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, $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                 if ( isset( $info['thumbDir'] ) ) {
  28                         $this->thumbDir =  $info['thumbDir'];
  29                 } else {
  30                         $this->thumbDir = "{$this->directory}/thumb";
  31                 }
  32                 if ( isset( $info['thumbUrl'] ) ) {
  33                         $this->thumbUrl = $info['thumbUrl'];
  34                 } else {
  35                         $this->thumbUrl = "{$this->url}/thumb";
  36                 }
  37         }
  38
  39         /**
  40          * Get the public root directory of the repository.
  41          */
  42         function getRootDirectory() {
  43                 return $this->directory;
  44         }
  45
  46         /**
  47          * Get the public root URL of the repository
  48          */
  49         function getRootUrl() {
  50                 return $this->url;
  51         }
  52
  53         /**
  54          * Returns true if the repository uses a multi-level directory structure
  55          */
  56         function isHashed() {
  57                 return (bool)$this->hashLevels;
  58         }
  59
  60         /**
  61          * Get the local directory corresponding to one of the three basic zones
  62          */
  63         function getZonePath( $zone ) {
  64                 switch ( $zone ) {
  65                         case 'public':
  66                                 return $this->directory;
  67                         case 'temp':
  68                                 return "{$this->directory}/temp";
  69                         case 'deleted':
  70                                 return $this->deletedDir;
  71                         case 'thumb':
  72                                 return $this->thumbDir;
  73                         default:
  74                                 return false;
  75                 }
  76         }
  77
  78         /**
  79          * @see FileRepo::getZoneUrl()
  80          */
  81         function getZoneUrl( $zone ) {
  82                 switch ( $zone ) {
  83                         case 'public':
  84                                 return $this->url;
  85                         case 'temp':
  86                                 return "{$this->url}/temp";
  87                         case 'deleted':
  88                                 return parent::getZoneUrl( $zone ); // no public URL
  89                         case 'thumb':
  90                                 return $this->thumbUrl;
  91                         default:
  92                                 return parent::getZoneUrl( $zone );
  93                 }
  94         }
  95
  96         /**
  97          * Get a URL referring to this repository, with the private mwrepo protocol.
  98          * The suffix, if supplied, is considered to be unencoded, and will be
  99          * URL-encoded before being returned.
 100          */
 101         function getVirtualUrl( $suffix = false ) {
 102                 $path = 'mwrepo://' . $this->name;
 103                 if ( $suffix !== false ) {
 104                         $path .= '/' . rawurlencode( $suffix );
 105                 }
 106                 return $path;
 107         }
 108
 109         /**
 110          * Get the local path corresponding to a virtual URL
 111          */
 112         function resolveVirtualUrl( $url ) {
 113                 if ( substr( $url, 0, 9 ) != 'mwrepo://' ) {
 114                         throw new MWException( __METHOD__.': unknown protoocl' );
 115                 }
 116
 117                 $bits = explode( '/', substr( $url, 9 ), 3 );
 118                 if ( count( $bits ) != 3 ) {
 119                         throw new MWException( __METHOD__.": invalid mwrepo URL: $url" );
 120                 }
 121                 list( $repo, $zone, $rel ) = $bits;
 122                 if ( $repo !== $this->name ) {
 123                         throw new MWException( __METHOD__.": fetching from a foreign repo is not supported" );
 124                 }
 125                 $base = $this->getZonePath( $zone );
 126                 if ( !$base ) {
 127                         throw new MWException( __METHOD__.": invalid zone: $zone" );
 128                 }
 129                 return $base . '/' . rawurldecode( $rel );
 130         }
 131
 132         /**
 133          * Store a batch of files
 134          *
 135          * @param array $triplets (src,zone,dest) triplets as per store()
 136          * @param integer $flags Bitwise combination of the following flags:
 137          *     self::DELETE_SOURCE     Delete the source file after upload
 138          *     self::OVERWRITE         Overwrite an existing destination file instead of failing
 139          *     self::OVERWRITE_SAME    Overwrite the file if the destination exists and has the
 140          *                             same contents as the source
 141          */
 142         function storeBatch( $triplets, $flags = 0 ) {
 143                 if ( !wfMkdirParents( $this->directory ) ) {
 144                         return $this->newFatal( 'upload_directory_missing', $this->directory );
 145                 }
 146                 if ( !is_writable( $this->directory ) ) {
 147                         return $this->newFatal( 'upload_directory_read_only', $this->directory );
 148                 }
 149                 $status = $this->newGood();
 150                 foreach ( $triplets as $i => $triplet ) {
 151                         list( $srcPath, $dstZone, $dstRel ) = $triplet;
 152
 153                         $root = $this->getZonePath( $dstZone );
