thumb.php

Go to the documentation of this file.

<?php
use MediaWiki\Logger\LoggerFactory;
use MediaWiki\MediaWikiServices;
 
define( 'MW_NO_OUTPUT_COMPRESSION', 1 );
require __DIR__ . '/includes/WebStart.php';
 
// Don't use fancy MIME detection, just check the file extension for jpg/gif/png
$wgTrivialMimeDetection = true;
 
if ( defined( 'THUMB_HANDLER' ) ) {
    // Called from thumb_handler.php via 404; extract params from the URI...
    wfThumbHandle404();
} else {
    // Called directly, use $_GET params
    wfStreamThumb( $_GET );
}
 
$mediawiki = new MediaWiki();
$mediawiki->doPostOutputShutdown( 'fast' );
 
// --------------------------------------------------------------------------
 
function wfThumbHandle404() {
    global $wgArticlePath;
 
    # Set action base paths so that WebRequest::getPathInfo()
    # recognizes the "X" as the 'title' in ../thumb_handler.php/X urls.
    # Note: If Custom per-extension repo paths are set, this may break.
    $repo = RepoGroup::singleton()->getLocalRepo();
    $oldArticlePath = $wgArticlePath;
    $wgArticlePath = $repo->getZoneUrl( 'thumb' ) . '/$1';
 
    $matches = WebRequest::getPathInfo();
 
    $wgArticlePath = $oldArticlePath;
 
    if ( !isset( $matches['title'] ) ) {
        wfThumbError( 404, 'Could not determine the name of the requested thumbnail.' );
        return;
    }
 
    $params = wfExtractThumbRequestInfo( $matches['title'] ); // basic wiki URL param extracting
    if ( $params == null ) {
        wfThumbError( 400, 'The specified thumbnail parameters are not recognized.' );
        return;
    }
 
    wfStreamThumb( $params ); // stream the thumbnail
}
 
function wfStreamThumb( array $params ) {
    global $wgVaryOnXFP;
 
    $headers = []; // HTTP headers to send
 
    $fileName = isset( $params['f'] ) ? $params['f'] : '';
 
    // Backwards compatibility parameters
    if ( isset( $params['w'] ) ) {
        $params['width'] = $params['w'];
        unset( $params['w'] );
    }
    if ( isset( $params['width'] ) && substr( $params['width'], -2 ) == 'px' ) {
        // strip the px (pixel) suffix, if found
        $params['width'] = substr( $params['width'], 0, -2 );
    }
    if ( isset( $params['p'] ) ) {
        $params['page'] = $params['p'];
    }
 
    // Is this a thumb of an archived file?
    $isOld = ( isset( $params['archived'] ) && $params['archived'] );
    unset( $params['archived'] ); // handlers don't care
 
    // Is this a thumb of a temp file?
    $isTemp = ( isset( $params['temp'] ) && $params['temp'] );
    unset( $params['temp'] ); // handlers don't care
 
    // Some basic input validation
    $fileName = strtr( $fileName, '\\/', '__' );
 
    // Actually fetch the image. Method depends on whether it is archived or not.
    if ( $isTemp ) {
        $repo = RepoGroup::singleton()->getLocalRepo()->getTempRepo();
        $img = new UnregisteredLocalFile( null, $repo,
            # Temp files are hashed based on the name without the timestamp.
            # The thumbnails will be hashed based on the entire name however.
            # @todo fix this convention to actually be reasonable.
            $repo->getZonePath( 'public' ) . '/' . $repo->getTempHashPath( $fileName ) . $fileName
        );
    } elseif ( $isOld ) {
        // Format is <timestamp>!<name>
        $bits = explode( '!', $fileName, 2 );
        if ( count( $bits ) != 2 ) {
            wfThumbError( 404, wfMessage( 'badtitletext' )->parse() );
            return;
        }
        $title = Title::makeTitleSafe( NS_FILE, $bits[1] );
        if ( !$title ) {
            wfThumbError( 404, wfMessage( 'badtitletext' )->parse() );
            return;
        }
        $img = RepoGroup::singleton()->getLocalRepo()->newFromArchiveName( $title, $fileName );
    } else {
        $img = wfLocalFile( $fileName );
    }
 
    // Check the source file title
    if ( !$img ) {
        wfThumbError( 404, wfMessage( 'badtitletext' )->parse() );
        return;
    }
 
