BacklinkCache.php

Go to the documentation of this file.

<?php
class BacklinkCache {
    protected static $cache;

    protected $partitionCache = array();

    protected $fullResultCache = array();

    protected $db;

    protected $title;

    const CACHE_EXPIRY = 3600;

    public function __construct( Title $title ) {
        $this->title = $title;
    }

    public static function get( Title $title ) {
        if ( !self::$cache ) { // init cache
            self::$cache = new ProcessCacheLRU( 1 );
        }
        $dbKey = $title->getPrefixedDBkey();
        if ( !self::$cache->has( $dbKey, 'obj', 3600 ) ) {
            self::$cache->set( $dbKey, 'obj', new self( $title ) );
        }
        return self::$cache->get( $dbKey, 'obj' );
    }

    function __sleep() {
        return array( 'partitionCache', 'fullResultCache', 'title' );
    }

    public function clear() {
        $this->partitionCache = array();
        $this->fullResultCache = array();
        unset( $this->db );
    }

    public function setDB( $db ) {
        $this->db = $db;
    }

    protected function getDB() {
        if ( !isset( $this->db ) ) {
            $this->db = wfGetDB( DB_SLAVE );
        }
        return $this->db;
    }

    public function getLinks( $table, $startId = false, $endId = false, $max = INF ) {
        return TitleArray::newFromResult( $this->queryLinks( $table, $startId, $endId, $max ) );
    }

    protected function queryLinks( $table, $startId, $endId, $max ) {
        wfProfileIn( __METHOD__ );

        $fromField = $this->getPrefix( $table ) . '_from';

        if ( !$startId && !$endId && is_infinite( $max )
            && isset( $this->fullResultCache[$table] ) )
        {
            wfDebug( __METHOD__ . ": got results from cache\n" );
            $res = $this->fullResultCache[$table];
        } else {
            wfDebug( __METHOD__ . ": got results from DB\n" );
            $conds = $this->getConditions( $table );
            // Use the from field in the condition rather than the joined page_id,
            // because databases are stupid and don't necessarily propagate indexes.
            if ( $startId ) {
                $conds[] = "$fromField >= " . intval( $startId );
            }
            if ( $endId ) {
                $conds[] = "$fromField <= " . intval( $endId );
            }
            $options = array( 'STRAIGHT_JOIN', 'ORDER BY' => $fromField );
            if ( is_finite( $max ) && $max > 0 ) {
                $options['LIMIT'] = $max;
            }

            $res = $this->getDB()->select(
                array( $table, 'page' ),
                array( 'page_namespace', 'page_title', 'page_id' ),
                $conds,
                __METHOD__,
                $options
            );

            if ( !$startId && !$endId && $res->numRows() < $max ) {
                // The full results fit within the limit, so cache them
                $this->fullResultCache[$table] = $res;
            } else {
                wfDebug( __METHOD__ . ": results from DB were uncacheable\n" );
            }
        }

        wfProfileOut( __METHOD__ );
        return $res;
    }

    protected function getPrefix( $table ) {
        static $prefixes = array(
            'pagelinks' => 'pl',
            'imagelinks' => 'il',
            'categorylinks' => 'cl',
            'templatelinks' => 'tl',
            'redirect' => 'rd',
        );

        if ( isset( $prefixes[$table] ) ) {
            return $prefixes[$table];
        } else {
            $prefix = null;
            wfRunHooks( 'BacklinkCacheGetPrefix', array( $table, &$prefix ) );
            if ( $prefix ) {
                return $prefix;
            } else {
                throw new MWException( "Invalid table \"$table\" in " . __CLASS__ );
            }
        }
    }

    protected function getConditions( $table ) {
        $prefix = $this->getPrefix( $table );

        // @todo FIXME: imagelinks and categorylinks do not rely on getNamespace,
        // they could be moved up for nicer case statements
        switch ( $table ) {
            case 'pagelinks':
            case 'templatelinks':
                $conds = array(
                    "{$prefix}_namespace" => $this->title->getNamespace(),
                    "{$prefix}_title" => $this->title->getDBkey(),
                    "page_id={$prefix}_from"
                );
                break;
            case 'redirect':
                $conds = array(
                    "{$prefix}_namespace" => $this->title->getNamespace(),
                    "{$prefix}_title" => $this->title->getDBkey(),
                    $this->getDb()->makeList( array(
                        "{$prefix}_interwiki" => '',
                        "{$prefix}_interwiki IS NULL",
                    ), LIST_OR ),
                    "page_id={$prefix}_from"
                );
                break;
            case 'imagelinks':
                $conds = array(
                    'il_to' => $this->title->getDBkey(),
                    'page_id=il_from'
                );
                break;
            case 'categorylinks':
                $conds = array(
                    'cl_to' => $this->title->getDBkey(),
                    'page_id=cl_from',
                );
                break;
            default:
                $conds = null;
                wfRunHooks( 'BacklinkCacheGetConditions', array( $table, $this->title, &$conds ) );
                if ( !$conds ) {
                    throw new MWException( "Invalid table \"$table\" in " . __CLASS__ );
                }
        }

        return $conds;
    }

    public function hasLinks( $table ) {
        return ( $this->getNumLinks( $table, 1 ) > 0 );
    }

    public function getNumLinks( $table, $max = INF ) {
        global $wgMemc;

