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, $url, $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          * Get the URL corresponding to one of the three basic zones
  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 false; // no public URL
  89                         case 'thumb':
  90                                 return $this->thumbUrl;
  91                         default:
  92                                 return false;
  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         function append( $srcPath, $toAppendPath ) {
 231                 $status = $this->newGood();
 232
 233                 // Resolve the virtual URL
 234                 if ( self::isVirtualUrl( $srcPath ) ) {
 235                         $srcPath = $this->resolveVirtualUrl( $srcPath );
 236                 }
 237                 // Make sure the files are there
 238                 if ( !is_file( $srcPath ) )
 239                         $status->fatal( 'append-src-filenotfound', $srcPath );
 240
 241                 if ( !is_file( $toAppendPath ) )
 242                         $status->fatal( 'append-toappend-filenotfound', $toAppendPath );
 243
 244                 // Do the append
 245                 if( file_put_contents( $srcPath, file_get_contents( $toAppendPath ), FILE_APPEND ) ) {
 246                         $status->value = $srcPath;
 247                 } else {
 248                         $status->fatal( 'fileappenderror', $toAppendPath,  $srcPath);
 249                 }
 250
 251                 // Remove the source file
 252                 unlink( $toAppendPath );
 253
 254                 return $status;
 255         }
 256
 257         /**
 258          * Checks existence of specified array of files.
 259          *
 260          * @param array $files URLs of files to check
 261          * @param integer $flags Bitwise combination of the following flags:
 262          *     self::FILES_ONLY     Mark file as existing only if it is a file (not directory)
 263          * @return Either array of files and existence flags, or false
 264          */
 265         function fileExistsBatch( $files, $flags = 0 ) {
 266                 if ( !file_exists( $this->directory ) || !is_readable( $this->directory ) ) {
 267                         return false;
 268                 }
 269                 $result = array();
 270                 foreach ( $files as $key => $file ) {
 271                         if ( self::isVirtualUrl( $file ) ) {
 272                                 $file = $this->resolveVirtualUrl( $file );
 273                         }
 274                         if( $flags & self::FILES_ONLY ) {
 275                                 $result[$key] = is_file( $file );
 276                         } else {
 277                                 $result[$key] = file_exists( $file );
 278                         }
 279                 }
 280
 281                 return $result;
 282         }
 283
 284         /**
 285          * Take all available measures to prevent web accessibility of new deleted
 286          * directories, in case the user has not configured offline storage
 287          */
 288         protected function initDeletedDir( $dir ) {
 289                 // Add a .htaccess file to the root of the deleted zone
 290                 $root = $this->getZonePath( 'deleted' );
 291                 if ( !file_exists( "$root/.htaccess" ) ) {
 292                         file_put_contents( "$root/.htaccess", "Deny from all\n" );
 293                 }
 294                 // Seed new directories with a blank index.html, to prevent crawling
 295                 file_put_contents( "$dir/index.html", '' );
 296         }
 297
 298         /**
 299          * Pick a random name in the temp zone and store a file to it.
 300          * @param string $originalName The base name of the file as specified
 301          *     by the user. The file extension will be maintained.
 302          * @param string $srcPath The current location of the file.
 303          * @return FileRepoStatus object with the URL in the value.
