MediaWiki.php

Go to the documentation of this file.

<?php
use Liuggio\StatsdClient\Sender\SocketSender;
use MediaWiki\HookContainer\ProtectedHookAccessorTrait;
use MediaWiki\Logger\LoggerFactory;
use MediaWiki\MediaWikiServices;
use Psr\Log\LoggerInterface;
use Wikimedia\AtEase;
use Wikimedia\Rdbms\ChronologyProtector;
use Wikimedia\Rdbms\DBConnectionError;
use Wikimedia\Rdbms\ILBFactory;
 
class MediaWiki {
    use ProtectedHookAccessorTrait;
 
    private $context;
    private $config;
 
    private $action;
    private $postSendStrategy;
 
    private const DEFER_FASTCGI_FINISH_REQUEST = 1;
    private const DEFER_SET_LENGTH_AND_FLUSH = 2;
 
    public function __construct( IContextSource $context = null ) {
        $this->context = $context ?: RequestContext::getMain();
        $this->config = $this->context->getConfig();
        if ( function_exists( 'fastcgi_finish_request' ) ) {
            $this->postSendStrategy = self::DEFER_FASTCGI_FINISH_REQUEST;
        } else {
            $this->postSendStrategy = self::DEFER_SET_LENGTH_AND_FLUSH;
        }
    }
 
    private function parseTitle() {
        $request = $this->context->getRequest();
        $curid = $request->getInt( 'curid' );
        $title = $request->getVal( 'title' );
        $action = $request->getVal( 'action' );
 
        if ( $request->getCheck( 'search' ) ) {
            // Compatibility with old search URLs which didn't use Special:Search
            // Just check for presence here, so blank requests still
            // show the search page when using ugly URLs (T10054).
            $ret = SpecialPage::getTitleFor( 'Search' );
        } elseif ( $curid ) {
            // URLs like this are generated by RC, because rc_title isn't always accurate
            $ret = Title::newFromID( $curid );
        } else {
            $ret = Title::newFromURL( $title );
            // Alias NS_MEDIA page URLs to NS_FILE...we only use NS_MEDIA
            // in wikitext links to tell Parser to make a direct file link
            if ( $ret !== null && $ret->getNamespace() === NS_MEDIA ) {
                $ret = Title::makeTitle( NS_FILE, $ret->getDBkey() );
            }
            $contLang = MediaWikiServices::getInstance()->getContentLanguage();
            // Check variant links so that interwiki links don't have to worry
            // about the possible different language variants
            if (
                $contLang->hasVariants() && $ret !== null && $ret->getArticleID() == 0
            ) {
                $contLang->findVariantLink( $title, $ret );
            }
        }
 
        // If title is not provided, always allow oldid and diff to set the title.
        // If title is provided, allow oldid and diff to override the title, unless
        // we are talking about a special page which might use these parameters for
        // other purposes.
        if ( $ret === null || !$ret->isSpecialPage() ) {
            // We can have urls with just ?diff=,?oldid= or even just ?diff=
            $oldid = $request->getInt( 'oldid' );
            $oldid = $oldid ?: $request->getInt( 'diff' );
            // Allow oldid to override a changed or missing title
            if ( $oldid ) {
                $revRecord = MediaWikiServices::getInstance()
                    ->getRevisionLookup()
                    ->getRevisionById( $oldid );
                if ( $revRecord ) {
                    $ret = Title::newFromLinkTarget(
                        $revRecord->getPageAsLinkTarget()
                    );
                }
            }
        }
 
        // Use the main page as default title if nothing else has been provided
        if ( $ret === null
            && strval( $title ) === ''
            && !$request->getCheck( 'curid' )
            && $action !== 'delete'
        ) {
            $ret = Title::newMainPage();
        }
 
        if ( $ret === null || ( $ret->getDBkey() == '' && !$ret->isExternal() ) ) {
            // If we get here, we definitely don't have a valid title; throw an exception.
            // Try to get detailed invalid title exception first, fall back to MalformedTitleException.
            Title::newFromTextThrow( $title );
            throw new MalformedTitleException( 'badtitletext', $title );
        }
 
        return $ret;
    }
 
    public function getTitle() {
        if ( !$this->context->hasTitle() ) {
            try {
                $this->context->setTitle( $this->parseTitle() );
            } catch ( MalformedTitleException $ex ) {
                $this->context->setTitle( SpecialPage::getTitleFor( 'Badtitle' ) );
            }
        }
        return $this->context->getTitle();
    }
 
    public function getAction() : string {
        if ( $this->action === null ) {
            $this->action = Action::getActionName( $this->context );
        }
 
        return $this->action;
    }
 
    private function performRequest() {
        global $wgTitle;
 
        $request = $this->context->getRequest();
        $requestTitle = $title = $this->context->getTitle();
        $output = $this->context->getOutput();
        $user = $this->context->getUser();
 
        if ( $request->getVal( 'printable' ) === 'yes' ) {
            $output->setPrintable();
        }
 
        $this->getHookRunner()->onBeforeInitialize( $title, null, $output, $user, $request, $this );
 
        // Invalid titles. T23776: The interwikis must redirect even if the page name is empty.
        if ( $title === null || ( $title->getDBkey() == '' && !$title->isExternal() )
            || $title->isSpecial( 'Badtitle' )
        ) {
            $this->context->setTitle( SpecialPage::getTitleFor( 'Badtitle' ) );
            try {
                $this->parseTitle();
            } catch ( MalformedTitleException $ex ) {
                throw new BadTitleError( $ex );
            }
            throw new BadTitleError();
        }
 
        // Check user's permissions to read this page.
        // We have to check here to catch special pages etc.
        // We will check again in Article::view().
        $permErrors = $title->isSpecial( 'RunJobs' )
            ? [] // relies on HMAC key signature alone
            : MediaWikiServices::getInstance()->getPermissionManager()
                ->getPermissionErrors( 'read', $user, $title );
        if ( count( $permErrors ) ) {
            // T34276: allowing the skin to generate output with $wgTitle or
            // $this->context->title set to the input title would allow anonymous users to
            // determine whether a page exists, potentially leaking private data. In fact, the
            // curid and oldid request  parameters would allow page titles to be enumerated even
            // when they are not guessable. So we reset the title to Special:Badtitle before the
            // permissions error is displayed.
 
