LinksUpdate.php

Go to the documentation of this file.

<?php
use MediaWiki\MediaWikiServices;
use Wikimedia\ScopedCallback;

class LinksUpdate extends DataUpdate implements EnqueueableDataUpdate {
    // @todo make members protected, but make sure extensions don't break

    public $mId;

    public $mTitle;

    public $mParserOutput;

    public $mLinks;

    public $mImages;

    public $mTemplates;

    public $mExternals;

    public $mCategories;

    public $mInterlangs;

    public $mInterwikis;

    public $mProperties;

    public $mRecursive;

    private $mRevision;

    private $linkInsertions = null;

    private $linkDeletions = null;

    private $propertyInsertions = null;

    private $propertyDeletions = null;

    private $user;

    private $db;

    function __construct( Title $title, ParserOutput $parserOutput, $recursive = true ) {
        parent::__construct();

        $this->mTitle = $title;
        $this->mId = $title->getArticleID( Title::GAID_FOR_UPDATE );

        if ( !$this->mId ) {
            throw new InvalidArgumentException(
                "The Title object yields no ID. Perhaps the page doesn't exist?"
            );
        }

        $this->mParserOutput = $parserOutput;

        $this->mLinks = $parserOutput->getLinks();
        $this->mImages = $parserOutput->getImages();
        $this->mTemplates = $parserOutput->getTemplates();
        $this->mExternals = $parserOutput->getExternalLinks();
        $this->mCategories = $parserOutput->getCategories();
        $this->mProperties = $parserOutput->getProperties();
        $this->mInterwikis = $parserOutput->getInterwikiLinks();

        # Convert the format of the interlanguage links
        # I didn't want to change it in the ParserOutput, because that array is passed all
        # the way back to the skin, so either a skin API break would be required, or an
        # inefficient back-conversion.
        $ill = $parserOutput->getLanguageLinks();
        $this->mInterlangs = [];
        foreach ( $ill as $link ) {
            list( $key, $title ) = explode( ':', $link, 2 );
            $this->mInterlangs[$key] = $title;
        }

        foreach ( $this->mCategories as &$sortkey ) {
            # If the sortkey is longer then 255 bytes,
            # it truncated by DB, and then doesn't get
            # matched when comparing existing vs current
            # categories, causing bug 25254.
            # Also. substr behaves weird when given "".
            if ( $sortkey !== '' ) {
                $sortkey = substr( $sortkey, 0, 255 );
            }
        }

        $this->mRecursive = $recursive;

        Hooks::run( 'LinksUpdateConstructed', [ &$this ] );
    }

    public function doUpdate() {
        if ( $this->ticket ) {
            // Make sure all links update threads see the changes of each other.
            // This handles the case when updates have to batched into several COMMITs.
            $scopedLock = self::acquirePageLock( $this->getDB(), $this->mId );
        }

        Hooks::run( 'LinksUpdate', [ &$this ] );
        $this->doIncrementalUpdate();

        // Commit and release the lock (if set)
        ScopedCallback::consume( $scopedLock );
        // Run post-commit hooks without DBO_TRX
        $this->getDB()->onTransactionIdle(
            function () {
                Hooks::run( 'LinksUpdateComplete', [ &$this, $this->ticket ] );
            },
            __METHOD__
        );
    }

    public static function acquirePageLock( IDatabase $dbw, $pageId, $why = 'atomicity' ) {
        $key = "LinksUpdate:$why:pageid:$pageId";
        $scopedLock = $dbw->getScopedLockAndFlush( $key, __METHOD__, 15 );
        if ( !$scopedLock ) {
            throw new RuntimeException( "Could not acquire lock '$key'." );
        }

        return $scopedLock;
    }

    protected function doIncrementalUpdate() {
        # Page links
        $existing = $this->getExistingLinks();
        $this->linkDeletions = $this->getLinkDeletions( $existing );
        $this->linkInsertions = $this->getLinkInsertions( $existing );
        $this->incrTableUpdate( 'pagelinks', 'pl', $this->linkDeletions, $this->linkInsertions );

        # Image links
        $existing = $this->getExistingImages();
        $imageDeletes = $this->getImageDeletions( $existing );
        $this->incrTableUpdate( 'imagelinks', 'il', $imageDeletes,
            $this->getImageInsertions( $existing ) );

