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