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, $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 ( !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                                 // In the deleted zone, seed new directories with a blank
 153                                 // index.html, to prevent crawling
 154                                 if ( $dstZone == 'deleted' ) {
 155                                         file_put_contents( "$dstDir/index.html", '' );
 156                                 }
 157                         }
 158
 159                         if ( self::isVirtualUrl( $srcPath ) ) {
 160                                 $srcPath = $triplets[$i][0] = $this->resolveVirtualUrl( $srcPath );
 161                         }
 162                         if ( !is_file( $srcPath ) ) {
 163                                 // Make a list of files that don't exist for return to the caller
 164                                 $status->fatal( 'filenotfound', $srcPath );
 165                                 continue;
 166                         }
 167                         if ( !( $flags & self::OVERWRITE ) && file_exists( $dstPath ) ) {
 168                                 if ( $flags & self::OVERWRITE_SAME ) {
 169                                         $hashSource = sha1_file( $srcPath );
 170                                         $hashDest = sha1_file( $dstPath );
 171                                         if ( $hashSource != $hashDest ) {
 172                                                 $status->fatal( 'fileexistserror', $dstPath );
 173                                         }
 174                                 } else {
 175                                         $status->fatal( 'fileexistserror', $dstPath );
 176                                 }
 177                         }
 178                 }
 179
 180                 $deleteDest = wfIsWindows() && ( $flags & self::OVERWRITE );
 181
 182                 // Abort now on failure
 183                 if ( !$status->ok ) {
 184                         return $status;
 185                 }
 186
 187                 foreach ( $triplets as $triplet ) {
 188                         list( $srcPath, $dstZone, $dstRel ) = $triplet;
 189                         $root = $this->getZonePath( $dstZone );
 190                         $dstPath = "$root/$dstRel";
 191                         $good = true;
 192
 193                         if ( $flags & self::DELETE_SOURCE ) {
 194                                 if ( $deleteDest ) {
 195                                         unlink( $dstPath );
 196                                 }
 197                                 if ( !rename( $srcPath, $dstPath ) ) {
 198                                         $status->error( 'filerenameerror', $srcPath, $dstPath );
 199                                         $good = false;
 200                                 }
 201                         } else {
 202                                 if ( !copy( $srcPath, $dstPath ) ) {
 203                                         $status->error( 'filecopyerror', $srcPath, $dstPath );
 204                                         $good = false;
 205                                 }
 206                         }
 207                         if ( $good ) {
 208                                 chmod( $dstPath, 0644 );
 209                                 $status->successCount++;
 210                         } else {
 211                                 $status->failCount++;
 212                         }
 213                 }
 214                 return $status;
 215         }
 216
 217         /**
 218          * Pick a random name in the temp zone and store a file to it.
 219          * @param string $originalName The base name of the file as specified
 220          *     by the user. The file extension will be maintained.
 221          * @param string $srcPath The current location of the file.
 222          * @return FileRepoStatus object with the URL in the value.
 223          */
 224         function storeTemp( $originalName, $srcPath ) {
 225                 $date = gmdate( "YmdHis" );
 226                 $hashPath = $this->getHashPath( $originalName );
 227                 $dstRel = "$hashPath$date!$originalName";
 228                 $dstUrlRel = $hashPath . $date . '!' . rawurlencode( $originalName );
 229
 230                 $result = $this->store( $srcPath, 'temp', $dstRel );
 231                 $result->value = $this->getVirtualUrl( 'temp' ) . '/' . $dstUrlRel;
 232                 return $result;
 233         }
 234
 235         /**
 236          * Remove a temporary file or mark it for garbage collection
 237          * @param string $virtualUrl The virtual URL returned by storeTemp
 238          * @return boolean True on success, false on failure
 239          */
 240         function freeTemp( $virtualUrl ) {
 241                 $temp = "mwrepo://{$this->name}/temp";
 242                 if ( substr( $virtualUrl, 0, strlen( $temp ) ) != $temp ) {
 243                         wfDebug( __METHOD__.": Invalid virtual URL\n" );
 244                         return false;
 245                 }
 246                 $path = $this->resolveVirtualUrl( $virtualUrl );
 247                 wfSuppressWarnings();
 248                 $success = unlink( $path );
 249                 wfRestoreWarnings();
 250                 return $success;
 251         }
 252
 253         /**
 254          * Publish a batch of files
 255          * @param array $triplets (source,dest,archive) triplets as per publish()
 256          * @param integer $flags Bitfield, may be FileRepo::DELETE_SOURCE to indicate
 257          *        that the source files should be deleted if possible
 258          */
 259         function publishBatch( $triplets, $flags = 0 ) {
 260                 // Perform initial checks
 261                 if ( !wfMkdirParents( $this->directory ) ) {
 262                         return $this->newFatal( 'upload_directory_missing', $this->directory );
 263                 }
 264                 if ( !is_writable( $this->directory ) ) {
 265                         return $this->newFatal( 'upload_directory_read_only', $this->directory );
 266                 }
 267                 $status = $this->newGood( array() );
 268                 foreach ( $triplets as $i => $triplet ) {