        # Invalidate all image description pages which had links added or removed
        $imageUpdates = $imageDeletes + array_diff_key( $this->mImages, $existing );
        $this->invalidateImageDescriptions( $imageUpdates );

        # External links
        $existing = $this->getExistingExternals();
        $this->incrTableUpdate( 'externallinks', 'el', $this->getExternalDeletions( $existing ),
            $this->getExternalInsertions( $existing ) );

        # Language links
        $existing = $this->getExistingInterlangs();
        $this->incrTableUpdate( 'langlinks', 'll', $this->getInterlangDeletions( $existing ),
            $this->getInterlangInsertions( $existing ) );

        # Inline interwiki links
        $existing = $this->getExistingInterwikis();
        $this->incrTableUpdate( 'iwlinks', 'iwl', $this->getInterwikiDeletions( $existing ),
            $this->getInterwikiInsertions( $existing ) );

        # Template links
        $existing = $this->getExistingTemplates();
        $this->incrTableUpdate( 'templatelinks', 'tl', $this->getTemplateDeletions( $existing ),
            $this->getTemplateInsertions( $existing ) );

        # Category links
        $existing = $this->getExistingCategories();
        $categoryDeletes = $this->getCategoryDeletions( $existing );
        $this->incrTableUpdate( 'categorylinks', 'cl', $categoryDeletes,
            $this->getCategoryInsertions( $existing ) );

        # Invalidate all categories which were added, deleted or changed (set symmetric difference)
        $categoryInserts = array_diff_assoc( $this->mCategories, $existing );
        $categoryUpdates = $categoryInserts + $categoryDeletes;
        $this->invalidateCategories( $categoryUpdates );
        $this->updateCategoryCounts( $categoryInserts, $categoryDeletes );

        # Page properties
        $existing = $this->getExistingProperties();
        $this->propertyDeletions = $this->getPropertyDeletions( $existing );
        $this->incrTableUpdate( 'page_props', 'pp', $this->propertyDeletions,
            $this->getPropertyInsertions( $existing ) );

        # Invalidate the necessary pages
        $this->propertyInsertions = array_diff_assoc( $this->mProperties, $existing );
        $changed = $this->propertyDeletions + $this->propertyInsertions;
        $this->invalidateProperties( $changed );

        # Refresh links of all pages including this page
        # This will be in a separate transaction
        if ( $this->mRecursive ) {
            $this->queueRecursiveJobs();
        }

        # Update the links table freshness for this title
        $this->updateLinksTimestamp();
    }

    protected function queueRecursiveJobs() {
        self::queueRecursiveJobsForTable( $this->mTitle, 'templatelinks' );
        if ( $this->mTitle->getNamespace() == NS_FILE ) {
            // Process imagelinks in case the title is or was a redirect
            self::queueRecursiveJobsForTable( $this->mTitle, 'imagelinks' );
        }

        $bc = $this->mTitle->getBacklinkCache();
        // Get jobs for cascade-protected backlinks for a high priority queue.
        // If meta-templates change to using a new template, the new template
        // should be implicitly protected as soon as possible, if applicable.
        // These jobs duplicate a subset of the above ones, but can run sooner.
        // Which ever runs first generally no-ops the other one.
        $jobs = [];
        foreach ( $bc->getCascadeProtectedLinks() as $title ) {
            $jobs[] = RefreshLinksJob::newPrioritized( $title, [] );
        }
        JobQueueGroup::singleton()->push( $jobs );
    }

    public static function queueRecursiveJobsForTable( Title $title, $table ) {
        if ( $title->getBacklinkCache()->hasLinks( $table ) ) {
            $job = new RefreshLinksJob(
                $title,
                [
                    'table' => $table,
                    'recursive' => true,
                ] + Job::newRootJobParams( // "overall" refresh links job info
                    "refreshlinks:{$table}:{$title->getPrefixedText()}"
                )
            );

