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  */
   7
   8 class FSRepo extends FileRepo {
   9         var $directory, $deletedDir, $url, $hashLevels, $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 ( !is_writable( $this->directory ) ) {
 129                         return $this->newFatal( 'upload_directory_read_only', $this->directory );
 130                 }
 131                 $status = $this->newGood();
 132                 foreach ( $triplets as $i => $triplet ) {
 133                         list( $srcPath, $dstZone, $dstRel ) = $triplet;
 134
 135                         $root = $this->getZonePath( $dstZone );
 136                         if ( !$root ) {
 137                                 throw new MWException( "Invalid zone: $dstZone" );
 138                         }
 139                         if ( !$this->validateFilename( $dstRel ) ) {
 140                                 throw new MWException( 'Validation error in $dstRel' );
 141                         }
 142                         $dstPath = "$root/$dstRel";
 143                         $dstDir = dirname( $dstPath );
 144
 145                         if ( !is_dir( $dstDir ) ) {
 146                                 if ( !wfMkdirParents( $dstDir ) ) {
 147                                         return $this->newFatal( 'directorycreateerror', $dstDir );
 148                                 }
 149                                 // In the deleted zone, seed new directories with a blank
 150                                 // index.html, to prevent crawling
 151                                 if ( $dstZone == 'deleted' ) {
 152                                         file_put_contents( "$dstDir/index.html", '' );
 153                                 }
 154                         }
 155
 156                         if ( self::isVirtualUrl( $srcPath ) ) {
 157                                 $srcPath = $triplets[$i][0] = $this->resolveVirtualUrl( $srcPath );
 158                         }
 159                         if ( !is_file( $srcPath ) ) {
 160                                 // Make a list of files that don't exist for return to the caller
 161                                 $status->fatal( 'filenotfound', $srcPath );
 162                                 continue;
 163                         }
 164                         if ( !( $flags & self::OVERWRITE ) && file_exists( $dstPath ) ) {
 165                                 if ( $flags & self::OVERWRITE_SAME ) {
 166                                         $hashSource = sha1_file( $srcPath );
 167                                         $hashDest = sha1_file( $dstPath );
 168                                         if ( $hashSource != $hashDest ) {
 169                                                 $status->fatal( 'fileexistserror', $dstPath );
 170                                         }
 171                                 } else {
 172                                         $status->fatal( 'fileexistserror', $dstPath );
 173                                 }
 174                         }
 175                 }
 176
 177                 $deleteDest = wfIsWindows() && ( $flags & self::OVERWRITE );
 178
 179                 // Abort now on failure
 180                 if ( !$status->ok ) {
 181                         return $status;
 182                 }
 183
 184                 foreach ( $triplets as $triplet ) {
 185                         list( $srcPath, $dstZone, $dstRel ) = $triplet;
 186                         $root = $this->getZonePath( $dstZone );
 187                         $dstPath = "$root/$dstRel";
 188                         $good = true;
 189
 190                         if ( $flags & self::DELETE_SOURCE ) {
 191                                 if ( $deleteDest ) {
 192                                         unlink( $dstPath );
 193                                 }
 194                                 if ( !rename( $srcPath, $dstPath ) ) {
 195                                         $status->error( 'filerenameerror', $srcPath, $dstPath );
 196                                         $good = false;
 197                                 }
 198                         } else {
 199                                 if ( !copy( $srcPath, $dstPath ) ) {
 200                                         $status->error( 'filecopyerror', $srcPath, $dstPath );
 201                                         $good = false;
 202                                 }
 203                         }
 204                         if ( $good ) {
 205                                 chmod( $dstPath, 0644 );
 206                                 $status->successCount++;
 207                         } else {
 208                                 $status->failCount++;
 209                         }
 210                 }
 211                 return $status;
 212         }
 213
 214         /**
 215          * Pick a random name in the temp zone and store a file to it.
 216          * @param string $originalName The base name of the file as specified
 217          *     by the user. The file extension will be maintained.
 218          * @param string $srcPath The current location of the file.
 219          * @return FileRepoStatus object with the URL in the value.
