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  * on 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          * @return bool|string
 348          */
 349         private function getWikiId() {
 350                 // TODO: get from RevisionStore
 351                 return false;
 352         }
 353
 354         /**
 355          * Checks whether this DerivedPageDataUpdater can be re-used for running updates targeting
 356          * the given revision.
 357          *
 358          * @param UserIdentity|null $user The user creating the revision in question
 359          * @param RevisionRecord|null $revision New revision (after save, if already saved)
 360          * @param RevisionSlotsUpdate|null $slotsUpdate New content (before PST)
 361          * @param null|int $parentId Parent revision of the edit (use 0 for page creation)
 362          *
 363          * @return bool
 364          */
 365         public function isReusableFor(
 366                 UserIdentity $user = null,
 367                 RevisionRecord $revision = null,
 368                 RevisionSlotsUpdate $slotsUpdate = null,
 369                 $parentId = null
 370         ) {
 371                 if ( $revision
 372                         && $parentId
 373                         && $revision->getParentId() !== $parentId
 374                 ) {
 375                         throw new InvalidArgumentException( '$parentId should match the parent of $revision' );
 376                 }
 377
 378                 // NOTE: For null revisions, $user may be different from $this->revision->getUser
 379                 // and also from $revision->getUser.
 380                 // But $user should always match $this->user.
 381                 if ( $user && $this->user && $user->getName() !== $this->user->getName() ) {
 382                         return false;
 383                 }
 384
 385                 if ( $revision && $this->revision && $this->revision->getId()
 386                         && $this->revision->getId() !== $revision->getId()
 387                 ) {
 388                         return false;
 389                 }
 390
 391                 if ( $this->pageState
 392                         && $revision
 393                         && $revision->getParentId() !== null
 394                         && $this->pageState['oldId'] !== $revision->getParentId()
 395                 ) {
 396                         return false;
 397                 }
 398
 399                 if ( $this->pageState
 400                         && $parentId !== null
 401                         && $this->pageState['oldId'] !== $parentId
 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                         return $this->revision->isDeleted( RevisionRecord::DELETED_TEXT );
 584                 } else {
 585                         // If the content has not been saved yet, it cannot have been deleted yet.
 586                         return false;
 587                 }
 588         }
 589
 590         /**
 591          * Returns the slot, modified or inherited, after PST, with no audience checks applied.
 592          *
 593          * @param string $role slot role name
 594          *
 595          * @throws PageUpdateException If the slot is neither set for update nor inherited from the
 596          *        parent revision.
 597          * @return SlotRecord
 598          */
 599         public function getRawSlot( $role ) {
 600                 return $this->getSlots()->getSlot( $role );
 601         }
 602
 603         /**
 604          * Returns the content of the given slot, with no audience checks.
 605          *
 606          * @throws PageUpdateException If the slot is neither set for update nor inherited from the
 607          *        parent revision.
 608          * @param string $role slot role name
 609          * @return Content
 610          */
 611         public function getRawContent( $role ) {
 612                 return $this->getRawSlot( $role )->getContent();
 613         }
 614
 615         /**
 616          * Returns the content model of the given slot
 617          *
 618          * @param string $role slot role name
 619          * @return string
 620          */
 621         private function getContentModel( $role ) {
 622                 return $this->getRawSlot( $role )->getModel();
 623         }
 624
 625         /**
 626          * @param string $role slot role name
 627          * @return ContentHandler
 628          */
 629         private function getContentHandler( $role ) {
 630                 // TODO: inject something like a ContentHandlerRegistry
 631                 return ContentHandler::getForModelID( $this->getContentModel( $role ) );
 632         }
 633
 634         private function useMaster() {
 635                 // TODO: can we just set a flag to true in prepareContent()?
 636                 return $this->wikiPage->wasLoadedFrom( self::READ_LATEST );
 637         }
 638
 639         /**
 640          * @return bool
 641          */
 642         public function isCountable() {
 643                 // NOTE: Keep in sync with WikiPage::isCountable.
 644
 645                 if ( !$this->getTitle()->isContentPage() ) {
 646                         return false;
 647                 }
 648
 649                 if ( $this->isContentDeleted() ) {
 650                         // This should be irrelevant: countability only applies to the current revision,
 651                         // and the current revision is never suppressed.
 652                         return false;
 653                 }
 654
 655                 if ( $this->isRedirect() ) {
 656                         return false;
 657                 }
 658
 659                 $hasLinks = null;
 660
 661                 if ( $this->articleCountMethod === 'link' ) {
 662                         // NOTE: it would be more appropriate to determine for each slot separately
 663                         // whether it has links, and use that information with that slot's
 664                         // isCountable() method. However, that would break parity with
 665                         // WikiPage::isCountable, which uses the pagelinks table to determine
 666                         // whether the current revision has links.
 667                         $hasLinks = (bool)count( $this->getCanonicalParserOutput()->getLinks() );
 668                 }
 669
 670                 foreach ( $this->getModifiedSlotRoles() as $role ) {
 671                         $roleHandler = $this->slotRoleRegistry->getRoleHandler( $role );
 672                         if ( $roleHandler->supportsArticleCount() ) {
 673                                 $content = $this->getRawContent( $role );
 674
 675                                 if ( $content->isCountable( $hasLinks ) ) {
 676                                         return true;
 677                                 }
 678                         }
 679                 }
 680
 681                 return false;
 682         }
 683
 684         /**
 685          * @return bool
 686          */
 687         public function isRedirect() {
 688                 // NOTE: main slot determines redirect status
 689                 // TODO: MCR: this should be controlled by a PageTypeHandler
 690                 $mainContent = $this->getRawContent( SlotRecord::MAIN );
 691
 692                 return $mainContent->isRedirect();
 693         }
 694
 695         /**
 696          * @param RevisionRecord $rev
 697          *
 698          * @return bool
 699          */
 700         private function revisionIsRedirect( RevisionRecord $rev ) {
 701                 // NOTE: main slot determines redirect status
 702                 $mainContent = $rev->getContent( SlotRecord::MAIN, RevisionRecord::RAW );
 703
 704                 return $mainContent->isRedirect();
 705         }
 706
 707         /**
 708          * Prepare updates based on an update which has not yet been saved.
