DerivedPageDataUpdater.php

Go to the documentation of this file.

<?php
namespace MediaWiki\Storage;

use CategoryMembershipChangeJob;
use Content;
use ContentHandler;
use DataUpdate;
use DeferrableUpdate;
use DeferredUpdates;
use Hooks;
use IDBAccessObject;
use InvalidArgumentException;
use JobQueueGroup;
use Language;
use LinksDeletionUpdate;
use LinksUpdate;
use LogicException;
use MediaWiki\Edit\PreparedEdit;
use MediaWiki\MediaWikiServices;
use MediaWiki\Revision\MutableRevisionRecord;
use MediaWiki\Revision\RenderedRevision;
use MediaWiki\Revision\RevisionRecord;
use MediaWiki\Revision\RevisionRenderer;
use MediaWiki\Revision\RevisionSlots;
use MediaWiki\Revision\RevisionStore;
use MediaWiki\Revision\SlotRoleRegistry;
use MediaWiki\Revision\SlotRecord;
use MediaWiki\User\UserIdentity;
use MessageCache;
use ParserCache;
use ParserOptions;
use ParserOutput;
use RecentChangesUpdateJob;
use ResourceLoaderWikiModule;
use Revision;
use SearchUpdate;
use SiteStatsUpdate;
use Title;
use User;
use Wikimedia\Assert\Assert;
use Wikimedia\Rdbms\LBFactory;
use WikiPage;

class DerivedPageDataUpdater implements IDBAccessObject {

    private $user = null;

    private $wikiPage;

    private $parserCache;

    private $revisionStore;

    private $contLang;

    private $jobQueueGroup;

    private $messageCache;

    private $loadbalancerFactory;

    private $articleCountMethod;

    private $rcWatchCategoryMembership = false;

    private $options = [
        'changed' => true,
        // newrev is true if prepareUpdate is handling the creation of a new revision,
        // as opposed to a null edit or a forced update.
        'newrev' => false,
        'created' => false,
        'moved' => false,
        'restored' => false,
        'oldrevision' => null,
        'oldcountable' => null,
        'oldredirect' => null,
        'triggeringUser' => null,
        // causeAction/causeAgent default to 'unknown' but that's handled where it's read,
        // to make the life of prepareUpdate() callers easier.
        'causeAction' => null,
        'causeAgent' => null,
    ];

    private $pageState = null;

    private $slotsUpdate = null;

    private $parentRevision = null;

    private $revision = null;

    private $renderedRevision = null;

    private $revisionRenderer;

    private $slotRoleRegistry;

    private $stage = 'new';

    private static $transitions = [
        'new' => [
            'new' => true,
            'knows-current' => true,
            'has-content' => true,
            'has-revision' => true,
        ],
        'knows-current' => [
            'knows-current' => true,
            'has-content' => true,
            'has-revision' => true,
        ],
        'has-content' => [
            'has-content' => true,
            'has-revision' => true,
        ],
        'has-revision' => [
            'has-revision' => true,
            'done' => true,
        ],
    ];

    public function __construct(
        WikiPage $wikiPage,
        RevisionStore $revisionStore,
        RevisionRenderer $revisionRenderer,
        SlotRoleRegistry $slotRoleRegistry,
        ParserCache $parserCache,
        JobQueueGroup $jobQueueGroup,
        MessageCache $messageCache,
        Language $contLang,
        LBFactory $loadbalancerFactory
    ) {
        $this->wikiPage = $wikiPage;

        $this->parserCache = $parserCache;
        $this->revisionStore = $revisionStore;
        $this->revisionRenderer = $revisionRenderer;
        $this->slotRoleRegistry = $slotRoleRegistry;
        $this->jobQueueGroup = $jobQueueGroup;
        $this->messageCache = $messageCache;
        $this->contLang = $contLang;
        // XXX only needed for waiting for replicas to catch up; there should be a narrower
        // interface for that.
        $this->loadbalancerFactory = $loadbalancerFactory;
    }

    private function doTransition( $newStage ) {
        $this->assertTransition( $newStage );

        $oldStage = $this->stage;
        $this->stage = $newStage;

        return $oldStage;
    }

    private function assertTransition( $newStage ) {
        if ( empty( self::$transitions[$this->stage][$newStage] ) ) {
            throw new LogicException( "Cannot transition from {$this->stage} to $newStage" );
        }
    }

    private function getWikiId() {
        // TODO: get from RevisionStore
        return false;
    }

    public function isReusableFor(
        UserIdentity $user = null,
        RevisionRecord $revision = null,
        RevisionSlotsUpdate $slotsUpdate = null,
        $parentId = null
    ) {
        if ( $revision
            && $parentId
            && $revision->getParentId() !== $parentId
        ) {
            throw new InvalidArgumentException( '$parentId should match the parent of $revision' );
        }