            JobQueueGroup::singleton()->push( $job );
        }
    }

    function invalidateCategories( $cats ) {
        PurgeJobUtils::invalidatePages( $this->getDB(), NS_CATEGORY, array_keys( $cats ) );
    }

    function updateCategoryCounts( $added, $deleted ) {
        $a = WikiPage::factory( $this->mTitle );
        $a->updateCategoryCounts(
            array_keys( $added ), array_keys( $deleted )
        );
    }

    function invalidateImageDescriptions( $images ) {
        PurgeJobUtils::invalidatePages( $this->getDB(), NS_FILE, array_keys( $images ) );
    }

    private function incrTableUpdate( $table, $prefix, $deletions, $insertions ) {
        $services = MediaWikiServices::getInstance();
        $bSize = $services->getMainConfig()->get( 'UpdateRowsPerQuery' );
        $factory = $services->getDBLoadBalancerFactory();

        if ( $table === 'page_props' ) {
            $fromField = 'pp_page';
        } else {
            $fromField = "{$prefix}_from";
        }

        $deleteWheres = []; // list of WHERE clause arrays for each DB delete() call
        if ( $table === 'pagelinks' || $table === 'templatelinks' || $table === 'iwlinks' ) {
            $baseKey =  ( $table === 'iwlinks' ) ? 'iwl_prefix' : "{$prefix}_namespace";

            $curBatchSize = 0;
            $curDeletionBatch = [];
            $deletionBatches = [];
            foreach ( $deletions as $ns => $dbKeys ) {
                foreach ( $dbKeys as $dbKey => $unused ) {
                    $curDeletionBatch[$ns][$dbKey] = 1;
                    if ( ++$curBatchSize >= $bSize ) {
                        $deletionBatches[] = $curDeletionBatch;
                        $curDeletionBatch = [];
                        $curBatchSize = 0;
                    }
                }
            }
            if ( $curDeletionBatch ) {
                $deletionBatches[] = $curDeletionBatch;
            }

            foreach ( $deletionBatches as $deletionBatch ) {
                $deleteWheres[] = [
                    $fromField => $this->mId,
                    $this->getDB()->makeWhereFrom2d( $deletionBatch, $baseKey, "{$prefix}_title" )
                ];
            }
        } else {
            if ( $table === 'langlinks' ) {
                $toField = 'll_lang';
            } elseif ( $table === 'page_props' ) {
                $toField = 'pp_propname';
            } else {
                $toField = $prefix . '_to';
            }

            $deletionBatches = array_chunk( array_keys( $deletions ), $bSize );
            foreach ( $deletionBatches as $deletionBatch ) {
                $deleteWheres[] = [ $fromField => $this->mId, $toField => $deletionBatch ];
            }
        }

        foreach ( $deleteWheres as $deleteWhere ) {
            $this->getDB()->delete( $table, $deleteWhere, __METHOD__ );
            $factory->commitAndWaitForReplication(
                __METHOD__, $this->ticket, [ 'wiki' => $this->getDB()->getWikiID() ]
            );
        }

        $insertBatches = array_chunk( $insertions, $bSize );
        foreach ( $insertBatches as $insertBatch ) {
            $this->getDB()->insert( $table, $insertBatch, __METHOD__, 'IGNORE' );
            $factory->commitAndWaitForReplication(
                __METHOD__, $this->ticket, [ 'wiki' => $this->getDB()->getWikiID() ]
            );
        }

        if ( count( $insertions ) ) {
            Hooks::run( 'LinksUpdateAfterInsert', [ $this, $table, $insertions ] );
        }
    }

    private function getLinkInsertions( $existing = [] ) {
        $arr = [];
        foreach ( $this->mLinks as $ns => $dbkeys ) {
            $diffs = isset( $existing[$ns] )
                ? array_diff_key( $dbkeys, $existing[$ns] )
                : $dbkeys;
            foreach ( $diffs as $dbk => $id ) {
                $arr[] = [
                    'pl_from' => $this->mId,
                    'pl_from_namespace' => $this->mTitle->getNamespace(),
                    'pl_namespace' => $ns,
                    'pl_title' => $dbk
                ];
            }
        }

        return $arr;
    }

    private function getTemplateInsertions( $existing = [] ) {
        $arr = [];
        foreach ( $this->mTemplates as $ns => $dbkeys ) {
            $diffs = isset( $existing[$ns] ) ? array_diff_key( $dbkeys, $existing[$ns] ) : $dbkeys;
            foreach ( $diffs as $dbk => $id ) {
                $arr[] = [
                    'tl_from' => $this->mId,
                    'tl_from_namespace' => $this->mTitle->getNamespace(),
                    'tl_namespace' => $ns,
                    'tl_title' => $dbk
                ];
            }
        }

