UploadStash.php

Go to the documentation of this file.

<?php
class UploadStash {
    // Format of the key for files -- has to be suitable as a filename itself (e.g. ab12cd34ef.jpg)
    const KEY_FORMAT_REGEX = '/^[\w\-\.]+\.\w*$/';
    const MAX_US_PROPS_SIZE = 65535;
 
    public $repo;
 
    // array of initialized repo objects
    protected $files = [];
 
    // cache of the file metadata that's stored in the database
    protected $fileMetadata = [];
 
    // fileprops cache
    protected $fileProps = [];
 
    // current user
    protected $user, $userId, $isLoggedIn;
 
    public function __construct( FileRepo $repo, $user = null ) {
        // this might change based on wiki's configuration.
        $this->repo = $repo;
 
        // if a user was passed, use it. otherwise, attempt to use the global.
        // this keeps FileRepo from breaking when it creates an UploadStash object
        if ( $user ) {
            $this->user = $user;
        } else {
            global $wgUser;
            $this->user = $wgUser;
        }
 
        if ( is_object( $this->user ) ) {
            $this->userId = $this->user->getId();
            $this->isLoggedIn = $this->user->isLoggedIn();
        }
    }
 
    public function getFile( $key, $noAuth = false ) {
        if ( !preg_match( self::KEY_FORMAT_REGEX, $key ) ) {
            throw new UploadStashBadPathException(
                wfMessage( 'uploadstash-bad-path-bad-format', $key )
            );
        }
 
        if ( !$noAuth && !$this->isLoggedIn ) {
            throw new UploadStashNotLoggedInException(
                wfMessage( 'uploadstash-not-logged-in' )
            );
        }
 
        if ( !isset( $this->fileMetadata[$key] ) ) {
            if ( !$this->fetchFileMetadata( $key ) ) {
                // If nothing was received, it's likely due to replication lag.
                // Check the master to see if the record is there.
                $this->fetchFileMetadata( $key, DB_MASTER );
            }
 
            if ( !isset( $this->fileMetadata[$key] ) ) {
                throw new UploadStashFileNotFoundException(
                    wfMessage( 'uploadstash-file-not-found', $key )
                );
            }
 
            // create $this->files[$key]
            $this->initFile( $key );
 
            // fetch fileprops
            if ( strlen( $this->fileMetadata[$key]['us_props'] ) ) {
                $this->fileProps[$key] = unserialize( $this->fileMetadata[$key]['us_props'] );
            } else { // b/c for rows with no us_props
                wfDebug( __METHOD__ . " fetched props for $key from file\n" );
                $path = $this->fileMetadata[$key]['us_path'];
                $this->fileProps[$key] = $this->repo->getFileProps( $path );
            }
        }
 
        if ( !$this->files[$key]->exists() ) {
            wfDebug( __METHOD__ . " tried to get file at $key, but it doesn't exist\n" );
            // @todo Is this not an UploadStashFileNotFoundException case?
            throw new UploadStashBadPathException(
                wfMessage( 'uploadstash-bad-path' )
            );
        }
 
        if ( !$noAuth && $this->fileMetadata[$key]['us_user'] != $this->userId ) {
            throw new UploadStashWrongOwnerException(
                wfMessage( 'uploadstash-wrong-owner', $key )
            );
        }
 
        return $this->files[$key];
    }
 
    public function getMetadata( $key ) {
        $this->getFile( $key );
 
        return $this->fileMetadata[$key];
    }
 
    public function getFileProps( $key ) {
        $this->getFile( $key );
 
        return $this->fileProps[$key];
    }
 
    public function stashFile( $path, $sourceType = null ) {
        if ( !is_file( $path ) ) {
            wfDebug( __METHOD__ . " tried to stash file at '$path', but it doesn't exist\n" );
            throw new UploadStashBadPathException(
                wfMessage( 'uploadstash-bad-path' )
            );
        }
 
        $mwProps = new MWFileProps( MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer() );
        $fileProps = $mwProps->getPropsFromPath( $path, true );
        wfDebug( __METHOD__ . " stashing file at '$path'\n" );
 
