includes/upload/UploadStash.php

   1 <?php
   2 /**
   3  * UploadStash is intended to accomplish a few things:
   4  *   - enable applications to temporarily stash files without publishing them to the wiki.
   5  *      - Several parts of MediaWiki do this in similar ways: UploadBase, UploadWizard, and FirefoggChunkedExtension
   6  *        And there are several that reimplement stashing from scratch, in idiosyncratic ways. The idea is to unify them all here.
   7  *        Mostly all of them are the same except for storing some custom fields, which we subsume into the data array.
   8  *   - enable applications to find said files later, as long as the session or temp files haven't been purged.
   9  *   - enable the uploading user (and *ONLY* the uploading user) to access said files, and thumbnails of said files, via a URL.
  10  *     We accomplish this by making the session serve as a URL->file mapping, on the assumption that nobody else can access
  11  *     the session, even the uploading user. See SpecialUploadStash, which implements a web interface to some files stored this way.
  12  *
  13  */
  14 class UploadStash {
  15
  16         // Format of the key for files -- has to be suitable as a filename itself (e.g. ab12cd34ef.jpg)
  17         const KEY_FORMAT_REGEX = '/^[\w-]+\.\w*$/';
  18
  19         /**
  20          * repository that this uses to store temp files
  21          * public because we sometimes need to get a LocalFile within the same repo.
  22          *
  23          * @var LocalRepo
  24          */
  25         public $repo;
  26
  27         // array of initialized objects obtained from session (lazily initialized upon getFile())
  28         private $files = array();
  29
  30         // TODO: Once UploadBase starts using this, switch to use these constants rather than UploadBase::SESSION*
  31         // const SESSION_VERSION = 2;
  32         // const SESSION_KEYNAME = 'wsUploadData';
  33
  34         /**
  35          * Represents the session which contains temporarily stored files.
  36          * Designed to be compatible with the session stashing code in UploadBase (should replace it eventually)
  37          *
  38          * @param $repo FileRepo
  39          */
  40         public function __construct( $repo ) {
  41
  42                 // this might change based on wiki's configuration.
  43                 $this->repo = $repo;
  44
  45                 if ( ! isset( $_SESSION ) ) {
  46                         throw new UploadStashNotAvailableException( 'no session variable' );
  47                 }
  48
  49                 if ( !isset( $_SESSION[UploadBase::SESSION_KEYNAME] ) ) {
  50                         $_SESSION[UploadBase::SESSION_KEYNAME] = array();
  51                 }
  52
  53         }
  54
  55         /**
  56          * Get a file and its metadata from the stash.
  57          * May throw exception if session data cannot be parsed due to schema change, or key not found.
  58          *
  59          * @param $key Integer: key
  60          * @throws UploadStashFileNotFoundException
  61          * @throws UploadStashBadVersionException
  62          * @return UploadStashFile
  63          */
  64         public function getFile( $key ) {
  65                 if ( ! preg_match( self::KEY_FORMAT_REGEX, $key ) ) {
  66                         throw new UploadStashBadPathException( "key '$key' is not in a proper format" );
  67                 }
  68
  69                 if ( !isset( $this->files[$key] ) ) {
  70                         if ( !isset( $_SESSION[UploadBase::SESSION_KEYNAME][$key] ) ) {
  71                                 throw new UploadStashFileNotFoundException( "key '$key' not found in stash" );
  72                         }
  73
  74                         $data = $_SESSION[UploadBase::SESSION_KEYNAME][$key];
  75                         // guards against PHP class changing while session data doesn't
  76                         if ($data['version'] !== UploadBase::SESSION_VERSION ) {
  77                                 throw new UploadStashBadVersionException( $data['version'] . " does not match current version " . UploadBase::SESSION_VERSION );
  78                         }
  79
  80                         // separate the stashData into the path, and then the rest of the data
  81                         $path = $data['mTempPath'];
  82                         unset( $data['mTempPath'] );
  83
  84                         $file = new UploadStashFile( $this, $this->repo, $path, $key, $data );
  85                         if ( $file->getSize() === 0 ) {
  86                                 throw new UploadStashZeroLengthFileException( "File is zero length" );
  87                         }
  88                         $this->files[$key] = $file;
  89
  90                 }
  91                 return $this->files[$key];
  92         }
  93
  94         /**
  95          * Stash a file in a temp directory and record that we did this in the session, along with other metadata.
