includes/filerepo/FSRepo.php

   1 <?php
   2
   3 /**
   4  * A repository for files accessible via the local filesystem. Does not support
   5  * database access or registration.
   6  * @ingroup FileRepo
   7  */
   8 class FSRepo extends FileRepo {
   9         var $directory, $deletedDir, $deletedHashLevels, $fileMode;
  10         var $fileFactory = array( 'UnregisteredLocalFile', 'newFromTitle' );
  11         var $oldFileFactory = false;
  12         var $pathDisclosureProtection = 'simple';
  13
  14         function __construct( $info ) {
  15                 parent::__construct( $info );
  16
  17                 // Required settings
  18                 $this->directory = $info['directory'];
  19                 $this->url = $info['url'];
  20
  21                 // Optional settings
  22                 $this->hashLevels = isset( $info['hashLevels'] ) ? $info['hashLevels'] : 2;
  23                 $this->deletedHashLevels = isset( $info['deletedHashLevels'] ) ?
  24                         $info['deletedHashLevels'] : $this->hashLevels;
  25                 $this->deletedDir = isset( $info['deletedDir'] ) ? $info['deletedDir'] : false;
  26                 $this->fileMode = isset( $info['fileMode'] ) ? $info['fileMode'] : 0644;
  27                 if ( isset( $info['thumbDir'] ) ) {
  28                         $this->thumbDir =  $info['thumbDir'];
  29                 } else {
  30                         $this->thumbDir = "{$this->directory}/thumb";
  31                 }
  32                 if ( isset( $info['thumbUrl'] ) ) {
  33                         $this->thumbUrl = $info['thumbUrl'];
  34                 } else {
  35                         $this->thumbUrl = "{$this->url}/thumb";
  36                 }
  37         }
  38
  39         /**
  40          * Get the public root directory of the repository.
  41          */
  42         function getRootDirectory() {
  43                 return $this->directory;
  44         }
  45
  46         /**
  47          * Get the public root URL of the repository
  48          */
  49         function getRootUrl() {
  50                 return $this->url;
  51         }
  52
  53         /**
  54          * Returns true if the repository uses a multi-level directory structure
  55          */
  56         function isHashed() {
  57                 return (bool)$this->hashLevels;
  58         }
  59
  60         /**
  61          * Get the local directory corresponding to one of the three basic zones
  62          */
  63         function getZonePath( $zone ) {
  64                 switch ( $zone ) {
  65                         case 'public':
  66                                 return $this->directory;
  67                         case 'temp':
  68                                 return "{$this->directory}/temp";
  69                         case 'deleted':
  70                                 return $this->deletedDir;
  71                         case 'thumb':
  72                                 return $this->thumbDir;
  73                         default:
  74                                 return false;
  75                 }
  76         }
  77
  78         /**
  79          * @see FileRepo::getZoneUrl()
  80          */
  81         function getZoneUrl( $zone ) {
  82                 switch ( $zone ) {
  83                         case 'public':
  84                                 return $this->url;
  85                         case 'temp':
  86                                 return "{$this->url}/temp";
  87                         case 'deleted':
  88                                 return parent::getZoneUrl( $zone ); // no public URL
  89                         case 'thumb':
  90                                 return $this->thumbUrl;
  91                         default:
  92                                 return parent::getZoneUrl( $zone );
  93                 }
  94         }
  95
  96         /**
  97          * Get a URL referring to this repository, with the private mwrepo protocol.
  98          * The suffix, if supplied, is considered to be unencoded, and will be
  99          * URL-encoded before being returned.
