PageEditStash.php

Go to the documentation of this file.

<?php
namespace MediaWiki\Storage;
 
use BagOStuff;
use Content;
use Liuggio\StatsdClient\Factory\StatsdDataFactoryInterface;
use MediaWiki\HookContainer\HookContainer;
use MediaWiki\HookContainer\HookRunner;
use MediaWiki\Page\PageIdentity;
use MediaWiki\Page\WikiPageFactory;
use MediaWiki\Parser\ParserOutput;
use MediaWiki\Parser\ParserOutputFlags;
use MediaWiki\Revision\SlotRecord;
use MediaWiki\Storage\Hook\ParserOutputStashForEditHook;
use MediaWiki\User\UserEditTracker;
use MediaWiki\User\UserFactory;
use MediaWiki\User\UserIdentity;
use Psr\Log\LoggerInterface;
use stdClass;
use Wikimedia\Rdbms\IConnectionProvider;
use Wikimedia\ScopedCallback;
use WikiPage;
 
class PageEditStash {
    private $cache;
    private $dbProvider;
    private $logger;
    private $stats;
    private $hookRunner;
    private $userEditTracker;
    private $userFactory;
    private $wikiPageFactory;
    private $initiator;
 
    public const ERROR_NONE = 'stashed';
    public const ERROR_PARSE = 'error_parse';
    public const ERROR_CACHE = 'error_cache';
    public const ERROR_UNCACHEABLE = 'uncacheable';
    public const ERROR_BUSY = 'busy';
 
    public const PRESUME_FRESH_TTL_SEC = 30;
    public const MAX_CACHE_TTL = 300; // 5 minutes
    public const MAX_SIGNATURE_TTL = 60;
 
    private const MAX_CACHE_RECENT = 2;
 
    public const INITIATOR_USER = 1;
    public const INITIATOR_JOB_OR_CLI = 2;
 
    public function __construct(
        BagOStuff $cache,
        IConnectionProvider $dbProvider,
        LoggerInterface $logger,
        StatsdDataFactoryInterface $stats,
        UserEditTracker $userEditTracker,
        UserFactory $userFactory,
        WikiPageFactory $wikiPageFactory,
        HookContainer $hookContainer,
        $initiator
    ) {
        $this->cache = $cache;
        $this->dbProvider = $dbProvider;
        $this->logger = $logger;
        $this->stats = $stats;
        $this->userEditTracker = $userEditTracker;
        $this->userFactory = $userFactory;
        $this->wikiPageFactory = $wikiPageFactory;
        $this->hookRunner = new HookRunner( $hookContainer );
        $this->initiator = $initiator;
    }
 
    public function parseAndCache( $pageUpdater, Content $content, UserIdentity $user, string $summary ) {
        $logger = $this->logger;
 
        if ( $pageUpdater instanceof WikiPage ) {
            wfDeprecated( __METHOD__ . ' with WikiPage instance', '1.42' );
            $pageUpdater = $pageUpdater->newPageUpdater( $user );
        }
 
        $page = $pageUpdater->getPage();
        $key = $this->getStashKey( $page, $this->getContentHash( $content ), $user );
        $fname = __METHOD__;
 
        // Use the primary DB to allow for fast blocking locks on the "save path" where this
        // value might actually be used to complete a page edit. If the edit submission request
        // happens before this edit stash requests finishes, then the submission will block until
        // the stash request finishes parsing. For the lock acquisition below, there is not much
        // need to duplicate parsing of the same content/user/summary bundle, so try to avoid
        // blocking at all here.
        $dbw = $this->dbProvider->getPrimaryDatabase();
        if ( !$dbw->lock( $key, $fname, 0 ) ) {
            // De-duplicate requests on the same key
            return self::ERROR_BUSY;
        }
        $unlocker = new ScopedCallback( static function () use ( $dbw, $key, $fname ) {
            $dbw->unlock( $key, $fname );
        } );
 
        $cutoffTime = time() - self::PRESUME_FRESH_TTL_SEC;
 
