ApiStashEdit.php

Go to the documentation of this file.

<?php
use MediaWiki\Logger\LoggerFactory;
use MediaWiki\MediaWikiServices;
use Wikimedia\ScopedCallback;

class ApiStashEdit extends ApiBase {
    const ERROR_NONE = 'stashed';
    const ERROR_PARSE = 'error_parse';
    const ERROR_CACHE = 'error_cache';
    const ERROR_UNCACHEABLE = 'uncacheable';
    const ERROR_BUSY = 'busy';

    const PRESUME_FRESH_TTL_SEC = 30;
    const MAX_CACHE_TTL = 300; // 5 minutes

    public function execute() {
        $user = $this->getUser();
        $params = $this->extractRequestParams();

        if ( $user->isBot() ) { // sanity
            $this->dieUsage( 'This interface is not supported for bots', 'botsnotsupported' );
        }

        $cache = ObjectCache::getLocalClusterInstance();
        $page = $this->getTitleOrPageId( $params );
        $title = $page->getTitle();

        if ( !ContentHandler::getForModelID( $params['contentmodel'] )
            ->isSupportedFormat( $params['contentformat'] )
        ) {
            $this->dieUsage( 'Unsupported content model/format', 'badmodelformat' );
        }

        $text = null;
        $textHash = null;
        if ( strlen( $params['stashedtexthash'] ) ) {
            // Load from cache since the client indicates the text is the same as last stash
            $textHash = $params['stashedtexthash'];
            $textKey = $cache->makeKey( 'stashedit', 'text', $textHash );
            $text = $cache->get( $textKey );
            if ( !is_string( $text ) ) {
                $this->dieUsage( 'No stashed text found with the given hash', 'missingtext' );
            }
        } elseif ( $params['text'] !== null ) {
            // Trim and fix newlines so the key SHA1's match (see WebRequest::getText())
            $text = rtrim( str_replace( "\r\n", "\n", $params['text'] ) );
            $textHash = sha1( $text );
        } else {
            $this->dieUsage(
                'The text or stashedtexthash parameter must be given', 'missingtextparam' );
        }

        $textContent = ContentHandler::makeContent(
            $text, $title, $params['contentmodel'], $params['contentformat'] );

        $page = WikiPage::factory( $title );
        if ( $page->exists() ) {
            // Page exists: get the merged content with the proposed change
            $baseRev = Revision::newFromPageId( $page->getId(), $params['baserevid'] );
            if ( !$baseRev ) {
                $this->dieUsage( "No revision ID {$params['baserevid']}", 'missingrev' );
            }
            $currentRev = $page->getRevision();
            if ( !$currentRev ) {
                $this->dieUsage( "No current revision of page ID {$page->getId()}", 'missingrev' );
            }
            // Merge in the new version of the section to get the proposed version
            $editContent = $page->replaceSectionAtRev(
                $params['section'],
                $textContent,
                $params['sectiontitle'],
                $baseRev->getId()
            );
            if ( !$editContent ) {
                $this->dieUsage( 'Could not merge updated section.', 'replacefailed' );
            }
            if ( $currentRev->getId() == $baseRev->getId() ) {
                // Base revision was still the latest; nothing to merge
                $content = $editContent;
            } else {
                // Merge the edit into the current version
                $baseContent = $baseRev->getContent();
                $currentContent = $currentRev->getContent();
                if ( !$baseContent || !$currentContent ) {
                    $this->dieUsage( "Missing content for page ID {$page->getId()}", 'missingrev' );
                }
                $handler = ContentHandler::getForModelID( $baseContent->getModel() );
                $content = $handler->merge3( $baseContent, $editContent, $currentContent );
            }
        } else {
            // New pages: use the user-provided content model
            $content = $textContent;
        }

        if ( !$content ) { // merge3() failed
            $this->getResult()->addValue( null,
                $this->getModuleName(), [ 'status' => 'editconflict' ] );
            return;
        }

        // The user will abort the AJAX request by pressing "save", so ignore that
        ignore_user_abort( true );

        if ( $user->pingLimiter( 'stashedit' ) ) {
            $status = 'ratelimited';
        } else {
            $status = self::parseAndStash( $page, $content, $user, $params['summary'] );
            $textKey = $cache->makeKey( 'stashedit', 'text', $textHash );
            $cache->set( $textKey, $text, self::MAX_CACHE_TTL );
        }

        $stats = MediaWikiServices::getInstance()->getStatsdDataFactory();
        $stats->increment( "editstash.cache_stores.$status" );