 220          */
 221         function storeTemp( $originalName, $srcPath ) {
 222                 $date = gmdate( "YmdHis" );
 223                 $hashPath = $this->getHashPath( $originalName );
 224                 $dstRel = "$hashPath$date!$originalName";
 225                 $dstUrlRel = $hashPath . $date . '!' . rawurlencode( $originalName );
 226
 227                 $result = $this->store( $srcPath, 'temp', $dstRel );
 228                 $result->value = $this->getVirtualUrl( 'temp' ) . '/' . $dstUrlRel;
 229                 return $result;
 230         }
 231
 232         /**
 233          * Remove a temporary file or mark it for garbage collection
 234          * @param string $virtualUrl The virtual URL returned by storeTemp
 235          * @return boolean True on success, false on failure
 236          */
 237         function freeTemp( $virtualUrl ) {
 238                 $temp = "mwrepo://{$this->name}/temp";
 239                 if ( substr( $virtualUrl, 0, strlen( $temp ) ) != $temp ) {
 240                         wfDebug( __METHOD__.": Invalid virtual URL\n" );
 241                         return false;
 242                 }
 243                 $path = $this->resolveVirtualUrl( $virtualUrl );
 244                 wfSuppressWarnings();
 245                 $success = unlink( $path );
 246                 wfRestoreWarnings();
 247                 return $success;
 248         }
 249
 250         /**
 251          * Publish a batch of files
 252          * @param array $triplets (source,dest,archive) triplets as per publish()
 253          * @param integer $flags Bitfield, may be FileRepo::DELETE_SOURCE to indicate
 254          *        that the source files should be deleted if possible
 255          */
 256         function publishBatch( $triplets, $flags = 0 ) {
 257                 // Perform initial checks
 258                 if ( !is_writable( $this->directory ) ) {
 259                         return $this->newFatal( 'upload_directory_read_only', $this->directory );
 260                 }
 261                 $status = $this->newGood( array() );
 262                 foreach ( $triplets as $i => $triplet ) {
 263                         list( $srcPath, $dstRel, $archiveRel ) = $triplet;
 264
 265                         if ( substr( $srcPath, 0, 9 ) == 'mwrepo://' ) {
 266                                 $triplets[$i][0] = $srcPath = $this->resolveVirtualUrl( $srcPath );
 267                         }
 268                         if ( !$this->validateFilename( $dstRel ) ) {
 269                                 throw new MWException( 'Validation error in $dstRel' );
 270                         }
 271                         if ( !$this->validateFilename( $archiveRel ) ) {
 272                                 throw new MWException( 'Validation error in $archiveRel' );
 273                         }
 274                         $dstPath = "{$this->directory}/$dstRel";
 275                         $archivePath = "{$this->directory}/$archiveRel";
 276
 277                         $dstDir = dirname( $dstPath );
 278                         $archiveDir = dirname( $archivePath );
 279                         // Abort immediately on directory creation errors since they're likely to be repetitive
 280                         if ( !is_dir( $dstDir ) && !wfMkdirParents( $dstDir ) ) {
 281                                 return $this->newFatal( 'directorycreateerror', $dstDir );
 282                         }
 283                         if ( !is_dir( $archiveDir ) && !wfMkdirParents( $archiveDir ) ) {
 284                                 return $this->newFatal( 'directorycreateerror', $archiveDir );
 285                         }
 286                         if ( !is_file( $srcPath ) ) {
 287                                 // Make a list of files that don't exist for return to the caller
 288                                 $status->fatal( 'filenotfound', $srcPath );
 289                         }
 290                 }
 291
 292                 if ( !$status->ok ) {
 293                         return $status;
 294                 }
 295
 296                 foreach ( $triplets as $i => $triplet ) {
 297                         list( $srcPath, $dstRel, $archiveRel ) = $triplet;
 298                         $dstPath = "{$this->directory}/$dstRel";
 299                         $archivePath = "{$this->directory}/$archiveRel";
 300
 301                         // Archive destination file if it exists
 302                         if( is_file( $dstPath ) ) {
 303                                 // Check if the archive file exists
 304                                 // This is a sanity check to avoid data loss. In UNIX, the rename primitive
 305                                 // unlinks the destination file if it exists. DB-based synchronisation in
 306                                 // publishBatch's caller should prevent races. In Windows there's no
 307                                 // problem because the rename primitive fails if the destination exists.