 100          */
 101         function getVirtualUrl( $suffix = false ) {
 102                 $path = 'mwrepo://' . $this->name;
 103                 if ( $suffix !== false ) {
 104                         $path .= '/' . rawurlencode( $suffix );
 105                 }
 106                 return $path;
 107         }
 108
 109         /**
 110          * Get the local path corresponding to a virtual URL
 111          */
 112         function resolveVirtualUrl( $url ) {
 113                 if ( substr( $url, 0, 9 ) != 'mwrepo://' ) {
 114                         throw new MWException( __METHOD__.': unknown protoocl' );
 115                 }
 116
 117                 $bits = explode( '/', substr( $url, 9 ), 3 );
 118                 if ( count( $bits ) != 3 ) {
 119                         throw new MWException( __METHOD__.": invalid mwrepo URL: $url" );
 120                 }
 121                 list( $repo, $zone, $rel ) = $bits;
 122                 if ( $repo !== $this->name ) {
 123                         throw new MWException( __METHOD__.": fetching from a foreign repo is not supported" );
 124                 }
 125                 $base = $this->getZonePath( $zone );
 126                 if ( !$base ) {
 127                         throw new MWException( __METHOD__.": invalid zone: $zone" );
 128                 }
 129                 return $base . '/' . rawurldecode( $rel );
 130         }
 131
 132         /**
 133          * Store a batch of files
 134          *
 135          * @param array $triplets (src,zone,dest) triplets as per store()
 136          * @param integer $flags Bitwise combination of the following flags:
 137          *     self::DELETE_SOURCE     Delete the source file after upload
 138          *     self::OVERWRITE         Overwrite an existing destination file instead of failing
 139          *     self::OVERWRITE_SAME    Overwrite the file if the destination exists and has the
 140          *                             same contents as the source
 141          */
 142         function storeBatch( $triplets, $flags = 0 ) {
 143                 if ( !wfMkdirParents( $this->directory ) ) {
 144                         return $this->newFatal( 'upload_directory_missing', $this->directory );
 145                 }
 146                 if ( !is_writable( $this->directory ) ) {
 147                         return $this->newFatal( 'upload_directory_read_only', $this->directory );
 148                 }
 149                 $status = $this->newGood();
 150                 foreach ( $triplets as $i => $triplet ) {
 151                         list( $srcPath, $dstZone, $dstRel ) = $triplet;
 152
 153                         $root = $this->getZonePath( $dstZone );
 154                         if ( !$root ) {
 155                                 throw new MWException( "Invalid zone: $dstZone" );
 156                         }
 157                         if ( !$this->validateFilename( $dstRel ) ) {
 158                                 throw new MWException( 'Validation error in $dstRel' );
 159                         }
 160                         $dstPath = "$root/$dstRel";
 161                         $dstDir = dirname( $dstPath );
 162
 163                         if ( !is_dir( $dstDir ) ) {
 164                                 if ( !wfMkdirParents( $dstDir ) ) {
 165                                         return $this->newFatal( 'directorycreateerror', $dstDir );
 166                                 }
 167                                 if ( $dstZone == 'deleted' ) {
 168                                         $this->initDeletedDir( $dstDir );
 169                                 }
 170                         }
 171
 172                         if ( self::isVirtualUrl( $srcPath ) ) {
 173                                 $srcPath = $triplets[$i][0] = $this->resolveVirtualUrl( $srcPath );
 174                         }
 175                         if ( !is_file( $srcPath ) ) {
 176                                 // Make a list of files that don't exist for return to the caller
 177                                 $status->fatal( 'filenotfound', $srcPath );
 178                                 continue;
 179                         }
 180                         if ( !( $flags & self::OVERWRITE ) && file_exists( $dstPath ) ) {
 181                                 if ( $flags & self::OVERWRITE_SAME ) {
 182                                         $hashSource = sha1_file( $srcPath );
 183                                         $hashDest = sha1_file( $dstPath );
 184                                         if ( $hashSource != $hashDest ) {
 185                                                 $status->fatal( 'fileexistserror', $dstPath );
 186                                         }
 187                                 } else {
 188                                         $status->fatal( 'fileexistserror', $dstPath );
 189                                 }
 190                         }
 191                 }
 192
 193                 $deleteDest = wfIsWindows() && ( $flags & self::OVERWRITE );
 194
 195                 // Abort now on failure
 196                 if ( !$status->ok ) {
 197                         return $status;
 198                 }
 199
 200                 foreach ( $triplets as $triplet ) {
 201                         list( $srcPath, $dstZone, $dstRel ) = $triplet;
 202                         $root = $this->getZonePath( $dstZone );
 203                         $dstPath = "$root/$dstRel";
 204                         $good = true;
 205
 206                         if ( $flags & self::DELETE_SOURCE ) {
 207                                 if ( $deleteDest ) {
 208                                         unlink( $dstPath );
 209                                 }
 210                                 if ( !rename( $srcPath, $dstPath ) ) {
 211                                         $status->error( 'filerenameerror', $srcPath, $dstPath );
 212                                         $good = false;
 213                                 }
 214                         } else {
 215                                 if ( !copy( $srcPath, $dstPath ) ) {
 216                                         $status->error( 'filecopyerror', $srcPath, $dstPath );
 217                                         $good = false;
 218                                 }
 219                         }
 220                         if ( $good ) {
 221                                 $this->chmod( $dstPath );
 222                                 $status->successCount++;
 223                         } else {
 224                                 $status->failCount++;
 225                         }
 226                 }
 227                 return $status;
 228         }
 229
 230         /**
 231          * Checks existence of specified array of files.