        // NOTE: For null revisions, $user may be different from $this->revision->getUser
        // and also from $revision->getUser.
        // But $user should always match $this->user.
        if ( $user && $this->user && $user->getName() !== $this->user->getName() ) {
            return false;
        }

        if ( $revision && $this->revision && $this->revision->getId()
            && $this->revision->getId() !== $revision->getId()
        ) {
            return false;
        }

        if ( $this->pageState
            && $revision
            && $revision->getParentId() !== null
            && $this->pageState['oldId'] !== $revision->getParentId()
        ) {
            return false;
        }

        if ( $this->pageState
            && $parentId !== null
            && $this->pageState['oldId'] !== $parentId
        ) {
            return false;
        }

        // NOTE: this check is the primary reason for having the $this->slotsUpdate field!
        if ( $this->slotsUpdate
            && $slotsUpdate
            && !$this->slotsUpdate->hasSameUpdates( $slotsUpdate )
        ) {
            return false;
        }

        if ( $revision
            && $this->revision
            && !$this->revision->getSlots()->hasSameContent( $revision->getSlots() )
        ) {
            return false;
        }

        return true;
    }

    public function setArticleCountMethod( $articleCountMethod ) {
        $this->articleCountMethod = $articleCountMethod;
    }

    public function setRcWatchCategoryMembership( $rcWatchCategoryMembership ) {
        $this->rcWatchCategoryMembership = $rcWatchCategoryMembership;
    }

    private function getTitle() {
        // NOTE: eventually, we won't get a WikiPage passed into the constructor any more
        return $this->wikiPage->getTitle();
    }

    private function getWikiPage() {
        // NOTE: eventually, we won't get a WikiPage passed into the constructor any more
        return $this->wikiPage;
    }

    public function pageExisted() {
        $this->assertHasPageState( __METHOD__ );

        return $this->pageState['oldId'] > 0;
    }

    private function getParentRevision() {
        $this->assertPrepared( __METHOD__ );

        if ( $this->parentRevision ) {
            return $this->parentRevision;
        }

        if ( !$this->pageState['oldId'] ) {
            // If there was no current revision, there is no parent revision,
            // since the page didn't exist.
            return null;
        }

        $oldId = $this->revision->getParentId();
        $flags = $this->useMaster() ? RevisionStore::READ_LATEST : 0;
        $this->parentRevision = $oldId
            ? $this->revisionStore->getRevisionById( $oldId, $flags )
            : null;

        return $this->parentRevision;
    }

    public function grabCurrentRevision() {
        if ( $this->pageState ) {
            return $this->pageState['oldRevision'];
        }

        $this->assertTransition( 'knows-current' );

        // NOTE: eventually, we won't get a WikiPage passed into the constructor any more
        $wikiPage = $this->getWikiPage();

        // Do not call WikiPage::clear(), since the caller may already have caused page data
        // to be loaded with SELECT FOR UPDATE. Just assert it's loaded now.
        $wikiPage->loadPageData( self::READ_LATEST );
        $rev = $wikiPage->getRevision();
        $current = $rev ? $rev->getRevisionRecord() : null;

        $this->pageState = [
            'oldRevision' => $current,
            'oldId' => $rev ? $rev->getId() : 0,
            'oldIsRedirect' => $wikiPage->isRedirect(), // NOTE: uses page table
            'oldCountable' => $wikiPage->isCountable(), // NOTE: uses pagelinks table
        ];

        $this->doTransition( 'knows-current' );

        return $this->pageState['oldRevision'];
    }

    public function isContentPrepared() {
        return $this->revision !== null;
    }

    public function isUpdatePrepared() {
        return $this->revision !== null && $this->revision->getId() !== null;
    }

    private function getPageId() {
        // NOTE: eventually, we won't get a WikiPage passed into the constructor any more
        return $this->wikiPage->getId();
    }

    public function isContentDeleted() {
        if ( $this->revision ) {
            // XXX: if that revision is the current revision, this should be skipped
            return $this->revision->isDeleted( RevisionRecord::DELETED_TEXT );
        } else {
            // If the content has not been saved yet, it cannot have been deleted yet.
            return false;
        }
    }

    public function getRawSlot( $role ) {
        return $this->getSlots()->getSlot( $role );
    }

    public function getRawContent( $role ) {
        return $this->getRawSlot( $role )->getContent();
    }

    private function getContentModel( $role ) {
        return $this->getRawSlot( $role )->getModel();
    }

    private function getContentHandler( $role ) {
        // TODO: inject something like a ContentHandlerRegistry
        return ContentHandler::getForModelID( $this->getContentModel( $role ) );
    }

    private function useMaster() {
        // TODO: can we just set a flag to true in prepareContent()?
        return $this->wikiPage->wasLoadedFrom( self::READ_LATEST );
    }

    public function isCountable() {
        // NOTE: Keep in sync with WikiPage::isCountable.