 308                                 if ( is_file( $archivePath ) ) {
 309                                         $success = false;
 310                                 } else {
 311                                         wfSuppressWarnings();
 312                                         $success = rename( $dstPath, $archivePath );
 313                                         wfRestoreWarnings();
 314                                 }
 315
 316                                 if( !$success ) {
 317                                         $status->error( 'filerenameerror',$dstPath, $archivePath );
 318                                         $status->failCount++;
 319                                         continue;
 320                                 } else {
 321                                         wfDebug(__METHOD__.": moved file $dstPath to $archivePath\n");
 322                                 }
 323                                 $status->value[$i] = 'archived';
 324                         } else {
 325                                 $status->value[$i] = 'new';
 326                         }
 327
 328                         $good = true;
 329                         wfSuppressWarnings();
 330                         if ( $flags & self::DELETE_SOURCE ) {
 331                                 if ( !rename( $srcPath, $dstPath ) ) {
 332                                         $status->error( 'filerenameerror', $srcPath, $dstPath );
 333                                         $good = false;
 334                                 }
 335                         } else {
 336                                 if ( !copy( $srcPath, $dstPath ) ) {
 337                                         $status->error( 'filecopyerror', $srcPath, $dstPath );
 338                                         $good = false;
 339                                 }
 340                         }
 341                         wfRestoreWarnings();
 342
 343                         if ( $good ) {
 344                                 $status->successCount++;
 345                                 wfDebug(__METHOD__.": wrote tempfile $srcPath to $dstPath\n");
 346                                 // Thread-safe override for umask
 347                                 chmod( $dstPath, 0644 );
 348                         } else {
 349                                 $status->failCount++;
 350                         }
 351                 }
 352                 return $status;
 353         }
 354
 355         /**
 356          * Move a group of files to the deletion archive.
 357          * If no valid deletion archive is configured, this may either delete the
 358          * file or throw an exception, depending on the preference of the repository.
 359          *
 360          * @param array $sourceDestPairs Array of source/destination pairs. Each element
 361          *        is a two-element array containing the source file path relative to the
 362          *        public root in the first element, and the archive file path relative
 363          *        to the deleted zone root in the second element.
 364          * @return FileRepoStatus
 365          */
 366         function deleteBatch( $sourceDestPairs ) {
 367                 $status = $this->newGood();
 368                 if ( !$this->deletedDir ) {
 369                         throw new MWException( __METHOD__.': no valid deletion archive directory' );
 370                 }
 371
 372                 /**
 373                  * Validate filenames and create archive directories
 374                  */
 375                 foreach ( $sourceDestPairs as $pair ) {
 376                         list( $srcRel, $archiveRel ) = $pair;
 377                         if ( !$this->validateFilename( $srcRel ) ) {
 378                                 throw new MWException( __METHOD__.':Validation error in $srcRel' );
 379                         }
 380                         if ( !$this->validateFilename( $archiveRel ) ) {
 381                                 throw new MWException( __METHOD__.':Validation error in $archiveRel' );
 382                         }
 383                         $archivePath = "{$this->deletedDir}/$archiveRel";
 384                         $archiveDir = dirname( $archivePath );
 385                         if ( !is_dir( $archiveDir ) ) {
 386                                 if ( !wfMkdirParents( $archiveDir ) ) {
 387                                         $status->fatal( 'directorycreateerror', $archiveDir );
 388                                         continue;
 389                                 }
 390                                 // Seed new directories with a blank index.html, to prevent crawling
 391                                 file_put_contents( "$archiveDir/index.html", '' );
 392                         }
 393                         // Check if the archive directory is writable
 394                         // This doesn't appear to work on NTFS
 395                         if ( !is_writable( $archiveDir ) ) {
 396                                 $status->fatal( 'filedelete-archive-read-only', $archiveDir );
 397                         }
 398                 }
 399                 if ( !$status->ok ) {
 400                         // Abort early
 401                         return $status;
 402                 }
 403
 404                 /**
 405                  * Move the files
 406                  * We're now committed to returning an OK result, which will lead to
 407                  * the files being moved in the DB also.