            // The skin mostly uses $this->context->getTitle() these days, but some extensions
            // still use $wgTitle.
            $badTitle = SpecialPage::getTitleFor( 'Badtitle' );
            $this->context->setTitle( $badTitle );
            $wgTitle = $badTitle;
 
            throw new PermissionsError( 'read', $permErrors );
        }
 
        // Interwiki redirects
        if ( $title->isExternal() ) {
            $rdfrom = $request->getVal( 'rdfrom' );
            if ( $rdfrom ) {
                $url = $title->getFullURL( [ 'rdfrom' => $rdfrom ] );
            } else {
                $query = $request->getValues();
                unset( $query['title'] );
                $url = $title->getFullURL( $query );
            }
            // Check for a redirect loop
            if ( !preg_match( '/^' . preg_quote( $this->config->get( 'Server' ), '/' ) . '/', $url )
                && $title->isLocal()
            ) {
                // 301 so google et al report the target as the actual url.
                $output->redirect( $url, 301 );
            } else {
                $this->context->setTitle( SpecialPage::getTitleFor( 'Badtitle' ) );
                try {
                    $this->parseTitle();
                } catch ( MalformedTitleException $ex ) {
                    throw new BadTitleError( $ex );
                }
                throw new BadTitleError();
            }
        // Handle any other redirects.
        // Redirect loops, titleless URL, $wgUsePathInfo URLs, and URLs with a variant
        } elseif ( !$this->tryNormaliseRedirect( $title ) ) {
            // Prevent information leak via Special:MyPage et al (T109724)
            $spFactory = MediaWikiServices::getInstance()->getSpecialPageFactory();
            if ( $title->isSpecialPage() ) {
                $specialPage = $spFactory->getPage( $title->getDBkey() );
                if ( $specialPage instanceof RedirectSpecialPage ) {
                    $specialPage->setContext( $this->context );
                    if ( $this->config->get( 'HideIdentifiableRedirects' )
                        && $specialPage->personallyIdentifiableTarget()
                    ) {
                        list( , $subpage ) = $spFactory->resolveAlias( $title->getDBkey() );
                        $target = $specialPage->getRedirect( $subpage );
                        // Target can also be true. We let that case fall through to normal processing.
                        if ( $target instanceof Title ) {
                            if ( $target->isExternal() ) {
                                // Handle interwiki redirects
                                $target = SpecialPage::getTitleFor(
                                    'GoToInterwiki',
                                    'force/' . $target->getPrefixedDBkey()
                                );
                            }
 
                            $query = $specialPage->getRedirectQuery( $subpage ) ?: [];
                            $request = new DerivativeRequest( $this->context->getRequest(), $query );
                            $request->setRequestURL( $this->context->getRequest()->getRequestURL() );
                            $this->context->setRequest( $request );
                            // Do not varnish cache these. May vary even for anons
                            $this->context->getOutput()->lowerCdnMaxage( 0 );
                            $this->context->setTitle( $target );
                            $wgTitle = $target;
                            // Reset action type cache. (Special pages have only view)
                            $this->action = null;
                            $title = $target;
                            $output->addJsConfigVars( [
                                'wgInternalRedirectTargetUrl' => $target->getLinkURL( $query ),
                            ] );
                            $output->addModules( 'mediawiki.action.view.redirect' );
                        }
                    }
                }
            }
 
            // Special pages ($title may have changed since if statement above)
            if ( $title->isSpecialPage() ) {
                // Actions that need to be made when we have a special pages
                $spFactory->executePath( $title, $this->context );
            } else {
                // ...otherwise treat it as an article view. The article
                // may still be a wikipage redirect to another article or URL.
                $article = $this->initializeArticle();
                if ( is_object( $article ) ) {
                    $this->performAction( $article, $requestTitle );
                } elseif ( is_string( $article ) ) {
                    $output->redirect( $article );
                } else {
                    throw new MWException( "Shouldn't happen: MediaWiki::initializeArticle()"
                        . " returned neither an object nor a URL" );
                }
            }
            $output->considerCacheSettingsFinal();
        }
    }
 
    private function tryNormaliseRedirect( Title $title ) {
        $request = $this->context->getRequest();
        $output = $this->context->getOutput();
 
        if ( $request->getVal( 'action', 'view' ) != 'view'
            || $request->wasPosted()
            || ( $request->getCheck( 'title' )
                && $title->getPrefixedDBkey() == $request->getVal( 'title' ) )
            || count( $request->getValueNames( [ 'action', 'title' ] ) )
            || !$this->getHookRunner()->onTestCanonicalRedirect( $request, $title, $output )
        ) {
            return false;
        }
 
        if ( $this->config->get( 'MainPageIsDomainRoot' ) && $request->getRequestURL() === '/' ) {
            return false;
        }
 