        if ( !$this->getTitle()->isContentPage() ) {
            return false;
        }

        if ( $this->isContentDeleted() ) {
            // This should be irrelevant: countability only applies to the current revision,
            // and the current revision is never suppressed.
            return false;
        }

        if ( $this->isRedirect() ) {
            return false;
        }

        $hasLinks = null;

        if ( $this->articleCountMethod === 'link' ) {
            // NOTE: it would be more appropriate to determine for each slot separately
            // whether it has links, and use that information with that slot's
            // isCountable() method. However, that would break parity with
            // WikiPage::isCountable, which uses the pagelinks table to determine
            // whether the current revision has links.
            $hasLinks = (bool)count( $this->getCanonicalParserOutput()->getLinks() );
        }

        foreach ( $this->getModifiedSlotRoles() as $role ) {
            $roleHandler = $this->slotRoleRegistry->getRoleHandler( $role );
            if ( $roleHandler->supportsArticleCount() ) {
                $content = $this->getRawContent( $role );

                if ( $content->isCountable( $hasLinks ) ) {
                    return true;
                }
            }
        }

        return false;
    }

    public function isRedirect() {
        // NOTE: main slot determines redirect status
        // TODO: MCR: this should be controlled by a PageTypeHandler
        $mainContent = $this->getRawContent( SlotRecord::MAIN );

        return $mainContent->isRedirect();
    }

    private function revisionIsRedirect( RevisionRecord $rev ) {
        // NOTE: main slot determines redirect status
        $mainContent = $rev->getContent( SlotRecord::MAIN, RevisionRecord::RAW );

        return $mainContent->isRedirect();
    }

    public function prepareContent(
        User $user,
        RevisionSlotsUpdate $slotsUpdate,
        $useStash = true
    ) {
        if ( $this->slotsUpdate ) {
            if ( !$this->user ) {
                throw new LogicException(
                    'Unexpected state: $this->slotsUpdate was initialized, '
                    . 'but $this->user was not.'
                );
            }

            if ( $this->user->getName() !== $user->getName() ) {
                throw new LogicException( 'Can\'t call prepareContent() again for different user! '
                    . 'Expected ' . $this->user->getName() . ', got ' . $user->getName()
                );
            }

            if ( !$this->slotsUpdate->hasSameUpdates( $slotsUpdate ) ) {
                throw new LogicException(
                    'Can\'t call prepareContent() again with different slot content!'
                );
            }

            return; // prepareContent() already done, nothing to do
        }

        $this->assertTransition( 'has-content' );

        $wikiPage = $this->getWikiPage(); // TODO: use only for legacy hooks!
        $title = $this->getTitle();

        $parentRevision = $this->grabCurrentRevision();

        // The edit may have already been prepared via api.php?action=stashedit
        $stashedEdit = false;

        // TODO: MCR: allow output for all slots to be stashed.
        if ( $useStash && $slotsUpdate->isModifiedSlot( SlotRecord::MAIN ) ) {
            $editStash = MediaWikiServices::getInstance()->getPageEditStash();
            $stashedEdit = $editStash->checkCache(
                $title,
                $slotsUpdate->getModifiedSlot( SlotRecord::MAIN )->getContent(),
                User::newFromIdentity( $user )
            );
        }

        $userPopts = ParserOptions::newFromUserAndLang( $user, $this->contLang );
        Hooks::run( 'ArticlePrepareTextForEdit', [ $wikiPage, $userPopts ] );

        $this->user = $user;
        $this->slotsUpdate = $slotsUpdate;

        if ( $parentRevision ) {
            $this->revision = MutableRevisionRecord::newFromParentRevision( $parentRevision );
        } else {
            $this->revision = new MutableRevisionRecord( $title );
        }

        // NOTE: user and timestamp must be set, so they can be used for
        // {{subst:REVISIONUSER}} and {{subst:REVISIONTIMESTAMP}} in PST!
        $this->revision->setTimestamp( wfTimestampNow() );
        $this->revision->setUser( $user );

        // Set up ParserOptions to operate on the new revision
        $oldCallback = $userPopts->getCurrentRevisionCallback();
        $userPopts->setCurrentRevisionCallback(
            function ( Title $parserTitle, $parser = false ) use ( $title, $oldCallback ) {
                if ( $parserTitle->equals( $title ) ) {
                    $legacyRevision = new Revision( $this->revision );
                    return $legacyRevision;
                } else {
                    return call_user_func( $oldCallback, $parserTitle, $parser );
                }
            }
        );

        $pstContentSlots = $this->revision->getSlots();

        foreach ( $slotsUpdate->getModifiedRoles() as $role ) {
            $slot = $slotsUpdate->getModifiedSlot( $role );

            if ( $slot->isInherited() ) {
                // No PST for inherited slots! Note that "modified" slots may still be inherited
                // from an earlier version, e.g. for rollbacks.
                $pstSlot = $slot;
            } elseif ( $role === SlotRecord::MAIN && $stashedEdit ) {
                // TODO: MCR: allow PST content for all slots to be stashed.
                $pstSlot = SlotRecord::newUnsaved( $role, $stashedEdit->pstContent );
            } else {
                $content = $slot->getContent();
                $pstContent = $content->preSaveTransform( $title, $this->user, $userPopts );
                $pstSlot = SlotRecord::newUnsaved( $role, $pstContent );
            }

            $pstContentSlots->setSlot( $pstSlot );
        }

        foreach ( $slotsUpdate->getRemovedRoles() as $role ) {
            $pstContentSlots->removeSlot( $role );
        }

        $this->options['created'] = ( $parentRevision === null );
        $this->options['changed'] = ( $parentRevision === null
            || !$pstContentSlots->hasSameContent( $parentRevision->getSlots() ) );

        $this->doTransition( 'has-content' );

        if ( !$this->options['changed'] ) {
            // null-edit!