 709          *
 710          * This may be used to create derived data that is needed when creating a new revision;
 711          * particularly, this makes available the slots of the new revision via the getSlots()
 712          * method, after applying PST and slot inheritance.
 713          *
 714          * The derived data prepared for revision creation may then later be re-used by doUpdates(),
 715          * without the need to re-calculate.
 716          *
 717          * @see docs/pageupdater.txt for more information on when thie method can and should be called.
 718          *
 719          * @note Calling this method more than once with the same $slotsUpdate
 720          * has no effect. Calling this method multiple times with different content will cause
 721          * an exception.
 722          *
 723          * @note Calling this method after prepareUpdate() has been called will cause an exception.
 724          *
 725          * @param User $user The user to act as context for pre-save transformation (PST).
 726          *        Type hint should be reduced to UserIdentity at some point.
 727          * @param RevisionSlotsUpdate $slotsUpdate The new content of the slots to be updated
 728          *        by this edit, before PST.
 729          * @param bool $useStash Whether to use stashed ParserOutput
 730          */
 731         public function prepareContent(
 732                 User $user,
 733                 RevisionSlotsUpdate $slotsUpdate,
 734                 $useStash = true
 735         ) {
 736                 if ( $this->slotsUpdate ) {
 737                         if ( !$this->user ) {
 738                                 throw new LogicException(
 739                                         'Unexpected state: $this->slotsUpdate was initialized, '
 740                                         . 'but $this->user was not.'
 741                                 );
 742                         }
 743
 744                         if ( $this->user->getName() !== $user->getName() ) {
 745                                 throw new LogicException( 'Can\'t call prepareContent() again for different user! '
 746                                         . 'Expected ' . $this->user->getName() . ', got ' . $user->getName()
 747                                 );
 748                         }
 749
 750                         if ( !$this->slotsUpdate->hasSameUpdates( $slotsUpdate ) ) {
 751                                 throw new LogicException(
 752                                         'Can\'t call prepareContent() again with different slot content!'
 753                                 );
 754                         }
 755
 756                         return; // prepareContent() already done, nothing to do
 757                 }
 758
 759                 $this->assertTransition( 'has-content' );
 760
 761                 $wikiPage = $this->getWikiPage(); // TODO: use only for legacy hooks!
 762                 $title = $this->getTitle();
 763
 764                 $parentRevision = $this->grabCurrentRevision();
 765
 766                 // The edit may have already been prepared via api.php?action=stashedit
 767                 $stashedEdit = false;
 768
 769                 // TODO: MCR: allow output for all slots to be stashed.
 770                 if ( $useStash && $slotsUpdate->isModifiedSlot( SlotRecord::MAIN ) ) {
 771                         $editStash = MediaWikiServices::getInstance()->getPageEditStash();
 772                         $stashedEdit = $editStash->checkCache(
 773                                 $title,
 774                                 $slotsUpdate->getModifiedSlot( SlotRecord::MAIN )->getContent(),
 775                                 User::newFromIdentity( $user )
 776                         );
 777                 }
 778
 779                 $userPopts = ParserOptions::newFromUserAndLang( $user, $this->contLang );
 780                 Hooks::run( 'ArticlePrepareTextForEdit', [ $wikiPage, $userPopts ] );
 781
 782                 $this->user = $user;
 783                 $this->slotsUpdate = $slotsUpdate;
 784
 785                 if ( $parentRevision ) {
 786                         $this->revision = MutableRevisionRecord::newFromParentRevision( $parentRevision );
 787                 } else {
 788                         $this->revision = new MutableRevisionRecord( $title );
 789                 }
 790
 791                 // NOTE: user and timestamp must be set, so they can be used for
 792                 // {{subst:REVISIONUSER}} and {{subst:REVISIONTIMESTAMP}} in PST!
 793                 $this->revision->setTimestamp( wfTimestampNow() );
 794                 $this->revision->setUser( $user );
 795
 796                 // Set up ParserOptions to operate on the new revision
 797                 $oldCallback = $userPopts->getCurrentRevisionCallback();
 798                 $userPopts->setCurrentRevisionCallback(
 799                         function ( Title $parserTitle, $parser = false ) use ( $title, $oldCallback ) {
 800                                 if ( $parserTitle->equals( $title ) ) {
 801                                         $legacyRevision = new Revision( $this->revision );
 802                                         return $legacyRevision;
 803                                 } else {
 804                                         return call_user_func( $oldCallback, $parserTitle, $parser );
 805                                 }
 806                         }
 807                 );
 808
 809                 $pstContentSlots = $this->revision->getSlots();
 810
 811                 foreach ( $slotsUpdate->getModifiedRoles() as $role ) {
 812                         $slot = $slotsUpdate->getModifiedSlot( $role );
 813
 814                         if ( $slot->isInherited() ) {
 815                                 // No PST for inherited slots! Note that "modified" slots may still be inherited
 816                                 // from an earlier version, e.g. for rollbacks.
 817                                 $pstSlot = $slot;
 818                         } elseif ( $role === SlotRecord::MAIN && $stashedEdit ) {
 819                                 // TODO: MCR: allow PST content for all slots to be stashed.
 820                                 $pstSlot = SlotRecord::newUnsaved( $role, $stashedEdit->pstContent );
 821                         } else {
 822                                 $content = $slot->getContent();
 823                                 $pstContent = $content->preSaveTransform( $title, $this->user, $userPopts );
 824                                 $pstSlot = SlotRecord::newUnsaved( $role, $pstContent );
 825                         }
 826
 827                         $pstContentSlots->setSlot( $pstSlot );
 828                 }
 829
 830                 foreach ( $slotsUpdate->getRemovedRoles() as $role ) {
 831                         $pstContentSlots->removeSlot( $role );
 832                 }
 833
 834                 $this->options['created'] = ( $parentRevision === null );
 835                 $this->options['changed'] = ( $parentRevision === null
 836                         || !$pstContentSlots->hasSameContent( $parentRevision->getSlots() ) );
 837
 838                 $this->doTransition( 'has-content' );
 839
 840                 if ( !$this->options['changed'] ) {
 841                         // null-edit!
