QueryPage.php

Go to the documentation of this file.

<?php
use MediaWiki\MediaWikiServices;
use Wikimedia\Rdbms\IResultWrapper;
use Wikimedia\Rdbms\IDatabase;
use Wikimedia\Rdbms\DBError;

abstract class QueryPage extends SpecialPage {
    protected $listoutput = false;

    protected $offset = 0;

    protected $limit = 0;

    protected $numRows;

    protected $cachedTimestamp = null;

    protected $shownavigation = true;

    public static function getPages() {
        static $qp = null;

        if ( $qp === null ) {
            // QueryPage subclass, Special page name
            $qp = [
                [ AncientPagesPage::class, 'Ancientpages' ],
                [ BrokenRedirectsPage::class, 'BrokenRedirects' ],
                [ DeadendPagesPage::class, 'Deadendpages' ],
                [ DoubleRedirectsPage::class, 'DoubleRedirects' ],
                [ FileDuplicateSearchPage::class, 'FileDuplicateSearch' ],
                [ ListDuplicatedFilesPage::class, 'ListDuplicatedFiles' ],
                [ LinkSearchPage::class, 'LinkSearch' ],
                [ ListredirectsPage::class, 'Listredirects' ],
                [ LonelyPagesPage::class, 'Lonelypages' ],
                [ LongPagesPage::class, 'Longpages' ],
                [ MediaStatisticsPage::class, 'MediaStatistics' ],
                [ MIMEsearchPage::class, 'MIMEsearch' ],
                [ MostcategoriesPage::class, 'Mostcategories' ],
                [ MostimagesPage::class, 'Mostimages' ],
                [ MostinterwikisPage::class, 'Mostinterwikis' ],
                [ MostlinkedCategoriesPage::class, 'Mostlinkedcategories' ],
                [ MostlinkedTemplatesPage::class, 'Mostlinkedtemplates' ],
                [ MostlinkedPage::class, 'Mostlinked' ],
                [ MostrevisionsPage::class, 'Mostrevisions' ],
                [ FewestrevisionsPage::class, 'Fewestrevisions' ],
                [ ShortPagesPage::class, 'Shortpages' ],
                [ UncategorizedCategoriesPage::class, 'Uncategorizedcategories' ],
                [ UncategorizedPagesPage::class, 'Uncategorizedpages' ],
                [ UncategorizedImagesPage::class, 'Uncategorizedimages' ],
                [ UncategorizedTemplatesPage::class, 'Uncategorizedtemplates' ],
                [ UnusedCategoriesPage::class, 'Unusedcategories' ],
                [ UnusedimagesPage::class, 'Unusedimages' ],
                [ WantedCategoriesPage::class, 'Wantedcategories' ],
                [ WantedFilesPage::class, 'Wantedfiles' ],
                [ WantedPagesPage::class, 'Wantedpages' ],
                [ WantedTemplatesPage::class, 'Wantedtemplates' ],
                [ UnwatchedpagesPage::class, 'Unwatchedpages' ],
                [ UnusedtemplatesPage::class, 'Unusedtemplates' ],
                [ WithoutInterwikiPage::class, 'Withoutinterwiki' ],
            ];
            Hooks::run( 'wgQueryPages', [ &$qp ] );
        }

        return $qp;
    }

    function setListoutput( $bool ) {
        $this->listoutput = $bool;
    }

    public function getQueryInfo() {
        return null;
    }

    function getSQL() {
        /* Implement getQueryInfo() instead */
        throw new MWException( "Bug in a QueryPage: doesn't implement getQueryInfo() nor "
            . "getQuery() properly" );
    }

    function getOrderFields() {
        return [ 'value' ];
    }

    public function usesTimestamps() {
        return false;
    }

    function sortDescending() {
        return true;
    }

    public function isExpensive() {
        return $this->getConfig()->get( 'DisableQueryPages' );
    }

    public function isCacheable() {
        return true;
    }

    public function isCached() {
        return $this->isExpensive() && $this->getConfig()->get( 'MiserMode' );
    }

    function isSyndicated() {
        return true;
    }

    abstract function formatResult( $skin, $result );

    function getPageHeader() {
        return '';
    }

    protected function showEmptyText() {
        $this->getOutput()->addWikiMsg( 'specialpage-empty' );
    }

    function linkParameters() {
        return [];
    }

    function tryLastResult() {
        return false;
    }

    public function recache( $limit, $ignoreErrors = true ) {
        if ( !$this->isCacheable() ) {
            return 0;
        }