        // we will be initializing from some tmpnam files that don't have extensions.
        // most of MediaWiki assumes all uploaded files have good extensions. So, we fix this.
        $extension = self::getExtensionForPath( $path );
        if ( !preg_match( "/\\.\\Q$extension\\E$/", $path ) ) {
            $pathWithGoodExtension = "$path.$extension";
        } else {
            $pathWithGoodExtension = $path;
        }
 
        // If no key was supplied, make one.  a mysql insertid would be totally
        // reasonable here, except that for historical reasons, the key is this
        // random thing instead.  At least it's not guessable.
        // Some things that when combined will make a suitably unique key.
        // see: http://www.jwz.org/doc/mid.html
        list( $usec, $sec ) = explode( ' ', microtime() );
        $usec = substr( $usec, 2 );
        $key = Wikimedia\base_convert( $sec . $usec, 10, 36 ) . '.' .
            Wikimedia\base_convert( mt_rand(), 10, 36 ) . '.' .
            $this->userId . '.' .
            $extension;
 
        $this->fileProps[$key] = $fileProps;
 
        if ( !preg_match( self::KEY_FORMAT_REGEX, $key ) ) {
            throw new UploadStashBadPathException(
                wfMessage( 'uploadstash-bad-path-bad-format', $key )
            );
        }
 
        wfDebug( __METHOD__ . " key for '$path': $key\n" );
 
        // if not already in a temporary area, put it there
        $storeStatus = $this->repo->storeTemp( basename( $pathWithGoodExtension ), $path );
 
        if ( !$storeStatus->isOK() ) {
            // It is a convention in MediaWiki to only return one error per API
            // exception, even if multiple errors are available. We use reset()
            // to pick the "first" thing that was wrong, preferring errors to
            // warnings. This is a bit lame, as we may have more info in the
            // $storeStatus and we're throwing it away, but to fix it means
            // redesigning API errors significantly.
            // $storeStatus->value just contains the virtual URL (if anything)
            // which is probably useless to the caller.
            $error = $storeStatus->getErrorsArray();
            $error = reset( $error );
            if ( !count( $error ) ) {
                $error = $storeStatus->getWarningsArray();
                $error = reset( $error );
                if ( !count( $error ) ) {
                    $error = [ 'unknown', 'no error recorded' ];
                }
            }
            // At this point, $error should contain the single "most important"
            // error, plus any parameters.
            $errorMsg = array_shift( $error );
            throw new UploadStashFileException( wfMessage( $errorMsg, $error ) );
        }
        $stashPath = $storeStatus->value;
 
        // fetch the current user ID
        if ( !$this->isLoggedIn ) {
            throw new UploadStashNotLoggedInException(
                wfMessage( 'uploadstash-not-logged-in' )
            );
        }
 
        // insert the file metadata into the db.
        wfDebug( __METHOD__ . " inserting $stashPath under $key\n" );
        $dbw = $this->repo->getMasterDB();
 
        $serializedFileProps = serialize( $fileProps );
        if ( strlen( $serializedFileProps ) > self::MAX_US_PROPS_SIZE ) {
            // Database is going to truncate this and make the field invalid.
            // Prioritize important metadata over file handler metadata.
            // File handler should be prepared to regenerate invalid metadata if needed.
            $fileProps['metadata'] = false;
            $serializedFileProps = serialize( $fileProps );
        }
 
        $this->fileMetadata[$key] = [
            'us_user' => $this->userId,
            'us_key' => $key,
            'us_orig_path' => $path,
            'us_path' => $stashPath, // virtual URL
            'us_props' => $dbw->encodeBlob( $serializedFileProps ),
            'us_size' => $fileProps['size'],
            'us_sha1' => $fileProps['sha1'],
            'us_mime' => $fileProps['mime'],
            'us_media_type' => $fileProps['media_type'],
            'us_image_width' => $fileProps['width'],
            'us_image_height' => $fileProps['height'],
            'us_image_bits' => $fileProps['bits'],
            'us_source_type' => $sourceType,
            'us_timestamp' => $dbw->timestamp(),
            'us_status' => 'finished'
        ];
 
        $dbw->insert(
            'uploadstash',
            $this->fileMetadata[$key],
            __METHOD__
        );
 
        // store the insertid in the class variable so immediate retrieval
        // (possibly laggy) isn't necessary.
        $this->fileMetadata[$key]['us_id'] = $dbw->insertId();
 