        // 1) try partition cache ...
        if ( isset( $this->partitionCache[$table] ) ) {
            $entry = reset( $this->partitionCache[$table] );
            return min( $max, $entry['numRows'] );
        }

        // 2) ... then try full result cache ...
        if ( isset( $this->fullResultCache[$table] ) ) {
            return min( $max, $this->fullResultCache[$table]->numRows() );
        }

        $memcKey = wfMemcKey( 'numbacklinks', md5( $this->title->getPrefixedDBkey() ), $table );

        // 3) ... fallback to memcached ...
        $count = $wgMemc->get( $memcKey );
        if ( $count ) {
            return min( $max, $count );
        }

        // 4) fetch from the database ...
        $count = $this->getLinks( $table, false, false, $max )->count();
        if ( $count < $max ) { // full count
            $wgMemc->set( $memcKey, $count, self::CACHE_EXPIRY );
        }

        return min( $max, $count );
    }

    public function partition( $table, $batchSize ) {
        global $wgMemc;

        // 1) try partition cache ...
        if ( isset( $this->partitionCache[$table][$batchSize] ) ) {
            wfDebug( __METHOD__ . ": got from partition cache\n" );
            return $this->partitionCache[$table][$batchSize]['batches'];
        }

        $this->partitionCache[$table][$batchSize] = false;
        $cacheEntry =& $this->partitionCache[$table][$batchSize];

        // 2) ... then try full result cache ...
        if ( isset( $this->fullResultCache[$table] ) ) {
            $cacheEntry = $this->partitionResult( $this->fullResultCache[$table], $batchSize );
            wfDebug( __METHOD__ . ": got from full result cache\n" );
            return $cacheEntry['batches'];
        }

        $memcKey = wfMemcKey(
            'backlinks',
            md5( $this->title->getPrefixedDBkey() ),
            $table,
            $batchSize
        );

        // 3) ... fallback to memcached ...
        $memcValue = $wgMemc->get( $memcKey );
        if ( is_array( $memcValue ) ) {
            $cacheEntry = $memcValue;
            wfDebug( __METHOD__ . ": got from memcached $memcKey\n" );
            return $cacheEntry['batches'];
        }

        // 4) ... finally fetch from the slow database :(
        $cacheEntry = array( 'numRows' => 0, 'batches' => array() ); // final result
        // Do the selects in batches to avoid client-side OOMs (bug 43452).
        // Use a LIMIT that plays well with $batchSize to keep equal sized partitions.
        $selectSize = max( $batchSize, 200000 - ( 200000 % $batchSize ) );
        $start = false;
        do {
            $res = $this->queryLinks( $table, $start, false, $selectSize );
            $partitions = $this->partitionResult( $res, $batchSize, false );
            // Merge the link count and range partitions for this chunk
            $cacheEntry['numRows'] += $partitions['numRows'];
            $cacheEntry['batches'] = array_merge( $cacheEntry['batches'], $partitions['batches'] );
            if ( count( $partitions['batches'] ) ) {
                list( $lStart, $lEnd ) = end( $partitions['batches'] );
                $start = $lEnd + 1; // pick up after this inclusive range
            }
        } while ( $partitions['numRows'] >= $selectSize );
        // Make sure the first range has start=false and the last one has end=false
        if ( count( $cacheEntry['batches'] ) ) {
            $cacheEntry['batches'][0][0] = false;
            $cacheEntry['batches'][count( $cacheEntry['batches'] ) - 1][1] = false;
        }

        // Save partitions to memcached
        $wgMemc->set( $memcKey, $cacheEntry, self::CACHE_EXPIRY );

        // Save backlink count to memcached
        $memcKey = wfMemcKey( 'numbacklinks', md5( $this->title->getPrefixedDBkey() ), $table );
        $wgMemc->set( $memcKey, $cacheEntry['numRows'], self::CACHE_EXPIRY );

        wfDebug( __METHOD__ . ": got from database\n" );
        return $cacheEntry['batches'];
    }

    protected function partitionResult( $res, $batchSize, $isComplete = true ) {
        $batches = array();
        $numRows = $res->numRows();
        $numBatches = ceil( $numRows / $batchSize );

        for ( $i = 0; $i < $numBatches; $i++ ) {
            if ( $i == 0 && $isComplete ) {
                $start = false;
            } else {
                $rowNum = $i * $batchSize;
                $res->seek( $rowNum );
                $row = $res->fetchObject();
                $start = (int)$row->page_id;
            }

            if ( $i == ( $numBatches - 1 ) && $isComplete ) {
                $end = false;
            } else {
                $rowNum = min( $numRows - 1, ( $i + 1 ) * $batchSize - 1 );
                $res->seek( $rowNum );
                $row = $res->fetchObject();
                $end = (int)$row->page_id;
            }

            # Sanity check order
            if ( $start && $end && $start > $end ) {
                throw new MWException( __METHOD__ . ': Internal error: query result out of order' );
            }

            $batches[] = array( $start, $end );
        }

        return array( 'numRows' => $numRows, 'batches' => $batches );
    }
}