        if ( $title->isSpecialPage() ) {
            list( $name, $subpage ) = MediaWikiServices::getInstance()->getSpecialPageFactory()->
                resolveAlias( $title->getDBkey() );
            if ( $name ) {
                $title = SpecialPage::getTitleFor( $name, $subpage );
            }
        }
        // Redirect to canonical url, make it a 301 to allow caching
        $targetUrl = wfExpandUrl( $title->getFullURL(), PROTO_CURRENT );
        if ( $targetUrl == $request->getFullRequestURL() ) {
            $message = "Redirect loop detected!\n\n" .
                "This means the wiki got confused about what page was " .
                "requested; this sometimes happens when moving a wiki " .
                "to a new server or changing the server configuration.\n\n";
 
            if ( $this->config->get( 'UsePathInfo' ) ) {
                $message .= "The wiki is trying to interpret the page " .
                    "title from the URL path portion (PATH_INFO), which " .
                    "sometimes fails depending on the web server. Try " .
                    "setting \"\$wgUsePathInfo = false;\" in your " .
                    "LocalSettings.php, or check that \$wgArticlePath " .
                    "is correct.";
            } else {
                $message .= "Your web server was detected as possibly not " .
                    "supporting URL path components (PATH_INFO) correctly; " .
                    "check your LocalSettings.php for a customized " .
                    "\$wgArticlePath setting and/or toggle \$wgUsePathInfo " .
                    "to true.";
            }
            throw new HttpError( 500, $message );
        }
        $output->setCdnMaxage( 1200 );
        $output->redirect( $targetUrl, '301' );
        return true;
    }
 
    private function initializeArticle() {
        $title = $this->context->getTitle();
        if ( $this->context->canUseWikiPage() ) {
            // Try to use request context wiki page, as there
            // is already data from db saved in per process
            // cache there from this->getAction() call.
            $page = $this->context->getWikiPage();
        } else {
            // This case should not happen, but just in case.
            // @TODO: remove this or use an exception
            $page = WikiPage::factory( $title );
            $this->context->setWikiPage( $page );
            wfWarn( "RequestContext::canUseWikiPage() returned false" );
        }
 
        // Make GUI wrapper for the WikiPage
        $article = Article::newFromWikiPage( $page, $this->context );
 
        // Skip some unnecessary code if the content model doesn't support redirects
        if ( !MediaWikiServices::getInstance()
                ->getContentHandlerFactory()
                ->getContentHandler( $title->getContentModel() )
                ->supportsRedirects()
        ) {
            return $article;
        }
 
        $request = $this->context->getRequest();
 
        // Namespace might change when using redirects
        // Check for redirects ...
        $action = $request->getVal( 'action', 'view' );
        $file = ( $page instanceof WikiFilePage ) ? $page->getFile() : null;
        if ( ( $action == 'view' || $action == 'render' ) // ... for actions that show content
            && !$request->getVal( 'oldid' ) // ... and are not old revisions
            && !$request->getVal( 'diff' ) // ... and not when showing diff
            && $request->getVal( 'redirect' ) != 'no' // ... unless explicitly told not to
            // ... and the article is not a non-redirect image page with associated file
            && !( is_object( $file ) && $file->exists() && !$file->getRedirected() )
        ) {
            // Give extensions a change to ignore/handle redirects as needed
            $ignoreRedirect = $target = false;
 
            $this->getHookRunner()->onInitializeArticleMaybeRedirect( $title, $request,
                $ignoreRedirect, $target, $article );
            $page = $article->getPage(); // reflect any hook changes
 
            // Follow redirects only for... redirects.
            // If $target is set, then a hook wanted to redirect.
            if ( !$ignoreRedirect && ( $target || $page->isRedirect() ) ) {
                // Is the target already set by an extension?
                $target = $target ?: $page->followRedirect();
                if ( is_string( $target ) && !$this->config->get( 'DisableHardRedirects' ) ) {
                    // we'll need to redirect
                    return $target;
                }
                if ( is_object( $target ) ) {
                    // Rewrite environment to redirected article
                    $rpage = WikiPage::factory( $target );
                    $rpage->loadPageData();
                    if ( $rpage->exists() || ( is_object( $file ) && !$file->isLocal() ) ) {
                        $rarticle = Article::newFromWikiPage( $rpage, $this->context );
                        $rarticle->setRedirectedFrom( $title );
 
                        $article = $rarticle;
                        $this->context->setTitle( $target );
                        $this->context->setWikiPage( $article->getPage() );
                    }
                }
            } else {
                // Article may have been changed by hook
                $this->context->setTitle( $article->getTitle() );
                $this->context->setWikiPage( $article->getPage() );
            }
        }
 
        return $article;
    }
 
    private function performAction( Article $article, Title $requestTitle ) {
        $request = $this->context->getRequest();
        $output = $this->context->getOutput();
        $title = $this->context->getTitle();
        $user = $this->context->getUser();
 
        if ( !$this->getHookRunner()->onMediaWikiPerformAction(
            $output, $article, $title, $user, $request, $this )
        ) {
            return;
        }
 
        $act = $this->getAction();
        $action = Action::factory( $act, $article, $this->context );
 
        if ( $action instanceof Action ) {
            // Narrow DB query expectations for this HTTP request
            $trxLimits = $this->config->get( 'TrxProfilerLimits' );
            $trxProfiler = Profiler::instance()->getTransactionProfiler();
            if ( $request->wasPosted() && !$action->doesWrites() ) {
                $trxProfiler->setExpectations( $trxLimits['POST-nonwrite'], __METHOD__ );
                $request->markAsSafeRequest();
            }
 