        # create the UploadStashFile object for this file.
        $this->initFile( $key );
 
        return $this->getFile( $key );
    }
 
    public function clear() {
        if ( !$this->isLoggedIn ) {
            throw new UploadStashNotLoggedInException(
                wfMessage( 'uploadstash-not-logged-in' )
            );
        }
 
        wfDebug( __METHOD__ . ' clearing all rows for user ' . $this->userId . "\n" );
        $dbw = $this->repo->getMasterDB();
        $dbw->delete(
            'uploadstash',
            [ 'us_user' => $this->userId ],
            __METHOD__
        );
 
        # destroy objects.
        $this->files = [];
        $this->fileMetadata = [];
 
        return true;
    }
 
    public function removeFile( $key ) {
        if ( !$this->isLoggedIn ) {
            throw new UploadStashNotLoggedInException(
                wfMessage( 'uploadstash-not-logged-in' )
            );
        }
 
        $dbw = $this->repo->getMasterDB();
 
        // this is a cheap query. it runs on the master so that this function
        // still works when there's lag. It won't be called all that often.
        $row = $dbw->selectRow(
            'uploadstash',
            'us_user',
            [ 'us_key' => $key ],
            __METHOD__
        );
 
        if ( !$row ) {
            throw new UploadStashNoSuchKeyException(
                wfMessage( 'uploadstash-no-such-key', $key )
            );
        }
 
        if ( $row->us_user != $this->userId ) {
            throw new UploadStashWrongOwnerException(
                wfMessage( 'uploadstash-wrong-owner', $key )
            );
        }
 
        return $this->removeFileNoAuth( $key );
    }
 
    public function removeFileNoAuth( $key ) {
        wfDebug( __METHOD__ . " clearing row $key\n" );
 
        // Ensure we have the UploadStashFile loaded for this key
        $this->getFile( $key, true );
 
        $dbw = $this->repo->getMasterDB();
 
        $dbw->delete(
            'uploadstash',
            [ 'us_key' => $key ],
            __METHOD__
        );
 
        $this->files[$key]->remove();
 
        unset( $this->files[$key] );
        unset( $this->fileMetadata[$key] );
 
        return true;
    }
 
    public function listFiles() {
        if ( !$this->isLoggedIn ) {
            throw new UploadStashNotLoggedInException(
                wfMessage( 'uploadstash-not-logged-in' )
            );
        }
 
        $dbr = $this->repo->getReplicaDB();
        $res = $dbr->select(
            'uploadstash',
            'us_key',
            [ 'us_user' => $this->userId ],
            __METHOD__
        );
 
        if ( !is_object( $res ) || $res->numRows() == 0 ) {
            // nothing to do.
            return false;
        }
 
        // finish the read before starting writes.
        $keys = [];
        foreach ( $res as $row ) {
            array_push( $keys, $row->us_key );
        }
 
        return $keys;
    }
 
    public static function getExtensionForPath( $path ) {
        global $wgFileBlacklist;
        // Does this have an extension?
        $n = strrpos( $path, '.' );
        $extension = null;
        if ( $n !== false ) {
            $extension = $n ? substr( $path, $n + 1 ) : '';
        } else {
            // If not, assume that it should be related to the MIME type of the original file.
            $magic = MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer();
            $mimeType = $magic->guessMimeType( $path );
            $extensions = explode( ' ', $magic->getExtensionsForType( $mimeType ) );
            if ( count( $extensions ) ) {
                $extension = $extensions[0];
            }
        }
 
        if ( is_null( $extension ) ) {
            throw new UploadStashFileException(
                wfMessage( 'uploadstash-no-extension' )
            );
        }
 
        $extension = File::normalizeExtension( $extension );
        if ( in_array( $extension, $wgFileBlacklist ) ) {
            // The file should already be checked for being evil.
            // However, if somehow we got here, we definitely
            // don't want to give it an extension of .php and
            // put it in a web accesible directory.
            return '';
        }
 
        return $extension;
    }
 
    protected function fetchFileMetadata( $key, $readFromDB = DB_REPLICA ) {
        // populate $fileMetadata[$key]
        $dbr = null;
        if ( $readFromDB === DB_MASTER ) {
            // sometimes reading from the master is necessary, if there's replication lag.
            $dbr = $this->repo->getMasterDB();
        } else {
            $dbr = $this->repo->getReplicaDB();
        }
 