    // Check permissions if there are read restrictions
    $varyHeader = [];
    if ( !in_array( 'read', User::getGroupPermissions( [ '*' ] ), true ) ) {
        if ( !$img->getTitle() || !$img->getTitle()->userCan( 'read' ) ) {
            wfThumbError( 403, 'Access denied. You do not have permission to access ' .
                'the source file.' );
            return;
        }
        $headers[] = 'Cache-Control: private';
        $varyHeader[] = 'Cookie';
    }
 
    // Check if the file is hidden
    if ( $img->isDeleted( File::DELETED_FILE ) ) {
        wfThumbErrorText( 404, "The source file '$fileName' does not exist." );
        return;
    }
 
    // Do rendering parameters extraction from thumbnail name.
    if ( isset( $params['thumbName'] ) ) {
        $params = wfExtractThumbParams( $img, $params );
    }
    if ( $params == null ) {
        wfThumbError( 400, 'The specified thumbnail parameters are not recognized.' );
        return;
    }
 
    // Check the source file storage path
    if ( !$img->exists() ) {
        $redirectedLocation = false;
        if ( !$isTemp ) {
            // Check for file redirect
            // Since redirects are associated with pages, not versions of files,
            // we look for the most current version to see if its a redirect.
            $possRedirFile = RepoGroup::singleton()->getLocalRepo()->findFile( $img->getName() );
            if ( $possRedirFile && !is_null( $possRedirFile->getRedirected() ) ) {
                $redirTarget = $possRedirFile->getName();
                $targetFile = wfLocalFile( Title::makeTitleSafe( NS_FILE, $redirTarget ) );
                if ( $targetFile->exists() ) {
                    $newThumbName = $targetFile->thumbName( $params );
                    if ( $isOld ) {
                        $newThumbUrl = $targetFile->getArchiveThumbUrl(
                            $bits[0] . '!' . $targetFile->getName(), $newThumbName );
                    } else {
                        $newThumbUrl = $targetFile->getThumbUrl( $newThumbName );
                    }
                    $redirectedLocation = wfExpandUrl( $newThumbUrl, PROTO_CURRENT );
                }
            }
        }
 
        if ( $redirectedLocation ) {
            // File has been moved. Give redirect.
            $response = RequestContext::getMain()->getRequest()->response();
            $response->statusHeader( 302 );
            $response->header( 'Location: ' . $redirectedLocation );
            $response->header( 'Expires: ' .
                gmdate( 'D, d M Y H:i:s', time() + 12 * 3600 ) . ' GMT' );
            if ( $wgVaryOnXFP ) {
                $varyHeader[] = 'X-Forwarded-Proto';
            }
            if ( count( $varyHeader ) ) {
                $response->header( 'Vary: ' . implode( ', ', $varyHeader ) );
            }
            $response->header( 'Content-Length: 0' );
            return;
        }
 
        // If its not a redirect that has a target as a local file, give 404.
        wfThumbErrorText( 404, "The source file '$fileName' does not exist." );
        return;
    } elseif ( $img->getPath() === false ) {
        wfThumbErrorText( 400, "The source file '$fileName' is not locally accessible." );
        return;
    }
 
    // Check IMS against the source file
    // This means that clients can keep a cached copy even after it has been deleted on the server
    if ( !empty( $_SERVER['HTTP_IF_MODIFIED_SINCE'] ) ) {
        // Fix IE brokenness
        $imsString = preg_replace( '/;.*$/', '', $_SERVER["HTTP_IF_MODIFIED_SINCE"] );
        // Calculate time
        MediaWiki\suppressWarnings();
        $imsUnix = strtotime( $imsString );
        MediaWiki\restoreWarnings();
        if ( wfTimestamp( TS_UNIX, $img->getTimestamp() ) <= $imsUnix ) {
            HttpStatus::header( 304 );
            return;
        }
    }
 
    $rel404 = isset( $params['rel404'] ) ? $params['rel404'] : null;
    unset( $params['r'] ); // ignore 'r' because we unconditionally pass File::RENDER
    unset( $params['f'] ); // We're done with 'f' parameter.
    unset( $params['rel404'] ); // moved to $rel404
 