            # Let CDN cache things if we can purge them.
            if ( $this->config->get( 'UseCdn' ) &&
                in_array(
                    // Use PROTO_INTERNAL because that's what getCdnUrls() uses
                    wfExpandUrl( $request->getRequestURL(), PROTO_INTERNAL ),
                    $requestTitle->getCdnUrls()
                )
            ) {
                $output->setCdnMaxage( $this->config->get( 'CdnMaxAge' ) );
            }
 
            $action->show();
            return;
        }
 
        // If we've not found out which action it is by now, it's unknown
        $output->setStatusCode( 404 );
        $output->showErrorPage( 'nosuchaction', 'nosuchactiontext' );
    }
 
    public function run() {
        try {
            $this->setDBProfilingAgent();
            try {
                $this->main();
            } catch ( ErrorPageError $e ) {
                $out = $this->context->getOutput();
                // TODO: Should ErrorPageError::report accept a OutputPage parameter?
                $e->report( ErrorPageError::STAGE_OUTPUT );
                $out->considerCacheSettingsFinal();
 
                // T64091: while exceptions are convenient to bubble up GUI errors,
                // they are not internal application faults. As with normal requests, this
                // should commit, print the output, do deferred updates, jobs, and profiling.
                $this->doPreOutputCommit();
                $out->output(); // display the GUI error
            }
        } catch ( Exception $e ) {
            $context = $this->context;
            $action = $context->getRequest()->getVal( 'action', 'view' );
            if (
                $e instanceof DBConnectionError &&
                $context->hasTitle() &&
                $context->getTitle()->canExist() &&
                in_array( $action, [ 'view', 'history' ], true ) &&
                HTMLFileCache::useFileCache( $context, HTMLFileCache::MODE_OUTAGE )
            ) {
                // Try to use any (even stale) file during outages...
                $cache = new HTMLFileCache( $context->getTitle(), $action );
                if ( $cache->isCached() ) {
                    $cache->loadFromFileCache( $context, HTMLFileCache::MODE_OUTAGE );
                    print MWExceptionRenderer::getHTML( $e );
                    exit;
                }
            }
            // GUI-ify and stash the page output in MediaWiki::doPreOutputCommit() while
            // ChronologyProtector synchronizes DB positions or replicas across all datacenters.
            MWExceptionHandler::handleException( $e, MWExceptionHandler::CAUGHT_BY_ENTRYPOINT );
        } catch ( Throwable $e ) {
            // Type errors and such: at least handle it now and clean up the LBFactory state
            MWExceptionHandler::handleException( $e, MWExceptionHandler::CAUGHT_BY_ENTRYPOINT );
        }
 
        $this->doPostOutputShutdown();
    }
 
    private function setDBProfilingAgent() {
        $services = MediaWikiServices::getInstance();
        $name = $this->context->getUser()->getName();
        $services->getDBLoadBalancerFactory()->setAgentName(
            mb_strlen( $name ) > 15 ? mb_substr( $name, 0, 15 ) . '...' : $name
        );
    }
 
    private function schedulePostSendJobs() {
        $jobRunRate = $this->config->get( 'JobRunRate' );
        if (
            // Recursion guard
            $this->getTitle()->isSpecial( 'RunJobs' ) ||
            // Short circuit if there is nothing to do
            ( $jobRunRate <= 0 || wfReadOnly() ) ||
            // Avoid blocking the client on stock apache; see doPostOutputShutdown()
            (
                $this->context->getRequest()->getMethod() === 'HEAD' ||
                $this->context->getRequest()->getHeader( 'If-Modified-Since' )
            )
        ) {
            return;
        }
 
        if ( $jobRunRate < 1 ) {
            $max = mt_getrandmax();
            if ( mt_rand( 0, $max ) > $max * $jobRunRate ) {
                return; // the higher the job run rate, the less likely we return here
            }
            $n = 1;
        } else {
            $n = intval( $jobRunRate );
        }
 
        // Note that DeferredUpdates will catch and log an errors (T88312)
        DeferredUpdates::addUpdate( new TransactionRoundDefiningUpdate( function () use ( $n ) {
            $logger = LoggerFactory::getInstance( 'runJobs' );
            if ( $this->config->get( 'RunJobsAsync' ) ) {
                // Send an HTTP request to the job RPC entry point if possible
                $invokedWithSuccess = $this->triggerAsyncJobs( $n, $logger );
                if ( !$invokedWithSuccess ) {
                    // Fall back to blocking on running the job(s)
                    $logger->warning( "Jobs switched to blocking; Special:RunJobs disabled" );
                    $this->triggerSyncJobs( $n );
                }
            } else {
                $this->triggerSyncJobs( $n );
            }
        }, __METHOD__ ) );
    }
 
    public function doPreOutputCommit( callable $postCommitWork = null ) {
        self::preOutputCommit( $this->context, $postCommitWork );
    }
 
    public static function preOutputCommit(
        IContextSource $context, callable $postCommitWork = null
    ) {
        $config = $context->getConfig();
        $request = $context->getRequest();
        $output = $context->getOutput();
        $services = MediaWikiServices::getInstance();
        $lbFactory = $services->getDBLoadBalancerFactory();
 
        // Try to make sure that all RDBMs, session, and other storage updates complete
        ignore_user_abort( true );
 
        // Commit all RDBMs changes from the main transaction round
        $lbFactory->commitMasterChanges(
            __METHOD__,
            // Abort if any transaction was too big
            [ 'maxWriteDuration' => $config->get( 'MaxUserDBWriteDuration' ) ]
        );
        wfDebug( __METHOD__ . ': primary transaction round committed' );
 