 232          *
 233          * @param array $files URLs of files to check
 234          * @param integer $flags Bitwise combination of the following flags:
 235          *     self::FILES_ONLY     Mark file as existing only if it is a file (not directory)
 236          * @return Either array of files and existence flags, or false
 237          */
 238         function fileExistsBatch( $files, $flags = 0 ) {
 239                 if ( !file_exists( $this->directory ) || !is_readable( $this->directory ) ) {
 240                         return false;
 241                 }
 242                 $result = array();
 243                 foreach ( $files as $key => $file ) {
 244                         if ( self::isVirtualUrl( $file ) ) {
 245                                 $file = $this->resolveVirtualUrl( $file );
 246                         }
 247                         if( $flags & self::FILES_ONLY ) {
 248                                 $result[$key] = is_file( $file );
 249                         } else {
 250                                 $result[$key] = file_exists( $file );
 251                         }
 252                 }
 253
 254                 return $result;
 255         }
 256
 257         /**
 258          * Take all available measures to prevent web accessibility of new deleted
 259          * directories, in case the user has not configured offline storage
 260          */
 261         protected function initDeletedDir( $dir ) {
 262                 // Add a .htaccess file to the root of the deleted zone
 263                 $root = $this->getZonePath( 'deleted' );
 264                 if ( !file_exists( "$root/.htaccess" ) ) {
 265                         file_put_contents( "$root/.htaccess", "Deny from all\n" );
 266                 }
 267                 // Seed new directories with a blank index.html, to prevent crawling
 268                 file_put_contents( "$dir/index.html", '' );
 269         }
 270
 271         /**
 272          * Pick a random name in the temp zone and store a file to it.
 273          * @param string $originalName The base name of the file as specified
 274          *     by the user. The file extension will be maintained.
 275          * @param string $srcPath The current location of the file.
 276          * @return FileRepoStatus object with the URL in the value.
