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