 304          */
 305         function storeTemp( $originalName, $srcPath ) {
 306                 $date = gmdate( "YmdHis" );
 307                 $hashPath = $this->getHashPath( $originalName );
 308                 $dstRel = "$hashPath$date!$originalName";
 309                 $dstUrlRel = $hashPath . $date . '!' . rawurlencode( $originalName );
 310
 311                 $result = $this->store( $srcPath, 'temp', $dstRel );
 312                 $result->value = $this->getVirtualUrl( 'temp' ) . '/' . $dstUrlRel;
 313                 return $result;
 314         }
 315
 316         /**
 317          * Remove a temporary file or mark it for garbage collection
 318          * @param string $virtualUrl The virtual URL returned by storeTemp
 319          * @return boolean True on success, false on failure
 320          */
 321         function freeTemp( $virtualUrl ) {
 322                 $temp = "mwrepo://{$this->name}/temp";
 323                 if ( substr( $virtualUrl, 0, strlen( $temp ) ) != $temp ) {
 324                         wfDebug( __METHOD__.": Invalid virtual URL\n" );
 325                         return false;
 326                 }
 327                 $path = $this->resolveVirtualUrl( $virtualUrl );
 328                 wfSuppressWarnings();
 329                 $success = unlink( $path );
 330                 wfRestoreWarnings();
 331                 return $success;
 332         }
 333
 334         /**
 335          * Publish a batch of files
 336          * @param array $triplets (source,dest,archive) triplets as per publish()
 337          * @param integer $flags Bitfield, may be FileRepo::DELETE_SOURCE to indicate
 338          *        that the source files should be deleted if possible
 339          */
 340         function publishBatch( $triplets, $flags = 0 ) {
 341                 // Perform initial checks
 342                 if ( !wfMkdirParents( $this->directory ) ) {
 343                         return $this->newFatal( 'upload_directory_missing', $this->directory );
 344                 }
 345                 if ( !is_writable( $this->directory ) ) {
 346                         return $this->newFatal( 'upload_directory_read_only', $this->directory );
 347                 }
 348                 $status = $this->newGood( array() );
 349                 foreach ( $triplets as $i => $triplet ) {
 350                         list( $srcPath, $dstRel, $archiveRel ) = $triplet;
 351
 352                         if ( substr( $srcPath, 0, 9 ) == 'mwrepo://' ) {
 353                                 $triplets[$i][0] = $srcPath = $this->resolveVirtualUrl( $srcPath );
 354                         }
 355                         if ( !$this->validateFilename( $dstRel ) ) {
 356                                 throw new MWException( 'Validation error in $dstRel' );
 357                         }
 358                         if ( !$this->validateFilename( $archiveRel ) ) {
 359                                 throw new MWException( 'Validation error in $archiveRel' );
 360                         }
 361                         $dstPath = "{$this->directory}/$dstRel";
 362                         $archivePath = "{$this->directory}/$archiveRel";
 363
 364                         $dstDir = dirname( $dstPath );
 365                         $archiveDir = dirname( $archivePath );
 366                         // Abort immediately on directory creation errors since they're likely to be repetitive
 367                         if ( !is_dir( $dstDir ) && !wfMkdirParents( $dstDir ) ) {
 368                                 return $this->newFatal( 'directorycreateerror', $dstDir );
 369                         }
 370                         if ( !is_dir( $archiveDir ) && !wfMkdirParents( $archiveDir ) ) {
 371                                 return $this->newFatal( 'directorycreateerror', $archiveDir );
 372                         }
 373                         if ( !is_file( $srcPath ) ) {
 374                                 // Make a list of files that don't exist for return to the caller
 375                                 $status->fatal( 'filenotfound', $srcPath );
 376                         }
 377                 }
 378
 379                 if ( !$status->ok ) {
 380                         return $status;
 381                 }
 382
 383                 foreach ( $triplets as $i => $triplet ) {
 384                         list( $srcPath, $dstRel, $archiveRel ) = $triplet;
 385                         $dstPath = "{$this->directory}/$dstRel";
 386                         $archivePath = "{$this->directory}/$archiveRel";
 387
 388                         // Archive destination file if it exists
 389                         if( is_file( $dstPath ) ) {
 390                                 // Check if the archive file exists
 391                                 // This is a sanity check to avoid data loss. In UNIX, the rename primitive
 392                                 // unlinks the destination file if it exists. DB-based synchronisation in
 393                                 // publishBatch's caller should prevent races. In Windows there's no
 394                                 // problem because the rename primitive fails if the destination exists.