 842
 843                         // TODO: move this into MutableRevisionRecord
 844                         // TODO: This needs to behave differently for a forced dummy edit!
 845                         $this->revision->setId( $parentRevision->getId() );
 846                         $this->revision->setTimestamp( $parentRevision->getTimestamp() );
 847                         $this->revision->setPageId( $parentRevision->getPageId() );
 848                         $this->revision->setParentId( $parentRevision->getParentId() );
 849                         $this->revision->setUser( $parentRevision->getUser( RevisionRecord::RAW ) );
 850                         $this->revision->setComment( $parentRevision->getComment( RevisionRecord::RAW ) );
 851                         $this->revision->setMinorEdit( $parentRevision->isMinor() );
 852                         $this->revision->setVisibility( $parentRevision->getVisibility() );
 853
 854                         // prepareUpdate() is redundant for null-edits
 855                         $this->doTransition( 'has-revision' );
 856                 } else {
 857                         $this->parentRevision = $parentRevision;
 858                 }
 859
 860                 $renderHints = [ 'use-master' => $this->useMaster(), 'audience' => RevisionRecord::RAW ];
 861
 862                 if ( $stashedEdit ) {
 863                         /** @var ParserOutput $output */
 864                         $output = $stashedEdit->output;
 865                         // TODO: this should happen when stashing the ParserOutput, not now!
 866                         $output->setCacheTime( $stashedEdit->timestamp );
 867
 868                         $renderHints['known-revision-output'] = $output;
 869
 870                         $this->logger->debug( __METHOD__ . ': using stashed edit output...' );
 871                 }
 872
 873                 // NOTE: we want a canonical rendering, so don't pass $this->user or ParserOptions
 874                 // NOTE: the revision is either new or current, so we can bypass audience checks.
 875                 $this->renderedRevision = $this->revisionRenderer->getRenderedRevision(
 876                         $this->revision,
 877                         null,
 878                         null,
 879                         $renderHints
 880                 );
 881         }
 882
 883         /**
 884          * Returns the update's target revision - that is, the revision that will be the current
 885          * revision after the update.
 886          *
 887          * @note Callers must treat the returned RevisionRecord's content as immutable, even
 888          * if it is a MutableRevisionRecord instance. Other aspects of a MutableRevisionRecord
 889          * returned from here, such as the user or the comment, may be changed, but may not
 890          * be reflected in ParserOutput until after prepareUpdate() has been called.
 891          *
 892          * @todo This is currently used by PageUpdater::makeNewRevision() to construct an unsaved
 893          * MutableRevisionRecord instance. Introduce something like an UnsavedRevisionFactory service
 894          * for that purpose instead!
 895          *
 896          * @return RevisionRecord
 897          */
 898         public function getRevision() {
 899                 $this->assertPrepared( __METHOD__ );
 900                 return $this->revision;
 901         }
 902
 903         /**
 904          * @return RenderedRevision
 905          */
 906         public function getRenderedRevision() {
 907                 $this->assertPrepared( __METHOD__ );
 908
 909                 return $this->renderedRevision;
 910         }
 911
 912         private function assertHasPageState( $method ) {
 913                 if ( !$this->pageState ) {
 914                         throw new LogicException(
 915                                 'Must call grabCurrentRevision() or prepareContent() '
 916                                 . 'or prepareUpdate() before calling ' . $method
 917                         );
 918                 }
 919         }
 920
 921         private function assertPrepared( $method ) {
 922                 if ( !$this->revision ) {
 923                         throw new LogicException(
 924                                 'Must call prepareContent() or prepareUpdate() before calling ' . $method
 925                         );
 926                 }
 927         }
 928
 929         private function assertHasRevision( $method ) {
 930                 if ( !$this->revision->getId() ) {
 931                         throw new LogicException(
 932                                 'Must call prepareUpdate() before calling ' . $method
 933                         );
 934                 }
 935         }
 936
 937         /**
 938          * Whether the edit creates the page.
 939          *
 940          * @return bool
 941          */
 942         public function isCreation() {
 943                 $this->assertPrepared( __METHOD__ );
 944                 return $this->options['created'];
 945         }
 946
 947         /**
 948          * Whether the edit created, or should create, a new revision (that is, it's not a null-edit).
 949          *
 950          * @warning at present, "null-revisions" that do not change content but do have a revision
 951          * record would return false after prepareContent(), but true after prepareUpdate()!
 952          * This should probably be fixed.
 953          *
 954          * @return bool
 955          */
 956         public function isChange() {
 957                 $this->assertPrepared( __METHOD__ );
 958                 return $this->options['changed'];
 959         }
 960
 961         /**
 962          * Whether the page was a redirect before the edit.
 963          *
 964          * @return bool
 965          */
 966         public function wasRedirect() {
 967                 $this->assertHasPageState( __METHOD__ );
 968
 969                 if ( $this->pageState['oldIsRedirect'] === null ) {
 970                         /** @var RevisionRecord $rev */
 971                         $rev = $this->pageState['oldRevision'];
 972                         if ( $rev ) {
 973                                 $this->pageState['oldIsRedirect'] = $this->revisionIsRedirect( $rev );
 974                         } else {
 975                                 $this->pageState['oldIsRedirect'] = false;
 976                         }
 977                 }
 978
 979                 return $this->pageState['oldIsRedirect'];
 980         }
 981
 982         /**
 983          * Returns the slots of the target revision, after PST.
 984          *
 985          * @note Callers must treat the returned RevisionSlots instance as immutable, even
 986          * if it is a MutableRevisionSlots instance.
 987          *
 988          * @return RevisionSlots
 989          */
 990         public function getSlots() {
 991                 $this->assertPrepared( __METHOD__ );
 992                 return $this->revision->getSlots();
 993         }
 994
 995         /**
 996          * Returns the RevisionSlotsUpdate for this updater.
