includes/upload/UploadStash.php

   1 <?php
   2 /**
   3  * Temporary storage for uploaded files.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  */
  22
  23 /**
  24  * UploadStash is intended to accomplish a few things:
  25  *   - Enable applications to temporarily stash files without publishing them to
  26  *     the wiki.
  27  *      - Several parts of MediaWiki do this in similar ways: UploadBase,
  28  *        UploadWizard, and FirefoggChunkedExtension.
  29  *        And there are several that reimplement stashing from scratch, in
  30  *        idiosyncratic ways. The idea is to unify them all here.
  31  *        Mostly all of them are the same except for storing some custom fields,
  32  *        which we subsume into the data array.
  33  *   - Enable applications to find said files later, as long as the db table or
  34  *     temp files haven't been purged.
  35  *   - Enable the uploading user (and *ONLY* the uploading user) to access said
  36  *     files, and thumbnails of said files, via a URL. We accomplish this using
  37  *     a database table, with ownership checking as you might expect. See
  38  *     SpecialUploadStash, which implements a web interface to some files stored
  39  *     this way.
  40  *
  41  * UploadStash right now is *mostly* intended to show you one user's slice of
  42  * the entire stash. The user parameter is only optional because there are few
  43  * cases where we clean out the stash from an automated script. In the future we
  44  * might refactor this.
  45  *
  46  * UploadStash represents the entire stash of temporary files.
  47  * UploadStashFile is a filestore for the actual physical disk files.
  48  * UploadFromStash extends UploadBase, and represents a single stashed file as
  49  * it is moved from the stash to the regular file repository
  50  *
  51  * @ingroup Upload
  52  */
  53 class UploadStash {
  54         // Format of the key for files -- has to be suitable as a filename itself (e.g. ab12cd34ef.jpg)
  55         const KEY_FORMAT_REGEX = '/^[\w\-\.]+\.\w*$/';
  56         const MAX_US_PROPS_SIZE = 65535;
  57
  58         /**
  59          * repository that this uses to store temp files
  60          * public because we sometimes need to get a LocalFile within the same repo.
  61          *
  62          * @var LocalRepo
  63          */
  64         public $repo;
  65
  66         // array of initialized repo objects
  67         protected $files = [];
  68
  69         // cache of the file metadata that's stored in the database
  70         protected $fileMetadata = [];
  71
  72         // fileprops cache
  73         protected $fileProps = [];
  74
  75         // current user
  76         protected $user, $userId, $isLoggedIn;
  77
  78         /**
  79          * Represents a temporary filestore, with metadata in the database.
  80          * Designed to be compatible with the session stashing code in UploadBase
  81          * (should replace it eventually).
  82          *
  83          * @param FileRepo $repo
  84          * @param User|null $user
  85          */
  86         public function __construct( FileRepo $repo, $user = null ) {
  87                 // this might change based on wiki's configuration.
  88                 $this->repo = $repo;
  89
  90                 // if a user was passed, use it. otherwise, attempt to use the global.
  91                 // this keeps FileRepo from breaking when it creates an UploadStash object
  92                 global $wgUser;
  93                 $this->user = $user ?: $wgUser;
  94
  95                 if ( is_object( $this->user ) ) {
  96                         $this->userId = $this->user->getId();
  97                         $this->isLoggedIn = $this->user->isLoggedIn();
  98                 }
  99         }
 100
 101         /**
 102          * Get a file and its metadata from the stash.
 103          * The noAuth param is a bit janky but is required for automated scripts
 104          * which clean out the stash.
 105          *
 106          * @param string $key Key under which file information is stored
 107          * @param bool $noAuth (optional) Don't check authentication. Used by maintenance scripts.
 108          * @throws UploadStashFileNotFoundException
 109          * @throws UploadStashNotLoggedInException
 110          * @throws UploadStashWrongOwnerException
 111          * @throws UploadStashBadPathException
 112          * @return UploadStashFile
 113          */
 114         public function getFile( $key, $noAuth = false ) {
 115                 if ( !preg_match( self::KEY_FORMAT_REGEX, $key ) ) {
 116                         throw new UploadStashBadPathException(
 117                                 wfMessage( 'uploadstash-bad-path-bad-format', $key )
 118                         );
 119                 }
 120
 121                 if ( !$noAuth && !$this->isLoggedIn ) {
 122                         throw new UploadStashNotLoggedInException(
 123                                 wfMessage( 'uploadstash-not-logged-in' )
 124                         );
 125                 }
 126
 127                 if ( !isset( $this->fileMetadata[$key] ) ) {
 128                         if ( !$this->fetchFileMetadata( $key ) ) {
 129                                 // If nothing was received, it's likely due to replication lag.