  96          * We store data in a flat key-val namespace because that's how UploadBase did it. This also means we have to
  97          * ensure that the key-val pairs in $data do not overwrite other required fields.
  98          *
  99          * @param $path String: path to file you want stashed
 100          * @param $data Array: optional, other data you want associated with the file. Do not use 'mTempPath', 'mFileProps', 'mFileSize', or 'version' as keys here
 101          * @param $key String: optional, unique key for this file in this session. Used for directory hashing when storing, otherwise not important
 102          * @throws UploadStashBadPathException
 103          * @throws UploadStashFileException
 104          * @return UploadStashFile: file, or null on failure
 105          */
 106         public function stashFile( $path, $data = array(), $key = null ) {
 107                 if ( ! file_exists( $path ) ) {
 108                         wfDebug( "UploadStash: tried to stash file at '$path', but it doesn't exist\n" );
 109                         throw new UploadStashBadPathException( "path doesn't exist" );
 110                 }
 111                 $fileProps = File::getPropsFromPath( $path );
 112
 113                 // we will be initializing from some tmpnam files that don't have extensions.
 114                 // most of MediaWiki assumes all uploaded files have good extensions. So, we fix this.
 115                 $extension = self::getExtensionForPath( $path );
 116                 if ( ! preg_match( "/\\.\\Q$extension\\E$/", $path ) ) {
 117                         $pathWithGoodExtension = "$path.$extension";
 118                         if ( ! rename( $path, $pathWithGoodExtension ) ) {
 119                                 throw new UploadStashFileException( "couldn't rename $path to have a better extension at $pathWithGoodExtension" );
 120                         }
 121                         $path = $pathWithGoodExtension;
 122                 }
 123
 124                 // If no key was supplied, use content hash. Also has the nice property of collapsing multiple identical files
 125                 // uploaded this session, which could happen if uploads had failed.
 126                 if ( is_null( $key ) ) {
 127                         $key = $fileProps['sha1'] . "." . $extension;
 128                 }
 129
 130                 if ( ! preg_match( self::KEY_FORMAT_REGEX, $key ) ) {
 131                         throw new UploadStashBadPathException( "key '$key' is not in a proper format" );
 132                 }
 133
 134                 // if not already in a temporary area, put it there
 135                 $status = $this->repo->storeTemp( basename( $path ), $path );
 136
 137                 if( ! $status->isOK() ) {
 138                         // It is a convention in MediaWiki to only return one error per API exception, even if multiple errors
 139                         // are available. We use reset() to pick the "first" thing that was wrong, preferring errors to warnings.
 140                         // This is a bit lame, as we may have more info in the $status and we're throwing it away, but to fix it means
 141                         // redesigning API errors significantly.
 142                         // $status->value just contains the virtual URL (if anything) which is probably useless to the caller
 143                         $error = $status->getErrorsArray();
 144                         $error = reset( $error );
 145                         if ( ! count( $error ) ) {
 146                                 $error = $status->getWarningsArray();
 147                                 $error = reset( $error );
 148                                 if ( ! count( $error ) ) {
 149                                         $error = array( 'unknown', 'no error recorded' );
 150                                 }
 151                         }
 152                         throw new UploadStashFileException( "error storing file in '$path': " . implode( '; ', $error ) );
 153                 }
 154                 $stashPath = $status->value;
 155
 156                 // required info we always store. Must trump any other application info in $data
 157                 // 'mTempPath', 'mFileSize', and 'mFileProps' are arbitrary names
 158                 // chosen for compatibility with UploadBase's way of doing this.
 159                 $requiredData = array(
 160                         'mTempPath' => $stashPath,
 161                         'mFileSize' => $fileProps['size'],
 162                         'mFileProps' => $fileProps,
 163                         'version' => UploadBase::SESSION_VERSION
 164                 );
 165
 166                 // now, merge required info and extra data into the session. (The extra data changes from application to application.
 167                 // UploadWizard wants different things than say FirefoggChunkedUpload.)
 168                 wfDebug( __METHOD__ . " storing under $key\n" );
 169                 $_SESSION[UploadBase::SESSION_KEYNAME][$key] = array_merge( $data, $requiredData );
 170
 171                 return $this->getFile( $key );
 172         }
 173
 174         /**
 175          * Remove all files from the stash.
 176          * Does not clean up files in the repo, just the record of them.
 177          * @return boolean: success
 178          */
 179         public function clear() {
 180                 $_SESSION[UploadBase::SESSION_KEYNAME] = array();
 181                 return true;
 182         }
 183
 184
 185         /**
 186          * Remove a particular file from the stash.