 997          *
 998          * @return RevisionSlotsUpdate
 999          */
1000         private function getRevisionSlotsUpdate() {
1001                 $this->assertPrepared( __METHOD__ );
1002
1003                 if ( !$this->slotsUpdate ) {
1004                         $old = $this->getParentRevision();
1005                         $this->slotsUpdate = RevisionSlotsUpdate::newFromRevisionSlots(
1006                                 $this->revision->getSlots(),
1007                                 $old ? $old->getSlots() : null
1008                         );
1009                 }
1010                 return $this->slotsUpdate;
1011         }
1012
1013         /**
1014          * Returns the role names of the slots touched by the new revision,
1015          * including removed roles.
1016          *
1017          * @return string[]
1018          */
1019         public function getTouchedSlotRoles() {
1020                 return $this->getRevisionSlotsUpdate()->getTouchedRoles();
1021         }
1022
1023         /**
1024          * Returns the role names of the slots modified by the new revision,
1025          * not including removed roles.
1026          *
1027          * @return string[]
1028          */
1029         public function getModifiedSlotRoles() {
1030                 return $this->getRevisionSlotsUpdate()->getModifiedRoles();
1031         }
1032
1033         /**
1034          * Returns the role names of the slots removed by the new revision.
1035          *
1036          * @return string[]
1037          */
1038         public function getRemovedSlotRoles() {
1039                 return $this->getRevisionSlotsUpdate()->getRemovedRoles();
1040         }
1041
1042         /**
1043          * Prepare derived data updates targeting the given Revision.
1044          *
1045          * Calling this method requires the given revision to be present in the database.
1046          * This may be right after a new revision has been created, or when re-generating
1047          * derived data e.g. in ApiPurge, RefreshLinksJob, and the refreshLinks
1048          * script.
1049          *
1050          * @see docs/pageupdater.txt for more information on when thie method can and should be called.
1051          *
1052          * @note Calling this method more than once with the same revision has no effect.
1053          * $options are only used for the first call. Calling this method multiple times with
1054          * different revisions will cause an exception.
1055          *
1056          * @note If grabCurrentRevision() (or prepareContent()) has been called before
1057          * calling this method, $revision->getParentRevision() has to refer to the revision that
1058          * was the current revision at the time grabCurrentRevision() was called.
1059          *
1060          * @param RevisionRecord $revision
1061          * @param array $options Array of options, following indexes are used:
1062          * - changed: bool, whether the revision changed the content (default true)
1063          * - created: bool, whether the revision created the page (default false)
1064          * - moved: bool, whether the page was moved (default false)
1065          * - restored: bool, whether the page was undeleted (default false)
1066          * - oldrevision: Revision object for the pre-update revision (default null)
1067          * - triggeringUser: The user triggering the update (UserIdentity, defaults to the
1068          *   user who created the revision)
1069          * - oldredirect: bool, null, or string 'no-change' (default null):
1070          *    - bool: whether the page was counted as a redirect before that
1071          *      revision, only used in changed is true and created is false
1072          *    - null or 'no-change': don't update the redirect status.
1073          * - oldcountable: bool, null, or string 'no-change' (default null):
1074          *    - bool: whether the page was counted as an article before that
1075          *      revision, only used in changed is true and created is false
1076          *    - null: if created is false, don't update the article count; if created
1077          *      is true, do update the article count
1078          *    - 'no-change': don't update the article count, ever
1079          *    When set to null, pageState['oldCountable'] will be used instead if available.
1080          *  - causeAction: an arbitrary string identifying the reason for the update.
1081          *    See DataUpdate::getCauseAction(). (default 'unknown')
1082          *  - causeAgent: name of the user who caused the update. See DataUpdate::getCauseAgent().
1083          *    (string, default 'unknown')
1084          *  - known-revision-output: a combined canonical ParserOutput for the revision, perhaps
1085          *    from some cache. The caller is responsible for ensuring that the ParserOutput indeed
1086          *    matched the $rev and $options. This mechanism is intended as a temporary stop-gap,
1087          *    for the time until caches have been changed to store RenderedRevision states instead
1088          *    of ParserOutput objects. (default: null) (since 1.33)
1089          */
1090         public function prepareUpdate( RevisionRecord $revision, array $options = [] ) {
1091                 Assert::parameter(
1092                         !isset( $options['oldrevision'] )
1093                         || $options['oldrevision'] instanceof Revision
1094                         || $options['oldrevision'] instanceof RevisionRecord,
1095                         '$options["oldrevision"]',
1096                         'must be a RevisionRecord (or Revision)'
1097                 );
1098                 Assert::parameter(
1099                         !isset( $options['triggeringUser'] )
1100                         || $options['triggeringUser'] instanceof UserIdentity,
1101                         '$options["triggeringUser"]',
1102                         'must be a UserIdentity'
1103                 );
1104
1105                 if ( !$revision->getId() ) {
1106                         throw new InvalidArgumentException(
1107                                 'Revision must have an ID set for it to be used with prepareUpdate()!'
1108                         );
1109                 }
1110
1111                 if ( $this->revision && $this->revision->getId() ) {
1112                         if ( $this->revision->getId() === $revision->getId() ) {
1113                                 return; // nothing to do!
1114                         } else {
1115                                 throw new LogicException(
1116                                         'Trying to re-use DerivedPageDataUpdater with revision '
1117                                         . $revision->getId()
1118                                         . ', but it\'s already bound to revision '
1119                                         . $this->revision->getId()
1120                                 );
1121                         }
1122                 }
1123
1124                 if ( $this->revision
1125                         && !$this->revision->getSlots()->hasSameContent( $revision->getSlots() )
1126                 ) {
1127                         throw new LogicException(
1128                                 'The Revision provided has mismatching content!'
1129                         );
1130                 }
1131
1132                 // Override fields defined in $this->options with values from $options.
1133                 $this->options = array_intersect_key( $options, $this->options ) + $this->options;
1134
1135                 if ( $this->revision ) {
1136                         $oldId = $this->pageState['oldId'] ?? 0;
1137                         $this->options['newrev'] = ( $revision->getId() !== $oldId );
1138                 } elseif ( isset( $this->options['oldrevision'] ) ) {
1139                         /** @var Revision|RevisionRecord $oldRev */
1140                         $oldRev = $this->options['oldrevision'];
1141                         $oldId = $oldRev->getId();
1142                         $this->options['newrev'] = ( $revision->getId() !== $oldId );
1143                 } else {
1144                         $oldId = $revision->getParentId();
1145                 }
1146
1147                 if ( $oldId !== null ) {
1148                         // XXX: what if $options['changed'] disagrees?