 130                                 // Check the master to see if the record is there.
 131                                 $this->fetchFileMetadata( $key, DB_MASTER );
 132                         }
 133
 134                         if ( !isset( $this->fileMetadata[$key] ) ) {
 135                                 throw new UploadStashFileNotFoundException(
 136                                         wfMessage( 'uploadstash-file-not-found', $key )
 137                                 );
 138                         }
 139
 140                         // create $this->files[$key]
 141                         $this->initFile( $key );
 142
 143                         // fetch fileprops
 144                         if ( strlen( $this->fileMetadata[$key]['us_props'] ) ) {
 145                                 $this->fileProps[$key] = unserialize( $this->fileMetadata[$key]['us_props'] );
 146                         } else { // b/c for rows with no us_props
 147                                 wfDebug( __METHOD__ . " fetched props for $key from file\n" );
 148                                 $path = $this->fileMetadata[$key]['us_path'];
 149                                 $this->fileProps[$key] = $this->repo->getFileProps( $path );
 150                         }
 151                 }
 152
 153                 if ( !$this->files[$key]->exists() ) {
 154                         wfDebug( __METHOD__ . " tried to get file at $key, but it doesn't exist\n" );
 155                         // @todo Is this not an UploadStashFileNotFoundException case?
 156                         throw new UploadStashBadPathException(
 157                                 wfMessage( 'uploadstash-bad-path' )
 158                         );
 159                 }
 160
 161                 if ( !$noAuth && $this->fileMetadata[$key]['us_user'] != $this->userId ) {
 162                         throw new UploadStashWrongOwnerException(
 163                                 wfMessage( 'uploadstash-wrong-owner', $key )
 164                         );
 165                 }
 166
 167                 return $this->files[$key];
 168         }
 169
 170         /**
 171          * Getter for file metadata.
 172          *
 173          * @param string $key Key under which file information is stored
 174          * @return array
 175          */
 176         public function getMetadata( $key ) {
 177                 $this->getFile( $key );
 178
 179                 return $this->fileMetadata[$key];
 180         }
 181
 182         /**
 183          * Getter for fileProps
 184          *
 185          * @param string $key Key under which file information is stored
 186          * @return array
 187          */
 188         public function getFileProps( $key ) {
 189                 $this->getFile( $key );
 190
 191                 return $this->fileProps[$key];
 192         }
 193
 194         /**
 195          * Stash a file in a temp directory and record that we did this in the
 196          * database, along with other metadata.
 197          *
 198          * @param string $path Path to file you want stashed
 199          * @param string|null $sourceType The type of upload that generated this file
 200          *   (currently, I believe, 'file' or null)
 201          * @throws UploadStashBadPathException
 202          * @throws UploadStashFileException
 203          * @throws UploadStashNotLoggedInException
 204          * @return UploadStashFile|null File, or null on failure
 205          */
 206         public function stashFile( $path, $sourceType = null ) {
 207                 if ( !is_file( $path ) ) {
 208                         wfDebug( __METHOD__ . " tried to stash file at '$path', but it doesn't exist\n" );
 209                         throw new UploadStashBadPathException(
 210                                 wfMessage( 'uploadstash-bad-path' )
 211                         );
 212                 }
 213
 214                 $mwProps = new MWFileProps( MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer() );
 215                 $fileProps = $mwProps->getPropsFromPath( $path, true );
 216                 wfDebug( __METHOD__ . " stashing file at '$path'\n" );
 217
 218                 // we will be initializing from some tmpnam files that don't have extensions.
 219                 // most of MediaWiki assumes all uploaded files have good extensions. So, we fix this.
 220                 $extension = self::getExtensionForPath( $path );
 221                 if ( !preg_match( "/\\.\\Q$extension\\E$/", $path ) ) {
 222                         $pathWithGoodExtension = "$path.$extension";
 223                 } else {
 224                         $pathWithGoodExtension = $path;
 225                 }
 226
 227                 // If no key was supplied, make one.  a mysql insertid would be totally
 228                 // reasonable here, except that for historical reasons, the key is this
 229                 // random thing instead.  At least it's not guessable.
