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