        return $arr;
    }

    private function getImageInsertions( $existing = [] ) {
        $arr = [];
        $diffs = array_diff_key( $this->mImages, $existing );
        foreach ( $diffs as $iname => $dummy ) {
            $arr[] = [
                'il_from' => $this->mId,
                'il_from_namespace' => $this->mTitle->getNamespace(),
                'il_to' => $iname
            ];
        }

        return $arr;
    }

    private function getExternalInsertions( $existing = [] ) {
        $arr = [];
        $diffs = array_diff_key( $this->mExternals, $existing );
        foreach ( $diffs as $url => $dummy ) {
            foreach ( wfMakeUrlIndexes( $url ) as $index ) {
                $arr[] = [
                    'el_id' => $this->getDB()->nextSequenceValue( 'externallinks_el_id_seq' ),
                    'el_from' => $this->mId,
                    'el_to' => $url,
                    'el_index' => $index,
                ];
            }
        }

        return $arr;
    }

    private function getCategoryInsertions( $existing = [] ) {
        global $wgContLang, $wgCategoryCollation;
        $diffs = array_diff_assoc( $this->mCategories, $existing );
        $arr = [];
        foreach ( $diffs as $name => $prefix ) {
            $nt = Title::makeTitleSafe( NS_CATEGORY, $name );
            $wgContLang->findVariantLink( $name, $nt, true );

            if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
                $type = 'subcat';
            } elseif ( $this->mTitle->getNamespace() == NS_FILE ) {
                $type = 'file';
            } else {
                $type = 'page';
            }

            # Treat custom sortkeys as a prefix, so that if multiple
            # things are forced to sort as '*' or something, they'll
            # sort properly in the category rather than in page_id
            # order or such.
            $sortkey = Collation::singleton()->getSortKey(
                $this->mTitle->getCategorySortkey( $prefix ) );

            $arr[] = [
                'cl_from' => $this->mId,
                'cl_to' => $name,
                'cl_sortkey' => $sortkey,
                'cl_timestamp' => $this->getDB()->timestamp(),
                'cl_sortkey_prefix' => $prefix,
                'cl_collation' => $wgCategoryCollation,
                'cl_type' => $type,
            ];
        }

        return $arr;
    }

    private function getInterlangInsertions( $existing = [] ) {
        $diffs = array_diff_assoc( $this->mInterlangs, $existing );
        $arr = [];
        foreach ( $diffs as $lang => $title ) {
            $arr[] = [
                'll_from' => $this->mId,
                'll_lang' => $lang,
                'll_title' => $title
            ];
        }

        return $arr;
    }

    function getPropertyInsertions( $existing = [] ) {
        $diffs = array_diff_assoc( $this->mProperties, $existing );

        $arr = [];
        foreach ( array_keys( $diffs ) as $name ) {
            $arr[] = $this->getPagePropRowData( $name );
        }

        return $arr;
    }

    private function getPagePropRowData( $prop ) {
        global $wgPagePropsHaveSortkey;

        $value = $this->mProperties[$prop];

        $row = [
            'pp_page' => $this->mId,
            'pp_propname' => $prop,
            'pp_value' => $value,
        ];

        if ( $wgPagePropsHaveSortkey ) {
            $row['pp_sortkey'] = $this->getPropertySortKeyValue( $value );
        }

        return $row;
    }

    private function getPropertySortKeyValue( $value ) {
        if ( is_int( $value ) || is_float( $value ) || is_bool( $value ) ) {
            return floatval( $value );
        }

        return null;
    }

    private function getInterwikiInsertions( $existing = [] ) {
        $arr = [];
        foreach ( $this->mInterwikis as $prefix => $dbkeys ) {
            $diffs = isset( $existing[$prefix] )
                ? array_diff_key( $dbkeys, $existing[$prefix] )
                : $dbkeys;

            foreach ( $diffs as $dbk => $id ) {
                $arr[] = [
                    'iwl_from' => $this->mId,
                    'iwl_prefix' => $prefix,
                    'iwl_title' => $dbk
                ];
            }
        }