 395                                 if ( is_file( $archivePath ) ) {
 396                                         $success = false;
 397                                 } else {
 398                                         wfSuppressWarnings();
 399                                         $success = rename( $dstPath, $archivePath );
 400                                         wfRestoreWarnings();
 401                                 }
 402
 403                                 if( !$success ) {
 404                                         $status->error( 'filerenameerror',$dstPath, $archivePath );
 405                                         $status->failCount++;
 406                                         continue;
 407                                 } else {
 408                                         wfDebug(__METHOD__.": moved file $dstPath to $archivePath\n");
 409                                 }
 410                                 $status->value[$i] = 'archived';
 411                         } else {
 412                                 $status->value[$i] = 'new';
 413                         }
 414
 415                         $good = true;
 416                         wfSuppressWarnings();
 417                         if ( $flags & self::DELETE_SOURCE ) {
 418                                 if ( !rename( $srcPath, $dstPath ) ) {
 419                                         $status->error( 'filerenameerror', $srcPath, $dstPath );
 420                                         $good = false;
 421                                 }
 422                         } else {
 423                                 if ( !copy( $srcPath, $dstPath ) ) {
 424                                         $status->error( 'filecopyerror', $srcPath, $dstPath );
 425                                         $good = false;
 426                                 }
 427                         }
 428                         wfRestoreWarnings();
 429
 430                         if ( $good ) {
 431                                 $status->successCount++;
 432                                 wfDebug(__METHOD__.": wrote tempfile $srcPath to $dstPath\n");
 433                                 // Thread-safe override for umask
 434                                 $this->chmod( $dstPath );
 435                         } else {
 436                                 $status->failCount++;
 437                         }
 438                 }
 439                 return $status;
 440         }
 441
 442         /**
 443          * Move a group of files to the deletion archive.
 444          * If no valid deletion archive is configured, this may either delete the
 445          * file or throw an exception, depending on the preference of the repository.
 446          *
 447          * @param array $sourceDestPairs Array of source/destination pairs. Each element
 448          *        is a two-element array containing the source file path relative to the
 449          *        public root in the first element, and the archive file path relative
 450          *        to the deleted zone root in the second element.
 451          * @return FileRepoStatus
 452          */
 453         function deleteBatch( $sourceDestPairs ) {
 454                 $status = $this->newGood();
 455                 if ( !$this->deletedDir ) {
 456                         throw new MWException( __METHOD__.': no valid deletion archive directory' );
 457                 }
 458
 459                 /**
 460                  * Validate filenames and create archive directories
 461                  */
 462                 foreach ( $sourceDestPairs as $pair ) {
 463                         list( $srcRel, $archiveRel ) = $pair;
 464                         if ( !$this->validateFilename( $srcRel ) ) {
 465                                 throw new MWException( __METHOD__.':Validation error in $srcRel' );
 466                         }
 467                         if ( !$this->validateFilename( $archiveRel ) ) {
 468                                 throw new MWException( __METHOD__.':Validation error in $archiveRel' );
 469                         }
 470                         $archivePath = "{$this->deletedDir}/$archiveRel";
 471                         $archiveDir = dirname( $archivePath );
 472                         if ( !is_dir( $archiveDir ) ) {
 473                                 if ( !wfMkdirParents( $archiveDir ) ) {
 474                                         $status->fatal( 'directorycreateerror', $archiveDir );
 475                                         continue;
 476                                 }
 477                                 $this->initDeletedDir( $archiveDir );
 478                         }
 479                         // Check if the archive directory is writable
 480                         // This doesn't appear to work on NTFS
 481                         if ( !is_writable( $archiveDir ) ) {
 482                                 $status->fatal( 'filedelete-archive-read-only', $archiveDir );
 483                         }
 484                 }
 485                 if ( !$status->ok ) {
 486                         // Abort early
 487                         return $status;
 488                 }
 489
 490                 /**
 491                  * Move the files
 492                  * We're now committed to returning an OK result, which will lead to
 493                  * the files being moved in the DB also.
