RevisionOutputCache.php

Go to the documentation of this file.

<?php
namespace MediaWiki\Parser;
 
use InvalidArgumentException;
use JsonException;
use MediaWiki\Json\JsonCodec;
use MediaWiki\Revision\RevisionRecord;
use MediaWiki\Revision\SlotRecord;
use MediaWiki\Utils\MWTimestamp;
use Psr\Log\LoggerInterface;
use Wikimedia\ObjectCache\WANObjectCache;
use Wikimedia\Stats\StatsFactory;
use Wikimedia\UUID\GlobalIdGenerator;
 
class RevisionOutputCache {
 
    private $name;
 
    private $cache;
 
    private $cacheEpoch;
 
    private $cacheExpiry;
 
    private $jsonCodec;
 
    private $stats;
 
    private $logger;
 
    private GlobalIdGenerator $globalIdGenerator;
 
    public function __construct(
        string $name,
        WANObjectCache $cache,
        int $cacheExpiry,
        string $cacheEpoch,
        JsonCodec $jsonCodec,
        StatsFactory $stats,
        LoggerInterface $logger,
        GlobalIdGenerator $globalIdGenerator
    ) {
        $this->name = $name;
        $this->cache = $cache;
        $this->cacheExpiry = $cacheExpiry;
        $this->cacheEpoch = $cacheEpoch;
        $this->jsonCodec = $jsonCodec;
        $this->stats = $stats;
        $this->logger = $logger;
        $this->globalIdGenerator = $globalIdGenerator;
    }
 
    private function getContentModelFromRevision( RevisionRecord $revision ) {
        if ( !$revision->hasSlot( SlotRecord::MAIN ) ) {
            return 'missing';
        }
        return str_replace( '.', '_', $revision->getMainContentModel() );
    }
 
    private function incrementStats( RevisionRecord $revision, string $status, ?string $reason = null ) {
        $contentModel = $this->getContentModelFromRevision( $revision );
        $metricSuffix = $reason ? "{$status}_{$reason}" : $status;
 
        $this->stats->getCounter( 'RevisionOutputCache_operation_total' )
            ->setLabel( 'name', $this->name )
            ->setLabel( 'contentModel', $contentModel )
            ->setLabel( 'status', $status )
            ->setLabel( 'reason', $reason ?: 'n/a' )
            ->copyToStatsdAt( "RevisionOutputCache.{$this->name}.{$metricSuffix}" )
            ->increment();
    }
 
    private function incrementRenderReasonStats( RevisionRecord $revision, $renderReason ) {
        $contentModel = $this->getContentModelFromRevision( $revision );
        $renderReason = preg_replace( '/\W+/', '_', $renderReason );
 
        $this->stats->getCounter( 'RevisionOutputCache_render_total' )
            ->setLabel( 'name', $this->name )
            ->setLabel( 'contentModel', $contentModel )
            ->setLabel( 'reason', $renderReason )
            ->increment();
    }
 
    public function makeParserOutputKey(
        RevisionRecord $revision,
        ParserOptions $options,
        ?array $usedOptions = null
    ): string {
        $usedOptions = ParserOptions::allCacheVaryingOptions();
 
        $revId = $revision->getId();
        if ( !$revId ) {
            // If RevId is null, this would probably be unsafe to use as a cache key.
            throw new InvalidArgumentException( "Revision must have an id number" );
        }
        $hash = $options->optionsHash( $usedOptions );
        return $this->cache->makeKey( $this->name, $revId, $hash );
    }
 
    public function makeParserOutputKeyOptionalRevId(
        RevisionRecord $revision,
        ParserOptions $options,
        ?array $usedOptions = null
    ): string {
        $usedOptions = ParserOptions::allCacheVaryingOptions();
 
        // revId may be null.
        $revId = (string)$revision->getId();
        $hash = $options->optionsHash( $usedOptions );
        return $this->cache->makeKey( $this->name, $revId, $hash );
    }
 
    public function get( RevisionRecord $revision, ParserOptions $parserOptions ) {
        if ( $this->cacheExpiry <= 0 ) {
            // disabled
            return false;
        }
 
        if ( !$parserOptions->isSafeToCache() ) {
            $this->incrementStats( $revision, 'miss', 'unsafe' );
            return false;
        }
 