1149                         // MovePage creates a dummy revision with changed = false!
1150                         // We may want to explicitly distinguish between "no new revision" (null-edit)
1151                         // and "new revision without new content" (dummy revision).
1152
1153                         if ( $oldId === $revision->getParentId() ) {
1154                                 // NOTE: this may still be a NullRevision!
1155                                 // New revision!
1156                                 $this->options['changed'] = true;
1157                         } elseif ( $oldId === $revision->getId() ) {
1158                                 // Null-edit!
1159                                 $this->options['changed'] = false;
1160                         } else {
1161                                 // This indicates that calling code has given us the wrong Revision object
1162                                 throw new LogicException(
1163                                         'The Revision mismatches old revision ID: '
1164                                         . 'Old ID is ' . $oldId
1165                                         . ', parent ID is ' . $revision->getParentId()
1166                                         . ', revision ID is ' . $revision->getId()
1167                                 );
1168                         }
1169                 }
1170
1171                 // If prepareContent() was used to generate the PST content (which is indicated by
1172                 // $this->slotsUpdate being set), and this is not a null-edit, then the given
1173                 // revision must have the acting user as the revision author. Otherwise, user
1174                 // signatures generated by PST would mismatch the user in the revision record.
1175                 if ( $this->user !== null && $this->options['changed'] && $this->slotsUpdate ) {
1176                         $user = $revision->getUser();
1177                         if ( !$this->user->equals( $user ) ) {
1178                                 throw new LogicException(
1179                                         'The Revision provided has a mismatching actor: expected '
1180                                         . $this->user->getName()
1181                                         . ', got '
1182                                         . $user->getName()
1183                                 );
1184                         }
1185                 }
1186
1187                 // If $this->pageState was not yet initialized by grabCurrentRevision or prepareContent,
1188                 // emulate the state of the page table before the edit, as good as we can.
1189                 if ( !$this->pageState ) {
1190                         $this->pageState = [
1191                                 'oldIsRedirect' => isset( $this->options['oldredirect'] )
1192                                         && is_bool( $this->options['oldredirect'] )
1193                                                 ? $this->options['oldredirect']
1194                                                 : null,
1195                                 'oldCountable' => isset( $this->options['oldcountable'] )
1196                                         && is_bool( $this->options['oldcountable'] )
1197                                                 ? $this->options['oldcountable']
1198                                                 : null,
1199                         ];
1200
1201                         if ( $this->options['changed'] ) {
1202                                 // The edit created a new revision
1203                                 $this->pageState['oldId'] = $revision->getParentId();
1204
1205                                 if ( isset( $this->options['oldrevision'] ) ) {
1206                                         $rev = $this->options['oldrevision'];
1207                                         $this->pageState['oldRevision'] = $rev instanceof Revision
1208                                                 ? $rev->getRevisionRecord()
1209                                                 : $rev;
1210                                 }
1211                         } else {
1212                                 // This is a null-edit, so the old revision IS the new revision!
1213                                 $this->pageState['oldId'] = $revision->getId();
1214                                 $this->pageState['oldRevision'] = $revision;
1215                         }
1216                 }
1217
1218                 // "created" is forced here
1219                 $this->options['created'] = ( $this->pageState['oldId'] === 0 );
1220
1221                 $this->revision = $revision;
1222
1223                 $this->doTransition( 'has-revision' );
1224
1225                 // NOTE: in case we have a User object, don't override with a UserIdentity.
1226                 // We already checked that $revision->getUser() mathces $this->user;
1227                 if ( !$this->user ) {
1228                         $this->user = $revision->getUser( RevisionRecord::RAW );
1229                 }
1230
1231                 // Prune any output that depends on the revision ID.
1232                 if ( $this->renderedRevision ) {
1233                         $this->renderedRevision->updateRevision( $revision );
1234                 } else {
1235                         // NOTE: we want a canonical rendering, so don't pass $this->user or ParserOptions
1236                         // NOTE: the revision is either new or current, so we can bypass audience checks.
1237                         $this->renderedRevision = $this->revisionRenderer->getRenderedRevision(
1238                                 $this->revision,
1239                                 null,
1240                                 null,
1241                                 [
1242                                         'use-master' => $this->useMaster(),
1243                                         'audience' => RevisionRecord::RAW,
1244                                         'known-revision-output' => $options['known-revision-output'] ?? null
1245                                 ]
1246                         );
1247
1248                         // XXX: Since we presumably are dealing with the current revision,
1249                         // we could try to get the ParserOutput from the parser cache.
1250                 }
1251
1252                 // TODO: optionally get ParserOutput from the ParserCache here.
1253                 // Move the logic used by RefreshLinksJob here!
1254         }
1255
1256         /**
1257          * @deprecated This only exists for B/C, use the getters on DerivedPageDataUpdater directly!
1258          * @return PreparedEdit
1259          */
1260         public function getPreparedEdit() {
1261                 $this->assertPrepared( __METHOD__ );
1262
1263                 $slotsUpdate = $this->getRevisionSlotsUpdate();
1264                 $preparedEdit = new PreparedEdit();
1265
1266                 $preparedEdit->popts = $this->getCanonicalParserOptions();
1267                 $preparedEdit->parserOutputCallback = [ $this, 'getCanonicalParserOutput' ];
1268                 $preparedEdit->pstContent = $this->revision->getContent( SlotRecord::MAIN );
1269                 $preparedEdit->newContent =
1270                         $slotsUpdate->isModifiedSlot( SlotRecord::MAIN )
1271                         ? $slotsUpdate->getModifiedSlot( SlotRecord::MAIN )->getContent()
1272                         : $this->revision->getContent( SlotRecord::MAIN ); // XXX: can we just remove this?