 494                  */
 495                 foreach ( $sourceDestPairs as $pair ) {
 496                         list( $srcRel, $archiveRel ) = $pair;
 497                         $srcPath = "{$this->directory}/$srcRel";
 498                         $archivePath = "{$this->deletedDir}/$archiveRel";
 499                         $good = true;
 500                         if ( file_exists( $archivePath ) ) {
 501                                 # A file with this content hash is already archived
 502                                 if ( !@unlink( $srcPath ) ) {
 503                                         $status->error( 'filedeleteerror', $srcPath );
 504                                         $good = false;
 505                                 }
 506                         } else{
 507                                 if ( !@rename( $srcPath, $archivePath ) ) {
 508                                         $status->error( 'filerenameerror', $srcPath, $archivePath );
 509                                         $good = false;
 510                                 } else {
 511                                         $this->chmod( $archivePath );
 512                                 }
 513                         }
 514                         if ( $good ) {
 515                                 $status->successCount++;
 516                         } else {
 517                                 $status->failCount++;
 518                         }
 519                 }
 520                 return $status;
 521         }
 522
 523         /**
 524          * Get a relative path for a deletion archive key,
 525          * e.g. s/z/a/ for sza251lrxrc1jad41h5mgilp8nysje52.jpg
 526          */
 527         function getDeletedHashPath( $key ) {
 528                 $path = '';
 529                 for ( $i = 0; $i < $this->deletedHashLevels; $i++ ) {
 530                         $path .= $key[$i] . '/';
 531                 }
 532                 return $path;
 533         }
 534
 535         /**
 536          * Call a callback function for every file in the repository.
 537          * Uses the filesystem even in child classes.
 538          */
 539         function enumFilesInFS( $callback ) {
 540                 $numDirs = 1 << ( $this->hashLevels * 4 );
 541                 for ( $flatIndex = 0; $flatIndex < $numDirs; $flatIndex++ ) {
 542                         $hexString = sprintf( "%0{$this->hashLevels}x", $flatIndex );
 543                         $path = $this->directory;
 544                         for ( $hexPos = 0; $hexPos < $this->hashLevels; $hexPos++ ) {
 545                                 $path .= '/' . substr( $hexString, 0, $hexPos + 1 );
 546                         }
 547                         if ( !file_exists( $path ) || !is_dir( $path ) ) {
 548                                 continue;
 549                         }
 550                         $dir = opendir( $path );
 551                         while ( false !== ( $name = readdir( $dir ) ) ) {
 552                                 call_user_func( $callback, $path . '/' . $name );
 553                         }
 554                 }
 555         }
 556
 557         /**
 558          * Call a callback function for every file in the repository
 559          * May use either the database or the filesystem
 560          */
 561         function enumFiles( $callback ) {
 562                 $this->enumFilesInFS( $callback );
 563         }
 564
 565         /**
 566          * Get properties of a file with a given virtual URL
 567          * The virtual URL must refer to this repo
 568          */
 569         function getFileProps( $virtualUrl ) {
 570                 $path = $this->resolveVirtualUrl( $virtualUrl );
 571                 return File::getPropsFromPath( $path );
 572         }
 573
 574         /**
 575          * Path disclosure protection functions
 576          *
 577          * Get a callback function to use for cleaning error message parameters
 578          */
 579         function getErrorCleanupFunction() {
 580                 switch ( $this->pathDisclosureProtection ) {
 581                         case 'simple':
 582                                 $callback = array( $this, 'simpleClean' );
 583                                 break;
 584                         default:
 585                                 $callback = parent::getErrorCleanupFunction();
 586                 }
 587                 return $callback;
 588         }
 589
 590         function simpleClean( $param ) {
 591                 if ( !isset( $this->simpleCleanPairs ) ) {
 592                         global $IP;
 593                         $this->simpleCleanPairs = array(
 594                                 $this->directory => 'public',
 595                                 "{$this->directory}/temp" => 'temp',
 596                                 $IP => '$IP',
 597                                 dirname( __FILE__ ) => '$IP/extensions/WebStore',
 598                         );
 599                         if ( $this->deletedDir ) {
 600                                 $this->simpleCleanPairs[$this->deletedDir] = 'deleted';
 601                         }
 602                 }
 603                 return strtr( $param, $this->simpleCleanPairs );
 604         }
 605
 606         /**
 607          * Chmod a file, supressing the warnings.
 608          * @param String $path The path to change
 609          */
 610         protected function chmod( $path ) {
 611                 wfSuppressWarnings();
 612                 chmod( $path, $this->fileMode );
 613                 wfRestoreWarnings();
 614         }
 615
 616 }