        // Reuse any freshly build matching edit stash cache
        $editInfo = $this->getStashValue( $key );
        if ( $editInfo && (int)wfTimestamp( TS_UNIX, $editInfo->timestamp ) >= $cutoffTime ) {
            $alreadyCached = true;
        } else {
            $pageUpdater->setContent( SlotRecord::MAIN, $content );
 
            $update = $pageUpdater->prepareUpdate( EDIT_INTERNAL ); // applies pre-safe transform
            $output = $update->getCanonicalParserOutput(); // causes content to be parsed
            $output->setCacheTime( $update->getRevision()->getTimestamp() );
 
            // emulate a cache value that kind of looks like a PreparedEdit, for use below
            $editInfo = (object)[
                'pstContent' => $update->getRawContent( SlotRecord::MAIN ),
                'output'     => $output,
                'timestamp'  => $output->getCacheTime()
            ];
 
            $alreadyCached = false;
        }
 
        $logContext = [ 'cachekey' => $key, 'title' => (string)$page ];
 
        if ( $editInfo->output ) {
            // Let extensions add ParserOutput metadata or warm other caches
            $legacyUser = $this->userFactory->newFromUserIdentity( $user );
            $legacyPage = $this->wikiPageFactory->newFromTitle( $page );
            $this->hookRunner->onParserOutputStashForEdit(
                $legacyPage, $content, $editInfo->output, $summary, $legacyUser );
 
            if ( $alreadyCached ) {
                $logger->debug( "Parser output for key '{cachekey}' already cached.", $logContext );
 
                return self::ERROR_NONE;
            }
 
            $code = $this->storeStashValue(
                $key,
                $editInfo->pstContent,
                $editInfo->output,
                $editInfo->timestamp,
                $user
            );
 
            if ( $code === true ) {
                $logger->debug( "Cached parser output for key '{cachekey}'.", $logContext );
 
                return self::ERROR_NONE;
            } elseif ( $code === 'uncacheable' ) {
                $logger->info(
                    "Uncacheable parser output for key '{cachekey}' [{code}].",
                    $logContext + [ 'code' => $code ]
                );
 
                return self::ERROR_UNCACHEABLE;
            } else {
                $logger->error(
                    "Failed to cache parser output for key '{cachekey}'.",
                    $logContext + [ 'code' => $code ]
                );
 
                return self::ERROR_CACHE;
            }
        }
 
        return self::ERROR_PARSE;
    }
 
    public function checkCache( PageIdentity $page, Content $content, UserIdentity $user ) {
        $legacyUser = $this->userFactory->newFromUserIdentity( $user );
        if (
            // The context is not an HTTP POST request
            !$legacyUser->getRequest()->wasPosted() ||
            // The context is a CLI script or a job runner HTTP POST request
            $this->initiator !== self::INITIATOR_USER ||
            // The editor account is a known bot
            $legacyUser->isBot()
        ) {
            // Avoid wasted queries and statsd pollution
            return false;
        }
 
        $logger = $this->logger;
 
        $key = $this->getStashKey( $page, $this->getContentHash( $content ), $user );
        $logContext = [
            'key' => $key,
            'title' => (string)$page,
            'user' => $user->getName()
        ];
 
        $editInfo = $this->getAndWaitForStashValue( $key );
        if ( !is_object( $editInfo ) || !$editInfo->output ) {
            $this->incrStatsByContent( 'cache_misses.no_stash', $content );
            if ( $this->recentStashEntryCount( $user ) > 0 ) {
                $logger->info( "Empty cache for key '{key}' but not for user.", $logContext );
            } else {
                $logger->debug( "Empty cache for key '{key}'.", $logContext );
            }
 
            return false;
        }
 
        $age = time() - (int)wfTimestamp( TS_UNIX, $editInfo->output->getCacheTime() );
        $logContext['age'] = $age;
 