1273                 $preparedEdit->oldContent = null; // unused. // XXX: could get this from the parent revision
1274                 $preparedEdit->revid = $this->revision ? $this->revision->getId() : null;
1275                 $preparedEdit->timestamp = $preparedEdit->output->getCacheTime();
1276                 $preparedEdit->format = $preparedEdit->pstContent->getDefaultFormat();
1277
1278                 return $preparedEdit;
1279         }
1280
1281         /**
1282          * @param string $role
1283          * @param bool $generateHtml
1284          * @return ParserOutput
1285          */
1286         public function getSlotParserOutput( $role, $generateHtml = true ) {
1287                 return $this->getRenderedRevision()->getSlotParserOutput(
1288                         $role,
1289                         [ 'generate-html' => $generateHtml ]
1290                 );
1291         }
1292
1293         /**
1294          * @return ParserOutput
1295          */
1296         public function getCanonicalParserOutput() {
1297                 return $this->getRenderedRevision()->getRevisionParserOutput();
1298         }
1299
1300         /**
1301          * @return ParserOptions
1302          */
1303         public function getCanonicalParserOptions() {
1304                 return $this->getRenderedRevision()->getOptions();
1305         }
1306
1307         /**
1308          * @param bool $recursive
1309          *
1310          * @return DeferrableUpdate[]
1311          */
1312         public function getSecondaryDataUpdates( $recursive = false ) {
1313                 if ( $this->isContentDeleted() ) {
1314                         // This shouldn't happen, since the current content is always public,
1315                         // and DataUpates are only needed for current content.
1316                         return [];
1317                 }
1318
1319                 $output = $this->getCanonicalParserOutput();
1320
1321                 // Construct a LinksUpdate for the combined canonical output.
1322                 $linksUpdate = new LinksUpdate(
1323                         $this->getTitle(),
1324                         $output,
1325                         $recursive
1326                 );
1327
1328                 $allUpdates = [ $linksUpdate ];
1329
1330                 // NOTE: Run updates for all slots, not just the modified slots! Otherwise,
1331                 // info for an inherited slot may end up being removed. This is also needed
1332                 // to ensure that purges are effective.
1333                 $renderedRevision = $this->getRenderedRevision();
1334                 foreach ( $this->getSlots()->getSlotRoles() as $role ) {
1335                         $slot = $this->getRawSlot( $role );
1336                         $content = $slot->getContent();
1337                         $handler = $content->getContentHandler();
1338
1339                         $updates = $handler->getSecondaryDataUpdates(
1340                                 $this->getTitle(),
1341                                 $content,
1342                                 $role,
1343                                 $renderedRevision
1344                         );
1345                         $allUpdates = array_merge( $allUpdates, $updates );
1346
1347                         // TODO: remove B/C hack in 1.32!
1348                         // NOTE: we assume that the combined output contains all relevant meta-data for
1349                         // all slots!
1350                         $legacyUpdates = $content->getSecondaryDataUpdates(
1351                                 $this->getTitle(),
1352                                 null,
1353                                 $recursive,
1354                                 $output
1355                         );
1356
1357                         // HACK: filter out redundant and incomplete LinksUpdates
1358                         $legacyUpdates = array_filter( $legacyUpdates, function ( $update ) {
1359                                 return !( $update instanceof LinksUpdate );
1360                         } );
1361
1362                         $allUpdates = array_merge( $allUpdates, $legacyUpdates );
1363                 }
1364
1365                 // XXX: if a slot was removed by an earlier edit, but deletion updates failed to run at
1366                 // that time, we don't know for which slots to run deletion updates when purging a page.
1367                 // We'd have to examine the entire history of the page to determine that. Perhaps there
1368                 // could be a "try extra hard" mode for that case that would run a DB query to find all
1369                 // roles/models ever used on the page. On the other hand, removing slots should be quite
1370                 // rare, so perhaps this isn't worth the trouble.
1371
1372                 // TODO: consolidate with similar logic in WikiPage::getDeletionUpdates()
1373                 $wikiPage = $this->getWikiPage();
1374                 $parentRevision = $this->getParentRevision();
1375                 foreach ( $this->getRemovedSlotRoles() as $role ) {
1376                         // HACK: we should get the content model of the removed slot from a SlotRoleHandler!
1377                         // For now, find the slot in the parent revision - if the slot was removed, it should
1378                         // always exist in the parent revision.
1379                         $parentSlot = $parentRevision->getSlot( $role, RevisionRecord::RAW );
1380                         $content = $parentSlot->getContent();
1381                         $handler = $content->getContentHandler();
1382
1383                         $updates = $handler->getDeletionUpdates(
1384                                 $this->getTitle(),
1385                                 $role
1386                         );
1387                         $allUpdates = array_merge( $allUpdates, $updates );
1388
1389                         // TODO: remove B/C hack in 1.32!
1390                         $legacyUpdates = $content->getDeletionUpdates( $wikiPage );
1391
1392                         // HACK: filter out redundant and incomplete LinksDeletionUpdate
1393                         $legacyUpdates = array_filter( $legacyUpdates, function ( $update ) {
1394                                 return !( $update instanceof LinksDeletionUpdate );
1395                         } );
1396
1397                         $allUpdates = array_merge( $allUpdates, $legacyUpdates );
1398                 }
1399
1400                 // TODO: hard deprecate SecondaryDataUpdates in favor of RevisionDataUpdates in 1.33!
1401                 Hooks::run(
1402                         'RevisionDataUpdates',
1403                         [ $this->getTitle(), $renderedRevision, &$allUpdates ]
1404                 );
1405
1406                 return $allUpdates;
1407         }
1408
1409         /**
1410          * Do standard updates after page edit, purge, or import.
1411          * Update links tables, site stats, search index, title cache, message cache, etc.
1412          * Purges pages that depend on this page when appropriate.
1413          * With a 10% chance, triggers pruning the recent changes table.
1414          *
1415          * @note prepareUpdate() must be called before calling this method!
1416          *
1417          * MCR migration note: this replaces WikiPage::doEditUpdates.
1418          */
1419         public function doUpdates() {
1420                 $this->assertTransition( 'done' );
1421
1422                 // TODO: move logic into a PageEventEmitter service
1423
1424                 $wikiPage = $this->getWikiPage(); // TODO: use only for legacy hooks!
