LinksUpdate.php

Go to the documentation of this file.

<?php
class LinksUpdate extends SqlDataUpdate {
    // @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 $mProperties;

    public $mDb;

    public $mOptions;

    public $mRecursive;

    private $linkInsertions = null;

    private $linkDeletions = null;

    function __construct( $title, $parserOutput, $recursive = true ) {
        parent::__construct( false ); // no implicit transaction

        if ( !( $title instanceof Title ) ) {
            throw new MWException( "The calling convention to LinksUpdate::LinksUpdate() has changed. " .
                "Please see Article::editUpdates() for an invocation example.\n" );
        }

        if ( !( $parserOutput instanceof ParserOutput ) ) {
            throw new MWException( "The calling convention to LinksUpdate::__construct() has changed. " .
                "Please see WikiPage::doEditUpdates() for an invocation example.\n" );
        }

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

        if ( !$this->mId ) {
            throw new MWException( "The Title object did not provide an article " .
                "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 = array();
        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;

        wfRunHooks( 'LinksUpdateConstructed', array( &$this ) );
    }

    public function doUpdate() {
        wfRunHooks( 'LinksUpdate', array( &$this ) );
        $this->doIncrementalUpdate();
        wfRunHooks( 'LinksUpdateComplete', array( &$this ) );
    }

    protected function doIncrementalUpdate() {
        wfProfileIn( __METHOD__ );

        # 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();

        $propertiesDeletes = $this->getPropertyDeletions( $existing );

        $this->incrTableUpdate( 'page_props', 'pp', $propertiesDeletes,
            $this->getPropertyInsertions( $existing ) );

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

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

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

        wfProfileOut( __METHOD__ );
    }

    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' );
        }
    }

    public static function queueRecursiveJobsForTable( Title $title, $table ) {
        wfProfileIn( __METHOD__ );
        if ( $title->getBacklinkCache()->hasLinks( $table ) ) {
            $job = new RefreshLinksJob(
                $title,
                array(
                    'table' => $table,
                    'recursive' => true,
                ) + Job::newRootJobParams( // "overall" refresh links job info
                    "refreshlinks:{$table}:{$title->getPrefixedText()}"
                )
            );
            JobQueueGroup::singleton()->push( $job );
            JobQueueGroup::singleton()->deduplicateRootJob( $job );
        }
        wfProfileOut( __METHOD__ );
    }

    function invalidateCategories( $cats ) {
        $this->invalidatePages( 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 ) {
        $this->invalidatePages( NS_FILE, array_keys( $images ) );
    }

    function incrTableUpdate( $table, $prefix, $deletions, $insertions ) {
        if ( $table == 'page_props' ) {
            $fromField = 'pp_page';
        } else {
            $fromField = "{$prefix}_from";
        }
        $where = array( $fromField => $this->mId );
        if ( $table == 'pagelinks' || $table == 'templatelinks' || $table == 'iwlinks' ) {
            if ( $table == 'iwlinks' ) {
                $baseKey = 'iwl_prefix';
            } else {
                $baseKey = "{$prefix}_namespace";
            }
            $clause = $this->mDb->makeWhereFrom2d( $deletions, $baseKey, "{$prefix}_title" );
            if ( $clause ) {
                $where[] = $clause;
            } else {
                $where = false;
            }
        } else {
            if ( $table == 'langlinks' ) {
                $toField = 'll_lang';
            } elseif ( $table == 'page_props' ) {
                $toField = 'pp_propname';
            } else {
                $toField = $prefix . '_to';
            }
            if ( count( $deletions ) ) {
                $where[$toField] = array_keys( $deletions );
            } else {
                $where = false;
            }
        }
        if ( $where ) {
            $this->mDb->delete( $table, $where, __METHOD__ );
        }
        if ( count( $insertions ) ) {
            $this->mDb->insert( $table, $insertions, __METHOD__, 'IGNORE' );
            wfRunHooks( 'LinksUpdateAfterInsert', array( $this, $table, $insertions ) );
        }
    }