        $isCacheUsable = true;
        if ( $age <= self::PRESUME_FRESH_TTL_SEC ) {
            // Assume nothing changed in this time
            $this->incrStatsByContent( 'cache_hits.presumed_fresh', $content );
            $logger->debug( "Timestamp-based cache hit for key '{key}'.", $logContext );
        } elseif ( !$user->isRegistered() ) {
            $lastEdit = $this->lastEditTime( $user );
            $cacheTime = $editInfo->output->getCacheTime();
            if ( $lastEdit < $cacheTime ) {
                // Logged-out user made no local upload/template edits in the meantime
                $this->incrStatsByContent( 'cache_hits.presumed_fresh', $content );
                $logger->debug( "Edit check based cache hit for key '{key}'.", $logContext );
            } else {
                $isCacheUsable = false;
                $this->incrStatsByContent( 'cache_misses.proven_stale', $content );
                $logger->info( "Stale cache for key '{key}' due to outside edits.", $logContext );
            }
        } else {
            if ( $editInfo->edits === $this->userEditTracker->getUserEditCount( $user ) ) {
                // Logged-in user made no local upload/template edits in the meantime
                $this->incrStatsByContent( 'cache_hits.presumed_fresh', $content );
                $logger->debug( "Edit count based cache hit for key '{key}'.", $logContext );
            } else {
                $isCacheUsable = false;
                $this->incrStatsByContent( 'cache_misses.proven_stale', $content );
                $logger->info( "Stale cache for key '{key}'due to outside edits.", $logContext );
            }
        }
 
        if ( !$isCacheUsable ) {
            return false;
        }
 
        if ( $editInfo->output->getOutputFlag( ParserOutputFlags::VARY_REVISION ) ) {
            // This can be used for the initial parse, e.g. for filters or doUserEditContent(),
            // but a second parse will be triggered in doEditUpdates() no matter what
            $logger->info(
                "Cache for key '{key}' has vary-revision; post-insertion parse inevitable.",
                $logContext
            );
        } else {
            static $flagsMaybeReparse = [
                // Similar to the above if we didn't guess the ID correctly
                ParserOutputFlags::VARY_REVISION_ID,
                // Similar to the above if we didn't guess the timestamp correctly
                ParserOutputFlags::VARY_REVISION_TIMESTAMP,
                // Similar to the above if we didn't guess the content correctly
                ParserOutputFlags::VARY_REVISION_SHA1,
                // Similar to the above if we didn't guess page ID correctly
                ParserOutputFlags::VARY_PAGE_ID,
            ];
            foreach ( $flagsMaybeReparse as $flag ) {
                if ( $editInfo->output->getOutputFlag( $flag ) ) {
                    $logger->debug(
                        "Cache for key '{key}' has $flag; post-insertion parse possible.",
                        $logContext
                    );
                }
            }
        }
 
        return $editInfo;
    }
 
    private function incrStatsByContent( $subkey, Content $content ) {
        $this->stats->increment( 'editstash.' . $subkey ); // overall for b/c
        $this->stats->increment( 'editstash_by_model.' . $content->getModel() . '.' . $subkey );
    }
 
    private function getAndWaitForStashValue( $key ) {
        $editInfo = $this->getStashValue( $key );
 
        if ( !$editInfo ) {
            $start = microtime( true );
            // We ignore user aborts and keep parsing. Block on any prior parsing
            // so as to use its results and make use of the time spent parsing.
            $dbw = $this->dbProvider->getPrimaryDatabase();
            if ( $dbw->lock( $key, __METHOD__, 30 ) ) {
                $editInfo = $this->getStashValue( $key );
                $dbw->unlock( $key, __METHOD__ );
            }
 
            $timeMs = 1000 * max( 0, microtime( true ) - $start );
            $this->stats->timing( 'editstash.lock_wait_time', $timeMs );
        }
 
        return $editInfo;
    }
 
    public function fetchInputText( $textHash ) {
        $textKey = $this->cache->makeKey( 'stashedit', 'text', $textHash );
 
        return $this->cache->get( $textKey );
    }
 
    public function stashInputText( $text, $textHash ) {
        $textKey = $this->cache->makeKey( 'stashedit', 'text', $textHash );
 
        return $this->cache->set(
            $textKey,
            $text,
            self::MAX_CACHE_TTL,
            BagOStuff::WRITE_ALLOW_SEGMENTS
        );
    }
 
    private function lastEditTime( UserIdentity $user ) {
        $time = $this->dbProvider->getReplicaDatabase()->newSelectQueryBuilder()
            ->select( 'MAX(rc_timestamp)' )
            ->from( 'recentchanges' )
            ->join( 'actor', null, 'actor_id=rc_actor' )
            ->where( [ 'actor_name' => $user->getName() ] )
            ->caller( __METHOD__ )
            ->fetchField();
 