        // Run updates that need to block the client or affect output (this is the last chance)
        DeferredUpdates::doUpdates( 'run', DeferredUpdates::PRESEND );
        wfDebug( __METHOD__ . ': pre-send deferred updates completed' );
        // Persist the session to avoid race conditions on subsequent requests by the client
        $request->getSession()->save(); // T214471
        wfDebug( __METHOD__ . ': session changes committed' );
 
        // Figure out whether to wait for DB replication now or to use some method that assures
        // that subsequent requests by the client will use the DB replication positions written
        // during the shutdown() call below; the later requires working around replication lag
        // of the store containing DB replication positions (e.g. dynomite, mcrouter).
        list( $flags, $strategy ) = self::getChronProtStrategy( $lbFactory, $output );
        // Record ChronologyProtector positions for DBs affected in this request at this point
        $cpIndex = null;
        $cpClientId = null;
        $lbFactory->shutdown( $flags, $postCommitWork, $cpIndex, $cpClientId );
        wfDebug( __METHOD__ . ': LBFactory shutdown completed' );
 
        $allowHeaders = !( $output->isDisabled() || headers_sent() );
        if ( $cpIndex > 0 ) {
            if ( $allowHeaders ) {
                $now = time();
                $expires = $now + ChronologyProtector::POSITION_COOKIE_TTL;
                $options = [ 'prefix' => '' ];
                $value = $lbFactory::makeCookieValueFromCPIndex( $cpIndex, $now, $cpClientId );
                $request->response()->setCookie( 'cpPosIndex', $value, $expires, $options );
            }
 
            if ( $strategy === 'cookie+url' ) {
                if ( $output->getRedirect() ) { // sanity
                    $safeUrl = $lbFactory->appendShutdownCPIndexAsQuery(
                        $output->getRedirect(),
                        $cpIndex
                    );
                    $output->redirect( $safeUrl );
                } else {
                    MWExceptionHandler::logException(
                        new LogicException( "No redirect; cannot append cpPosIndex parameter." ),
                        MWExceptionHandler::CAUGHT_BY_ENTRYPOINT
                    );
                }
            }
        }
 
        if ( $allowHeaders ) {
            // Set a cookie to tell all CDN edge nodes to "stick" the user to the DC that
            // handles this POST request (e.g. the "master" data center). Also have the user
            // briefly bypass CDN so ChronologyProtector works for cacheable URLs.
            if ( $request->wasPosted() && $lbFactory->hasOrMadeRecentMasterChanges() ) {
                $expires = time() + $config->get( 'DataCenterUpdateStickTTL' );
                $options = [ 'prefix' => '' ];
                $request->response()->setCookie( 'UseDC', 'master', $expires, $options );
                $request->response()->setCookie( 'UseCDNCache', 'false', $expires, $options );
            }
 
            // Avoid letting a few seconds of replica DB lag cause a month of stale data.
            // This logic is also intimately related to the value of $wgCdnReboundPurgeDelay.
            if ( $lbFactory->laggedReplicaUsed() ) {
                $maxAge = $config->get( 'CdnMaxageLagged' );
                $output->lowerCdnMaxage( $maxAge );
                $request->response()->header( "X-Database-Lagged: true" );
                wfDebugLog( 'replication',
                    "Lagged DB used; CDN cache TTL limited to $maxAge seconds" );
            }
 
            // Avoid long-term cache pollution due to message cache rebuild timeouts (T133069)
            if ( $services->getMessageCache()->isDisabled() ) {
                $maxAge = $config->get( 'CdnMaxageSubstitute' );
                $output->lowerCdnMaxage( $maxAge );
                $request->response()->header( "X-Response-Substitute: true" );
            }
 
            if ( !$output->couldBePublicCached() || $output->haveCacheVaryCookies() ) {
                // Autoblocks: If this user is autoblocked (and the cookie block feature is enabled
                // for autoblocks), then set a cookie to track this block.
                // This has to be done on all logged-in page loads (not just upon saving edits),
                // because an autoblocked editor might not edit again from the same IP address.
                //
                // IP blocks: For anons, if their IP is blocked (and cookie block feature is enabled
                // for IP blocks), we also want to set the cookie whenever it is safe to do.
                // Basically from any url that are definitely not publicly cacheable (like viewing
                // EditPage), or when the HTTP response is personalised for other reasons (e.g. viewing
                // articles within the same browsing session after making an edit).
                $user = $context->getUser();
                $services->getBlockManager()
                    ->trackBlockWithCookie( $user, $request->response() );
            }
        }
    }
 
    private static function getChronProtStrategy( ILBFactory $lbFactory, OutputPage $output ) {
        // Should the client return, their request should observe the new ChronologyProtector
        // DB positions. This request might be on a foreign wiki domain, so synchronously update
        // the DB positions in all datacenters to be safe. If this output is not a redirect,
        // then OutputPage::output() will be relatively slow, meaning that running it in
        // $postCommitWork should help mask the latency of those updates.
        $flags = $lbFactory::SHUTDOWN_CHRONPROT_SYNC;
        $strategy = 'cookie+sync';
 