 154                         if ( !$root ) {
 155                                 throw new MWException( "Invalid zone: $dstZone" );
 156                         }
 157                         if ( !$this->validateFilename( $dstRel ) ) {
 158                                 throw new MWException( 'Validation error in $dstRel' );
 159                         }
 160                         $dstPath = "$root/$dstRel";
 161                         $dstDir = dirname( $dstPath );
 162
 163                         if ( !is_dir( $dstDir ) ) {
 164                                 if ( !wfMkdirParents( $dstDir ) ) {
 165                                         return $this->newFatal( 'directorycreateerror', $dstDir );
 166                                 }
 167                                 if ( $dstZone == 'deleted' ) {
 168                                         $this->initDeletedDir( $dstDir );
 169                                 }
 170                         }
 171
 172                         if ( self::isVirtualUrl( $srcPath ) ) {
 173                                 $srcPath = $triplets[$i][0] = $this->resolveVirtualUrl( $srcPath );
 174                         }
 175                         if ( !is_file( $srcPath ) ) {
 176                                 // Make a list of files that don't exist for return to the caller
 177                                 $status->fatal( 'filenotfound', $srcPath );
 178                                 continue;
 179                         }
 180                         if ( !( $flags & self::OVERWRITE ) && file_exists( $dstPath ) ) {
 181                                 if ( $flags & self::OVERWRITE_SAME ) {
 182                                         $hashSource = sha1_file( $srcPath );
 183                                         $hashDest = sha1_file( $dstPath );
 184                                         if ( $hashSource != $hashDest ) {
 185                                                 $status->fatal( 'fileexistserror', $dstPath );
 186                                         }
 187                                 } else {
 188                                         $status->fatal( 'fileexistserror', $dstPath );
 189                                 }
 190                         }
 191                 }
 192
 193                 $deleteDest = wfIsWindows() && ( $flags & self::OVERWRITE );
 194
 195                 // Abort now on failure
 196                 if ( !$status->ok ) {
 197                         return $status;
 198                 }
 199
 200                 foreach ( $triplets as $triplet ) {
 201                         list( $srcPath, $dstZone, $dstRel ) = $triplet;
 202                         $root = $this->getZonePath( $dstZone );
 203                         $dstPath = "$root/$dstRel";
 204                         $good = true;
 205
 206                         if ( $flags & self::DELETE_SOURCE ) {
 207                                 if ( $deleteDest ) {
 208                                         unlink( $dstPath );
 209                                 }
 210                                 if ( !rename( $srcPath, $dstPath ) ) {
 211                                         $status->error( 'filerenameerror', $srcPath, $dstPath );
 212                                         $good = false;
 213                                 }
 214                         } else {
 215                                 if ( !copy( $srcPath, $dstPath ) ) {
 216                                         $status->error( 'filecopyerror', $srcPath, $dstPath );
 217                                         $good = false;
 218                                 }
 219                         }
 220                         if ( $good ) {
 221                                 $this->chmod( $dstPath );
 222                                 $status->successCount++;
 223                         } else {
 224                                 $status->failCount++;
 225                         }
 226                 }
 227                 return $status;
 228         }
 229
 230         function append( $srcPath, $toAppendPath, $flags = 0 ) {
 231                 $status = $this->newGood();
 232
 233                 // Resolve the virtual URL
 234                 if ( self::isVirtualUrl( $srcPath ) ) {
 235                         $srcPath = $this->resolveVirtualUrl( $srcPath );
 236                 }
 237                 // Make sure the files are there
 238                 if ( !is_file( $srcPath ) )
 239                         $status->fatal( 'filenotfound', $srcPath );
 240
 241                 if ( !is_file( $toAppendPath ) )
 242                         $status->fatal( 'filenotfound', $toAppendPath );
 243
 244                 if ( !$status->isOk() ) return $status;
 245
 246                 // Do the append
 247                 $chunk = file_get_contents( $toAppendPath );
 248                 if( $chunk === false ) {
 249                         $status->fatal( 'fileappenderrorread', $toAppendPath );
 250                 }
 251
 252                 if( $status->isOk() ) {
 253                         if ( file_put_contents( $srcPath, $chunk, FILE_APPEND ) ) {
 254                                 $status->value = $srcPath;
 255                         } else {
 256                                 $status->fatal( 'fileappenderror', $toAppendPath,  $srcPath);
 257                         }
 258                 }
 259
 260                 if ( $flags & self::DELETE_SOURCE ) {
 261                         unlink( $toAppendPath );
 262                 }
 263
 264                 return $status;
 265         }
 266
 267         /**
 268          * Checks existence of specified array of files.