    // Get the normalized thumbnail name from the parameters...
    try {
        $thumbName = $img->thumbName( $params );
        if ( !strlen( $thumbName ) ) { // invalid params?
            throw new MediaTransformInvalidParametersException(
                'Empty return from File::thumbName'
            );
        }
        $thumbName2 = $img->thumbName( $params, File::THUMB_FULL_NAME ); // b/c; "long" style
    } catch ( MediaTransformInvalidParametersException $e ) {
        wfThumbError(
            400,
            'The specified thumbnail parameters are not valid: ' . $e->getMessage()
        );
        return;
    } catch ( MWException $e ) {
        wfThumbError( 500, $e->getHTML(), 'Exception caught while extracting thumb name',
            [ 'exception' => $e ] );
        return;
    }
 
    // For 404 handled thumbnails, we only use the base name of the URI
    // for the thumb params and the parent directory for the source file name.
    // Check that the zone relative path matches up so squid caches won't pick
    // up thumbs that would not be purged on source file deletion (T36231).
    if ( $rel404 !== null ) { // thumbnail was handled via 404
        if ( rawurldecode( $rel404 ) === $img->getThumbRel( $thumbName ) ) {
            // Request for the canonical thumbnail name
        } elseif ( rawurldecode( $rel404 ) === $img->getThumbRel( $thumbName2 ) ) {
            // Request for the "long" thumbnail name; redirect to canonical name
            $response = RequestContext::getMain()->getRequest()->response();
            $response->statusHeader( 301 );
            $response->header( 'Location: ' .
                wfExpandUrl( $img->getThumbUrl( $thumbName ), PROTO_CURRENT ) );
            $response->header( 'Expires: ' .
                gmdate( 'D, d M Y H:i:s', time() + 7 * 86400 ) . ' GMT' );
            if ( $wgVaryOnXFP ) {
                $varyHeader[] = 'X-Forwarded-Proto';
            }
            if ( count( $varyHeader ) ) {
                $response->header( 'Vary: ' . implode( ', ', $varyHeader ) );
            }
            return;
        } else {
            wfThumbErrorText( 404, "The given path of the specified thumbnail is incorrect;
                expected '" . $img->getThumbRel( $thumbName ) . "' but got '" .
                rawurldecode( $rel404 ) . "'." );
            return;
        }
    }
 
    $dispositionType = isset( $params['download'] ) ? 'attachment' : 'inline';
 
    // Suggest a good name for users downloading this thumbnail
    $headers[] =
        "Content-Disposition: {$img->getThumbDisposition( $thumbName, $dispositionType )}";
 
    if ( count( $varyHeader ) ) {
        $headers[] = 'Vary: ' . implode( ', ', $varyHeader );
    }
 
    // Stream the file if it exists already...
    $thumbPath = $img->getThumbPath( $thumbName );
    if ( $img->getRepo()->fileExists( $thumbPath ) ) {
        $starttime = microtime( true );
        $status = $img->getRepo()->streamFileWithStatus( $thumbPath, $headers );
        $streamtime = microtime( true ) - $starttime;
 
        if ( $status->isOK() ) {
            MediaWikiServices::getInstance()->getStatsdDataFactory()->timing(
                'media.thumbnail.stream', $streamtime
            );
        } else {
            wfThumbError( 500, 'Could not stream the file', null, [ 'file' => $thumbName,
                'path' => $thumbPath, 'error' => $status->getWikiText( false, false, 'en' ) ] );
        }
        return;
    }
 
    $user = RequestContext::getMain()->getUser();
    if ( !wfThumbIsStandard( $img, $params ) && $user->pingLimiter( 'renderfile-nonstandard' ) ) {
        wfThumbError( 429, wfMessage( 'actionthrottledtext' )->parse() );
        return;
    } elseif ( $user->pingLimiter( 'renderfile' ) ) {
        wfThumbError( 429, wfMessage( 'actionthrottledtext' )->parse() );
        return;
    }
 
    list( $thumb, $errorMsg ) = wfGenerateThumbnail( $img, $params, $thumbName, $thumbPath );
 