            // TODO: move this into MutableRevisionRecord
            // TODO: This needs to behave differently for a forced dummy edit!
            $this->revision->setId( $parentRevision->getId() );
            $this->revision->setTimestamp( $parentRevision->getTimestamp() );
            $this->revision->setPageId( $parentRevision->getPageId() );
            $this->revision->setParentId( $parentRevision->getParentId() );
            $this->revision->setUser( $parentRevision->getUser( RevisionRecord::RAW ) );
            $this->revision->setComment( $parentRevision->getComment( RevisionRecord::RAW ) );
            $this->revision->setMinorEdit( $parentRevision->isMinor() );
            $this->revision->setVisibility( $parentRevision->getVisibility() );

            // prepareUpdate() is redundant for null-edits
            $this->doTransition( 'has-revision' );
        } else {
            $this->parentRevision = $parentRevision;
        }

        $renderHints = [ 'use-master' => $this->useMaster(), 'audience' => RevisionRecord::RAW ];

        if ( $stashedEdit ) {
            $output = $stashedEdit->output;

            // TODO: this should happen when stashing the ParserOutput, not now!
            $output->setCacheTime( $stashedEdit->timestamp );

            $renderHints['known-revision-output'] = $output;
        }

        // NOTE: we want a canonical rendering, so don't pass $this->user or ParserOptions
        // NOTE: the revision is either new or current, so we can bypass audience checks.
        $this->renderedRevision = $this->revisionRenderer->getRenderedRevision(
            $this->revision,
            null,
            null,
            $renderHints
        );
    }

    public function getRevision() {
        $this->assertPrepared( __METHOD__ );
        return $this->revision;
    }

    public function getRenderedRevision() {
        $this->assertPrepared( __METHOD__ );

        return $this->renderedRevision;
    }

    private function assertHasPageState( $method ) {
        if ( !$this->pageState ) {
            throw new LogicException(
                'Must call grabCurrentRevision() or prepareContent() '
                . 'or prepareUpdate() before calling ' . $method
            );
        }
    }

    private function assertPrepared( $method ) {
        if ( !$this->revision ) {
            throw new LogicException(
                'Must call prepareContent() or prepareUpdate() before calling ' . $method
            );
        }
    }

    private function assertHasRevision( $method ) {
        if ( !$this->revision->getId() ) {
            throw new LogicException(
                'Must call prepareUpdate() before calling ' . $method
            );
        }
    }

    public function isCreation() {
        $this->assertPrepared( __METHOD__ );
        return $this->options['created'];
    }

    public function isChange() {
        $this->assertPrepared( __METHOD__ );
        return $this->options['changed'];
    }

    public function wasRedirect() {
        $this->assertHasPageState( __METHOD__ );

        if ( $this->pageState['oldIsRedirect'] === null ) {
            $rev = $this->pageState['oldRevision'];
            if ( $rev ) {
                $this->pageState['oldIsRedirect'] = $this->revisionIsRedirect( $rev );
            } else {
                $this->pageState['oldIsRedirect'] = false;
            }
        }

        return $this->pageState['oldIsRedirect'];
    }

    public function getSlots() {
        $this->assertPrepared( __METHOD__ );
        return $this->revision->getSlots();
    }

    private function getRevisionSlotsUpdate() {
        $this->assertPrepared( __METHOD__ );

        if ( !$this->slotsUpdate ) {
            $old = $this->getParentRevision();
            $this->slotsUpdate = RevisionSlotsUpdate::newFromRevisionSlots(
                $this->revision->getSlots(),
                $old ? $old->getSlots() : null
            );
        }
        return $this->slotsUpdate;
    }

    public function getTouchedSlotRoles() {
        return $this->getRevisionSlotsUpdate()->getTouchedRoles();
    }

    public function getModifiedSlotRoles() {
        return $this->getRevisionSlotsUpdate()->getModifiedRoles();
    }

    public function getRemovedSlotRoles() {
        return $this->getRevisionSlotsUpdate()->getRemovedRoles();
    }

    public function prepareUpdate( RevisionRecord $revision, array $options = [] ) {
        Assert::parameter(
            !isset( $options['oldrevision'] )
            || $options['oldrevision'] instanceof Revision
            || $options['oldrevision'] instanceof RevisionRecord,
            '$options["oldrevision"]',
            'must be a RevisionRecord (or Revision)'
        );
        Assert::parameter(
            !isset( $options['triggeringUser'] )
            || $options['triggeringUser'] instanceof UserIdentity,
            '$options["triggeringUser"]',
            'must be a UserIdentity'
        );

        if ( !$revision->getId() ) {
            throw new InvalidArgumentException(
                'Revision must have an ID set for it to be used with prepareUpdate()!'
            );
        }

        if ( $this->revision && $this->revision->getId() ) {
            if ( $this->revision->getId() === $revision->getId() ) {
                return; // nothing to do!
            } else {
                throw new LogicException(
                    'Trying to re-use DerivedPageDataUpdater with revision '
                    . $revision->getId()
                    . ', but it\'s already bound to revision '
                    . $this->revision->getId()
                );
            }
        }

        if ( $this->revision
            && !$this->revision->getSlots()->hasSameContent( $revision->getSlots() )
        ) {
            throw new LogicException(
                'The Revision provided has mismatching content!'
            );
        }