 408                  */
 409                 foreach ( $sourceDestPairs as $pair ) {
 410                         list( $srcRel, $archiveRel ) = $pair;
 411                         $srcPath = "{$this->directory}/$srcRel";
 412                         $archivePath = "{$this->deletedDir}/$archiveRel";
 413                         $good = true;
 414                         if ( file_exists( $archivePath ) ) {
 415                                 # A file with this content hash is already archived
 416                                 if ( !@unlink( $srcPath ) ) {
 417                                         $status->error( 'filedeleteerror', $srcPath );
 418                                         $good = false;
 419                                 }
 420                         } else{
 421                                 if ( !@rename( $srcPath, $archivePath ) ) {
 422                                         $status->error( 'filerenameerror', $srcPath, $archivePath );
 423                                         $good = false;
 424                                 } else {
 425                                         chmod( $archivePath, 0644 );
 426                                 }
 427                         }
 428                         if ( $good ) {
 429                                 $status->successCount++;
 430                         } else {
 431                                 $status->failCount++;
 432                         }
 433                 }
 434                 return $status;
 435         }
 436
 437         /**
 438          * Get a relative path including trailing slash, e.g. f/fa/
 439          * If the repo is not hashed, returns an empty string
 440          */
 441         function getHashPath( $name ) {
 442                 return FileRepo::getHashPathForLevel( $name, $this->hashLevels );
 443         }
 444
 445         /**
 446          * Get a relative path for a deletion archive key,
 447          * e.g. s/z/a/ for sza251lrxrc1jad41h5mgilp8nysje52.jpg
 448          */
 449         function getDeletedHashPath( $key ) {
 450                 $path = '';
 451                 for ( $i = 0; $i < $this->deletedHashLevels; $i++ ) {
 452                         $path .= $key[$i] . '/';
 453                 }
 454                 return $path;
 455         }
 456
 457         /**
 458          * Call a callback function for every file in the repository.
 459          * Uses the filesystem even in child classes.
 460          */
 461         function enumFilesInFS( $callback ) {
 462                 $numDirs = 1 << ( $this->hashLevels * 4 );
 463                 for ( $flatIndex = 0; $flatIndex < $numDirs; $flatIndex++ ) {
 464                         $hexString = sprintf( "%0{$this->hashLevels}x", $flatIndex );
 465                         $path = $this->directory;
 466                         for ( $hexPos = 0; $hexPos < $this->hashLevels; $hexPos++ ) {
 467                                 $path .= '/' . substr( $hexString, 0, $hexPos + 1 );
 468                         }
 469                         if ( !file_exists( $path ) || !is_dir( $path ) ) {
 470                                 continue;
 471                         }
 472                         $dir = opendir( $path );
 473                         while ( false !== ( $name = readdir( $dir ) ) ) {
 474                                 call_user_func( $callback, $path . '/' . $name );
 475                         }
 476                 }
 477         }
 478
 479         /**
 480          * Call a callback function for every file in the repository
 481          * May use either the database or the filesystem
 482          */
 483         function enumFiles( $callback ) {
 484                 $this->enumFilesInFS( $callback );
 485         }
 486
 487         /**
 488          * Get properties of a file with a given virtual URL
 489          * The virtual URL must refer to this repo
 490          */
 491         function getFileProps( $virtualUrl ) {
 492                 $path = $this->resolveVirtualUrl( $virtualUrl );
 493                 return File::getPropsFromPath( $path );
 494         }
 495
 496         /**
 497          * Path disclosure protection functions
 498          *
 499          * Get a callback function to use for cleaning error message parameters
 500          */
 501         function getErrorCleanupFunction() {
 502                 switch ( $this->pathDisclosureProtection ) {
 503                         case 'simple':
 504                                 $callback = array( $this, 'simpleClean' );
 505                                 break;
 506                         default:
 507                                 $callback = parent::getErrorCleanupFunction();
 508                 }
 509                 return $callback;
 510         }
 511
 512         function simpleClean( $param ) {
 513                 if ( !isset( $this->simpleCleanPairs ) ) {
 514                         global $IP;
 515                         $this->simpleCleanPairs = array(
 516                                 $this->directory => 'public',
 517                                 "{$this->directory}/temp" => 'temp',
 518                                 $IP => '$IP',
 519                                 dirname( __FILE__ ) => '$IP/extensions/WebStore',
 520                         );
 521                         if ( $this->deletedDir ) {
 522                                 $this->simpleCleanPairs[$this->deletedDir] = 'deleted';
 523                         }
 524                 }
 525                 return strtr( $param, $this->simpleCleanPairs );
 526         }
 527
 528 }
 529
 530