        $fname = static::class . '::recache';
        $dbw = wfGetDB( DB_MASTER );
        if ( !$dbw ) {
            return false;
        }

        try {
            # Do query
            $res = $this->reallyDoQuery( $limit, false );
            $num = false;
            if ( $res ) {
                $num = $res->numRows();
                # Fetch results
                $vals = [];
                foreach ( $res as $row ) {
                    if ( isset( $row->value ) ) {
                        if ( $this->usesTimestamps() ) {
                            $value = wfTimestamp( TS_UNIX,
                                $row->value );
                        } else {
                            $value = intval( $row->value ); // T16414
                        }
                    } else {
                        $value = 0;
                    }

                    $vals[] = [
                        'qc_type' => $this->getName(),
                        'qc_namespace' => $row->namespace,
                        'qc_title' => $row->title,
                        'qc_value' => $value
                    ];
                }

                $dbw->doAtomicSection(
                    __METHOD__,
                    function ( IDatabase $dbw, $fname ) use ( $vals ) {
                        # Clear out any old cached data
                        $dbw->delete( 'querycache',
                            [ 'qc_type' => $this->getName() ],
                            $fname
                        );
                        # Save results into the querycache table on the master
                        if ( count( $vals ) ) {
                            $dbw->insert( 'querycache', $vals, $fname );
                        }
                        # Update the querycache_info record for the page
                        $dbw->delete( 'querycache_info',
                            [ 'qci_type' => $this->getName() ],
                            $fname
                        );
                        $dbw->insert( 'querycache_info',
                            [ 'qci_type' => $this->getName(),
                                'qci_timestamp' => $dbw->timestamp() ],
                            $fname
                        );
                    }
                );
            }
        } catch ( DBError $e ) {
            if ( !$ignoreErrors ) {
                throw $e; // report query error
            }
            $num = false; // set result to false to indicate error
        }

        return $num;
    }

    function getRecacheDB() {
        return wfGetDB( DB_REPLICA, [ $this->getName(), 'QueryPage::recache', 'vslow' ] );
    }

    public function reallyDoQuery( $limit, $offset = false ) {
        $fname = static::class . '::reallyDoQuery';
        $dbr = $this->getRecacheDB();
        $query = $this->getQueryInfo();
        $order = $this->getOrderFields();

        if ( $this->sortDescending() ) {
            foreach ( $order as &$field ) {
                $field .= ' DESC';
            }
        }

        if ( is_array( $query ) ) {
            $tables = isset( $query['tables'] ) ? (array)$query['tables'] : [];
            $fields = isset( $query['fields'] ) ? (array)$query['fields'] : [];
            $conds = isset( $query['conds'] ) ? (array)$query['conds'] : [];
            $options = isset( $query['options'] ) ? (array)$query['options'] : [];
            $join_conds = isset( $query['join_conds'] ) ? (array)$query['join_conds'] : [];

            if ( $order ) {
                $options['ORDER BY'] = $order;
            }

            if ( $limit !== false ) {
                $options['LIMIT'] = intval( $limit );
            }

            if ( $offset !== false ) {
                $options['OFFSET'] = intval( $offset );
            }

            $res = $dbr->select( $tables, $fields, $conds, $fname,
                    $options, $join_conds
            );
        } else {
            // Old-fashioned raw SQL style, deprecated
            $sql = $this->getSQL();
            $sql .= ' ORDER BY ' . implode( ', ', $order );
            $sql = $dbr->limitResult( $sql, $limit, $offset );
            $res = $dbr->query( $sql, $fname );
        }

        return $res;
    }

    public function doQuery( $offset = false, $limit = false ) {
        if ( $this->isCached() && $this->isCacheable() ) {
            return $this->fetchFromCache( $limit, $offset );
        } else {
            return $this->reallyDoQuery( $limit, $offset );
        }
    }

    public function fetchFromCache( $limit, $offset = false ) {
        $dbr = wfGetDB( DB_REPLICA );
        $options = [];

        if ( $limit !== false ) {
            $options['LIMIT'] = intval( $limit );
        }

        if ( $offset !== false ) {
            $options['OFFSET'] = intval( $offset );
        }

        $order = $this->getCacheOrderFields();
        if ( $this->sortDescending() ) {
            foreach ( $order as &$field ) {
                $field .= " DESC";
            }
        }
        if ( $order ) {
            $options['ORDER BY'] = $order;
        }

        return $dbr->select( 'querycache',
                [ 'qc_type',
                'namespace' => 'qc_namespace',
                'title' => 'qc_title',
                'value' => 'qc_value' ],
                [ 'qc_type' => $this->getName() ],
                __METHOD__,
                $options
        );
    }

    function getCacheOrderFields() {
        return [ 'value' ];
    }

    public function getCachedTimestamp() {
        if ( is_null( $this->cachedTimestamp ) ) {
            $dbr = wfGetDB( DB_REPLICA );
            $fname = static::class . '::getCachedTimestamp';
            $this->cachedTimestamp = $dbr->selectField( 'querycache_info', 'qci_timestamp',
                [ 'qci_type' => $this->getName() ], $fname );
        }
        return $this->cachedTimestamp;
    }