        // Override fields defined in $this->options with values from $options.
        $this->options = array_intersect_key( $options, $this->options ) + $this->options;

        if ( $this->revision ) {
            $oldId = $this->pageState['oldId'] ?? 0;
            $this->options['newrev'] = ( $revision->getId() !== $oldId );
        } elseif ( isset( $this->options['oldrevision'] ) ) {
            $oldRev = $this->options['oldrevision'];
            $oldId = $oldRev->getId();
            $this->options['newrev'] = ( $revision->getId() !== $oldId );
        } else {
            $oldId = $revision->getParentId();
        }

        if ( $oldId !== null ) {
            // XXX: what if $options['changed'] disagrees?
            // MovePage creates a dummy revision with changed = false!
            // We may want to explicitly distinguish between "no new revision" (null-edit)
            // and "new revision without new content" (dummy revision).

            if ( $oldId === $revision->getParentId() ) {
                // NOTE: this may still be a NullRevision!
                // New revision!
                $this->options['changed'] = true;
            } elseif ( $oldId === $revision->getId() ) {
                // Null-edit!
                $this->options['changed'] = false;
            } else {
                // This indicates that calling code has given us the wrong Revision object
                throw new LogicException(
                    'The Revision mismatches old revision ID: '
                    . 'Old ID is ' . $oldId
                    . ', parent ID is ' . $revision->getParentId()
                    . ', revision ID is ' . $revision->getId()
                );
            }
        }

        // If prepareContent() was used to generate the PST content (which is indicated by
        // $this->slotsUpdate being set), and this is not a null-edit, then the given
        // revision must have the acting user as the revision author. Otherwise, user
        // signatures generated by PST would mismatch the user in the revision record.
        if ( $this->user !== null && $this->options['changed'] && $this->slotsUpdate ) {
            $user = $revision->getUser();
            if ( !$this->user->equals( $user ) ) {
                throw new LogicException(
                    'The Revision provided has a mismatching actor: expected '
                    . $this->user->getName()
                    . ', got '
                    . $user->getName()
                );
            }
        }

        // If $this->pageState was not yet initialized by grabCurrentRevision or prepareContent,
        // emulate the state of the page table before the edit, as good as we can.
        if ( !$this->pageState ) {
            $this->pageState = [
                'oldIsRedirect' => isset( $this->options['oldredirect'] )
                    && is_bool( $this->options['oldredirect'] )
                        ? $this->options['oldredirect']
                        : null,
                'oldCountable' => isset( $this->options['oldcountable'] )
                    && is_bool( $this->options['oldcountable'] )
                        ? $this->options['oldcountable']
                        : null,
            ];

            if ( $this->options['changed'] ) {
                // The edit created a new revision
                $this->pageState['oldId'] = $revision->getParentId();

                if ( isset( $this->options['oldrevision'] ) ) {
                    $rev = $this->options['oldrevision'];
                    $this->pageState['oldRevision'] = $rev instanceof Revision
                        ? $rev->getRevisionRecord()
                        : $rev;
                }
            } else {
                // This is a null-edit, so the old revision IS the new revision!
                $this->pageState['oldId'] = $revision->getId();
                $this->pageState['oldRevision'] = $revision;
            }
        }

        // "created" is forced here
        $this->options['created'] = ( $this->pageState['oldId'] === 0 );

        $this->revision = $revision;

        $this->doTransition( 'has-revision' );

        // NOTE: in case we have a User object, don't override with a UserIdentity.
        // We already checked that $revision->getUser() mathces $this->user;
        if ( !$this->user ) {
            $this->user = $revision->getUser( RevisionRecord::RAW );
        }

        // Prune any output that depends on the revision ID.
        if ( $this->renderedRevision ) {
            $this->renderedRevision->updateRevision( $revision );
        } else {

            // NOTE: we want a canonical rendering, so don't pass $this->user or ParserOptions
            // NOTE: the revision is either new or current, so we can bypass audience checks.
            $this->renderedRevision = $this->revisionRenderer->getRenderedRevision(
                $this->revision,
                null,
                null,
                [ 'use-master' => $this->useMaster(), 'audience' => RevisionRecord::RAW ]
            );

            // XXX: Since we presumably are dealing with the current revision,
            // we could try to get the ParserOutput from the parser cache.
        }

        // TODO: optionally get ParserOutput from the ParserCache here.
        // Move the logic used by RefreshLinksJob here!
    }

    public function getPreparedEdit() {
        $this->assertPrepared( __METHOD__ );

        $slotsUpdate = $this->getRevisionSlotsUpdate();
        $preparedEdit = new PreparedEdit();