 277          */
 278         function storeTemp( $originalName, $srcPath ) {
 279                 $date = gmdate( "YmdHis" );
 280                 $hashPath = $this->getHashPath( $originalName );
 281                 $dstRel = "$hashPath$date!$originalName";
 282                 $dstUrlRel = $hashPath . $date . '!' . rawurlencode( $originalName );
 283
 284                 $result = $this->store( $srcPath, 'temp', $dstRel );
 285                 $result->value = $this->getVirtualUrl( 'temp' ) . '/' . $dstUrlRel;
 286                 return $result;
 287         }
 288
 289         /**
 290          * Remove a temporary file or mark it for garbage collection
 291          * @param string $virtualUrl The virtual URL returned by storeTemp
 292          * @return boolean True on success, false on failure
 293          */
 294         function freeTemp( $virtualUrl ) {
 295                 $temp = "mwrepo://{$this->name}/temp";
 296                 if ( substr( $virtualUrl, 0, strlen( $temp ) ) != $temp ) {
 297                         wfDebug( __METHOD__.": Invalid virtual URL\n" );
 298                         return false;
 299                 }
 300                 $path = $this->resolveVirtualUrl( $virtualUrl );
 301                 wfSuppressWarnings();
 302                 $success = unlink( $path );
 303                 wfRestoreWarnings();
 304                 return $success;
 305         }
 306
 307         /**
 308          * Publish a batch of files
 309          * @param array $triplets (source,dest,archive) triplets as per publish()
 310          * @param integer $flags Bitfield, may be FileRepo::DELETE_SOURCE to indicate
 311          *        that the source files should be deleted if possible
 312          */
 313         function publishBatch( $triplets, $flags = 0 ) {
 314                 // Perform initial checks
 315                 if ( !wfMkdirParents( $this->directory ) ) {
 316                         return $this->newFatal( 'upload_directory_missing', $this->directory );
 317                 }
 318                 if ( !is_writable( $this->directory ) ) {
 319                         return $this->newFatal( 'upload_directory_read_only', $this->directory );
 320                 }
 321                 $status = $this->newGood( array() );
 322                 foreach ( $triplets as $i => $triplet ) {
 323                         list( $srcPath, $dstRel, $archiveRel ) = $triplet;
 324
 325                         if ( substr( $srcPath, 0, 9 ) == 'mwrepo://' ) {
 326                                 $triplets[$i][0] = $srcPath = $this->resolveVirtualUrl( $srcPath );
 327                         }
 328                         if ( !$this->validateFilename( $dstRel ) ) {
 329                                 throw new MWException( 'Validation error in $dstRel' );
 330                         }
 331                         if ( !$this->validateFilename( $archiveRel ) ) {
 332                                 throw new MWException( 'Validation error in $archiveRel' );
 333                         }
 334                         $dstPath = "{$this->directory}/$dstRel";
 335                         $archivePath = "{$this->directory}/$archiveRel";
 336
 337                         $dstDir = dirname( $dstPath );
 338                         $archiveDir = dirname( $archivePath );
 339                         // Abort immediately on directory creation errors since they're likely to be repetitive
 340                         if ( !is_dir( $dstDir ) && !wfMkdirParents( $dstDir ) ) {
 341                                 return $this->newFatal( 'directorycreateerror', $dstDir );
 342                         }
 343                         if ( !is_dir( $archiveDir ) && !wfMkdirParents( $archiveDir ) ) {
 344                                 return $this->newFatal( 'directorycreateerror', $archiveDir );
 345                         }
 346                         if ( !is_file( $srcPath ) ) {
 347                                 // Make a list of files that don't exist for return to the caller
 348                                 $status->fatal( 'filenotfound', $srcPath );
 349                         }
 350                 }
 351
 352                 if ( !$status->ok ) {
 353                         return $status;
 354                 }
 355
 356                 foreach ( $triplets as $i => $triplet ) {
 357                         list( $srcPath, $dstRel, $archiveRel ) = $triplet;
 358                         $dstPath = "{$this->directory}/$dstRel";
 359                         $archivePath = "{$this->directory}/$archiveRel";
 360
 361                         // Archive destination file if it exists
 362                         if( is_file( $dstPath ) ) {
 363                                 // Check if the archive file exists
 364                                 // This is a sanity check to avoid data loss. In UNIX, the rename primitive
 365                                 // unlinks the destination file if it exists. DB-based synchronisation in
 366                                 // publishBatch's caller should prevent races. In Windows there's no
 367                                 // problem because the rename primitive fails if the destination exists.