        return $arr;
    }

    private function getLinkDeletions( $existing ) {
        $del = [];
        foreach ( $existing as $ns => $dbkeys ) {
            if ( isset( $this->mLinks[$ns] ) ) {
                $del[$ns] = array_diff_key( $existing[$ns], $this->mLinks[$ns] );
            } else {
                $del[$ns] = $existing[$ns];
            }
        }

        return $del;
    }

    private function getTemplateDeletions( $existing ) {
        $del = [];
        foreach ( $existing as $ns => $dbkeys ) {
            if ( isset( $this->mTemplates[$ns] ) ) {
                $del[$ns] = array_diff_key( $existing[$ns], $this->mTemplates[$ns] );
            } else {
                $del[$ns] = $existing[$ns];
            }
        }

        return $del;
    }

    private function getImageDeletions( $existing ) {
        return array_diff_key( $existing, $this->mImages );
    }

    private function getExternalDeletions( $existing ) {
        return array_diff_key( $existing, $this->mExternals );
    }

    private function getCategoryDeletions( $existing ) {
        return array_diff_assoc( $existing, $this->mCategories );
    }

    private function getInterlangDeletions( $existing ) {
        return array_diff_assoc( $existing, $this->mInterlangs );
    }

    function getPropertyDeletions( $existing ) {
        return array_diff_assoc( $existing, $this->mProperties );
    }

    private function getInterwikiDeletions( $existing ) {
        $del = [];
        foreach ( $existing as $prefix => $dbkeys ) {
            if ( isset( $this->mInterwikis[$prefix] ) ) {
                $del[$prefix] = array_diff_key( $existing[$prefix], $this->mInterwikis[$prefix] );
            } else {
                $del[$prefix] = $existing[$prefix];
            }
        }

        return $del;
    }

    private function getExistingLinks() {
        $res = $this->getDB()->select( 'pagelinks', [ 'pl_namespace', 'pl_title' ],
            [ 'pl_from' => $this->mId ], __METHOD__ );
        $arr = [];
        foreach ( $res as $row ) {
            if ( !isset( $arr[$row->pl_namespace] ) ) {
                $arr[$row->pl_namespace] = [];
            }
            $arr[$row->pl_namespace][$row->pl_title] = 1;
        }

        return $arr;
    }

    private function getExistingTemplates() {
        $res = $this->getDB()->select( 'templatelinks', [ 'tl_namespace', 'tl_title' ],
            [ 'tl_from' => $this->mId ], __METHOD__ );
        $arr = [];
        foreach ( $res as $row ) {
            if ( !isset( $arr[$row->tl_namespace] ) ) {
                $arr[$row->tl_namespace] = [];
            }
            $arr[$row->tl_namespace][$row->tl_title] = 1;
        }

        return $arr;
    }

    private function getExistingImages() {
        $res = $this->getDB()->select( 'imagelinks', [ 'il_to' ],
            [ 'il_from' => $this->mId ], __METHOD__ );
        $arr = [];
        foreach ( $res as $row ) {
            $arr[$row->il_to] = 1;
        }

        return $arr;
    }

    private function getExistingExternals() {
        $res = $this->getDB()->select( 'externallinks', [ 'el_to' ],
            [ 'el_from' => $this->mId ], __METHOD__ );
        $arr = [];
        foreach ( $res as $row ) {
            $arr[$row->el_to] = 1;
        }

        return $arr;
    }

    private function getExistingCategories() {
        $res = $this->getDB()->select( 'categorylinks', [ 'cl_to', 'cl_sortkey_prefix' ],
            [ 'cl_from' => $this->mId ], __METHOD__ );
        $arr = [];
        foreach ( $res as $row ) {
            $arr[$row->cl_to] = $row->cl_sortkey_prefix;
        }

        return $arr;
    }

    private function getExistingInterlangs() {
        $res = $this->getDB()->select( 'langlinks', [ 'll_lang', 'll_title' ],
            [ 'll_from' => $this->mId ], __METHOD__ );
        $arr = [];
        foreach ( $res as $row ) {
            $arr[$row->ll_lang] = $row->ll_title;
        }