        $allowHeaders = !( $output->isDisabled() || headers_sent() );
        if ( $output->getRedirect() && $lbFactory->hasOrMadeRecentMasterChanges( INF ) ) {
            // OutputPage::output() will be fast, so $postCommitWork is useless for masking
            // the latency of synchronously updating the DB positions in all datacenters.
            // Try to make use of the time the client spends following redirects instead.
            $domainDistance = self::getUrlDomainDistance( $output->getRedirect() );
            if ( $domainDistance === 'local' && $allowHeaders ) {
                $flags = $lbFactory::SHUTDOWN_CHRONPROT_ASYNC;
                $strategy = 'cookie'; // use same-domain cookie and keep the URL uncluttered
            } elseif ( $domainDistance === 'remote' ) {
                $flags = $lbFactory::SHUTDOWN_CHRONPROT_ASYNC;
                $strategy = 'cookie+url'; // cross-domain cookie might not work
            }
        }
 
        return [ $flags, $strategy ];
    }
 
    private static function getUrlDomainDistance( $url ) {
        $clusterWiki = WikiMap::getWikiFromUrl( $url );
        if ( WikiMap::isCurrentWikiId( $clusterWiki ) ) {
            return 'local'; // the current wiki
        }
        if ( $clusterWiki !== false ) {
            return 'remote'; // another wiki in this cluster/farm
        }
 
        return 'external';
    }
 
    public function doPostOutputShutdown() {
        // Record backend request timing
        $timing = $this->context->getTiming();
        $timing->mark( 'requestShutdown' );
 
        // Perform the last synchronous operations...
        try {
            // Show visible profiling data if enabled (which cannot be post-send)
            Profiler::instance()->logDataPageOutputOnly();
        } catch ( Throwable $e ) {
            // An error may already have been shown in run(), so just log it to be safe
            MWExceptionHandler::logException( $e, MWExceptionHandler::CAUGHT_BY_ENTRYPOINT );
        }
 
        // Disable WebResponse setters for post-send processing (T191537).
        WebResponse::disableForPostSend();
 
        // Defer everything else if possible...
        $callback = function () {
            try {
                $this->restInPeace();
            } catch ( Throwable $e ) {
                // If this is post-send, then displaying errors can cause broken HTML
                MWExceptionHandler::rollbackMasterChangesAndLog(
                    $e,
                    MWExceptionHandler::CAUGHT_BY_ENTRYPOINT
                );
            }
        };
 
        if ( $this->postSendStrategy === self::DEFER_FASTCGI_FINISH_REQUEST ) {
            fastcgi_finish_request();
            $callback();
        } else {
            // Flush PHP and web server output buffers
            if ( !$this->config->get( 'CommandLineMode' ) ) {
                AtEase\AtEase::suppressWarnings();
                if ( ob_get_status() ) {
                    ob_end_flush();
                }
                flush();
                AtEase\AtEase::restoreWarnings();
            }
            $callback();
        }
    }
 
    private function main() {
        global $wgTitle;
 
        $output = $this->context->getOutput();
        $request = $this->context->getRequest();
 
        // Send Ajax requests to the Ajax dispatcher.
        if ( $request->getVal( 'action' ) === 'ajax' ) {
            // Set a dummy title, because $wgTitle == null might break things
            $title = Title::makeTitle( NS_SPECIAL, 'Badtitle/performing an AJAX call in '
                . __METHOD__
            );
            $this->context->setTitle( $title );
            $wgTitle = $title;
 
            $dispatcher = new AjaxDispatcher( $this->config );
            $dispatcher->performAction( $this->context->getUser() );
 
            return;
        }
 
        // Get title from request parameters,
        // is set on the fly by parseTitle the first time.
        $title = $this->getTitle();
        $action = $this->getAction();
        $wgTitle = $title;
 
        // Set DB query expectations for this HTTP request
        $trxLimits = $this->config->get( 'TrxProfilerLimits' );
        $trxProfiler = Profiler::instance()->getTransactionProfiler();
        $trxProfiler->setLogger( LoggerFactory::getInstance( 'DBPerformance' ) );
        if ( $request->hasSafeMethod() ) {
            $trxProfiler->setExpectations( $trxLimits['GET'], __METHOD__ );
        } else {
            $trxProfiler->setExpectations( $trxLimits['POST'], __METHOD__ );
        }
 
        if ( $this->maybeDoHttpsRedirect() ) {
            return;
        }
 
        if ( $title->canExist() && HTMLFileCache::useFileCache( $this->context ) ) {
            // Try low-level file cache hit
            $cache = new HTMLFileCache( $title, $action );
            if ( $cache->isCacheGood( /* Assume up to date */ ) ) {
                // Check incoming headers to see if client has this cached
                $timestamp = $cache->cacheTimestamp();
                if ( !$output->checkLastModified( $timestamp ) ) {
                    $cache->loadFromFileCache( $this->context );
                }
                // Do any stats increment/watchlist stuff, assuming user is viewing the
                // latest revision (which should always be the case for file cache)
                $this->context->getWikiPage()->doViewUpdates( $this->context->getUser() );
                // Tell OutputPage that output is taken care of
                $output->disable();
 
                return;
            }
        }
 
        // Actually do the work of the request and build up any output
        $this->performRequest();
 
        // GUI-ify and stash the page output in MediaWiki::doPreOutputCommit() while
        // ChronologyProtector synchronizes DB positions or replicas across all datacenters.
        $buffer = null;
        $outputWork = function () use ( $output, &$buffer ) {
            if ( $buffer === null ) {
                $buffer = $output->output( true );
            }
 
            return $buffer;
        };
 
        // Commit any changes in the current transaction round so that:
        // a) the transaction is not rolled back after success output was already sent
        // b) error output is not jumbled together with success output in the response
        $this->doPreOutputCommit( $outputWork );
        // If needed, push a deferred update to run jobs after the output is send
        $this->schedulePostSendJobs();
        // If no exceptions occurred then send the output since it is safe now
        $this->outputResponsePayload( $outputWork() );
    }
 
    private function shouldDoHttpRedirect() {
        $request = $this->context->getRequest();
 