        $preparedEdit->popts = $this->getCanonicalParserOptions();
        $preparedEdit->parserOutputCallback = [ $this, 'getCanonicalParserOutput' ];
        $preparedEdit->pstContent = $this->revision->getContent( SlotRecord::MAIN );
        $preparedEdit->newContent =
            $slotsUpdate->isModifiedSlot( SlotRecord::MAIN )
            ? $slotsUpdate->getModifiedSlot( SlotRecord::MAIN )->getContent()
            : $this->revision->getContent( SlotRecord::MAIN ); // XXX: can we just remove this?
        $preparedEdit->oldContent = null; // unused. // XXX: could get this from the parent revision
        $preparedEdit->revid = $this->revision ? $this->revision->getId() : null;
        $preparedEdit->timestamp = $preparedEdit->output->getCacheTime();
        $preparedEdit->format = $preparedEdit->pstContent->getDefaultFormat();

        return $preparedEdit;
    }

    public function getSlotParserOutput( $role, $generateHtml = true ) {
        return $this->getRenderedRevision()->getSlotParserOutput(
            $role,
            [ 'generate-html' => $generateHtml ]
        );
    }

    public function getCanonicalParserOutput() {
        return $this->getRenderedRevision()->getRevisionParserOutput();
    }

    public function getCanonicalParserOptions() {
        return $this->getRenderedRevision()->getOptions();
    }

    public function getSecondaryDataUpdates( $recursive = false ) {
        if ( $this->isContentDeleted() ) {
            // This shouldn't happen, since the current content is always public,
            // and DataUpates are only needed for current content.
            return [];
        }

        $output = $this->getCanonicalParserOutput();

        // Construct a LinksUpdate for the combined canonical output.
        $linksUpdate = new LinksUpdate(
            $this->getTitle(),
            $output,
            $recursive
        );

        $allUpdates = [ $linksUpdate ];

        // NOTE: Run updates for all slots, not just the modified slots! Otherwise,
        // info for an inherited slot may end up being removed. This is also needed
        // to ensure that purges are effective.
        $renderedRevision = $this->getRenderedRevision();
        foreach ( $this->getSlots()->getSlotRoles() as $role ) {
            $slot = $this->getRawSlot( $role );
            $content = $slot->getContent();
            $handler = $content->getContentHandler();

            $updates = $handler->getSecondaryDataUpdates(
                $this->getTitle(),
                $content,
                $role,
                $renderedRevision
            );
            $allUpdates = array_merge( $allUpdates, $updates );

            // TODO: remove B/C hack in 1.32!
            // NOTE: we assume that the combined output contains all relevant meta-data for
            // all slots!
            $legacyUpdates = $content->getSecondaryDataUpdates(
                $this->getTitle(),
                null,
                $recursive,
                $output
            );

            // HACK: filter out redundant and incomplete LinksUpdates
            $legacyUpdates = array_filter( $legacyUpdates, function ( $update ) {
                return !( $update instanceof LinksUpdate );
            } );

            $allUpdates = array_merge( $allUpdates, $legacyUpdates );
        }

        // XXX: if a slot was removed by an earlier edit, but deletion updates failed to run at
        // that time, we don't know for which slots to run deletion updates when purging a page.
        // We'd have to examine the entire history of the page to determine that. Perhaps there
        // could be a "try extra hard" mode for that case that would run a DB query to find all
        // roles/models ever used on the page. On the other hand, removing slots should be quite
        // rare, so perhaps this isn't worth the trouble.

        // TODO: consolidate with similar logic in WikiPage::getDeletionUpdates()
        $wikiPage = $this->getWikiPage();
        $parentRevision = $this->getParentRevision();
        foreach ( $this->getRemovedSlotRoles() as $role ) {
            // HACK: we should get the content model of the removed slot from a SlotRoleHandler!
            // For now, find the slot in the parent revision - if the slot was removed, it should
            // always exist in the parent revision.
            $parentSlot = $parentRevision->getSlot( $role, RevisionRecord::RAW );
            $content = $parentSlot->getContent();
            $handler = $content->getContentHandler();

            $updates = $handler->getDeletionUpdates(
                $this->getTitle(),
                $role
            );
            $allUpdates = array_merge( $allUpdates, $updates );

            // TODO: remove B/C hack in 1.32!
            $legacyUpdates = $content->getDeletionUpdates( $wikiPage );

            // HACK: filter out redundant and incomplete LinksDeletionUpdate
            $legacyUpdates = array_filter( $legacyUpdates, function ( $update ) {
                return !( $update instanceof LinksDeletionUpdate );
            } );

            $allUpdates = array_merge( $allUpdates, $legacyUpdates );
        }

        // TODO: hard deprecate SecondaryDataUpdates in favor of RevisionDataUpdates in 1.33!
        Hooks::run(
            'RevisionDataUpdates',
            [ $this->getTitle(), $renderedRevision, &$allUpdates ]
        );

        return $allUpdates;
    }

    public function doUpdates() {
        $this->assertTransition( 'done' );

        // TODO: move logic into a PageEventEmitter service

        $wikiPage = $this->getWikiPage(); // TODO: use only for legacy hooks!