        $this->getResult()->addValue(
            null,
            $this->getModuleName(),
            [
                'status' => $status,
                'texthash' => $textHash
            ]
        );
    }

    public static function parseAndStash( WikiPage $page, Content $content, User $user, $summary ) {
        $cache = ObjectCache::getLocalClusterInstance();
        $logger = LoggerFactory::getInstance( 'StashEdit' );

        $title = $page->getTitle();
        $key = self::getStashKey( $title, self::getContentHash( $content ), $user );

        // Use the master DB for fast blocking locks
        $dbw = wfGetDB( DB_MASTER );
        if ( !$dbw->lock( $key, __METHOD__, 1 ) ) {
            // De-duplicate requests on the same key
            return self::ERROR_BUSY;
        }
        $unlocker = new ScopedCallback( function () use ( $dbw, $key ) {
            $dbw->unlock( $key, __METHOD__ );
        } );

        $cutoffTime = time() - self::PRESUME_FRESH_TTL_SEC;

        // Reuse any freshly build matching edit stash cache
        $editInfo = $cache->get( $key );
        if ( $editInfo && wfTimestamp( TS_UNIX, $editInfo->timestamp ) >= $cutoffTime ) {
            $alreadyCached = true;
        } else {
            $format = $content->getDefaultFormat();
            $editInfo = $page->prepareContentForEdit( $content, null, $user, $format, false );
            $alreadyCached = false;
        }

        if ( $editInfo && $editInfo->output ) {
            // Let extensions add ParserOutput metadata or warm other caches
            Hooks::run( 'ParserOutputStashForEdit',
                [ $page, $content, $editInfo->output, $summary, $user ] );

            if ( $alreadyCached ) {
                $logger->debug( "Already cached parser output for key '$key' ('$title')." );
                return self::ERROR_NONE;
            }

            list( $stashInfo, $ttl, $code ) = self::buildStashValue(
                $editInfo->pstContent,
                $editInfo->output,
                $editInfo->timestamp,
                $user
            );

            if ( $stashInfo ) {
                $ok = $cache->set( $key, $stashInfo, $ttl );
                if ( $ok ) {
                    $logger->debug( "Cached parser output for key '$key' ('$title')." );
                    return self::ERROR_NONE;
                } else {
                    $logger->error( "Failed to cache parser output for key '$key' ('$title')." );
                    return self::ERROR_CACHE;
                }
            } else {
                $logger->info( "Uncacheable parser output for key '$key' ('$title') [$code]." );
                return self::ERROR_UNCACHEABLE;
            }
        }

        return self::ERROR_PARSE;
    }

    public static function checkCache( Title $title, Content $content, User $user ) {
        if ( $user->isBot() ) {
            return false; // bots never stash - don't pollute stats
        }

        $cache = ObjectCache::getLocalClusterInstance();
        $logger = LoggerFactory::getInstance( 'StashEdit' );
        $stats = MediaWikiServices::getInstance()->getStatsdDataFactory();

        $key = self::getStashKey( $title, self::getContentHash( $content ), $user );
        $editInfo = $cache->get( $key );
        if ( !is_object( $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.
            // Skip this logic if there no master connection in case this method
            // is called on an HTTP GET request for some reason.
            $lb = MediaWikiServices::getInstance()->getDBLoadBalancer();
            $dbw = $lb->getAnyOpenConnection( $lb->getWriterIndex() );
            if ( $dbw && $dbw->lock( $key, __METHOD__, 30 ) ) {
                $editInfo = $cache->get( $key );
                $dbw->unlock( $key, __METHOD__ );
            }

            $timeMs = 1000 * max( 0, microtime( true ) - $start );
            $stats->timing( 'editstash.lock_wait_time', $timeMs );
        }

        if ( !is_object( $editInfo ) || !$editInfo->output ) {
            $stats->increment( 'editstash.cache_misses.no_stash' );
            $logger->debug( "Empty cache for key '$key' ('$title'); user '{$user->getName()}'." );
            return false;
        }

