
1568                 ResourceLoaderWikiModule::invalidateModuleCache(
1569                         $title,
1570                         $oldLegacyRevision,
1571                         $legacyRevision,
1572                         $this->loadbalancerFactory->getLocalDomainID()
1573                 );
1574
1575                 $this->doTransition( 'done' );
1576         }
1577
1578         /**
1579          * Do secondary data updates (such as updating link tables).
1580          *
1581          * MCR note: this method is temporarily exposed via WikiPage::doSecondaryDataUpdates.
1582          *
1583          * @param array $options
1584          *   - recursive: make the update recursive, i.e. also update pages which transclude the
1585          *     current page or otherwise depend on it (default: false)
1586          *   - defer: one of the DeferredUpdates constants, or false to run immediately after waiting
1587          *     for replication of the changes from the SecondaryDataUpdates hooks (default: false)
1588          * @since 1.32
1589          */
1590         public function doSecondaryDataUpdates( array $options = [] ) {
1591                 $this->assertHasRevision( __METHOD__ );
1592                 $options += [ 'recursive' => false, 'defer' => false ];
1593                 $deferValues = [ false, DeferredUpdates::PRESEND, DeferredUpdates::POSTSEND ];
1594                 if ( !in_array( $options['defer'], $deferValues, true ) ) {
1595                         throw new InvalidArgumentException( 'Invalid value for defer: ' . $options['defer'] );
1596                 }
1597                 $updates = $this->getSecondaryDataUpdates( $options['recursive'] );
1598
1599                 $triggeringUser = $this->options['triggeringUser'] ?? $this->user;
1600                 if ( !$triggeringUser instanceof User ) {
1601                         $triggeringUser = User::newFromIdentity( $triggeringUser );
1602                 }
1603                 $causeAction = $this->options['causeAction'] ?? 'unknown';
1604                 $causeAgent = $this->options['causeAgent'] ?? 'unknown';
1605                 $legacyRevision = new Revision( $this->revision );
1606
1607                 foreach ( $updates as $update ) {
1608                         if ( $update instanceof DataUpdate ) {
1609                                 $update->setCause( $causeAction, $causeAgent );
1610                         }
1611                         if ( $update instanceof LinksUpdate ) {
1612                                 $update->setRevision( $legacyRevision );
1613                                 $update->setTriggeringUser( $triggeringUser );
1614                         }
1615                 }
1616
1617                 if ( $options['defer'] === false ) {
1618                         // T221577: flush any transaction; each update needs outer transaction scope
1619                         $this->loadbalancerFactory->commitMasterChanges( __METHOD__ );
1620                         foreach ( $updates as $update ) {
1621                                 DeferredUpdates::attemptUpdate( $update, $this->loadbalancerFactory );
1622                         }
1623                 } else {
1624                         foreach ( $updates as $update ) {
1625                                 DeferredUpdates::addUpdate( $update, $options['defer'] );
1626                         }
1627                 }
1628         }
1629
1630         public function doParserCacheUpdate() {
1631                 $this->assertHasRevision( __METHOD__ );
1632
1633                 $wikiPage = $this->getWikiPage(); // TODO: ParserCache should accept a RevisionRecord instead
1634
1635                 // NOTE: this may trigger the first parsing of the new content after an edit (when not
1636                 // using pre-generated stashed output).
1637                 // XXX: we may want to use the PoolCounter here. This would perhaps allow the initial parse
1638                 // to be performed post-send. The client could already follow a HTTP redirect to the
1639                 // page view, but would then have to wait for a response until rendering is complete.
1640                 $output = $this->getCanonicalParserOutput();
1641
1642                 // Save it to the parser cache. Use the revision timestamp in the case of a
1643                 // freshly saved edit, as that matches page_touched and a mismatch would trigger an
1644                 // unnecessary reparse.
1645                 $timestamp = $this->options['newrev'] ? $this->revision->getTimestamp()
1646                         : $output->getCacheTime();
1647                 $this->parserCache->save(
1648                         $output, $wikiPage, $this->getCanonicalParserOptions(),
1649                         $timestamp, $this->revision->getId()
1650                 );
1651         }
1652
1653 }