        $legacyUser = User::newFromIdentity( $this->user );
        $legacyRevision = new Revision( $this->revision );

        $userParserOptions = ParserOptions::newFromUser( $legacyUser );
        // Decide whether to save the final canonical parser ouput based on the fact that
        // users are typically redirected to viewing pages right after they edit those pages.
        // Due to vary-revision-id, getting/saving that output here might require a reparse.
        if ( $userParserOptions->matchesForCacheKey( $this->getCanonicalParserOptions() ) ) {
            // Whether getting the final output requires a reparse or not, the user will
            // need canonical output anyway, since that is what their parser options use.
            // A reparse now at least has the benefit of various warm process caches.
            $this->doParserCacheUpdate();
        } else {
            // If the user does not have canonical parse options, then don't risk another parse
            // to make output they cannot use on the page refresh that typically occurs after
            // editing. Doing the parser output save post-send will still benefit *other* users.
            DeferredUpdates::addCallableUpdate( function () {
                $this->doParserCacheUpdate();
            } );
        }

        // Defer the getCannonicalParserOutput() call triggered by getSecondaryDataUpdates()
        DeferredUpdates::addCallableUpdate( function () {
            $this->doSecondaryDataUpdates( [
                // T52785 do not update any other pages on a null edit
                'recursive' => $this->options['changed']
            ] );
        } );

        // TODO: MCR: check if *any* changed slot supports categories!
        if ( $this->rcWatchCategoryMembership
            && $this->getContentHandler( SlotRecord::MAIN )->supportsCategories() === true
            && ( $this->options['changed'] || $this->options['created'] )
            && !$this->options['restored']
        ) {
            // Note: jobs are pushed after deferred updates, so the job should be able to see
            // the recent change entry (also done via deferred updates) and carry over any
            // bot/deletion/IP flags, ect.
            $this->jobQueueGroup->lazyPush(
                CategoryMembershipChangeJob::newSpec(
                    $this->getTitle(),
                    $this->revision->getTimestamp()
                )
            );
        }

        // TODO: replace legacy hook! Use a listener on PageEventEmitter instead!
        // @note: Extensions should *avoid* calling getCannonicalParserOutput() when using
        // this hook whenever possible in order to avoid unnecessary additional parses.
        $editInfo = $this->getPreparedEdit();
        Hooks::run( 'ArticleEditUpdates',
            [ &$wikiPage, &$editInfo, $this->options['changed'] ] );

        // TODO: replace legacy hook! Use a listener on PageEventEmitter instead!
        if ( Hooks::run( 'ArticleEditUpdatesDeleteFromRecentchanges', [ &$wikiPage ] ) ) {
            // Flush old entries from the `recentchanges` table
            if ( mt_rand( 0, 9 ) == 0 ) {
                $this->jobQueueGroup->lazyPush( RecentChangesUpdateJob::newPurgeJob() );
            }
        }

        $id = $this->getPageId();
        $title = $this->getTitle();
        $dbKey = $title->getPrefixedDBkey();
        $shortTitle = $title->getDBkey();

        if ( !$title->exists() ) {
            wfDebug( __METHOD__ . ": Page doesn't exist any more, bailing out\n" );

            $this->doTransition( 'done' );
            return;
        }

        if ( $this->options['oldcountable'] === 'no-change' ||
            ( !$this->options['changed'] && !$this->options['moved'] )
        ) {
            $good = 0;
        } elseif ( $this->options['created'] ) {
            $good = (int)$this->isCountable();
        } elseif ( $this->options['oldcountable'] !== null ) {
            $good = (int)$this->isCountable()
                - (int)$this->options['oldcountable'];
        } elseif ( isset( $this->pageState['oldCountable'] ) ) {
            $good = (int)$this->isCountable()
                - (int)$this->pageState['oldCountable'];
        } else {
            $good = 0;
        }
        $edits = $this->options['changed'] ? 1 : 0;
        $pages = $this->options['created'] ? 1 : 0;

        DeferredUpdates::addUpdate( SiteStatsUpdate::factory(
            [ 'edits' => $edits, 'articles' => $good, 'pages' => $pages ]
        ) );

        // TODO: make search infrastructure aware of slots!
        $mainSlot = $this->revision->getSlot( SlotRecord::MAIN );
        if ( !$mainSlot->isInherited() && !$this->isContentDeleted() ) {
            DeferredUpdates::addUpdate( new SearchUpdate( $id, $dbKey, $mainSlot->getContent() ) );
        }