        $row = $dbr->selectRow(
            'uploadstash',
            [
                'us_user', 'us_key', 'us_orig_path', 'us_path', 'us_props',
                'us_size', 'us_sha1', 'us_mime', 'us_media_type',
                'us_image_width', 'us_image_height', 'us_image_bits',
                'us_source_type', 'us_timestamp', 'us_status',
            ],
            [ 'us_key' => $key ],
            __METHOD__
        );
 
        if ( !is_object( $row ) ) {
            // key wasn't present in the database. this will happen sometimes.
            return false;
        }
 
        $this->fileMetadata[$key] = (array)$row;
        $this->fileMetadata[$key]['us_props'] = $dbr->decodeBlob( $row->us_props );
 
        return true;
    }
 
    protected function initFile( $key ) {
        $file = new UploadStashFile( $this->repo, $this->fileMetadata[$key]['us_path'], $key );
        if ( $file->getSize() === 0 ) {
            throw new UploadStashZeroLengthFileException(
                wfMessage( 'uploadstash-zero-length' )
            );
        }
        $this->files[$key] = $file;
 
        return true;
    }
}
 
class UploadStashFile extends UnregisteredLocalFile {
    private $fileKey;
    private $urlName;
    protected $url;
 
    public function __construct( $repo, $path, $key ) {
        $this->fileKey = $key;
 
        // resolve mwrepo:// urls
        if ( FileRepo::isVirtualUrl( $path ) ) {
            $path = $repo->resolveVirtualUrl( $path );
        } else {
            // check if path appears to be sane, no parent traversals,
            // and is in this repo's temp zone.
            $repoTempPath = $repo->getZonePath( 'temp' );
            if ( ( !$repo->validateFilename( $path ) ) ||
                ( strpos( $path, $repoTempPath ) !== 0 )
            ) {
                wfDebug( "UploadStash: tried to construct an UploadStashFile "
                    . "from a file that should already exist at '$path', but path is not valid\n" );
                throw new UploadStashBadPathException(
                    wfMessage( 'uploadstash-bad-path-invalid' )
                );
            }
 
            // check if path exists! and is a plain file.
            if ( !$repo->fileExists( $path ) ) {
                wfDebug( "UploadStash: tried to construct an UploadStashFile from "
                    . "a file that should already exist at '$path', but path is not found\n" );
                throw new UploadStashFileNotFoundException(
                    wfMessage( 'uploadstash-file-not-found-not-exists' )
                );
            }
        }
 
        parent::__construct( false, $repo, $path, false );
 
        $this->name = basename( $this->path );
    }
 
    public function getDescriptionUrl() {
        return $this->getUrl();
    }
 
    public function getThumbPath( $thumbName = false ) {
        $path = dirname( $this->path );
        if ( $thumbName !== false ) {
            $path .= "/$thumbName";
        }
 
        return $path;
    }
 
    function thumbName( $params, $flags = 0 ) {
        return $this->generateThumbName( $this->getUrlName(), $params );
    }
 
    private function getSpecialUrl( $subPage ) {
        return SpecialPage::getTitleFor( 'UploadStash', $subPage )->getLocalURL();
    }
 
    public function getThumbUrl( $thumbName = false ) {
        wfDebug( __METHOD__ . " getting for $thumbName \n" );
 
        return $this->getSpecialUrl( 'thumb/' . $this->getUrlName() . '/' . $thumbName );
    }
 
    public function getUrlName() {
        if ( !$this->urlName ) {
            $this->urlName = $this->fileKey;
        }
 
        return $this->urlName;
    }
 
    public function getUrl() {
        if ( !isset( $this->url ) ) {
            $this->url = $this->getSpecialUrl( 'file/' . $this->getUrlName() );
        }
 
        return $this->url;
    }
 
    public function getFullUrl() {
        return $this->getUrl();
    }
 
    public function getFileKey() {
        return $this->fileKey;
    }
 
    public function remove() {
        if ( !$this->repo->fileExists( $this->path ) ) {
            // Maybe the file's already been removed? This could totally happen in UploadBase.
            return true;
        }
 
        return $this->repo->freeTemp( $this->path );
    }
 
    public function exists() {
        return $this->repo->fileExists( $this->path );
    }
}