    protected function getLimitOffset() {
        list( $limit, $offset ) = $this->getRequest()->getLimitOffset();
        if ( $this->getConfig()->get( 'MiserMode' ) ) {
            $maxResults = $this->getMaxResults();
            // Can't display more than max results on a page
            $limit = min( $limit, $maxResults );
            // Can't skip over more than the end of $maxResults
            $offset = min( $offset, $maxResults + 1 );
        }
        return [ $limit, $offset ];
    }

    protected function getDBLimit( $uiLimit, $uiOffset ) {
        $maxResults = $this->getMaxResults();
        if ( $this->getConfig()->get( 'MiserMode' ) ) {
            $limit = min( $uiLimit + 1, $maxResults - $uiOffset );
            return max( $limit, 0 );
        } else {
            return $uiLimit + 1;
        }
    }

    protected function getMaxResults() {
        // Max of 10000, unless we store more than 10000 in query cache.
        return max( $this->getConfig()->get( 'QueryCacheLimit' ), 10000 );
    }

    public function execute( $par ) {
        $user = $this->getUser();
        if ( !$this->userCanExecute( $user ) ) {
            $this->displayRestrictionError();
            return;
        }

        $this->setHeaders();
        $this->outputHeader();

        $out = $this->getOutput();

        if ( $this->isCached() && !$this->isCacheable() ) {
            $out->addWikiMsg( 'querypage-disabled' );
            return;
        }

        $out->setSyndicated( $this->isSyndicated() );

        if ( $this->limit == 0 && $this->offset == 0 ) {
            list( $this->limit, $this->offset ) = $this->getLimitOffset();
        }
        $dbLimit = $this->getDBLimit( $this->limit, $this->offset );
        // @todo Use doQuery()
        if ( !$this->isCached() ) {
            # select one extra row for navigation
            $res = $this->reallyDoQuery( $dbLimit, $this->offset );
        } else {
            # Get the cached result, select one extra row for navigation
            $res = $this->fetchFromCache( $dbLimit, $this->offset );
            if ( !$this->listoutput ) {
                # Fetch the timestamp of this update
                $ts = $this->getCachedTimestamp();
                $lang = $this->getLanguage();
                $maxResults = $lang->formatNum( $this->getConfig()->get( 'QueryCacheLimit' ) );

                if ( $ts ) {
                    $updated = $lang->userTimeAndDate( $ts, $user );
                    $updateddate = $lang->userDate( $ts, $user );
                    $updatedtime = $lang->userTime( $ts, $user );
                    $out->addMeta( 'Data-Cache-Time', $ts );
                    $out->addJsConfigVars( 'dataCacheTime', $ts );
                    $out->addWikiMsg( 'perfcachedts', $updated, $updateddate, $updatedtime, $maxResults );
                } else {
                    $out->addWikiMsg( 'perfcached', $maxResults );
                }

                # If updates on this page have been disabled, let the user know
                # that the data set won't be refreshed for now
                if ( is_array( $this->getConfig()->get( 'DisableQueryPageUpdate' ) )
                    && in_array( $this->getName(), $this->getConfig()->get( 'DisableQueryPageUpdate' ) )
                ) {
                    $out->wrapWikiMsg(
                        "<div class=\"mw-querypage-no-updates\">\n$1\n</div>",
                        'querypage-no-updates'
                    );
                }
            }
        }

        $this->numRows = $res->numRows();

        $dbr = $this->getRecacheDB();
        $this->preprocessResults( $dbr, $res );

        $out->addHTML( Xml::openElement( 'div', [ 'class' => 'mw-spcontent' ] ) );

        # Top header and navigation
        if ( $this->shownavigation ) {
            $out->addHTML( $this->getPageHeader() );
            if ( $this->numRows > 0 ) {
                $out->addHTML( $this->msg( 'showingresultsinrange' )->numParams(
                    min( $this->numRows, $this->limit ), # do not show the one extra row, if exist
                    $this->offset + 1, ( min( $this->numRows, $this->limit ) + $this->offset ) )->parseAsBlock() );
                # Disable the "next" link when we reach the end
                $miserMaxResults = $this->getConfig()->get( 'MiserMode' )
                    && ( $this->offset + $this->limit >= $this->getMaxResults() );
                $atEnd = ( $this->numRows <= $this->limit ) || $miserMaxResults;
                $paging = $this->getLanguage()->viewPrevNext( $this->getPageTitle( $par ), $this->offset,
                    $this->limit, $this->linkParameters(), $atEnd );
                $out->addHTML( '<p>' . $paging . '</p>' );
            } else {
                # No results to show, so don't bother with "showing X of Y" etc.
                # -- just let the user know and give up now
                $this->showEmptyText();
                $out->addHTML( Xml::closeElement( 'div' ) );
                return;
            }
        }