        // If this is another user's talk page, update newtalk.
        // Don't do this if $options['changed'] = false (null-edits) nor if
        // it's a minor edit and the user making the edit doesn't generate notifications for those.
        if ( $this->options['changed']
            && $title->getNamespace() == NS_USER_TALK
            && $shortTitle != $legacyUser->getTitleKey()
            && !( $this->revision->isMinor() && $legacyUser->isAllowed( 'nominornewtalk' ) )
        ) {
            $recipient = User::newFromName( $shortTitle, false );
            if ( !$recipient ) {
                wfDebug( __METHOD__ . ": invalid username\n" );
            } else {
                // Allow extensions to prevent user notification
                // when a new message is added to their talk page
                // TODO: replace legacy hook!  Use a listener on PageEventEmitter instead!
                if ( Hooks::run( 'ArticleEditUpdateNewTalk', [ &$wikiPage, $recipient ] ) ) {
                    if ( User::isIP( $shortTitle ) ) {
                        // An anonymous user
                        $recipient->setNewtalk( true, $legacyRevision );
                    } elseif ( $recipient->isLoggedIn() ) {
                        $recipient->setNewtalk( true, $legacyRevision );
                    } else {
                        wfDebug( __METHOD__ . ": don't need to notify a nonexistent user\n" );
                    }
                }
            }
        }

        if ( $title->getNamespace() == NS_MEDIAWIKI
            && $this->getRevisionSlotsUpdate()->isModifiedSlot( SlotRecord::MAIN )
        ) {
            $mainContent = $this->isContentDeleted() ? null : $this->getRawContent( SlotRecord::MAIN );

            $this->messageCache->updateMessageOverride( $title, $mainContent );
        }

        // TODO: move onArticleCreate and onArticle into a PageEventEmitter service
        if ( $this->options['created'] ) {
            WikiPage::onArticleCreate( $title );
        } elseif ( $this->options['changed'] ) { // T52785
            WikiPage::onArticleEdit( $title, $legacyRevision, $this->getTouchedSlotRoles() );
        }

        $oldRevision = $this->getParentRevision();
        $oldLegacyRevision = $oldRevision ? new Revision( $oldRevision ) : null;

        // TODO: In the wiring, register a listener for this on the new PageEventEmitter
        ResourceLoaderWikiModule::invalidateModuleCache(
            $title, $oldLegacyRevision, $legacyRevision, $this->getWikiId() ?: wfWikiID()
        );

        $this->doTransition( 'done' );
    }

    public function doSecondaryDataUpdates( array $options = [] ) {
        $this->assertHasRevision( __METHOD__ );
        $options += [
            'recursive' => false,
            'defer' => false,
            'transactionTicket' => null,
        ];
        $deferValues = [ false, DeferredUpdates::PRESEND, DeferredUpdates::POSTSEND ];
        if ( !in_array( $options['defer'], $deferValues, true ) ) {
            throw new InvalidArgumentException( 'invalid value for defer: ' . $options['defer'] );
        }
        Assert::parameterType( 'integer|null', $options['transactionTicket'],
            '$options[\'transactionTicket\']' );

        $updates = $this->getSecondaryDataUpdates( $options['recursive'] );

        $triggeringUser = $this->options['triggeringUser'] ?? $this->user;
        if ( !$triggeringUser instanceof User ) {
            $triggeringUser = User::newFromIdentity( $triggeringUser );
        }
        $causeAction = $this->options['causeAction'] ?? 'unknown';
        $causeAgent = $this->options['causeAgent'] ?? 'unknown';
        $legacyRevision = new Revision( $this->revision );

        if ( $options['defer'] === false && $options['transactionTicket'] !== null ) {
            // For legacy hook handlers doing updates via LinksUpdateConstructed, make sure
            // any pending writes they made get flushed before the doUpdate() calls below.
            // This avoids snapshot-clearing errors in LinksUpdate::acquirePageLock().
            $this->loadbalancerFactory->commitAndWaitForReplication(
                __METHOD__, $options['transactionTicket']
            );
        }

        foreach ( $updates as $update ) {
            if ( $update instanceof DataUpdate ) {
                $update->setCause( $causeAction, $causeAgent );
            }
            if ( $update instanceof LinksUpdate ) {
                $update->setRevision( $legacyRevision );
                $update->setTriggeringUser( $triggeringUser );
            }

            if ( $options['defer'] === false ) {
                if ( $update instanceof DataUpdate && $options['transactionTicket'] !== null ) {
                    $update->setTransactionTicket( $options['transactionTicket'] );
                }
                $update->doUpdate();
            } else {
                DeferredUpdates::addUpdate( $update, $options['defer'] );
            }
        }
    }

    public function doParserCacheUpdate() {
        $this->assertHasRevision( __METHOD__ );

        $wikiPage = $this->getWikiPage(); // TODO: ParserCache should accept a RevisionRecord instead

        // NOTE: this may trigger the first parsing of the new content after an edit (when not
        // using pre-generated stashed output).
        // XXX: we may want to use the PoolCounter here. This would perhaps allow the initial parse
        // to be performed post-send. The client could already follow a HTTP redirect to the
        // page view, but would then have to wait for a response until rendering is complete.
        $output = $this->getCanonicalParserOutput();

        // Save it to the parser cache. Use the revision timestamp in the case of a
        // freshly saved edit, as that matches page_touched and a mismatch would trigger an
        // unnecessary reparse.
        $timestamp = $this->options['newrev'] ? $this->revision->getTimestamp()
            : $output->getCacheTime();
        $this->parserCache->save(
            $output, $wikiPage, $this->getCanonicalParserOptions(),
            $timestamp, $this->revision->getId()
        );
    }

}