    // Check for thumbnail generation errors...
    $msg = wfMessage( 'thumbnail_error' );
    $errorCode = 500;
 
    if ( !$thumb ) {
        $errorMsg = $errorMsg ?: $msg->rawParams( 'File::transform() returned false' )->escaped();
        if ( $errorMsg instanceof MessageSpecifier &&
            $errorMsg->getKey() === 'thumbnail_image-failure-limit'
        ) {
            $errorCode = 429;
        }
    } elseif ( $thumb->isError() ) {
        $errorMsg = $thumb->getHtmlMsg();
        $errorCode = $thumb->getHttpStatusCode();
    } elseif ( !$thumb->hasFile() ) {
        $errorMsg = $msg->rawParams( 'No path supplied in thumbnail object' )->escaped();
    } elseif ( $thumb->fileIsSource() ) {
        $errorMsg = $msg
            ->rawParams( 'Image was not scaled, is the requested width bigger than the source?' )
            ->escaped();
        $errorCode = 400;
    }
 
    if ( $errorMsg !== false ) {
        wfThumbError( $errorCode, $errorMsg, null, [ 'file' => $thumbName, 'path' => $thumbPath ] );
    } else {
        // Stream the file if there were no errors
        $status = $thumb->streamFileWithStatus( $headers );
        if ( !$status->isOK() ) {
            wfThumbError( 500, 'Could not stream the file', null, [
                'file' => $thumbName, 'path' => $thumbPath,
                'error' => $status->getWikiText( false, false, 'en' ) ] );
        }
    }
}
 
function wfGenerateThumbnail( File $file, array $params, $thumbName, $thumbPath ) {
    global $wgAttemptFailureEpoch;
 
    $cache = ObjectCache::getLocalClusterInstance();
    $key = $cache->makeKey(
        'attempt-failures',
        $wgAttemptFailureEpoch,
        $file->getRepo()->getName(),
        $file->getSha1(),
        md5( $thumbName )
    );
 
    // Check if this file keeps failing to render
    if ( $cache->get( $key ) >= 4 ) {
        return [ false, wfMessage( 'thumbnail_image-failure-limit', 4 ) ];
    }
 
    $done = false;
    // Record failures on PHP fatals in addition to caching exceptions
    register_shutdown_function( function () use ( $cache, &$done, $key ) {
        if ( !$done ) { // transform() gave a fatal
            // Randomize TTL to reduce stampedes
            $cache->incrWithInit( $key, $cache::TTL_HOUR + mt_rand( 0, 300 ) );
        }
    } );
 
    $thumb = false;
    $errorHtml = false;
 
    // guard thumbnail rendering with PoolCounter to avoid stampedes
    // expensive files use a separate PoolCounter config so it is possible
    // to set up a global limit on them
    if ( $file->isExpensiveToThumbnail() ) {
        $poolCounterType = 'FileRenderExpensive';
    } else {
        $poolCounterType = 'FileRender';
    }
 
    // Thumbnail isn't already there, so create the new thumbnail...
    try {
        $work = new PoolCounterWorkViaCallback( $poolCounterType, sha1( $file->getName() ),
            [
                'doWork' => function () use ( $file, $params ) {
                    return $file->transform( $params, File::RENDER_NOW );
                },
                'doCachedWork' => function () use ( $file, $params, $thumbPath ) {
                    // If the worker that finished made this thumbnail then use it.
                    // Otherwise, it probably made a different thumbnail for this file.
                    return $file->getRepo()->fileExists( $thumbPath )
                        ? $file->transform( $params, File::RENDER_NOW )
                        : false; // retry once more in exclusive mode
                },
                'error' => function ( Status $status ) {
                    return wfMessage( 'generic-pool-error' )->parse() . '<hr>' . $status->getHTML();
                }
            ]
        );
        $result = $work->execute();
        if ( $result instanceof MediaTransformOutput ) {
            $thumb = $result;
        } elseif ( is_string( $result ) ) { // error
            $errorHtml = $result;
        }
    } catch ( Exception $e ) {
        // Tried to select a page on a non-paged file?
    }
 
    $done = true; // no PHP fatal occured
 
    if ( !$thumb || $thumb->isError() ) {
        // Randomize TTL to reduce stampedes
        $cache->incrWithInit( $key, $cache::TTL_HOUR + mt_rand( 0, 300 ) );
    }
 
    return [ $thumb, $errorHtml ];
}
 
function wfExtractThumbRequestInfo( $thumbRel ) {
    $repo = RepoGroup::singleton()->getLocalRepo();
 
    $hashDirReg = $subdirReg = '';
    $hashLevels = $repo->getHashLevels();
    for ( $i = 0; $i < $hashLevels; $i++ ) {
        $subdirReg .= '[0-9a-f]';
        $hashDirReg .= "$subdirReg/";
    }
 