 269          *
 270          * @param array $files URLs of files to check
 271          * @param integer $flags Bitwise combination of the following flags:
 272          *     self::FILES_ONLY     Mark file as existing only if it is a file (not directory)
 273          * @return Either array of files and existence flags, or false
 274          */
 275         function fileExistsBatch( $files, $flags = 0 ) {
 276                 if ( !file_exists( $this->directory ) || !is_readable( $this->directory ) ) {
 277                         return false;
 278                 }
 279                 $result = array();
 280                 foreach ( $files as $key => $file ) {
 281                         if ( self::isVirtualUrl( $file ) ) {
 282                                 $file = $this->resolveVirtualUrl( $file );
 283                         }
 284                         if( $flags & self::FILES_ONLY ) {
 285                                 $result[$key] = is_file( $file );
 286                         } else {
 287                                 $result[$key] = file_exists( $file );
 288                         }
 289                 }
 290
 291                 return $result;
 292         }
 293
 294         /**
 295          * Take all available measures to prevent web accessibility of new deleted
 296          * directories, in case the user has not configured offline storage
 297          */
 298         protected function initDeletedDir( $dir ) {
 299                 // Add a .htaccess file to the root of the deleted zone
 300                 $root = $this->getZonePath( 'deleted' );
 301                 if ( !file_exists( "$root/.htaccess" ) ) {
 302                         file_put_contents( "$root/.htaccess", "Deny from all\n" );
 303                 }
 304                 // Seed new directories with a blank index.html, to prevent crawling
 305                 file_put_contents( "$dir/index.html", '' );
 306         }
 307
 308         /**
 309          * Pick a random name in the temp zone and store a file to it.
 310          * @param string $originalName The base name of the file as specified
 311          *     by the user. The file extension will be maintained.
 312          * @param string $srcPath The current location of the file.
 313          * @return FileRepoStatus object with the URL in the value.