 187          * Does not clean up file in the repo, just the record of it.
 188          * @return boolean: success
 189          */
 190         public function removeFile( $key ) {
 191                 unset ( $_SESSION[UploadBase::SESSION_KEYNAME][$key] );
 192                 return true;
 193         }
 194
 195         /**
 196          * List all files in the stash.
 197          */
 198         public function listFiles() {
 199                 return array_keys( $_SESSION[UploadBase::SESSION_KEYNAME] );
 200         }
 201
 202         /**
 203          * Find or guess extension -- ensuring that our extension matches our mime type.
 204          * Since these files are constructed from php tempnames they may not start off
 205          * with an extension.
 206          * XXX this is somewhat redundant with the checks that ApiUpload.php does with incoming
 207          * uploads versus the desired filename. Maybe we can get that passed to us...
 208          */
 209         public static function getExtensionForPath( $path ) {
 210                 // Does this have an extension?
 211                 $n = strrpos( $path, '.' );
 212                 $extension = null;
 213                 if ( $n !== false ) {
 214                         $extension = $n ? substr( $path, $n + 1 ) : '';
 215                 } else {
 216                         // If not, assume that it should be related to the mime type of the original file.
 217                         $magic = MimeMagic::singleton();
 218                         $mimeType = $magic->guessMimeType( $path );
 219                         $extensions = explode( ' ', MimeMagic::singleton()->getExtensionsForType( $mimeType ) );
 220                         if ( count( $extensions ) ) {
 221                                 $extension = $extensions[0];
 222                         }
 223                 }
 224
 225                 if ( is_null( $extension ) ) {
 226                         throw new UploadStashFileException( "extension is null" );
 227                 }
 228
 229                 return File::normalizeExtension( $extension );
 230         }
 231
 232 }
 233
 234 class UploadStashFile extends UnregisteredLocalFile {
 235         private $sessionStash;
 236         private $sessionKey;
 237         private $sessionData;
 238         private $urlName;
 239         protected $url;
 240
 241         /**
 242          * A LocalFile wrapper around a file that has been temporarily stashed, so we can do things like create thumbnails for it
 243          * Arguably UnregisteredLocalFile should be handling its own file repo but that class is a bit retarded currently
 244          *
 245          * @param $stash UploadStash: useful for obtaining config, stashing transformed files
 246          * @param $repo FSRepo: repository where we should find the path
 247          * @param $path String: path to file
 248          * @param $key String: key to store the path and any stashed data under
 249          * @param $data String: any other data we want stored with this file
 250          * @throws UploadStashBadPathException
 251          * @throws UploadStashFileNotFoundException
 252          */
 253         public function __construct( $stash, $repo, $path, $key, $data ) {
 254                 $this->sessionStash = $stash;
 255                 $this->sessionKey = $key;
 256                 $this->sessionData = $data;
 257
 258                 // resolve mwrepo:// urls
 259                 if ( $repo->isVirtualUrl( $path ) ) {
 260                         $path = $repo->resolveVirtualUrl( $path );
 261                 }
 262
 263                 // check if path appears to be sane, no parent traversals, and is in this repo's temp zone.
 264                 $repoTempPath = $repo->getZonePath( 'temp' );
 265                 if ( ( ! $repo->validateFilename( $path ) ) ||
 266                                 ( strpos( $path, $repoTempPath ) !== 0 ) ) {
 267                         wfDebug( "UploadStash: tried to construct an UploadStashFile from a file that should already exist at '$path', but path is not valid\n" );
 268                         throw new UploadStashBadPathException( 'path is not valid' );
 269                 }
 270
 271                 // check if path exists! and is a plain file.