    // Check if this is a thumbnail of an original in the local file repo
    if ( preg_match( "!^((archive/)?$hashDirReg([^/]*)/([^/]*))$!", $thumbRel, $m ) ) {
        list( /*all*/, $rel, $archOrTemp, $filename, $thumbname ) = $m;
    // Check if this is a thumbnail of an temp file in the local file repo
    } elseif ( preg_match( "!^(temp/)($hashDirReg([^/]*)/([^/]*))$!", $thumbRel, $m ) ) {
        list( /*all*/, $archOrTemp, $rel, $filename, $thumbname ) = $m;
    } else {
        return null; // not a valid looking thumbnail request
    }
 
    $params = [ 'f' => $filename, 'rel404' => $rel ];
    if ( $archOrTemp === 'archive/' ) {
        $params['archived'] = 1;
    } elseif ( $archOrTemp === 'temp/' ) {
        $params['temp'] = 1;
    }
 
    $params['thumbName'] = $thumbname;
    return $params;
}
 
function wfExtractThumbParams( $file, $params ) {
    if ( !isset( $params['thumbName'] ) ) {
        throw new InvalidArgumentException( "No thumbnail name passed to wfExtractThumbParams" );
    }
 
    $thumbname = $params['thumbName'];
    unset( $params['thumbName'] );
 
    // FIXME: Files in the temp zone don't set a MIME type, which means
    // they don't have a handler. Which means we can't parse the param
    // string. However, not a big issue as what good is a param string
    // if you have no handler to make use of the param string and
    // actually generate the thumbnail.
    $handler = $file->getHandler();
 
    // Based on UploadStash::parseKey
    $fileNamePos = strrpos( $thumbname, $params['f'] );
    if ( $fileNamePos === false ) {
        // Maybe using a short filename? (see FileRepo::nameForThumb)
        $fileNamePos = strrpos( $thumbname, 'thumbnail' );
    }
 
    if ( $handler && $fileNamePos !== false ) {
        $paramString = substr( $thumbname, 0, $fileNamePos - 1 );
        $extraParams = $handler->parseParamString( $paramString );
        if ( $extraParams !== false ) {
            return $params + $extraParams;
        }
    }
 
    // As a last ditch fallback, use the traditional common parameters
    if ( preg_match( '!^(page(\d*)-)*(\d*)px-[^/]*$!', $thumbname, $matches ) ) {
        list( /* all */, /* pagefull */, $pagenum, $size ) = $matches;
        $params['width'] = $size;
        if ( $pagenum ) {
            $params['page'] = $pagenum;
        }
        return $params; // valid thumbnail URL
    }
    return null;
}
 
function wfThumbErrorText( $status, $msgText ) {
    wfThumbError( $status, htmlspecialchars( $msgText ) );
}
 
function wfThumbError( $status, $msgHtml, $msgText = null, $context = [] ) {
    global $wgShowHostnames;
 
    header( 'Cache-Control: no-cache' );
    header( 'Content-Type: text/html; charset=utf-8' );
    if ( $status == 400 || $status == 404 || $status == 429 ) {
        HttpStatus::header( $status );
    } elseif ( $status == 403 ) {
        HttpStatus::header( 403 );
        header( 'Vary: Cookie' );
    } else {
        LoggerFactory::getInstance( 'thumb' )->error( $msgText ?: $msgHtml, $context );
        HttpStatus::header( 500 );
    }
    if ( $wgShowHostnames ) {
        header( 'X-MW-Thumbnail-Renderer: ' . wfHostname() );
        $url = htmlspecialchars(
            isset( $_SERVER['REQUEST_URI'] ) ? $_SERVER['REQUEST_URI'] : ''
        );
        $hostname = htmlspecialchars( wfHostname() );
        $debug = "<!-- $url -->\n<!-- $hostname -->\n";
    } else {
        $debug = '';
    }
    $content = <<<EOT
<!DOCTYPE html>
<html><head>
<meta charset="UTF-8" />
<title>Error generating thumbnail</title>
</head>
<body>
<h1>Error generating thumbnail</h1>
<p>
$msgHtml
</p>
$debug
</body>
</html>
 
EOT;
    header( 'Content-Length: ' . strlen( $content ) );
    echo $content;
}