ExternalStoreDB.php

Go to the documentation of this file.

<?php
use MediaWiki\MediaWikiServices;
use Wikimedia\Rdbms\ILoadBalancer;
use Wikimedia\Rdbms\IDatabase;
use Wikimedia\Rdbms\DBConnRef;
use Wikimedia\Rdbms\MaintainableDBConnRef;
use Wikimedia\Rdbms\DatabaseDomain;
 
class ExternalStoreDB extends ExternalStoreMedium {
    public function fetchFromURL( $url ) {
        list( $cluster, $id, $itemID ) = $this->parseURL( $url );
        $ret = $this->fetchBlob( $cluster, $id, $itemID );
 
        if ( $itemID !== false && $ret !== false ) {
            return $ret->getItem( $itemID );
        }
 
        return $ret;
    }
 
    public function batchFetchFromURLs( array $urls ) {
        $batched = $inverseUrlMap = [];
        foreach ( $urls as $url ) {
            list( $cluster, $id, $itemID ) = $this->parseURL( $url );
            $batched[$cluster][$id][] = $itemID;
            // false $itemID gets cast to int, but should be ok
            // since we do === from the $itemID in $batched
            $inverseUrlMap[$cluster][$id][$itemID] = $url;
        }
        $ret = [];
        foreach ( $batched as $cluster => $batchByCluster ) {
            $res = $this->batchFetchBlobs( $cluster, $batchByCluster );
            foreach ( $res as $id => $blob ) {
                foreach ( $batchByCluster[$id] as $itemID ) {
                    $url = $inverseUrlMap[$cluster][$id][$itemID];
                    if ( $itemID === false ) {
                        $ret[$url] = $blob;
                    } else {
                        $ret[$url] = $blob->getItem( $itemID );
                    }
                }
            }
        }
 
        return $ret;
    }
 
    public function store( $location, $data ) {
        $dbw = $this->getMaster( $location );
        $dbw->insert( $this->getTable( $dbw ),
            [ 'blob_text' => $data ],
            __METHOD__ );
        $id = $dbw->insertId();
        if ( !$id ) {
            throw new MWException( __METHOD__ . ': no insert ID' );
        }
 
        return "DB://$location/$id";
    }
 
    public function isReadOnly( $location ) {
        $lb = $this->getLoadBalancer( $location );
        $domainId = $this->getDomainId( $lb->getServerInfo( $lb->getWriterIndex() ) );
        return ( $lb->getReadOnlyReason( $domainId ) !== false );
    }
 
    private function getLoadBalancer( $cluster ) {
        $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
        return $lbFactory->getExternalLB( $cluster );
    }
 
    public function getSlave( $cluster ) {
        global $wgDefaultExternalStore;
 
        $lb = $this->getLoadBalancer( $cluster );
        $domainId = $this->getDomainId( $lb->getServerInfo( $lb->getWriterIndex() ) );
 
        if ( !in_array( "DB://" . $cluster, (array)$wgDefaultExternalStore ) ) {
            wfDebug( "read only external store\n" );
            $lb->allowLagged( true );
        } else {
            wfDebug( "writable external store\n" );
        }
 
        $db = $lb->getConnectionRef( DB_REPLICA, [], $domainId );
        $db->clearFlag( DBO_TRX ); // sanity
 
        return $db;
    }
 
    public function getMaster( $cluster ) {
        $lb = $this->getLoadBalancer( $cluster );
        $domainId = $this->getDomainId( $lb->getServerInfo( $lb->getWriterIndex() ) );
 
        $db = $lb->getMaintenanceConnectionRef( DB_MASTER, [], $domainId );
        $db->clearFlag( DBO_TRX ); // sanity
 
        return $db;
    }
 
    private function getDomainId( array $server ) {
        if ( isset( $this->params['wiki'] ) && $this->params['wiki'] !== false ) {
            return $this->params['wiki']; // explicit domain
        }
 
        if ( isset( $server['dbname'] ) ) {
            // T200471: for b/c, treat any "dbname" field as forcing which database to use.
            // MediaWiki/LoadBalancer previously did not enforce any concept of a local DB
            // domain, but rather assumed that the LB server configuration matched $wgDBname.
            // This check is useful when the external storage DB for this cluster does not use
            // the same name as the corresponding "main" DB(s) for wikis.
            $domain = new DatabaseDomain(
                $server['dbname'],
                $server['schema'] ?? null,
                $server['tablePrefix'] ?? ''
            );
 
            return $domain->getId();
        }
 
        return false; // local LB domain
    }
 
    public function getTable( $db ) {
        $table = $db->getLBInfo( 'blobs table' );
        if ( is_null( $table ) ) {
            $table = 'blobs';
        }
 
        return $table;
    }
 
    private function fetchBlob( $cluster, $id, $itemID ) {
        static $externalBlobCache = [];
 