        // Don't redirect if we're already on HTTPS
        if ( $request->getProtocol() !== 'http' ) {
            return false;
        }
 
        $force = $this->config->get( 'ForceHTTPS' );
 
        // Don't redirect if $wgServer is explicitly HTTP. We test for this here
        // by checking whether wfExpandUrl() is able to force HTTPS.
        if ( !preg_match( '#^https://#', wfExpandUrl( $request->getRequestURL(), PROTO_HTTPS ) ) ) {
            if ( $force ) {
                throw new RuntimeException( '$wgForceHTTPS is true but the server is not HTTPS' );
            }
            return false;
        }
 
        // Configured $wgForceHTTPS overrides the remaining conditions
        if ( $force ) {
            return true;
        }
 
        // Check if HTTPS is required by the session or user preferences
        return $request->getSession()->shouldForceHTTPS() ||
            // Check the cookie manually, for paranoia
            $request->getCookie( 'forceHTTPS', '' ) ||
            // Avoid checking the user and groups unless it's enabled.
            (
                $this->context->getUser()->isLoggedIn()
                && $this->context->getUser()->requiresHTTPS()
            );
    }
 
    private function maybeDoHttpsRedirect() {
        if ( !$this->shouldDoHttpRedirect() ) {
            return false;
        }
 
        $request = $this->context->getRequest();
        $oldUrl = $request->getFullRequestURL();
        $redirUrl = preg_replace( '#^http://#', 'https://', $oldUrl );
 
        // ATTENTION: This hook is likely to be removed soon due to overall design of the system.
        if ( !$this->getHookRunner()->onBeforeHttpsRedirect( $this->context, $redirUrl ) ) {
            return false;
        }
 
        if ( $request->wasPosted() ) {
            // This is weird and we'd hope it almost never happens. This
            // means that a POST came in via HTTP and policy requires us
            // redirecting to HTTPS. It's likely such a request is going
            // to fail due to post data being lost, but let's try anyway
            // and just log the instance.
 
            // @todo FIXME: See if we could issue a 307 or 308 here, need
            // to see how clients (automated & browser) behave when we do
            wfDebugLog( 'RedirectedPosts', "Redirected from HTTP to HTTPS: $oldUrl" );
        }
        // Setup dummy Title, otherwise OutputPage::redirect will fail
        $title = Title::newFromText( 'REDIR', NS_MAIN );
        $this->context->setTitle( $title );
        // Since we only do this redir to change proto, always send a vary header
        $output = $this->context->getOutput();
        $output->addVaryHeader( 'X-Forwarded-Proto' );
        $output->redirect( $redirUrl );
        $output->output();
 
        return true;
    }
 
    private function outputResponsePayload( $content ) {
        if (
            $this->postSendStrategy === self::DEFER_SET_LENGTH_AND_FLUSH &&
            DeferredUpdates::pendingUpdatesCount() &&
            !headers_sent()
        ) {
            $response = $this->context->getRequest()->response();
            // Make the browser indicate the page as "loaded" as soon as it gets all the content
            $response->header( 'Connection: close' );
            // The client should not be blocked on "post-send" updates. If apache or ob_* decide
            // that a response should be gzipped, the entire script will have to finish before
            // any data can be sent. Disable compression if there are any post-send updates.
            $response->header( 'Content-Encoding: none' );
            AtEase\AtEase::suppressWarnings();
            ini_set( 'zlib.output_compression', 0 );
            if ( function_exists( 'apache_setenv' ) ) {
                apache_setenv( 'no-gzip', '1' );
            }
            AtEase\AtEase::restoreWarnings();
            // Also set the Content-Length so that apache does not block waiting on PHP to finish.
            // If OutputPage is disabled, then either there is no body (e.g. HTTP 304) and thus no
            // Content-Length, or it was taken care of already.
            if ( !$this->context->getOutput()->isDisabled() ) {
                ob_start();
                print $content;
                $response->header( 'Content-Length: ' . ob_get_length() );
                ob_end_flush();
            }
            // @TODO: this still blocks on HEAD responses and 304 responses to GETs
        } else {
            print $content;
        }
    }
 
    public function restInPeace() {
        // Either all DB and deferred updates should happen or none.
        // The latter should not be cancelled due to client disconnect.
        ignore_user_abort( true );
 
        $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
        // Assure deferred updates are not in the main transaction
        $lbFactory->commitMasterChanges( __METHOD__ );
 
        // Loosen DB query expectations since the HTTP client is unblocked
        $trxProfiler = Profiler::instance()->getTransactionProfiler();
        $trxProfiler->redefineExpectations(
            $this->context->getRequest()->hasSafeMethod()
                ? $this->config->get( 'TrxProfilerLimits' )['PostSend-GET']
                : $this->config->get( 'TrxProfilerLimits' )['PostSend-POST'],
            __METHOD__
        );
 
        // Do any deferred jobs; preferring to run them now if a client will not wait on them
        DeferredUpdates::doUpdates( 'run' );
 
        // Log profiling data, e.g. in the database or UDP
        wfLogProfilingData();
 