 314          */
 315         function storeTemp( $originalName, $srcPath ) {
 316                 $date = gmdate( "YmdHis" );
 317                 $hashPath = $this->getHashPath( $originalName );
 318                 $dstRel = "$hashPath$date!$originalName";
 319                 $dstUrlRel = $hashPath . $date . '!' . rawurlencode( $originalName );
 320
 321                 $result = $this->store( $srcPath, 'temp', $dstRel );
 322                 $result->value = $this->getVirtualUrl( 'temp' ) . '/' . $dstUrlRel;
 323                 return $result;
 324         }
 325
 326         /**
 327          * Remove a temporary file or mark it for garbage collection
 328          * @param string $virtualUrl The virtual URL returned by storeTemp
 329          * @return boolean True on success, false on failure
 330          */
 331         function freeTemp( $virtualUrl ) {
 332                 $temp = "mwrepo://{$this->name}/temp";
 333                 if ( substr( $virtualUrl, 0, strlen( $temp ) ) != $temp ) {
 334                         wfDebug( __METHOD__.": Invalid virtual URL\n" );
 335                         return false;
 336                 }
 337                 $path = $this->resolveVirtualUrl( $virtualUrl );
 338                 wfSuppressWarnings();
 339                 $success = unlink( $path );
 340                 wfRestoreWarnings();
 341                 return $success;
 342         }
 343
 344         /**
 345          * Publish a batch of files
 346          * @param array $triplets (source,dest,archive) triplets as per publish()
 347          * @param integer $flags Bitfield, may be FileRepo::DELETE_SOURCE to indicate
 348          *        that the source files should be deleted if possible
 349          */
 350         function publishBatch( $triplets, $flags = 0 ) {
 351                 // Perform initial checks
 352                 if ( !wfMkdirParents( $this->directory ) ) {
 353                         return $this->newFatal( 'upload_directory_missing', $this->directory );
 354                 }
 355                 if ( !is_writable( $this->directory ) ) {
 356                         return $this->newFatal( 'upload_directory_read_only', $this->directory );
 357                 }
 358                 $status = $this->newGood( array() );
 359                 foreach ( $triplets as $i => $triplet ) {
 360                         list( $srcPath, $dstRel, $archiveRel ) = $triplet;
 361
 362                         if ( substr( $srcPath, 0, 9 ) == 'mwrepo://' ) {
 363                                 $triplets[$i][0] = $srcPath = $this->resolveVirtualUrl( $srcPath );
 364                         }
 365                         if ( !$this->validateFilename( $dstRel ) ) {
 366                                 throw new MWException( 'Validation error in $dstRel' );
 367                         }
 368                         if ( !$this->validateFilename( $archiveRel ) ) {
 369                                 throw new MWException( 'Validation error in $archiveRel' );
 370                         }
 371                         $dstPath = "{$this->directory}/$dstRel";
 372                         $archivePath = "{$this->directory}/$archiveRel";
 373
 374                         $dstDir = dirname( $dstPath );
 375                         $archiveDir = dirname( $archivePath );
 376                         // Abort immediately on directory creation errors since they're likely to be repetitive
 377                         if ( !is_dir( $dstDir ) && !wfMkdirParents( $dstDir ) ) {
 378                                 return $this->newFatal( 'directorycreateerror', $dstDir );
 379                         }
 380                         if ( !is_dir( $archiveDir ) && !wfMkdirParents( $archiveDir ) ) {
 381                                 return $this->newFatal( 'directorycreateerror', $archiveDir );
 382                         }
 383                         if ( !is_file( $srcPath ) ) {
 384                                 // Make a list of files that don't exist for return to the caller
 385                                 $status->fatal( 'filenotfound', $srcPath );
 386                         }
 387                 }
 388
 389                 if ( !$status->ok ) {
 390                         return $status;
 391                 }
 392
 393                 foreach ( $triplets as $i => $triplet ) {
 394                         list( $srcPath, $dstRel, $archiveRel ) = $triplet;
 395                         $dstPath = "{$this->directory}/$dstRel";
 396                         $archivePath = "{$this->directory}/$archiveRel";
 397
 398                         // Archive destination file if it exists
 399                         if( is_file( $dstPath ) ) {
 400                                 // Check if the archive file exists
 401                                 // This is a sanity check to avoid data loss. In UNIX, the rename primitive
 402                                 // unlinks the destination file if it exists. DB-based synchronisation in
 403                                 // publishBatch's caller should prevent races. In Windows there's no
 404                                 // problem because the rename primitive fails if the destination exists.
