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*$/';

        public $repo;

        // array of initialized repo objects
        protected $files = array();

        // cache of the file metadata that's stored in the database
        protected $fileMetadata = array();

        // fileprops cache
        protected $fileProps = array();

        // 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, $key ) ) {
                        throw new UploadStashBadPathException( "key '$key' is not in a proper format" );
                }

                if ( !$noAuth ) {
                        if ( !$this->isLoggedIn ) {
                                throw new UploadStashNotLoggedInException( __METHOD__ . ' No user is logged in, files must belong to users' );
                        }
                }

                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( "key '$key' not found in stash" );
                        }

                        // create $this->files[$key]
                        $this->initFile( $key );

                        // fetch fileprops
                        $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" );
                        throw new UploadStashBadPathException( "path doesn't exist" );
                }

                if ( !$noAuth ) {
                        if ( $this->fileMetadata[$key]['us_user'] != $this->userId ) {
                                throw new UploadStashWrongOwnerException( "This file ($key) doesn't belong to the current user." );
                        }
                }

                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 ( ! file_exists( $path ) ) {
                        wfDebug( __METHOD__ . " tried to stash file at '$path', but it doesn't exist\n" );
                        throw new UploadStashBadPathException( "path doesn't exist" );
                }
                $fileProps = FSFile::getPropsFromPath( $path );
                wfDebug( __METHOD__ . " stashing file at '$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";
                        if ( ! rename( $path, $pathWithGoodExtension ) ) {
                                throw new UploadStashFileException( "couldn't rename $path to have a better extension at $pathWithGoodExtension" );
                        }
                        $path = $pathWithGoodExtension;
                }

                // 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 = wfBaseConvert( $sec . $usec, 10, 36 ) . '.' .
                        wfBaseConvert( mt_rand(), 10, 36 ) . '.'.
                        $this->userId . '.' .
                        $extension;

                $this->fileProps[$key] = $fileProps;

                if ( ! preg_match( self::KEY_FORMAT_REGEX, $key ) ) {
                        throw new UploadStashBadPathException( "key '$key' is not in a proper format" );
                }

                wfDebug( __METHOD__ . " key for '$path': $key\n" );

                // if not already in a temporary area, put it there
                $storeStatus = $this->repo->storeTemp( basename( $path ), $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 = array( 'unknown', 'no error recorded' );
                                }
                        }
                        // at this point, $error should contain the single "most important" error, plus any parameters.
                        throw new UploadStashFileException( "Error storing file in '$path': " . wfMessage( $error )->text() );
                }
                $stashPath = $storeStatus->value;

                // fetch the current user ID
                if ( !$this->isLoggedIn ) {
                        throw new UploadStashNotLoggedInException( __METHOD__ . ' No user is logged in, files must belong to users' );
                }

                // insert the file metadata into the db.
                wfDebug( __METHOD__ . " inserting $stashPath under $key\n" );
                $dbw = $this->repo->getMasterDb();

                $this->fileMetadata[$key] = array(
                        'us_id' => $dbw->nextSequenceValue( 'uploadstash_us_id_seq' ),
                        'us_user' => $this->userId,
                        'us_key' => $key,
                        'us_orig_path' => $path,
                        'us_path' => $stashPath, // virtual URL
                        '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 necesary.
                $this->fileMetadata[$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( __METHOD__ . ' No user is logged in, files must belong to users' );
                }

                wfDebug( __METHOD__ . ' clearing all rows for user ' . $this->userId . "\n" );
                $dbw = $this->repo->getMasterDb();
                $dbw->delete(
                        'uploadstash',
                        array( 'us_user' => $this->userId ),
                        __METHOD__
                );

                # destroy objects.
                $this->files = array();
                $this->fileMetadata = array();

                return true;
        }

        public function removeFile( $key ) {
                if ( !$this->isLoggedIn ) {
                        throw new UploadStashNotLoggedInException( __METHOD__ . ' No user is logged in, files must belong to users' );
                }

                $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',
                        array( 'us_key' => $key ),
                        __METHOD__
                );

                if( !$row ) {
                        throw new UploadStashNoSuchKeyException( "No such key ($key), cannot remove" );
                }

                if ( $row->us_user != $this->userId ) {
                        throw new UploadStashWrongOwnerException( "Can't delete: the file ($key) doesn't belong to this user." );
                }

                return $this->removeFileNoAuth( $key );
        }


        public function removeFileNoAuth( $key ) {
                wfDebug( __METHOD__ . " clearing row $key\n" );

                $dbw = $this->repo->getMasterDb();

                // this gets its own transaction since it's called serially by the cleanupUploadStash maintenance script
                $dbw->begin( __METHOD__ );
                $dbw->delete(
                        'uploadstash',
                        array( 'us_key' => $key ),
                        __METHOD__
                );
                $dbw->commit( __METHOD__ );

                // TODO: look into UnregisteredLocalFile and find out why the rv here is sometimes wrong (false when file was removed)
                // for now, ignore.
                $this->files[$key]->remove();

                unset( $this->files[$key] );
                unset( $this->fileMetadata[$key] );

                return true;
        }

        public function listFiles() {
                if ( !$this->isLoggedIn ) {
                        throw new UploadStashNotLoggedInException( __METHOD__ . ' No user is logged in, files must belong to users' );
                }

                $dbr = $this->repo->getSlaveDb();
                $res = $dbr->select(
                        'uploadstash',
                        'us_key',
                        array( 'us_user' => $this->userId ),
                        __METHOD__
                );

                if ( !is_object( $res ) || $res->numRows() == 0 ) {
                        // nothing to do.
                        return false;
                }

                // finish the read before starting writes.
                $keys = array();
                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 = MimeMagic::singleton();
                        $mimeType = $magic->guessMimeType( $path );
                        $extensions = explode( ' ', MimeMagic::singleton()->getExtensionsForType( $mimeType ) );
                        if ( count( $extensions ) ) {
                                $extension = $extensions[0];
                        }
                }

                if ( is_null( $extension ) ) {
                        throw new UploadStashFileException( "extension is null" );
                }

                $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_SLAVE ) {
                // 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->getSlaveDb();
                }

                $row = $dbr->selectRow(
                        'uploadstash',
                        '*',
                        array( '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;

                return true;
        }

        protected function initFile( $key ) {
                $file = new UploadStashFile( $this->repo, $this->fileMetadata[$key]['us_path'], $key );
                if ( $file->getSize() === 0 ) {
                        throw new UploadStashZeroLengthFileException( "File is 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 ( $repo->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( 'path is not valid' );
                        }

                        // check if path exists! and is a plain file.
                        if ( ! $repo->fileExists( $path, FileRepo::FILES_ONLY ) ) {
                                wfDebug( "UploadStash: tried to construct an UploadStashFile from a file that should already exist at '$path', but path is not found\n" );
                                throw new UploadStashFileNotFoundException( 'cannot find path, or not a plain file' );
                        }
                }

                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 ) {
                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, FileRepo::FILES_ONLY ) ) {
                        // 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, FileRepo::FILES_ONLY );
        }

}

class UploadStashNotAvailableException extends MWException {};
class UploadStashFileNotFoundException extends MWException {};
class UploadStashBadPathException extends MWException {};
class UploadStashFileException extends MWException {};
class UploadStashZeroLengthFileException extends MWException {};
class UploadStashNotLoggedInException extends MWException {};
class UploadStashWrongOwnerException extends MWException {};
class UploadStashNoSuchKeyException extends MWException {};