        $age = time() - wfTimestamp( TS_UNIX, $editInfo->output->getCacheTime() );
        if ( $age <= self::PRESUME_FRESH_TTL_SEC ) {
            // Assume nothing changed in this time
            $stats->increment( 'editstash.cache_hits.presumed_fresh' );
            $logger->debug( "Timestamp-based cache hit for key '$key' (age: $age sec)." );
        } elseif ( isset( $editInfo->edits ) && $editInfo->edits === $user->getEditCount() ) {
            // Logged-in user made no local upload/template edits in the meantime
            $stats->increment( 'editstash.cache_hits.presumed_fresh' );
            $logger->debug( "Edit count based cache hit for key '$key' (age: $age sec)." );
        } elseif ( $user->isAnon()
            && self::lastEditTime( $user ) < $editInfo->output->getCacheTime()
        ) {
            // Logged-out user made no local upload/template edits in the meantime
            $stats->increment( 'editstash.cache_hits.presumed_fresh' );
            $logger->debug( "Edit check based cache hit for key '$key' (age: $age sec)." );
        } else {
            // User may have changed included content
            $editInfo = false;
        }

        if ( !$editInfo ) {
            $stats->increment( 'editstash.cache_misses.proven_stale' );
            $logger->info( "Stale cache for key '$key'; old key with outside edits. (age: $age sec)" );
        } elseif ( $editInfo->output->getFlag( 'vary-revision' ) ) {
            // This can be used for the initial parse, e.g. for filters or doEditContent(),
            // but a second parse will be triggered in doEditUpdates(). This is not optimal.
            $logger->info( "Cache for key '$key' ('$title') has vary_revision." );
        } elseif ( $editInfo->output->getFlag( 'vary-revision-id' ) ) {
            // Similar to the above if we didn't guess the ID correctly.
            $logger->info( "Cache for key '$key' ('$title') has vary_revision_id." );
        }

        return $editInfo;
    }

    private static function lastEditTime( User $user ) {
        $time = wfGetDB( DB_REPLICA )->selectField(
            'recentchanges',
            'MAX(rc_timestamp)',
            [ 'rc_user_text' => $user->getName() ],
            __METHOD__
        );

        return wfTimestampOrNull( TS_MW, $time );
    }

    private static function getContentHash( Content $content ) {
        return sha1( implode( "\n", [
            $content->getModel(),
            $content->getDefaultFormat(),
            $content->serialize( $content->getDefaultFormat() )
        ] ) );
    }

    private static function getStashKey( Title $title, $contentHash, User $user ) {
        return ObjectCache::getLocalClusterInstance()->makeKey(
            'prepared-edit',
            md5( $title->getPrefixedDBkey() ),
            // Account for the edit model/text
            $contentHash,
            // Account for user name related variables like signatures
            md5( $user->getId() . "\n" . $user->getName() )
        );
    }

    private static function buildStashValue(
        Content $pstContent, ParserOutput $parserOutput, $timestamp, User $user
    ) {
        // If an item is renewed, mind the cache TTL determined by config and parser functions.
        // Put an upper limit on the TTL for sanity to avoid extreme template/file staleness.
        $since = time() - wfTimestamp( TS_UNIX, $parserOutput->getTimestamp() );
        $ttl = min( $parserOutput->getCacheExpiry() - $since, self::MAX_CACHE_TTL );
        if ( $ttl <= 0 ) {
            return [ null, 0, 'no_ttl' ];
        }

        // Only store what is actually needed
        $stashInfo = (object)[
            'pstContent' => $pstContent,
            'output'     => $parserOutput,
            'timestamp'  => $timestamp,
            'edits'      => $user->getEditCount()
        ];

        return [ $stashInfo, $ttl, 'ok' ];
    }

    public function getAllowedParams() {
        return [
            'title' => [
                ApiBase::PARAM_TYPE => 'string',
                ApiBase::PARAM_REQUIRED => true
            ],
            'section' => [
                ApiBase::PARAM_TYPE => 'string',
            ],
            'sectiontitle' => [
                ApiBase::PARAM_TYPE => 'string'
            ],
            'text' => [
                ApiBase::PARAM_TYPE => 'text',
                ApiBase::PARAM_DFLT => null
            ],
            'stashedtexthash' => [
                ApiBase::PARAM_TYPE => 'string',
                ApiBase::PARAM_DFLT => null
            ],
            'summary' => [
                ApiBase::PARAM_TYPE => 'string',
            ],
            'contentmodel' => [
                ApiBase::PARAM_TYPE => ContentHandler::getContentModels(),
                ApiBase::PARAM_REQUIRED => true
            ],
            'contentformat' => [
                ApiBase::PARAM_TYPE => ContentHandler::getAllContentFormats(),
                ApiBase::PARAM_REQUIRED => true
            ],
            'baserevid' => [
                ApiBase::PARAM_TYPE => 'integer',
                ApiBase::PARAM_REQUIRED => true
            ]
        ];
    }

    public function needsToken() {
        return 'csrf';
    }

    public function mustBePosted() {
        return true;
    }

    public function isWriteMode() {
        return true;
    }

    public function isInternal() {
        return true;
    }
}