 230                 // Some things that when combined will make a suitably unique key.
 231                 // see: http://www.jwz.org/doc/mid.html
 232                 list( $usec, $sec ) = explode( ' ', microtime() );
 233                 $usec = substr( $usec, 2 );
 234                 $key = Wikimedia\base_convert( $sec . $usec, 10, 36 ) . '.' .
 235                         Wikimedia\base_convert( mt_rand(), 10, 36 ) . '.' .
 236                         $this->userId . '.' .
 237                         $extension;
 238
 239                 $this->fileProps[$key] = $fileProps;
 240
 241                 if ( !preg_match( self::KEY_FORMAT_REGEX, $key ) ) {
 242                         throw new UploadStashBadPathException(
 243                                 wfMessage( 'uploadstash-bad-path-bad-format', $key )
 244                         );
 245                 }
 246
 247                 wfDebug( __METHOD__ . " key for '$path': $key\n" );
 248
 249                 // if not already in a temporary area, put it there
 250                 $storeStatus = $this->repo->storeTemp( basename( $pathWithGoodExtension ), $path );
 251
 252                 if ( !$storeStatus->isOK() ) {
 253                         // It is a convention in MediaWiki to only return one error per API
 254                         // exception, even if multiple errors are available. We use reset()
 255                         // to pick the "first" thing that was wrong, preferring errors to
 256                         // warnings. This is a bit lame, as we may have more info in the
 257                         // $storeStatus and we're throwing it away, but to fix it means
 258                         // redesigning API errors significantly.
 259                         // $storeStatus->value just contains the virtual URL (if anything)
 260                         // which is probably useless to the caller.
 261                         $error = $storeStatus->getErrorsArray();
 262                         $error = reset( $error );
 263                         if ( !count( $error ) ) {
 264                                 $error = $storeStatus->getWarningsArray();
 265                                 $error = reset( $error );
 266                                 if ( !count( $error ) ) {
 267                                         $error = [ 'unknown', 'no error recorded' ];
 268                                 }
 269                         }
 270                         // At this point, $error should contain the single "most important"
 271                         // error, plus any parameters.
 272                         $errorMsg = array_shift( $error );
 273                         throw new UploadStashFileException( wfMessage( $errorMsg, $error ) );
 274                 }
 275                 $stashPath = $storeStatus->value;
 276
 277                 // fetch the current user ID
 278                 if ( !$this->isLoggedIn ) {
 279                         throw new UploadStashNotLoggedInException(
 280                                 wfMessage( 'uploadstash-not-logged-in' )
 281                         );
 282                 }
 283
 284                 // insert the file metadata into the db.
 285                 wfDebug( __METHOD__ . " inserting $stashPath under $key\n" );
 286                 $dbw = $this->repo->getMasterDB();
 287
 288                 $serializedFileProps = serialize( $fileProps );
 289                 if ( strlen( $serializedFileProps ) > self::MAX_US_PROPS_SIZE ) {
 290                         // Database is going to truncate this and make the field invalid.
 291                         // Prioritize important metadata over file handler metadata.
 292                         // File handler should be prepared to regenerate invalid metadata if needed.
 293                         $fileProps['metadata'] = false;
 294                         $serializedFileProps = serialize( $fileProps );
 295                 }
 296
 297                 $this->fileMetadata[$key] = [
 298                         'us_user' => $this->userId,
 299                         'us_key' => $key,
 300                         'us_orig_path' => $path,
 301                         'us_path' => $stashPath, // virtual URL
 302                         'us_props' => $dbw->encodeBlob( $serializedFileProps ),
 303                         'us_size' => $fileProps['size'],
 304                         'us_sha1' => $fileProps['sha1'],
 305                         'us_mime' => $fileProps['mime'],
 306                         'us_media_type' => $fileProps['media_type'],
 307                         'us_image_width' => $fileProps['width'],
 308                         'us_image_height' => $fileProps['height'],
 309                         'us_image_bits' => $fileProps['bits'],
 310                         'us_source_type' => $sourceType,
 311                         'us_timestamp' => $dbw->timestamp(),
 312                         'us_status' => 'finished'
 313                 ];
 314
 315                 $dbw->insert(
 316                         'uploadstash',
 317                         $this->fileMetadata[$key],
 318                         __METHOD__
 319                 );
 320
 321                 // store the insertid in the class variable so immediate retrieval
 322                 // (possibly laggy) isn't necessary.
