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          * Take all available measures to prevent web accessibility of new deleted
 217          * directories, in case the user has not configured offline storage
 218          */
 219         protected function initDeletedDir( $dir ) {
 220                 // Add a .htaccess file to the root of the deleted zone
 221                 $root = $this->getZonePath( 'deleted' );
 222                 if ( !file_exists( "$root/.htaccess" ) ) {
 223                         file_put_contents( "$root/.htaccess", "Deny from all\n" );
 224                 }
 225                 // Seed new directories with a blank index.html, to prevent crawling
 226                 file_put_contents( "$dir/index.html", '' );
 227         }
 228
 229         /**
 230          * Pick a random name in the temp zone and store a file to it.
 231          * @param string $originalName The base name of the file as specified
 232          *     by the user. The file extension will be maintained.
 233          * @param string $srcPath The current location of the file.
 234          * @return FileRepoStatus object with the URL in the value.
 235          */
 236         function storeTemp( $originalName, $srcPath ) {
 237                 $date = gmdate( "YmdHis" );
 238                 $hashPath = $this->getHashPath( $originalName );
 239                 $dstRel = "$hashPath$date!$originalName";
 240                 $dstUrlRel = $hashPath . $date . '!' . rawurlencode( $originalName );
 241
 242                 $result = $this->store( $srcPath, 'temp', $dstRel );
 243                 $result->value = $this->getVirtualUrl( 'temp' ) . '/' . $dstUrlRel;
 244                 return $result;
 245         }
 246
 247         /**
 248          * Remove a temporary file or mark it for garbage collection
 249          * @param string $virtualUrl The virtual URL returned by storeTemp
 250          * @return boolean True on success, false on failure
 251          */
 252         function freeTemp( $virtualUrl ) {
 253                 $temp = "mwrepo://{$this->name}/temp";
 254                 if ( substr( $virtualUrl, 0, strlen( $temp ) ) != $temp ) {
 255                         wfDebug( __METHOD__.": Invalid virtual URL\n" );
 256                         return false;
 257                 }
 258                 $path = $this->resolveVirtualUrl( $virtualUrl );
 259                 wfSuppressWarnings();
 260                 $success = unlink( $path );
 261                 wfRestoreWarnings();
 262                 return $success;
 263         }
 264
 265         /**
 266          * Publish a batch of files
 267          * @param array $triplets (source,dest,archive) triplets as per publish()
 268          * @param integer $flags Bitfield, may be FileRepo::DELETE_SOURCE to indicate
 269          *        that the source files should be deleted if possible
 270          */
 271         function publishBatch( $triplets, $flags = 0 ) {
 272                 // Perform initial checks
 273                 if ( !wfMkdirParents( $this->directory ) ) {
 274                         return $this->newFatal( 'upload_directory_missing', $this->directory );
 275                 }
 276                 if ( !is_writable( $this->directory ) ) {
 277                         return $this->newFatal( 'upload_directory_read_only', $this->directory );
 278                 }
 279                 $status = $this->newGood( array() );
 280                 foreach ( $triplets as $i => $triplet ) {
 281                         list( $srcPath, $dstRel, $archiveRel ) = $triplet;
 282
 283                         if ( substr( $srcPath, 0, 9 ) == 'mwrepo://' ) {
 284                                 $triplets[$i][0] = $srcPath = $this->resolveVirtualUrl( $srcPath );
 285                         }
 286                         if ( !$this->validateFilename( $dstRel ) ) {
 287                                 throw new MWException( 'Validation error in $dstRel' );
 288                         }
 289                         if ( !$this->validateFilename( $archiveRel ) ) {
 290                                 throw new MWException( 'Validation error in $archiveRel' );
 291                         }
 292                         $dstPath = "{$this->directory}/$dstRel";
 293                         $archivePath = "{$this->directory}/$archiveRel";
 294
 295                         $dstDir = dirname( $dstPath );
 296                         $archiveDir = dirname( $archivePath );
 297                         // Abort immediately on directory creation errors since they're likely to be repetitive
 298                         if ( !is_dir( $dstDir ) && !wfMkdirParents( $dstDir ) ) {
 299                                 return $this->newFatal( 'directorycreateerror', $dstDir );
 300                         }
 301                         if ( !is_dir( $archiveDir ) && !wfMkdirParents( $archiveDir ) ) {
 302                                 return $this->newFatal( 'directorycreateerror', $archiveDir );
 303                         }
 304                         if ( !is_file( $srcPath ) ) {
 305                                 // Make a list of files that don't exist for return to the caller
 306                                 $status->fatal( 'filenotfound', $srcPath );
 307                         }
 308                 }
 309
 310                 if ( !$status->ok ) {
 311                         return $status;
 312                 }
 313
 314                 foreach ( $triplets as $i => $triplet ) {
 315                         list( $srcPath, $dstRel, $archiveRel ) = $triplet;
 316                         $dstPath = "{$this->directory}/$dstRel";
 317                         $archivePath = "{$this->directory}/$archiveRel";
 318
 319                         // Archive destination file if it exists
 320                         if( is_file( $dstPath ) ) {
 321                                 // Check if the archive file exists
 322                                 // This is a sanity check to avoid data loss. In UNIX, the rename primitive
 323                                 // unlinks the destination file if it exists. DB-based synchronisation in
 324                                 // publishBatch's caller should prevent races. In Windows there's no
 325                                 // problem because the rename primitive fails if the destination exists.