        # The actual results; specialist subclasses will want to handle this
        # with more than a straight list, so we hand them the info, plus
        # an OutputPage, and let them get on with it
        $this->outputResults( $out,
            $this->getSkin(),
            $dbr, # Should use a ResultWrapper for this
            $res,
            min( $this->numRows, $this->limit ), # do not format the one extra row, if exist
            $this->offset );

        # Repeat the paging links at the bottom
        if ( $this->shownavigation ) {
            $out->addHTML( '<p>' . $paging . '</p>' );
        }

        $out->addHTML( Xml::closeElement( 'div' ) );
    }

    protected function outputResults( $out, $skin, $dbr, $res, $num, $offset ) {
        if ( $num > 0 ) {
            $html = [];
            if ( !$this->listoutput ) {
                $html[] = $this->openList( $offset );
            }

            # $res might contain the whole 1,000 rows, so we read up to
            # $num [should update this to use a Pager]
            // phpcs:ignore Generic.CodeAnalysis.ForLoopWithTestFunctionCall
            for ( $i = 0; $i < $num && $row = $res->fetchObject(); $i++ ) {
                $line = $this->formatResult( $skin, $row );
                if ( $line ) {
                    $html[] = $this->listoutput
                        ? $line
                        : "<li>{$line}</li>\n";
                }
            }

            # Flush the final result
            if ( $this->tryLastResult() ) {
                $row = null;
                $line = $this->formatResult( $skin, $row );
                if ( $line ) {
                    $html[] = $this->listoutput
                        ? $line
                        : "<li>{$line}</li>\n";
                }
            }

            if ( !$this->listoutput ) {
                $html[] = $this->closeList();
            }

            $html = $this->listoutput
                ? MediaWikiServices::getInstance()->getContentLanguage()->listToText( $html )
                : implode( '', $html );

            $out->addHTML( $html );
        }
    }

    function openList( $offset ) {
        return "\n<ol start='" . ( $offset + 1 ) . "' class='special'>\n";
    }

    function closeList() {
        return "</ol>\n";
    }

    function preprocessResults( $db, $res ) {
    }

    function doFeed( $class = '', $limit = 50 ) {
        if ( !$this->getConfig()->get( 'Feed' ) ) {
            $this->getOutput()->addWikiMsg( 'feed-unavailable' );
            return false;
        }

        $limit = min( $limit, $this->getConfig()->get( 'FeedLimit' ) );

        $feedClasses = $this->getConfig()->get( 'FeedClasses' );
        if ( isset( $feedClasses[$class] ) ) {
            $feed = new $feedClasses[$class](
                $this->feedTitle(),
                $this->feedDesc(),
                $this->feedUrl() );
            $feed->outHeader();

            $res = $this->reallyDoQuery( $limit, 0 );
            foreach ( $res as $obj ) {
                $item = $this->feedResult( $obj );
                if ( $item ) {
                    $feed->outItem( $item );
                }
            }

            $feed->outFooter();
            return true;
        } else {
            return false;
        }
    }

    function feedResult( $row ) {
        if ( !isset( $row->title ) ) {
            return null;
        }
        $title = Title::makeTitle( intval( $row->namespace ), $row->title );
        if ( $title ) {
            $date = $row->timestamp ?? '';
            $comments = '';
            if ( $title ) {
                $talkpage = $title->getTalkPage();
                $comments = $talkpage->getFullURL();
            }

            return new FeedItem(
                $title->getPrefixedText(),
                $this->feedItemDesc( $row ),
                $title->getFullURL(),
                $date,
                $this->feedItemAuthor( $row ),
                $comments );
        } else {
            return null;
        }
    }

    function feedItemDesc( $row ) {
        return isset( $row->comment ) ? htmlspecialchars( $row->comment ) : '';
    }

    function feedItemAuthor( $row ) {
        return $row->user_text ?? '';
    }

    function feedTitle() {
        $desc = $this->getDescription();
        $code = $this->getConfig()->get( 'LanguageCode' );
        $sitename = $this->getConfig()->get( 'Sitename' );
        return "$sitename - $desc [$code]";
    }

    function feedDesc() {
        return $this->msg( 'tagline' )->text();
    }

    function feedUrl() {
        return $this->getPageTitle()->getFullURL();
    }

    protected function executeLBFromResultWrapper( IResultWrapper $res, $ns = null ) {
        if ( !$res->numRows() ) {
            return;
        }

        $batch = new LinkBatch;
        foreach ( $res as $row ) {
            $batch->add( $ns ?? $row->namespace, $row->title );
        }
        $batch->execute();

        $res->seek( 0 );
    }
}