 323                 $this->fileMetadata[$key]['us_id'] = $dbw->insertId();
 324
 325                 # create the UploadStashFile object for this file.
 326                 $this->initFile( $key );
 327
 328                 return $this->getFile( $key );
 329         }
 330
 331         /**
 332          * Remove all files from the stash.
 333          * Does not clean up files in the repo, just the record of them.
 334          *
 335          * @throws UploadStashNotLoggedInException
 336          * @return bool Success
 337          */
 338         public function clear() {
 339                 if ( !$this->isLoggedIn ) {
 340                         throw new UploadStashNotLoggedInException(
 341                                 wfMessage( 'uploadstash-not-logged-in' )
 342                         );
 343                 }
 344
 345                 wfDebug( __METHOD__ . ' clearing all rows for user ' . $this->userId . "\n" );
 346                 $dbw = $this->repo->getMasterDB();
 347                 $dbw->delete(
 348                         'uploadstash',
 349                         [ 'us_user' => $this->userId ],
 350                         __METHOD__
 351                 );
 352
 353                 # destroy objects.
 354                 $this->files = [];
 355                 $this->fileMetadata = [];
 356
 357                 return true;
 358         }
 359
 360         /**
 361          * Remove a particular file from the stash.  Also removes it from the repo.
 362          *
 363          * @param string $key
 364          * @throws UploadStashNoSuchKeyException|UploadStashNotLoggedInException
 365          * @throws UploadStashWrongOwnerException
 366          * @return bool Success
 367          */
 368         public function removeFile( $key ) {
 369                 if ( !$this->isLoggedIn ) {
 370                         throw new UploadStashNotLoggedInException(
 371                                 wfMessage( 'uploadstash-not-logged-in' )
 372                         );
 373                 }
 374
 375                 $dbw = $this->repo->getMasterDB();
 376
 377                 // this is a cheap query. it runs on the master so that this function
 378                 // still works when there's lag. It won't be called all that often.
 379                 $row = $dbw->selectRow(
 380                         'uploadstash',
 381                         'us_user',
 382                         [ 'us_key' => $key ],
 383                         __METHOD__
 384                 );
 385
 386                 if ( !$row ) {
 387                         throw new UploadStashNoSuchKeyException(
 388                                 wfMessage( 'uploadstash-no-such-key', $key )
 389                         );
 390                 }
 391
 392                 if ( $row->us_user != $this->userId ) {
 393                         throw new UploadStashWrongOwnerException(
 394                                 wfMessage( 'uploadstash-wrong-owner', $key )
 395                         );
 396                 }
 397
 398                 return $this->removeFileNoAuth( $key );
 399         }
 400
 401         /**
 402          * Remove a file (see removeFile), but doesn't check ownership first.
 403          *
 404          * @param string $key
 405          * @return bool Success
 406          */
 407         public function removeFileNoAuth( $key ) {
 408                 wfDebug( __METHOD__ . " clearing row $key\n" );
 409
 410                 // Ensure we have the UploadStashFile loaded for this key
 411                 $this->getFile( $key, true );
 412
 413                 $dbw = $this->repo->getMasterDB();
 414
 415                 $dbw->delete(
 416                         'uploadstash',
 417                         [ 'us_key' => $key ],
 418                         __METHOD__
 419                 );
 420
 421                 /** @todo Look into UnregisteredLocalFile and find out why the rv here is
 422                  *  sometimes wrong (false when file was removed). For now, ignore.
 423                  */
 424                 $this->files[$key]->remove();
 425
 426                 unset( $this->files[$key] );
 427                 unset( $this->fileMetadata[$key] );
 428
 429                 return true;
 430         }
 431
 432         /**
 433          * List all files in the stash.
 434          *
 435          * @throws UploadStashNotLoggedInException
 436          * @return array|false
 437          */
 438         public function listFiles() {
 439                 if ( !$this->isLoggedIn ) {
 440                         throw new UploadStashNotLoggedInException(
 441                                 wfMessage( 'uploadstash-not-logged-in' )
 442                         );
 443                 }
 444
 445                 $dbr = $this->repo->getReplicaDB();
 446                 $res = $dbr->select(
 447                         'uploadstash',
 448                         'us_key',
 449                         [ 'us_user' => $this->userId ],
 450                         __METHOD__
 451                 );
 452
 453                 if ( !is_object( $res ) || $res->numRows() == 0 ) {
 454                         // nothing to do.
