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