 272                 if ( ! $repo->fileExists( $path, FileRepo::FILES_ONLY ) ) {
 273                         wfDebug( "UploadStash: tried to construct an UploadStashFile from a file that should already exist at '$path', but path is not found\n" );
 274                         throw new UploadStashFileNotFoundException( 'cannot find path, or not a plain file' );
 275                 }
 276
 277                 parent::__construct( false, $repo, $path, false );
 278
 279                 $this->name = basename( $this->path );
 280         }
 281
 282         /**
 283          * A method needed by the file transforming and scaling routines in File.php
 284          * We do not necessarily care about doing the description at this point
 285          * However, we also can't return the empty string, as the rest of MediaWiki demands this (and calls to imagemagick
 286          * convert require it to be there)
 287          *
 288          * @return String: dummy value
 289          */
 290         public function getDescriptionUrl() {
 291                 return $this->getUrl();
 292         }
 293
 294         /**
 295          * Get the path for the thumbnail (actually any transformation of this file)
 296          * The actual argument is the result of thumbName although we seem to have
 297          * buggy code elsewhere that expects a boolean 'suffix'
 298          *
 299          * @param $thumbName String: name of thumbnail (e.g. "120px-123456.jpg" ), or false to just get the path
 300          * @return String: path thumbnail should take on filesystem, or containing directory if thumbname is false
 301          */
 302         public function getThumbPath( $thumbName = false ) {
 303                 $path = dirname( $this->path );
 304                 if ( $thumbName !== false ) {
 305                         $path .= "/$thumbName";
 306                 }
 307                 return $path;
 308         }
 309
 310         /**
 311          * Return the file/url base name of a thumbnail with the specified parameters.
 312          * We override this because we want to use the pretty url name instead of the
 313          * ugly file name.
 314          *
 315          * @param $params Array: handler-specific parameters
 316          * @return String: base name for URL, like '120px-12345.jpg', or null if there is no handler
 317          */
 318         function thumbName( $params ) {
 319                 return $this->generateThumbName( $this->getUrlName(), $params );
 320         }
 321
 322         /**
 323          * Helper function -- given a 'subpage', return the local URL e.g. /wiki/Special:UploadStash/subpage
 324          * @param {String} $subPage
 325          * @return {String} local URL for this subpage in the Special:UploadStash space.
 326          */
 327         private function getSpecialUrl( $subPage ) {
 328                 return SpecialPage::getTitleFor( 'UploadStash', $subPage )->getLocalURL();
 329         }
 330
 331         /**
 332          * Get a URL to access the thumbnail
 333          * This is required because the model of how files work requires that
 334          * the thumbnail urls be predictable. However, in our model the URL is not based on the filename
 335          * (that's hidden in the session)
 336          *
 337          * @param $thumbName String: basename of thumbnail file -- however, we don't want to use the file exactly
 338          * @return String: URL to access thumbnail, or URL with partial path
 339          */
 340         public function getThumbUrl( $thumbName = false ) {
 341                 wfDebug( __METHOD__ . " getting for $thumbName \n" );
 342                 return $this->getSpecialUrl( 'thumb/' . $this->getUrlName() . '/' . $thumbName );
 343         }
 344
 345         /**
 346          * The basename for the URL, which we want to not be related to the filename.
 347          * Will also be used as the lookup key for a thumbnail file.
 348          *
 349          * @return String: base url name, like '120px-123456.jpg'
 350          */
 351         public function getUrlName() {
 352                 if ( ! $this->urlName ) {
 353                         $this->urlName = $this->sessionKey;
 354                 }
 355                 return $this->urlName;
 356         }
 357
 358         /**
 359          * Return the URL of the file, if for some reason we wanted to download it
 360          * We tend not to do this for the original file, but we do want thumb icons
 361          *
 362          * @return String: url
 363          */
 364         public function getUrl() {
 365                 if ( !isset( $this->url ) ) {
 366                         $this->url = $this->getSpecialUrl( 'file/' . $this->getUrlName() );
 367                 }
 368                 return $this->url;
 369         }
 370
 371         /**
 372          * Parent classes use this method, for no obvious reason, to return the path (relative to wiki root, I assume).
 373          * But with this class, the URL is unrelated to the path.
 374          *
 375          * @return String: url
 376          */
 377         public function getFullUrl() {
 378                 return $this->getUrl();
 379         }
 380
 381         /**
 382          * Getter for session key (the session-unique id by which this file's location & metadata is stored in the session)
 383          *
 384          * @return String: session key
 385          */
 386         public function getSessionKey() {
 387                 return $this->sessionKey;
 388         }
 389
 390         /**
 391          * Remove the associated temporary file
 392          * @return Status: success
 393          */
 394         public function remove() {
 395                 return $this->repo->freeTemp( $this->path );
 396         }
 397
 398 }
 399
 400 class UploadStashNotAvailableException extends MWException {};
 401 class UploadStashFileNotFoundException extends MWException {};
 402 class UploadStashBadPathException extends MWException {};
 403 class UploadStashBadVersionException extends MWException {};
 404 class UploadStashFileException extends MWException {};
 405 class UploadStashZeroLengthFileException extends MWException {};
 406