 455                         return false;
 456                 }
 457
 458                 // finish the read before starting writes.
 459                 $keys = [];
 460                 foreach ( $res as $row ) {
 461                         array_push( $keys, $row->us_key );
 462                 }
 463
 464                 return $keys;
 465         }
 466
 467         /**
 468          * Find or guess extension -- ensuring that our extension matches our MIME type.
 469          * Since these files are constructed from php tempnames they may not start off
 470          * with an extension.
 471          * XXX this is somewhat redundant with the checks that ApiUpload.php does with incoming
 472          * uploads versus the desired filename. Maybe we can get that passed to us...
 473          * @param string $path
 474          * @throws UploadStashFileException
 475          * @return string
 476          */
 477         public static function getExtensionForPath( $path ) {
 478                 global $wgFileBlacklist;
 479                 // Does this have an extension?
 480                 $n = strrpos( $path, '.' );
 481                 $extension = null;
 482                 if ( $n !== false ) {
 483                         $extension = $n ? substr( $path, $n + 1 ) : '';
 484                 } else {
 485                         // If not, assume that it should be related to the MIME type of the original file.
 486                         $magic = MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer();
 487                         $mimeType = $magic->guessMimeType( $path );
 488                         $extensions = explode( ' ', $magic->getExtensionsForType( $mimeType ) );
 489                         if ( count( $extensions ) ) {
 490                                 $extension = $extensions[0];
 491                         }
 492                 }
 493
 494                 if ( is_null( $extension ) ) {
 495                         throw new UploadStashFileException(
 496                                 wfMessage( 'uploadstash-no-extension' )
 497                         );
 498                 }
 499
 500                 $extension = File::normalizeExtension( $extension );
 501                 if ( in_array( $extension, $wgFileBlacklist ) ) {
 502                         // The file should already be checked for being evil.
 503                         // However, if somehow we got here, we definitely
 504                         // don't want to give it an extension of .php and
 505                         // put it in a web accesible directory.
 506                         return '';
 507                 }
 508
 509                 return $extension;
 510         }
 511
 512         /**
 513          * Helper function: do the actual database query to fetch file metadata.
 514          *
 515          * @param string $key
 516          * @param int $readFromDB Constant (default: DB_REPLICA)
 517          * @return bool
 518          */
 519         protected function fetchFileMetadata( $key, $readFromDB = DB_REPLICA ) {
 520                 // populate $fileMetadata[$key]
 521                 $dbr = null;
 522                 if ( $readFromDB === DB_MASTER ) {
 523                         // sometimes reading from the master is necessary, if there's replication lag.
 524                         $dbr = $this->repo->getMasterDB();
 525                 } else {
 526                         $dbr = $this->repo->getReplicaDB();
 527                 }
 528
 529                 $row = $dbr->selectRow(
 530                         'uploadstash',
 531                         [
 532                                 'us_user', 'us_key', 'us_orig_path', 'us_path', 'us_props',
 533                                 'us_size', 'us_sha1', 'us_mime', 'us_media_type',
 534                                 'us_image_width', 'us_image_height', 'us_image_bits',
 535                                 'us_source_type', 'us_timestamp', 'us_status',
 536                         ],
 537                         [ 'us_key' => $key ],
 538                         __METHOD__
 539                 );
 540
 541                 if ( !is_object( $row ) ) {
 542                         // key wasn't present in the database. this will happen sometimes.
 543                         return false;
 544                 }
 545
 546                 $this->fileMetadata[$key] = (array)$row;
 547                 $this->fileMetadata[$key]['us_props'] = $dbr->decodeBlob( $row->us_props );
 548
 549                 return true;
 550         }
 551
 552         /**
 553          * Helper function: Initialize the UploadStashFile for a given file.
 554          *
 555          * @param string $key Key under which to store the object
 556          * @throws UploadStashZeroLengthFileException
 557          * @return bool
 558          */
 559         protected function initFile( $key ) {
 560                 $file = new UploadStashFile( $this->repo, $this->fileMetadata[$key]['us_path'], $key );
 561                 if ( $file->getSize() === 0 ) {
 562                         throw new UploadStashZeroLengthFileException(
 563                                 wfMessage( 'uploadstash-zero-length' )
 564                         );
 565                 }
 566                 $this->files[$key] = $file;
 567
 568                 return true;
 569         }
 570 }