 368                                 if ( is_file( $archivePath ) ) {
 369                                         $success = false;
 370                                 } else {
 371                                         wfSuppressWarnings();
 372                                         $success = rename( $dstPath, $archivePath );
 373                                         wfRestoreWarnings();
 374                                 }
 375
 376                                 if( !$success ) {
 377                                         $status->error( 'filerenameerror',$dstPath, $archivePath );
 378                                         $status->failCount++;
 379                                         continue;
 380                                 } else {
 381                                         wfDebug(__METHOD__.": moved file $dstPath to $archivePath\n");
 382                                 }
 383                                 $status->value[$i] = 'archived';
 384                         } else {
 385                                 $status->value[$i] = 'new';
 386                         }
 387
 388                         $good = true;
 389                         wfSuppressWarnings();
 390                         if ( $flags & self::DELETE_SOURCE ) {
 391                                 if ( !rename( $srcPath, $dstPath ) ) {
 392                                         $status->error( 'filerenameerror', $srcPath, $dstPath );
 393                                         $good = false;
 394                                 }
 395                         } else {
 396                                 if ( !copy( $srcPath, $dstPath ) ) {
 397                                         $status->error( 'filecopyerror', $srcPath, $dstPath );
 398                                         $good = false;
 399                                 }
 400                         }
 401                         wfRestoreWarnings();
 402
 403                         if ( $good ) {
 404                                 $status->successCount++;
 405                                 wfDebug(__METHOD__.": wrote tempfile $srcPath to $dstPath\n");
 406                                 // Thread-safe override for umask
 407                                 $this->chmod( $dstPath );
 408                         } else {
 409                                 $status->failCount++;
 410                         }
 411                 }
 412                 return $status;
 413         }
 414
 415         /**
 416          * Move a group of files to the deletion archive.
 417          * If no valid deletion archive is configured, this may either delete the
 418          * file or throw an exception, depending on the preference of the repository.
 419          *
 420          * @param array $sourceDestPairs Array of source/destination pairs. Each element
 421          *        is a two-element array containing the source file path relative to the
 422          *        public root in the first element, and the archive file path relative
 423          *        to the deleted zone root in the second element.
 424          * @return FileRepoStatus
 425          */
 426         function deleteBatch( $sourceDestPairs ) {
 427                 $status = $this->newGood();
 428                 if ( !$this->deletedDir ) {
 429                         throw new MWException( __METHOD__.': no valid deletion archive directory' );
 430                 }
 431
 432                 /**
 433                  * Validate filenames and create archive directories
 434                  */
 435                 foreach ( $sourceDestPairs as $pair ) {
 436                         list( $srcRel, $archiveRel ) = $pair;
 437                         if ( !$this->validateFilename( $srcRel ) ) {
 438                                 throw new MWException( __METHOD__.':Validation error in $srcRel' );
 439                         }
 440                         if ( !$this->validateFilename( $archiveRel ) ) {
 441                                 throw new MWException( __METHOD__.':Validation error in $archiveRel' );
 442                         }
 443                         $archivePath = "{$this->deletedDir}/$archiveRel";
 444                         $archiveDir = dirname( $archivePath );
 445                         if ( !is_dir( $archiveDir ) ) {
 446                                 if ( !wfMkdirParents( $archiveDir ) ) {
 447                                         $status->fatal( 'directorycreateerror', $archiveDir );
 448                                         continue;
 449                                 }
 450                                 $this->initDeletedDir( $archiveDir );
 451                         }
 452                         // Check if the archive directory is writable
 453                         // This doesn't appear to work on NTFS
 454                         if ( !is_writable( $archiveDir ) ) {
 455                                 $status->fatal( 'filedelete-archive-read-only', $archiveDir );
 456                         }
 457                 }
 458                 if ( !$status->ok ) {
 459                         // Abort early
 460                         return $status;
 461                 }
 462
 463                 /**
 464                  * Move the files
 465                  * We're now committed to returning an OK result, which will lead to
 466                  * the files being moved in the DB also.