1425
1426                 $legacyUser = User::newFromIdentity( $this->user );
1427                 $legacyRevision = new Revision( $this->revision );
1428
1429                 $userParserOptions = ParserOptions::newFromUser( $legacyUser );
1430                 // Decide whether to save the final canonical parser ouput based on the fact that
1431                 // users are typically redirected to viewing pages right after they edit those pages.
1432                 // Due to vary-revision-id, getting/saving that output here might require a reparse.
1433                 if ( $userParserOptions->matchesForCacheKey( $this->getCanonicalParserOptions() ) ) {
1434                         // Whether getting the final output requires a reparse or not, the user will
1435                         // need canonical output anyway, since that is what their parser options use.
1436                         // A reparse now at least has the benefit of various warm process caches.
1437                         $this->doParserCacheUpdate();
1438                 } else {
1439                         // If the user does not have canonical parse options, then don't risk another parse
1440                         // to make output they cannot use on the page refresh that typically occurs after
1441                         // editing. Doing the parser output save post-send will still benefit *other* users.
1442                         DeferredUpdates::addCallableUpdate( function () {
1443                                 $this->doParserCacheUpdate();
1444                         } );
1445                 }
1446
1447                 // Defer the getCannonicalParserOutput() call triggered by getSecondaryDataUpdates()
1448                 // by wrapping the code that schedules the secondary updates in a callback itself
1449                 $wrapperUpdate = new MWCallableUpdate(
1450                         function () {
1451                                 $this->doSecondaryDataUpdates( [
1452                                         // T52785 do not update any other pages on a null edit
1453                                         'recursive' => $this->options['changed']
1454                                 ] );
1455                         },
1456                         __METHOD__
1457                 );
1458                 $wrapperUpdate->setTransactionRoundRequirement( $wrapperUpdate::TRX_ROUND_ABSENT );
1459                 DeferredUpdates::addUpdate( $wrapperUpdate );
1460
1461                 // TODO: MCR: check if *any* changed slot supports categories!
1462                 if ( $this->rcWatchCategoryMembership
1463                         && $this->getContentHandler( SlotRecord::MAIN )->supportsCategories() === true
1464                         && ( $this->options['changed'] || $this->options['created'] )
1465                         && !$this->options['restored']
1466                 ) {
1467                         // Note: jobs are pushed after deferred updates, so the job should be able to see
1468                         // the recent change entry (also done via deferred updates) and carry over any
1469                         // bot/deletion/IP flags, ect.
1470                         $this->jobQueueGroup->lazyPush(
1471                                 CategoryMembershipChangeJob::newSpec(
1472                                         $this->getTitle(),
1473                                         $this->revision->getTimestamp()
1474                                 )
1475                         );
1476                 }
1477
1478                 // TODO: replace legacy hook! Use a listener on PageEventEmitter instead!
1479                 // @note: Extensions should *avoid* calling getCannonicalParserOutput() when using
1480                 // this hook whenever possible in order to avoid unnecessary additional parses.
1481                 $editInfo = $this->getPreparedEdit();
1482                 Hooks::run( 'ArticleEditUpdates',
1483                         [ &$wikiPage, &$editInfo, $this->options['changed'] ] );
1484
1485                 // TODO: replace legacy hook! Use a listener on PageEventEmitter instead!
1486                 if ( Hooks::run( 'ArticleEditUpdatesDeleteFromRecentchanges', [ &$wikiPage ] ) ) {
1487                         // Flush old entries from the `recentchanges` table
1488                         if ( mt_rand( 0, 9 ) == 0 ) {
1489                                 $this->jobQueueGroup->lazyPush( RecentChangesUpdateJob::newPurgeJob() );
1490                         }
1491                 }
1492
1493                 $id = $this->getPageId();
1494                 $title = $this->getTitle();
1495                 $dbKey = $title->getPrefixedDBkey();
1496                 $shortTitle = $title->getDBkey();
1497
1498                 if ( !$title->exists() ) {
1499                         wfDebug( __METHOD__ . ": Page doesn't exist any more, bailing out\n" );
1500
1501                         $this->doTransition( 'done' );
1502                         return;
1503                 }
1504
1505                 DeferredUpdates::addCallableUpdate( function () {
1506                         if (
1507                                 $this->options['oldcountable'] === 'no-change' ||
1508                                 ( !$this->options['changed'] && !$this->options['moved'] )
1509                         ) {
1510                                 $good = 0;
1511                         } elseif ( $this->options['created'] ) {
1512                                 $good = (int)$this->isCountable();
1513                         } elseif ( $this->options['oldcountable'] !== null ) {
1514                                 $good = (int)$this->isCountable()
1515                                         - (int)$this->options['oldcountable'];
1516                         } elseif ( isset( $this->pageState['oldCountable'] ) ) {
1517                                 $good = (int)$this->isCountable()
1518                                         - (int)$this->pageState['oldCountable'];
1519                         } else {
1520                                 $good = 0;
1521                         }
1522                         $edits = $this->options['changed'] ? 1 : 0;
1523                         $pages = $this->options['created'] ? 1 : 0;
1524
1525                         DeferredUpdates::addUpdate( SiteStatsUpdate::factory(
1526                                 [ 'edits' => $edits, 'articles' => $good, 'pages' => $pages ]
1527                         ) );
1528                 } );
1529
1530                 // TODO: make search infrastructure aware of slots!
1531                 $mainSlot = $this->revision->getSlot( SlotRecord::MAIN );
1532                 if ( !$mainSlot->isInherited() && !$this->isContentDeleted() ) {
1533                         DeferredUpdates::addUpdate( new SearchUpdate( $id, $dbKey, $mainSlot->getContent() ) );
1534                 }
1535
1536                 // If this is another user's talk page, update newtalk.
1537                 // Don't do this if $options['changed'] = false (null-edits) nor if
1538                 // it's a minor edit and the user making the edit doesn't generate notifications for those.