        $cacheKey = $this->makeParserOutputKey( $revision, $parserOptions );
        $json = $this->cache->get( $cacheKey );
 
        if ( $json === false ) {
            $this->incrementStats( $revision, 'miss', 'absent' );
            return false;
        }
 
        $output = $this->restoreFromJson( $json, $cacheKey );
        if ( $output === null ) {
            $this->incrementStats( $revision, 'miss', 'unserialize' );
            return false;
        }
 
        $cacheTime = (int)MWTimestamp::convert( TS_UNIX, $output->getCacheTime() );
        $expiryTime = (int)MWTimestamp::convert( TS_UNIX, $this->cacheEpoch );
        $expiryTime = max( $expiryTime, (int)MWTimestamp::now( TS_UNIX ) - $this->cacheExpiry );
 
        if ( $cacheTime < $expiryTime ) {
            $this->incrementStats( $revision, 'miss', 'expired' );
            return false;
        }
 
        $this->logger->debug( 'old-revision cache hit' );
        $this->incrementStats( $revision, 'hit' );
        return $output;
    }
 
    public function save(
        ParserOutput $output,
        RevisionRecord $revision,
        ParserOptions $parserOptions,
        ?string $cacheTime = null
    ) {
        if ( !$output->hasText() ) {
            throw new InvalidArgumentException( 'Attempt to cache a ParserOutput with no text set!' );
        }
 
        if ( $this->cacheExpiry <= 0 ) {
            // disabled
            return;
        }
 
        $cacheKey = $this->makeParserOutputKey( $revision, $parserOptions );
 
        // Ensure cache properties are set in the ParserOutput
        // T350538: These should be turned into assertions that the
        // properties are already present (and the $cacheTime argument
        // removed).
        if ( $cacheTime ) {
            $output->setCacheTime( $cacheTime );
        } else {
            $cacheTime = $output->getCacheTime();
        }
        if ( !$output->getCacheRevisionId() ) {
            $output->setCacheRevisionId( $revision->getId() );
        }
        if ( !$output->getRenderId() ) {
            $output->setRenderId( $this->globalIdGenerator->newUUIDv1() );
        }
        if ( !$output->getRevisionTimestamp() ) {
            $output->setRevisionTimestamp( $revision->getTimestamp() );
        }
 
        $msg = "Saved in RevisionOutputCache with key $cacheKey" .
            " and timestamp $cacheTime" .
            " and revision id {$revision->getId()}.";
 
        $output->addCacheMessage( $msg );
 
        // The ParserOutput might be dynamic and have been marked uncacheable by the parser.
        $output->updateCacheExpiry( $this->cacheExpiry );
 
        $expiry = $output->getCacheExpiry();
        if ( $expiry <= 0 ) {
            $this->incrementStats( $revision, 'save', 'uncacheable' );
            return;
        }
 
        if ( !$parserOptions->isSafeToCache() ) {
            $this->incrementStats( $revision, 'save', 'unsafe' );
            return;
        }
 
        $json = $this->encodeAsJson( $output, $cacheKey );
        if ( $json === null ) {
            $this->incrementStats( $revision, 'save', 'nonserializable' );
            return;
        }
 
        $this->cache->set( $cacheKey, $json, $expiry );
        $this->incrementStats( $revision, 'save', 'success' );
        $this->incrementRenderReasonStats( $revision, $parserOptions->getRenderReason() );
    }
 
    private function restoreFromJson( string $jsonData, string $key ) {
        try {
            $obj = $this->jsonCodec->deserialize( $jsonData, ParserOutput::class );
            return $obj;
        } catch ( JsonException $e ) {
            $this->logger->error( 'Unable to deserialize JSON', [
                'name' => $this->name,
                'cache_key' => $key,
                'message' => $e->getMessage()
            ] );
            return null;
        }
    }
 
    private function encodeAsJson( CacheTime $obj, string $key ) {
        try {
            return $this->jsonCodec->serialize( $obj );
        } catch ( JsonException $e ) {
            $this->logger->error( 'Unable to serialize JSON', [
                'name' => $this->name,
                'cache_key' => $key,
                'message' => $e->getMessage(),
            ] );
            return null;
        }
    }
}