 326                                 if ( is_file( $archivePath ) ) {
 327                                         $success = false;
 328                                 } else {
 329                                         wfSuppressWarnings();
 330                                         $success = rename( $dstPath, $archivePath );
 331                                         wfRestoreWarnings();
 332                                 }
 333
 334                                 if( !$success ) {
 335                                         $status->error( 'filerenameerror',$dstPath, $archivePath );
 336                                         $status->failCount++;
 337                                         continue;
 338                                 } else {
 339                                         wfDebug(__METHOD__.": moved file $dstPath to $archivePath\n");
 340                                 }
 341                                 $status->value[$i] = 'archived';
 342                         } else {
 343                                 $status->value[$i] = 'new';
 344                         }
 345
 346                         $good = true;
 347                         wfSuppressWarnings();
 348                         if ( $flags & self::DELETE_SOURCE ) {
 349                                 if ( !rename( $srcPath, $dstPath ) ) {
 350                                         $status->error( 'filerenameerror', $srcPath, $dstPath );
 351                                         $good = false;
 352                                 }
 353                         } else {
 354                                 if ( !copy( $srcPath, $dstPath ) ) {
 355                                         $status->error( 'filecopyerror', $srcPath, $dstPath );
 356                                         $good = false;
 357                                 }
 358                         }
 359                         wfRestoreWarnings();
 360
 361                         if ( $good ) {
 362                                 $status->successCount++;
 363                                 wfDebug(__METHOD__.": wrote tempfile $srcPath to $dstPath\n");
 364                                 // Thread-safe override for umask
 365                                 chmod( $dstPath, 0644 );
 366                         } else {
 367                                 $status->failCount++;
 368                         }
 369                 }
 370                 return $status;
 371         }
 372
 373         /**
 374          * Move a group of files to the deletion archive.
 375          * If no valid deletion archive is configured, this may either delete the
 376          * file or throw an exception, depending on the preference of the repository.
 377          *
 378          * @param array $sourceDestPairs Array of source/destination pairs. Each element
 379          *        is a two-element array containing the source file path relative to the
 380          *        public root in the first element, and the archive file path relative
 381          *        to the deleted zone root in the second element.
 382          * @return FileRepoStatus
 383          */
 384         function deleteBatch( $sourceDestPairs ) {
 385                 $status = $this->newGood();
 386                 if ( !$this->deletedDir ) {
 387                         throw new MWException( __METHOD__.': no valid deletion archive directory' );
 388                 }
 389
 390                 /**
 391                  * Validate filenames and create archive directories
 392                  */
 393                 foreach ( $sourceDestPairs as $pair ) {
 394                         list( $srcRel, $archiveRel ) = $pair;
 395                         if ( !$this->validateFilename( $srcRel ) ) {
 396                                 throw new MWException( __METHOD__.':Validation error in $srcRel' );
 397                         }
 398                         if ( !$this->validateFilename( $archiveRel ) ) {
 399                                 throw new MWException( __METHOD__.':Validation error in $archiveRel' );
 400                         }
 401                         $archivePath = "{$this->deletedDir}/$archiveRel";
 402                         $archiveDir = dirname( $archivePath );
 403                         if ( !is_dir( $archiveDir ) ) {
 404                                 if ( !wfMkdirParents( $archiveDir ) ) {
 405                                         $status->fatal( 'directorycreateerror', $archiveDir );
 406                                         continue;
 407                                 }
 408                                 $this->initDeletedDir( $archiveDir );
 409                         }
 410                         // Check if the archive directory is writable
 411                         // This doesn't appear to work on NTFS
 412                         if ( !is_writable( $archiveDir ) ) {
 413                                 $status->fatal( 'filedelete-archive-read-only', $archiveDir );
 414                         }
 415                 }
 416                 if ( !$status->ok ) {
 417                         // Abort early
 418                         return $status;
 419                 }
 420
 421                 /**
 422                  * Move the files
 423                  * We're now committed to returning an OK result, which will lead to
 424                  * the files being moved in the DB also.