1539                 if ( $this->options['changed']
1540                         && $title->getNamespace() == NS_USER_TALK
1541                         && $shortTitle != $legacyUser->getTitleKey()
1542                         && !( $this->revision->isMinor() && $legacyUser->isAllowed( 'nominornewtalk' ) )
1543                 ) {
1544                         $recipient = User::newFromName( $shortTitle, false );
1545                         if ( !$recipient ) {
1546                                 wfDebug( __METHOD__ . ": invalid username\n" );
1547                         } else {
1548                                 // Allow extensions to prevent user notification
1549                                 // when a new message is added to their talk page
1550                                 // TODO: replace legacy hook!  Use a listener on PageEventEmitter instead!
1551                                 if ( Hooks::run( 'ArticleEditUpdateNewTalk', [ &$wikiPage, $recipient ] ) ) {
1552                                         if ( User::isIP( $shortTitle ) ) {
1553                                                 // An anonymous user
1554                                                 $recipient->setNewtalk( true, $legacyRevision );
1555                                         } elseif ( $recipient->isLoggedIn() ) {
1556                                                 $recipient->setNewtalk( true, $legacyRevision );
1557                                         } else {
1558                                                 wfDebug( __METHOD__ . ": don't need to notify a nonexistent user\n" );
1559                                         }
1560                                 }
1561                         }
1562                 }
1563
1564                 if ( $title->getNamespace() == NS_MEDIAWIKI
1565                         && $this->getRevisionSlotsUpdate()->isModifiedSlot( SlotRecord::MAIN )
1566                 ) {
1567                         $mainContent = $this->isContentDeleted() ? null : $this->getRawContent( SlotRecord::MAIN );
1568
1569                         $this->messageCache->updateMessageOverride( $title, $mainContent );
1570                 }
1571
1572                 // TODO: move onArticleCreate and onArticle into a PageEventEmitter service
1573                 if ( $this->options['created'] ) {
1574                         WikiPage::onArticleCreate( $title );
1575                 } elseif ( $this->options['changed'] ) { // T52785
1576                         WikiPage::onArticleEdit( $title, $legacyRevision, $this->getTouchedSlotRoles() );
1577                 }
1578
1579                 $oldRevision = $this->getParentRevision();
1580                 $oldLegacyRevision = $oldRevision ? new Revision( $oldRevision ) : null;
1581
1582                 // TODO: In the wiring, register a listener for this on the new PageEventEmitter
1583                 ResourceLoaderWikiModule::invalidateModuleCache(
1584                         $title, $oldLegacyRevision, $legacyRevision, $this->getWikiId() ?: wfWikiID()
1585                 );
1586
1587                 $this->doTransition( 'done' );
1588         }
1589
1590         /**
1591          * Do secondary data updates (such as updating link tables).
1592          *
1593          * MCR note: this method is temporarily exposed via WikiPage::doSecondaryDataUpdates.
1594          *
1595          * @param array $options
1596          *   - recursive: make the update recursive, i.e. also update pages which transclude the
1597          *     current page or otherwise depend on it (default: false)
1598          *   - defer: one of the DeferredUpdates constants, or false to run immediately after waiting
1599          *     for replication of the changes from the SecondaryDataUpdates hooks (default: false)
1600          * @since 1.32
1601          */
1602         public function doSecondaryDataUpdates( array $options = [] ) {
1603                 $this->assertHasRevision( __METHOD__ );
1604                 $options += [ 'recursive' => false, 'defer' => false ];
1605                 $deferValues = [ false, DeferredUpdates::PRESEND, DeferredUpdates::POSTSEND ];
1606                 if ( !in_array( $options['defer'], $deferValues, true ) ) {
1607                         throw new InvalidArgumentException( 'Invalid value for defer: ' . $options['defer'] );
1608                 }
1609                 $updates = $this->getSecondaryDataUpdates( $options['recursive'] );
1610
1611                 $triggeringUser = $this->options['triggeringUser'] ?? $this->user;
1612                 if ( !$triggeringUser instanceof User ) {
1613                         $triggeringUser = User::newFromIdentity( $triggeringUser );
1614                 }
1615                 $causeAction = $this->options['causeAction'] ?? 'unknown';
1616                 $causeAgent = $this->options['causeAgent'] ?? 'unknown';
1617                 $legacyRevision = new Revision( $this->revision );
1618
1619                 foreach ( $updates as $update ) {
1620                         if ( $update instanceof DataUpdate ) {
1621                                 $update->setCause( $causeAction, $causeAgent );
1622                         }
1623                         if ( $update instanceof LinksUpdate ) {
1624                                 $update->setRevision( $legacyRevision );
1625                                 $update->setTriggeringUser( $triggeringUser );
1626                         }
1627                 }
1628
1629                 if ( $options['defer'] === false ) {
1630                         // T221577: flush any transaction; each update needs outer transaction scope
1631                         $this->loadbalancerFactory->commitMasterChanges( __METHOD__ );
1632                         foreach ( $updates as $update ) {
1633                                 DeferredUpdates::attemptUpdate( $update, $this->loadbalancerFactory );
1634                         }
1635                 } else {
1636                         foreach ( $updates as $update ) {
1637                                 DeferredUpdates::addUpdate( $update, $options['defer'] );
1638                         }
1639                 }
1640         }
1641
1642         public function doParserCacheUpdate() {
1643                 $this->assertHasRevision( __METHOD__ );
1644
1645                 $wikiPage = $this->getWikiPage(); // TODO: ParserCache should accept a RevisionRecord instead
1646
1647                 // NOTE: this may trigger the first parsing of the new content after an edit (when not
1648                 // using pre-generated stashed output).
1649                 // XXX: we may want to use the PoolCounter here. This would perhaps allow the initial parse
1650                 // to be performed post-send. The client could already follow a HTTP redirect to the
1651                 // page view, but would then have to wait for a response until rendering is complete.
1652                 $output = $this->getCanonicalParserOutput();
1653
1654                 // Save it to the parser cache. Use the revision timestamp in the case of a
1655                 // freshly saved edit, as that matches page_touched and a mismatch would trigger an
1656                 // unnecessary reparse.
1657                 $timestamp = $this->options['newrev'] ? $this->revision->getTimestamp()
1658                         : $output->getCacheTime();
1659                 $this->parserCache->save(
1660                         $output, $wikiPage, $this->getCanonicalParserOptions(),
1661                         $timestamp, $this->revision->getId()
1662                 );
1663         }
1664
1665 }