        $cacheID = ( $itemID === false ) ? "$cluster/$id" : "$cluster/$id/";
 
        $wiki = $this->params['wiki'] ?? false;
        $cacheID = ( $wiki === false ) ? $cacheID : "$cacheID@$wiki";
 
        if ( isset( $externalBlobCache[$cacheID] ) ) {
            wfDebugLog( 'ExternalStoreDB-cache',
                "ExternalStoreDB::fetchBlob cache hit on $cacheID" );
 
            return $externalBlobCache[$cacheID];
        }
 
        wfDebugLog( 'ExternalStoreDB-cache',
            "ExternalStoreDB::fetchBlob cache miss on $cacheID" );
 
        $dbr = $this->getSlave( $cluster );
        $ret = $dbr->selectField( $this->getTable( $dbr ),
            'blob_text', [ 'blob_id' => $id ], __METHOD__ );
        if ( $ret === false ) {
            wfDebugLog( 'ExternalStoreDB',
                "ExternalStoreDB::fetchBlob master fallback on $cacheID" );
            // Try the master
            $dbw = $this->getMaster( $cluster );
            $ret = $dbw->selectField( $this->getTable( $dbw ),
                'blob_text', [ 'blob_id' => $id ], __METHOD__ );
            if ( $ret === false ) {
                wfDebugLog( 'ExternalStoreDB',
                    "ExternalStoreDB::fetchBlob master failed to find $cacheID" );
            }
        }
        if ( $itemID !== false && $ret !== false ) {
            // Unserialise object; caller extracts item
            $ret = unserialize( $ret );
        }
 
        $externalBlobCache = [ $cacheID => $ret ];
 
        return $ret;
    }
 
    private function batchFetchBlobs( $cluster, array $ids ) {
        $dbr = $this->getSlave( $cluster );
        $res = $dbr->select( $this->getTable( $dbr ),
            [ 'blob_id', 'blob_text' ], [ 'blob_id' => array_keys( $ids ) ], __METHOD__ );
        $ret = [];
        if ( $res !== false ) {
            $this->mergeBatchResult( $ret, $ids, $res );
        }
        if ( $ids ) {
            wfDebugLog( __CLASS__, __METHOD__ .
                " master fallback on '$cluster' for: " .
                implode( ',', array_keys( $ids ) ) );
            // Try the master
            $dbw = $this->getMaster( $cluster );
            $res = $dbw->select( $this->getTable( $dbr ),
                [ 'blob_id', 'blob_text' ],
                [ 'blob_id' => array_keys( $ids ) ],
                __METHOD__ );
            if ( $res === false ) {
                wfDebugLog( __CLASS__, __METHOD__ . " master failed on '$cluster'" );
            } else {
                $this->mergeBatchResult( $ret, $ids, $res );
            }
        }
        if ( $ids ) {
            wfDebugLog( __CLASS__, __METHOD__ .
                " master on '$cluster' failed locating items: " .
                implode( ',', array_keys( $ids ) ) );
        }
 
        return $ret;
    }
 
    private function mergeBatchResult( array &$ret, array &$ids, $res ) {
        foreach ( $res as $row ) {
            $id = $row->blob_id;
            $itemIDs = $ids[$id];
            unset( $ids[$id] ); // to track if everything is found
            if ( count( $itemIDs ) === 1 && reset( $itemIDs ) === false ) {
                // single result stored per blob
                $ret[$id] = $row->blob_text;
            } else {
                // multi result stored per blob
                $ret[$id] = unserialize( $row->blob_text );
            }
        }
    }
 
    protected function parseURL( $url ) {
        $path = explode( '/', $url );
 
        return [
            $path[2], // cluster
            $path[3], // id
            $path[4] ?? false // itemID
        ];
    }
}