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

        return $arr;
    }

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

        return $arr;
    }

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

        return $arr;
    }

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

        return $arr;
    }

    private function getCategoryInsertions( $existing = array() ) {
        global $wgContLang, $wgCategoryCollation;
        $diffs = array_diff_assoc( $this->mCategories, $existing );
        $arr = array();
        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[] = array(
                'cl_from' => $this->mId,
                'cl_to' => $name,
                'cl_sortkey' => $sortkey,
                'cl_timestamp' => $this->mDb->timestamp(),
                'cl_sortkey_prefix' => $prefix,
                'cl_collation' => $wgCategoryCollation,
                'cl_type' => $type,
            );
        }

        return $arr;
    }

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

        return $arr;
    }

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

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

        return $arr;
    }

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

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

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

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

        return $arr;
    }

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

        return $arr;
    }

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

        return $arr;
    }

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

        return $arr;
    }

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

        return $arr;
    }

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

        return $arr;
    }

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

        return $arr;
    }

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

        return $arr;
    }

    private function getExistingProperties() {
        $res = $this->mDb->select( 'page_props', array( 'pp_propname', 'pp_value' ),
            array( 'pp_page' => $this->mId ), __METHOD__, $this->mOptions );
        $arr = array();
        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;
    }

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

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

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

        return $result;
    }

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

class LinksDeletionUpdate extends SqlDataUpdate {
    protected $mPage;

    function __construct( WikiPage $page ) {
        parent::__construct( false ); // no implicit transaction

        $this->mPage = $page;

        if ( !$page->exists() ) {
            throw new MWException( "Page ID not known, perhaps the page doesn't exist?" );
        }
    }

    public function doUpdate() {
        $title = $this->mPage->getTitle();
        $id = $this->mPage->getId();

        # Delete restrictions for it
        $this->mDb->delete( 'page_restrictions', array( 'pr_page' => $id ), __METHOD__ );

        # Fix category table counts
        $cats = array();
        $res = $this->mDb->select( 'categorylinks', 'cl_to', array( 'cl_from' => $id ), __METHOD__ );

        foreach ( $res as $row ) {
            $cats[] = $row->cl_to;
        }

        $this->mPage->updateCategoryCounts( array(), $cats );

        # If using cascading deletes, we can skip some explicit deletes
        if ( !$this->mDb->cascadingDeletes() ) {
            # Delete outgoing links
            $this->mDb->delete( 'pagelinks', array( 'pl_from' => $id ), __METHOD__ );
            $this->mDb->delete( 'imagelinks', array( 'il_from' => $id ), __METHOD__ );
            $this->mDb->delete( 'categorylinks', array( 'cl_from' => $id ), __METHOD__ );
            $this->mDb->delete( 'templatelinks', array( 'tl_from' => $id ), __METHOD__ );
            $this->mDb->delete( 'externallinks', array( 'el_from' => $id ), __METHOD__ );
            $this->mDb->delete( 'langlinks', array( 'll_from' => $id ), __METHOD__ );
            $this->mDb->delete( 'iwlinks', array( 'iwl_from' => $id ), __METHOD__ );
            $this->mDb->delete( 'redirect', array( 'rd_from' => $id ), __METHOD__ );
            $this->mDb->delete( 'page_props', array( 'pp_page' => $id ), __METHOD__ );
        }

        # If using cleanup triggers, we can skip some manual deletes
        if ( !$this->mDb->cleanupTriggers() ) {
            # Clean up recentchanges entries...
            $this->mDb->delete( 'recentchanges',
                array( 'rc_type != ' . RC_LOG,
                    'rc_namespace' => $title->getNamespace(),
                    'rc_title' => $title->getDBkey() ),
                __METHOD__ );
            $this->mDb->delete( 'recentchanges',
                array( 'rc_type != ' . RC_LOG, 'rc_cur_id' => $id ),
                __METHOD__ );
        }
    }

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