 405                                 if ( is_file( $archivePath ) ) {
 406                                         $success = false;
 407                                 } else {
 408                                         wfSuppressWarnings();
 409                                         $success = rename( $dstPath, $archivePath );
 410                                         wfRestoreWarnings();
 411                                 }
 412
 413                                 if( !$success ) {
 414                                         $status->error( 'filerenameerror',$dstPath, $archivePath );
 415                                         $status->failCount++;
 416                                         continue;
 417                                 } else {
 418                                         wfDebug(__METHOD__.": moved file $dstPath to $archivePath\n");
 419                                 }
 420                                 $status->value[$i] = 'archived';
 421                         } else {
 422                                 $status->value[$i] = 'new';
 423                         }
 424
 425                         $good = true;
 426                         wfSuppressWarnings();
 427                         if ( $flags & self::DELETE_SOURCE ) {
 428                                 if ( !rename( $srcPath, $dstPath ) ) {
 429                                         $status->error( 'filerenameerror', $srcPath, $dstPath );
 430                                         $good = false;
 431                                 }
 432                         } else {
 433                                 if ( !copy( $srcPath, $dstPath ) ) {
 434                                         $status->error( 'filecopyerror', $srcPath, $dstPath );
 435                                         $good = false;
 436                                 }
 437                         }
 438                         wfRestoreWarnings();
 439
 440                         if ( $good ) {
 441                                 $status->successCount++;
 442                                 wfDebug(__METHOD__.": wrote tempfile $srcPath to $dstPath\n");
 443                                 // Thread-safe override for umask
 444                                 $this->chmod( $dstPath );
 445                         } else {
 446                                 $status->failCount++;
 447                         }
 448                 }
 449                 return $status;
 450         }
 451
 452         /**
 453          * Move a group of files to the deletion archive.
 454          * If no valid deletion archive is configured, this may either delete the
 455          * file or throw an exception, depending on the preference of the repository.
 456          *
 457          * @param array $sourceDestPairs Array of source/destination pairs. Each element
 458          *        is a two-element array containing the source file path relative to the
 459          *        public root in the first element, and the archive file path relative
 460          *        to the deleted zone root in the second element.
 461          * @return FileRepoStatus
 462          */
 463         function deleteBatch( $sourceDestPairs ) {
 464                 $status = $this->newGood();
 465                 if ( !$this->deletedDir ) {
 466                         throw new MWException( __METHOD__.': no valid deletion archive directory' );
 467                 }
 468
 469                 /**
 470                  * Validate filenames and create archive directories
 471                  */
 472                 foreach ( $sourceDestPairs as $pair ) {
 473                         list( $srcRel, $archiveRel ) = $pair;
 474                         if ( !$this->validateFilename( $srcRel ) ) {
 475                                 throw new MWException( __METHOD__.':Validation error in $srcRel' );
 476                         }
 477                         if ( !$this->validateFilename( $archiveRel ) ) {
 478                                 throw new MWException( __METHOD__.':Validation error in $archiveRel' );
 479                         }
 480                         $archivePath = "{$this->deletedDir}/$archiveRel";
 481                         $archiveDir = dirname( $archivePath );
 482                         if ( !is_dir( $archiveDir ) ) {
 483                                 if ( !wfMkdirParents( $archiveDir ) ) {
 484                                         $status->fatal( 'directorycreateerror', $archiveDir );
 485                                         continue;
 486                                 }
 487                                 $this->initDeletedDir( $archiveDir );
 488                         }
 489                         // Check if the archive directory is writable
 490                         // This doesn't appear to work on NTFS
 491                         if ( !is_writable( $archiveDir ) ) {
 492                                 $status->fatal( 'filedelete-archive-read-only', $archiveDir );
 493                         }
 494                 }
 495                 if ( !$status->ok ) {
 496                         // Abort early
 497                         return $status;
 498                 }
 499
 500                 /**
 501                  * Move the files
 502                  * We're now committed to returning an OK result, which will lead to
 503                  * the files being moved in the DB also.