        return wfTimestampOrNull( TS_MW, $time );
    }
 
    private function getContentHash( Content $content ) {
        return sha1( implode( "\n", [
            $content->getModel(),
            $content->getDefaultFormat(),
            $content->serialize( $content->getDefaultFormat() )
        ] ) );
    }
 
    private function getStashKey( PageIdentity $page, $contentHash, UserIdentity $user ) {
        return $this->cache->makeKey(
            'stashedit-info-v2',
            md5( "{$page->getNamespace()}\n{$page->getDBkey()}" ),
            // Account for the edit model/text
            $contentHash,
            // Account for user name related variables like signatures
            md5( "{$user->getId()}\n{$user->getName()}" )
        );
    }
 
    private function getStashValue( $key ) {
        $serial = $this->cache->get( $key );
 
        return $this->unserializeStashInfo( $serial );
    }
 
    private function storeStashValue(
        $key,
        Content $pstContent,
        ParserOutput $parserOutput,
        $timestamp,
        UserIdentity $user
    ) {
        // If an item is renewed, mind the cache TTL determined by config and parser functions.
        // Put an upper limit on the TTL to avoid extreme template/file staleness.
        $age = time() - (int)wfTimestamp( TS_UNIX, $parserOutput->getCacheTime() );
        $ttl = min( $parserOutput->getCacheExpiry() - $age, self::MAX_CACHE_TTL );
        // Avoid extremely stale user signature timestamps (T84843)
        if ( $parserOutput->getOutputFlag( ParserOutputFlags::USER_SIGNATURE ) ) {
            $ttl = min( $ttl, self::MAX_SIGNATURE_TTL );
        }
 
        if ( $ttl <= 0 ) {
            return 'uncacheable'; // low TTL due to a tag, magic word, or signature?
        }
 
        // Store what is actually needed and split the output into another key (T204742)
        $stashInfo = (object)[
            'pstContent' => $pstContent,
            'output'     => $parserOutput,
            'timestamp'  => $timestamp,
            'edits'      => $this->userEditTracker->getUserEditCount( $user ),
        ];
        $serial = $this->serializeStashInfo( $stashInfo );
        if ( $serial === false ) {
            return 'store_error';
        }
 
        $ok = $this->cache->set( $key, $serial, $ttl, BagOStuff::WRITE_ALLOW_SEGMENTS );
        if ( $ok ) {
            // These blobs can waste slots in low cardinality memcached slabs
            $this->pruneExcessStashedEntries( $user, $key );
        }
 
        return $ok ? true : 'store_error';
    }
 
    private function pruneExcessStashedEntries( UserIdentity $user, $newKey ) {
        $key = $this->cache->makeKey( 'stash-edit-recent', sha1( $user->getName() ) );
 
        $keyList = $this->cache->get( $key ) ?: [];
        if ( count( $keyList ) >= self::MAX_CACHE_RECENT ) {
            $oldestKey = array_shift( $keyList );
            $this->cache->delete( $oldestKey, BagOStuff::WRITE_PRUNE_SEGMENTS );
        }
 
        $keyList[] = $newKey;
        $this->cache->set( $key, $keyList, 2 * self::MAX_CACHE_TTL );
    }
 
    private function recentStashEntryCount( UserIdentity $user ) {
        $key = $this->cache->makeKey( 'stash-edit-recent', sha1( $user->getName() ) );
 
        return count( $this->cache->get( $key ) ?: [] );
    }
 
    private function serializeStashInfo( stdClass $stashInfo ) {
        // @todo: use JSON with ParserOutput and Content
        return serialize( $stashInfo );
    }
 
    private function unserializeStashInfo( $serial ) {
        if ( is_string( $serial ) ) {
            // @todo: use JSON with ParserOutput and Content
            $stashInfo = unserialize( $serial );
            if ( is_object( $stashInfo ) && $stashInfo->output instanceof ParserOutput ) {
                return $stashInfo;
            }
        }
 
        return false;
    }
}