        // Commit and close up!
        $lbFactory->commitMasterChanges( __METHOD__ );
        $lbFactory->shutdown( $lbFactory::SHUTDOWN_NO_CHRONPROT );
 
        wfDebug( "Request ended normally" );
    }
 
    public static function emitBufferedStatsdData(
        IBufferingStatsdDataFactory $stats, Config $config
    ) {
        if ( $config->get( 'StatsdServer' ) && $stats->hasData() ) {
            try {
                $statsdServer = explode( ':', $config->get( 'StatsdServer' ), 2 );
                $statsdHost = $statsdServer[0];
                $statsdPort = $statsdServer[1] ?? 8125;
                $statsdSender = new SocketSender( $statsdHost, $statsdPort );
                $statsdClient = new SamplingStatsdClient( $statsdSender, true, false );
                $statsdClient->setSamplingRates( $config->get( 'StatsdSamplingRates' ) );
                $statsdClient->send( $stats->getData() );
 
                $stats->clearData(); // empty buffer for the next round
            } catch ( Exception $e ) {
                MWExceptionHandler::logException( $e, MWExceptionHandler::CAUGHT_BY_ENTRYPOINT );
            }
        }
    }
 
    public function triggerJobs() {
        $jobRunRate = $this->config->get( 'JobRunRate' );
        if ( $this->getTitle()->isSpecial( 'RunJobs' ) ) {
            return; // recursion guard
        } elseif ( $jobRunRate <= 0 || wfReadOnly() ) {
            return;
        }
 
        if ( $jobRunRate < 1 ) {
            $max = mt_getrandmax();
            if ( mt_rand( 0, $max ) > $max * $jobRunRate ) {
                return; // the higher the job run rate, the less likely we return here
            }
            $n = 1;
        } else {
            $n = intval( $jobRunRate );
        }
 
        $logger = LoggerFactory::getInstance( 'runJobs' );
 
        try {
            if ( $this->config->get( 'RunJobsAsync' ) ) {
                // Send an HTTP request to the job RPC entry point if possible
                $invokedWithSuccess = $this->triggerAsyncJobs( $n, $logger );
                if ( !$invokedWithSuccess ) {
                    // Fall back to blocking on running the job(s)
                    $logger->warning( "Jobs switched to blocking; Special:RunJobs disabled" );
                    $this->triggerSyncJobs( $n );
                }
            } else {
                $this->triggerSyncJobs( $n );
            }
        } catch ( JobQueueError $e ) {
            // Do not make the site unavailable (T88312)
            MWExceptionHandler::logException( $e, MWExceptionHandler::CAUGHT_BY_ENTRYPOINT );
        }
    }
 
    private function triggerSyncJobs( $n ) {
        $trxProfiler = Profiler::instance()->getTransactionProfiler();
        $old = $trxProfiler->setSilenced( true );
        try {
            $runner = MediaWikiServices::getInstance()->getJobRunner();
            $runner->run( [ 'maxJobs' => $n ] );
        } finally {
            $trxProfiler->setSilenced( $old );
        }
    }
 
    private function triggerAsyncJobs( $n, LoggerInterface $runJobsLogger ) {
        // Do not send request if there are probably no jobs
        $group = JobQueueGroup::singleton();
        if ( !$group->queuesHaveJobs( JobQueueGroup::TYPE_DEFAULT ) ) {
            return true;
        }
 
        $query = [ 'title' => 'Special:RunJobs',
            'tasks' => 'jobs', 'maxjobs' => $n, 'sigexpiry' => time() + 5 ];
        $query['signature'] = SpecialRunJobs::getQuerySignature(
            $query, $this->config->get( 'SecretKey' ) );
 
        $errno = $errstr = null;
        $info = wfParseUrl( $this->config->get( 'CanonicalServer' ) );
        $host = $info ? $info['host'] : null;
        $port = 80;
        if ( isset( $info['scheme'] ) && $info['scheme'] == 'https' ) {
            $host = "tls://" . $host;
            $port = 443;
        }
        if ( isset( $info['port'] ) ) {
            $port = $info['port'];
        }
 
        Wikimedia\suppressWarnings();
        $sock = $host ? fsockopen(
            $host,
            $port,
            $errno,
            $errstr,
            // If it takes more than 100ms to connect to ourselves there is a problem...
            0.100
        ) : false;
        Wikimedia\restoreWarnings();
 
        $invokedWithSuccess = true;
        if ( $sock ) {
            $special = MediaWikiServices::getInstance()->getSpecialPageFactory()->
                getPage( 'RunJobs' );
            $url = $special->getPageTitle()->getCanonicalURL( $query );
            $req = (
                "POST $url HTTP/1.1\r\n" .
                "Host: {$info['host']}\r\n" .
                "Connection: Close\r\n" .
                "Content-Length: 0\r\n\r\n"
            );
 
            $runJobsLogger->info( "Running $n job(s) via '$url'" );
            // Send a cron API request to be performed in the background.
            // Give up if this takes too long to send (which should be rare).
            stream_set_timeout( $sock, 2 );
            $bytes = fwrite( $sock, $req );
            if ( $bytes !== strlen( $req ) ) {
                $invokedWithSuccess = false;
                $runJobsLogger->error( "Failed to start cron API (socket write error)" );
            } else {
                // Do not wait for the response (the script should handle client aborts).
                // Make sure that we don't close before that script reaches ignore_user_abort().
                $start = microtime( true );
                $status = fgets( $sock );
                $sec = microtime( true ) - $start;
                if ( !preg_match( '#^HTTP/\d\.\d 202 #', $status ) ) {
                    $invokedWithSuccess = false;
                    $runJobsLogger->error( "Failed to start cron API: received '$status' ($sec)" );
                }
            }
            fclose( $sock );
        } else {
            $invokedWithSuccess = false;
            $runJobsLogger->error( "Failed to start cron API (socket error $errno): $errstr" );
        }
 
        return $invokedWithSuccess;
    }
}