 504                  */
 505                 foreach ( $sourceDestPairs as $pair ) {
 506                         list( $srcRel, $archiveRel ) = $pair;
 507                         $srcPath = "{$this->directory}/$srcRel";
 508                         $archivePath = "{$this->deletedDir}/$archiveRel";
 509                         $good = true;
 510                         if ( file_exists( $archivePath ) ) {
 511                                 # A file with this content hash is already archived
 512                                 if ( !@unlink( $srcPath ) ) {
 513                                         $status->error( 'filedeleteerror', $srcPath );
 514                                         $good = false;
 515                                 }
 516                         } else{
 517                                 if ( !@rename( $srcPath, $archivePath ) ) {
 518                                         $status->error( 'filerenameerror', $srcPath, $archivePath );
 519                                         $good = false;
 520                                 } else {
 521                                         $this->chmod( $archivePath );
 522                                 }
 523                         }
 524                         if ( $good ) {
 525                                 $status->successCount++;
 526                         } else {
 527                                 $status->failCount++;
 528                         }
 529                 }
 530                 return $status;
 531         }
 532
 533         /**
 534          * Get a relative path for a deletion archive key,
 535          * e.g. s/z/a/ for sza251lrxrc1jad41h5mgilp8nysje52.jpg
 536          */
 537         function getDeletedHashPath( $key ) {
 538                 $path = '';
 539                 for ( $i = 0; $i < $this->deletedHashLevels; $i++ ) {
 540                         $path .= $key[$i] . '/';
 541                 }
 542                 return $path;
 543         }
 544
 545         /**
 546          * Call a callback function for every file in the repository.
 547          * Uses the filesystem even in child classes.
 548          */
 549         function enumFilesInFS( $callback ) {
 550                 $numDirs = 1 << ( $this->hashLevels * 4 );
 551                 for ( $flatIndex = 0; $flatIndex < $numDirs; $flatIndex++ ) {
 552                         $hexString = sprintf( "%0{$this->hashLevels}x", $flatIndex );
 553                         $path = $this->directory;
 554                         for ( $hexPos = 0; $hexPos < $this->hashLevels; $hexPos++ ) {
 555                                 $path .= '/' . substr( $hexString, 0, $hexPos + 1 );
 556                         }
 557                         if ( !file_exists( $path ) || !is_dir( $path ) ) {
 558                                 continue;
 559                         }
 560                         $dir = opendir( $path );
 561                         while ( false !== ( $name = readdir( $dir ) ) ) {
 562                                 call_user_func( $callback, $path . '/' . $name );
 563                         }
 564                 }
 565         }
 566
 567         /**
 568          * Call a callback function for every file in the repository
 569          * May use either the database or the filesystem
 570          */
 571         function enumFiles( $callback ) {
 572                 $this->enumFilesInFS( $callback );
 573         }
 574
 575         /**
 576          * Get properties of a file with a given virtual URL
 577          * The virtual URL must refer to this repo
 578          */
 579         function getFileProps( $virtualUrl ) {
 580                 $path = $this->resolveVirtualUrl( $virtualUrl );
 581                 return File::getPropsFromPath( $path );
 582         }
 583
 584         /**
 585          * Path disclosure protection functions
 586          *
 587          * Get a callback function to use for cleaning error message parameters
 588          */
 589         function getErrorCleanupFunction() {
 590                 switch ( $this->pathDisclosureProtection ) {
 591                         case 'simple':
 592                                 $callback = array( $this, 'simpleClean' );
 593                                 break;
 594                         default:
 595                                 $callback = parent::getErrorCleanupFunction();
 596                 }
 597                 return $callback;
 598         }
 599
 600         function simpleClean( $param ) {
 601                 if ( !isset( $this->simpleCleanPairs ) ) {
 602                         global $IP;
 603                         $this->simpleCleanPairs = array(
 604                                 $this->directory => 'public',
 605                                 "{$this->directory}/temp" => 'temp',
 606                                 $IP => '$IP',
 607                                 dirname( __FILE__ ) => '$IP/extensions/WebStore',
 608                         );
 609                         if ( $this->deletedDir ) {
 610                                 $this->simpleCleanPairs[$this->deletedDir] = 'deleted';
 611                         }
 612                 }
 613                 return strtr( $param, $this->simpleCleanPairs );
 614         }
 615
 616         /**
 617          * Chmod a file, supressing the warnings.
 618          * @param String $path The path to change
 619          */
 620         protected function chmod( $path ) {
 621                 wfSuppressWarnings();
 622                 chmod( $path, $this->fileMode );
 623                 wfRestoreWarnings();
 624         }
 625
 626 }