 467                  */
 468                 foreach ( $sourceDestPairs as $pair ) {
 469                         list( $srcRel, $archiveRel ) = $pair;
 470                         $srcPath = "{$this->directory}/$srcRel";
 471                         $archivePath = "{$this->deletedDir}/$archiveRel";
 472                         $good = true;
 473                         if ( file_exists( $archivePath ) ) {
 474                                 # A file with this content hash is already archived
 475                                 if ( !@unlink( $srcPath ) ) {
 476                                         $status->error( 'filedeleteerror', $srcPath );
 477                                         $good = false;
 478                                 }
 479                         } else{
 480                                 if ( !@rename( $srcPath, $archivePath ) ) {
 481                                         $status->error( 'filerenameerror', $srcPath, $archivePath );
 482                                         $good = false;
 483                                 } else {
 484                                         $this->chmod( $archivePath );
 485                                 }
 486                         }
 487                         if ( $good ) {
 488                                 $status->successCount++;
 489                         } else {
 490                                 $status->failCount++;
 491                         }
 492                 }
 493                 return $status;
 494         }
 495
 496         /**
 497          * Get a relative path for a deletion archive key,
 498          * e.g. s/z/a/ for sza251lrxrc1jad41h5mgilp8nysje52.jpg
 499          */
 500         function getDeletedHashPath( $key ) {
 501                 $path = '';
 502                 for ( $i = 0; $i < $this->deletedHashLevels; $i++ ) {
 503                         $path .= $key[$i] . '/';
 504                 }
 505                 return $path;
 506         }
 507
 508         /**
 509          * Call a callback function for every file in the repository.
 510          * Uses the filesystem even in child classes.
 511          */
 512         function enumFilesInFS( $callback ) {
 513                 $numDirs = 1 << ( $this->hashLevels * 4 );
 514                 for ( $flatIndex = 0; $flatIndex < $numDirs; $flatIndex++ ) {
 515                         $hexString = sprintf( "%0{$this->hashLevels}x", $flatIndex );
 516                         $path = $this->directory;
 517                         for ( $hexPos = 0; $hexPos < $this->hashLevels; $hexPos++ ) {
 518                                 $path .= '/' . substr( $hexString, 0, $hexPos + 1 );
 519                         }
 520                         if ( !file_exists( $path ) || !is_dir( $path ) ) {
 521                                 continue;
 522                         }
 523                         $dir = opendir( $path );
 524                         while ( false !== ( $name = readdir( $dir ) ) ) {
 525                                 call_user_func( $callback, $path . '/' . $name );
 526                         }
 527                 }
 528         }
 529
 530         /**
 531          * Call a callback function for every file in the repository
 532          * May use either the database or the filesystem
 533          */
 534         function enumFiles( $callback ) {
 535                 $this->enumFilesInFS( $callback );
 536         }
 537
 538         /**
 539          * Get properties of a file with a given virtual URL
 540          * The virtual URL must refer to this repo
 541          */
 542         function getFileProps( $virtualUrl ) {
 543                 $path = $this->resolveVirtualUrl( $virtualUrl );
 544                 return File::getPropsFromPath( $path );
 545         }
 546
 547         /**
 548          * Path disclosure protection functions
 549          *
 550          * Get a callback function to use for cleaning error message parameters
 551          */
 552         function getErrorCleanupFunction() {
 553                 switch ( $this->pathDisclosureProtection ) {
 554                         case 'simple':
 555                                 $callback = array( $this, 'simpleClean' );
 556                                 break;
 557                         default:
 558                                 $callback = parent::getErrorCleanupFunction();
 559                 }
 560                 return $callback;
 561         }
 562
 563         function simpleClean( $param ) {
 564                 if ( !isset( $this->simpleCleanPairs ) ) {
 565                         global $IP;
 566                         $this->simpleCleanPairs = array(
 567                                 $this->directory => 'public',
 568                                 "{$this->directory}/temp" => 'temp',
 569                                 $IP => '$IP',
 570                                 dirname( __FILE__ ) => '$IP/extensions/WebStore',
 571                         );
 572                         if ( $this->deletedDir ) {
 573                                 $this->simpleCleanPairs[$this->deletedDir] = 'deleted';
 574                         }
 575                 }
 576                 return strtr( $param, $this->simpleCleanPairs );
 577         }
 578
 579         /**
 580          * Chmod a file, supressing the warnings.
 581          * @param String $path The path to change
 582          */
 583         protected function chmod( $path ) {
 584                 wfSuppressWarnings();
 585                 chmod( $path, $this->fileMode );
 586                 wfRestoreWarnings();
 587         }
 588
 589 }