        return $arr;
    }

    private function getExistingInterwikis() {
        $res = $this->getDB()->select( 'iwlinks', [ 'iwl_prefix', 'iwl_title' ],
            [ 'iwl_from' => $this->mId ], __METHOD__ );
        $arr = [];
        foreach ( $res as $row ) {
            if ( !isset( $arr[$row->iwl_prefix] ) ) {
                $arr[$row->iwl_prefix] = [];
            }
            $arr[$row->iwl_prefix][$row->iwl_title] = 1;
        }

        return $arr;
    }

    private function getExistingProperties() {
        $res = $this->getDB()->select( 'page_props', [ 'pp_propname', 'pp_value' ],
            [ 'pp_page' => $this->mId ], __METHOD__ );
        $arr = [];
        foreach ( $res as $row ) {
            $arr[$row->pp_propname] = $row->pp_value;
        }

        return $arr;
    }

    public function getTitle() {
        return $this->mTitle;
    }

    public function getParserOutput() {
        return $this->mParserOutput;
    }

    public function getImages() {
        return $this->mImages;
    }

    public function setRevision( Revision $revision ) {
        $this->mRevision = $revision;
    }

    public function getRevision() {
        return $this->mRevision;
    }

    public function setTriggeringUser( User $user ) {
        $this->user = $user;
    }

    public function getTriggeringUser() {
        return $this->user;
    }

    private function invalidateProperties( $changed ) {
        global $wgPagePropLinkInvalidations;

        foreach ( $changed as $name => $value ) {
            if ( isset( $wgPagePropLinkInvalidations[$name] ) ) {
                $inv = $wgPagePropLinkInvalidations[$name];
                if ( !is_array( $inv ) ) {
                    $inv = [ $inv ];
                }
                foreach ( $inv as $table ) {
                    DeferredUpdates::addUpdate( new HTMLCacheUpdate( $this->mTitle, $table ) );
                }
            }
        }
    }

    public function getAddedLinks() {
        if ( $this->linkInsertions === null ) {
            return null;
        }
        $result = [];
        foreach ( $this->linkInsertions as $insertion ) {
            $result[] = Title::makeTitle( $insertion['pl_namespace'], $insertion['pl_title'] );
        }

        return $result;
    }

    public function getRemovedLinks() {
        if ( $this->linkDeletions === null ) {
            return null;
        }
        $result = [];
        foreach ( $this->linkDeletions as $ns => $titles ) {
            foreach ( $titles as $title => $unused ) {
                $result[] = Title::makeTitle( $ns, $title );
            }
        }

        return $result;
    }

    public function getAddedProperties() {
        return $this->propertyInsertions;
    }

    public function getRemovedProperties() {
        return $this->propertyDeletions;
    }

    private function updateLinksTimestamp() {
        if ( $this->mId ) {
            // The link updates made here only reflect the freshness of the parser output
            $timestamp = $this->mParserOutput->getCacheTime();
            $this->getDB()->update( 'page',
                [ 'page_links_updated' => $this->getDB()->timestamp( $timestamp ) ],
                [ 'page_id' => $this->mId ],
                __METHOD__
            );
        }
    }

    private function getDB() {
        if ( !$this->db ) {
            $this->db = wfGetDB( DB_MASTER );
        }

        return $this->db;
    }

    public function getAsJobSpecification() {
        if ( $this->user ) {
            $userInfo = [
                'userId' => $this->user->getId(),
                'userName' => $this->user->getName(),
            ];
        } else {
            $userInfo = false;
        }

        if ( $this->mRevision ) {
            $triggeringRevisionId = $this->mRevision->getId();
        } else {
            $triggeringRevisionId = false;
        }

        return [
            'wiki' => $this->getDB()->getWikiID(),
            'job'  => new JobSpecification(
                'refreshLinksPrioritized',
                [
                    // Reuse the parser cache if it was saved
                    'rootJobTimestamp' => $this->mParserOutput->getCacheTime(),
                    'useRecursiveLinksUpdate' => $this->mRecursive,
                    'triggeringUser' => $userInfo,
                    'triggeringRevisionId' => $triggeringRevisionId,
                ],
                [ 'removeDuplicates' => true ],
                $this->getTitle()
            )
        ];
    }
}