 425                  */
 426                 foreach ( $sourceDestPairs as $pair ) {
 427                         list( $srcRel, $archiveRel ) = $pair;
 428                         $srcPath = "{$this->directory}/$srcRel";
 429                         $archivePath = "{$this->deletedDir}/$archiveRel";
 430                         $good = true;
 431                         if ( file_exists( $archivePath ) ) {
 432                                 # A file with this content hash is already archived
 433                                 if ( !@unlink( $srcPath ) ) {
 434                                         $status->error( 'filedeleteerror', $srcPath );
 435                                         $good = false;
 436                                 }
 437                         } else{
 438                                 if ( !@rename( $srcPath, $archivePath ) ) {
 439                                         $status->error( 'filerenameerror', $srcPath, $archivePath );
 440                                         $good = false;
 441                                 } else {
 442                                         @chmod( $archivePath, 0644 );
 443                                 }
 444                         }
 445                         if ( $good ) {
 446                                 $status->successCount++;
 447                         } else {
 448                                 $status->failCount++;
 449                         }
 450                 }
 451                 return $status;
 452         }
 453
 454         /**
 455          * Get a relative path for a deletion archive key,
 456          * e.g. s/z/a/ for sza251lrxrc1jad41h5mgilp8nysje52.jpg
 457          */
 458         function getDeletedHashPath( $key ) {
 459                 $path = '';
 460                 for ( $i = 0; $i < $this->deletedHashLevels; $i++ ) {
 461                         $path .= $key[$i] . '/';
 462                 }
 463                 return $path;
 464         }
 465
 466         /**
 467          * Call a callback function for every file in the repository.
 468          * Uses the filesystem even in child classes.
 469          */
 470         function enumFilesInFS( $callback ) {
 471                 $numDirs = 1 << ( $this->hashLevels * 4 );
 472                 for ( $flatIndex = 0; $flatIndex < $numDirs; $flatIndex++ ) {
 473                         $hexString = sprintf( "%0{$this->hashLevels}x", $flatIndex );
 474                         $path = $this->directory;
 475                         for ( $hexPos = 0; $hexPos < $this->hashLevels; $hexPos++ ) {
 476                                 $path .= '/' . substr( $hexString, 0, $hexPos + 1 );
 477                         }
 478                         if ( !file_exists( $path ) || !is_dir( $path ) ) {
 479                                 continue;
 480                         }
 481                         $dir = opendir( $path );
 482                         while ( false !== ( $name = readdir( $dir ) ) ) {
 483                                 call_user_func( $callback, $path . '/' . $name );
 484                         }
 485                 }
 486         }
 487
 488         /**
 489          * Call a callback function for every file in the repository
 490          * May use either the database or the filesystem
 491          */
 492         function enumFiles( $callback ) {
 493                 $this->enumFilesInFS( $callback );
 494         }
 495
 496         /**
 497          * Get properties of a file with a given virtual URL
 498          * The virtual URL must refer to this repo
 499          */
 500         function getFileProps( $virtualUrl ) {
 501                 $path = $this->resolveVirtualUrl( $virtualUrl );
 502                 return File::getPropsFromPath( $path );
 503         }
 504
 505         /**
 506          * Path disclosure protection functions
 507          *
 508          * Get a callback function to use for cleaning error message parameters
 509          */
 510         function getErrorCleanupFunction() {
 511                 switch ( $this->pathDisclosureProtection ) {
 512                         case 'simple':
 513                                 $callback = array( $this, 'simpleClean' );
 514                                 break;
 515                         default:
 516                                 $callback = parent::getErrorCleanupFunction();
 517                 }
 518                 return $callback;
 519         }
 520
 521         function simpleClean( $param ) {
 522                 if ( !isset( $this->simpleCleanPairs ) ) {
 523                         global $IP;
 524                         $this->simpleCleanPairs = array(
 525                                 $this->directory => 'public',
 526                                 "{$this->directory}/temp" => 'temp',
 527                                 $IP => '$IP',
 528                                 dirname( __FILE__ ) => '$IP/extensions/WebStore',
 529                         );
 530                         if ( $this->deletedDir ) {
 531                                 $this->simpleCleanPairs[$this->deletedDir] = 'deleted';
 532                         }
 533                 }
 534                 return strtr( $param, $this->simpleCleanPairs );
 535         }
 536
 537 }