 269                         list( $srcPath, $dstRel, $archiveRel ) = $triplet;
 270
 271                         if ( substr( $srcPath, 0, 9 ) == 'mwrepo://' ) {
 272                                 $triplets[$i][0] = $srcPath = $this->resolveVirtualUrl( $srcPath );
 273                         }
 274                         if ( !$this->validateFilename( $dstRel ) ) {
 275                                 throw new MWException( 'Validation error in $dstRel' );
 276                         }
 277                         if ( !$this->validateFilename( $archiveRel ) ) {
 278                                 throw new MWException( 'Validation error in $archiveRel' );
 279                         }
 280                         $dstPath = "{$this->directory}/$dstRel";
 281                         $archivePath = "{$this->directory}/$archiveRel";
 282
 283                         $dstDir = dirname( $dstPath );
 284                         $archiveDir = dirname( $archivePath );
 285                         // Abort immediately on directory creation errors since they're likely to be repetitive
 286                         if ( !is_dir( $dstDir ) && !wfMkdirParents( $dstDir ) ) {
 287                                 return $this->newFatal( 'directorycreateerror', $dstDir );
 288                         }
 289                         if ( !is_dir( $archiveDir ) && !wfMkdirParents( $archiveDir ) ) {
 290                                 return $this->newFatal( 'directorycreateerror', $archiveDir );
 291                         }
 292                         if ( !is_file( $srcPath ) ) {
 293                                 // Make a list of files that don't exist for return to the caller
 294                                 $status->fatal( 'filenotfound', $srcPath );
 295                         }
 296                 }
 297
 298                 if ( !$status->ok ) {
 299                         return $status;
 300                 }
 301
 302                 foreach ( $triplets as $i => $triplet ) {
 303                         list( $srcPath, $dstRel, $archiveRel ) = $triplet;
 304                         $dstPath = "{$this->directory}/$dstRel";
 305                         $archivePath = "{$this->directory}/$archiveRel";
 306
 307                         // Archive destination file if it exists
 308                         if( is_file( $dstPath ) ) {
 309                                 // Check if the archive file exists
 310                                 // This is a sanity check to avoid data loss. In UNIX, the rename primitive
 311                                 // unlinks the destination file if it exists. DB-based synchronisation in
 312                                 // publishBatch's caller should prevent races. In Windows there's no
 313                                 // problem because the rename primitive fails if the destination exists.
 314                                 if ( is_file( $archivePath ) ) {
 315                                         $success = false;
 316                                 } else {
 317                                         wfSuppressWarnings();
 318                                         $success = rename( $dstPath, $archivePath );
 319                                         wfRestoreWarnings();
 320                                 }
 321
 322                                 if( !$success ) {
 323                                         $status->error( 'filerenameerror',$dstPath, $archivePath );
 324                                         $status->failCount++;
 325                                         continue;
 326                                 } else {
 327                                         wfDebug(__METHOD__.": moved file $dstPath to $archivePath\n");
 328                                 }
 329                                 $status->value[$i] = 'archived';
 330                         } else {
 331                                 $status->value[$i] = 'new';
 332                         }
 333
 334                         $good = true;
 335                         wfSuppressWarnings();
 336                         if ( $flags & self::DELETE_SOURCE ) {
 337                                 if ( !rename( $srcPath, $dstPath ) ) {
 338                                         $status->error( 'filerenameerror', $srcPath, $dstPath );
 339                                         $good = false;
 340                                 }
 341                         } else {
 342                                 if ( !copy( $srcPath, $dstPath ) ) {
 343                                         $status->error( 'filecopyerror', $srcPath, $dstPath );
 344                                         $good = false;
 345                                 }
 346                         }
 347                         wfRestoreWarnings();
 348
 349                         if ( $good ) {
 350                                 $status->successCount++;
 351                                 wfDebug(__METHOD__.": wrote tempfile $srcPath to $dstPath\n");
 352                                 // Thread-safe override for umask
 353                                 chmod( $dstPath, 0644 );
 354                         } else {
 355                                 $status->failCount++;
 356                         }
 357                 }
 358                 return $status;
 359         }
 360
 361         /**
 362          * Move a group of files to the deletion archive.
 363          * If no valid deletion archive is configured, this may either delete the
 364          * file or throw an exception, depending on the preference of the repository.
 365          *
 366          * @param array $sourceDestPairs Array of source/destination pairs. Each element
 367          *        is a two-element array containing the source file path relative to the
 368          *        public root in the first element, and the archive file path relative
 369          *        to the deleted zone root in the second element.
 370          * @return FileRepoStatus
 371          */
 372         function deleteBatch( $sourceDestPairs ) {
 373                 $status = $this->newGood();
 374                 if ( !$this->deletedDir ) {
 375                         throw new MWException( __METHOD__.': no valid deletion archive directory' );
 376                 }
 377
 378                 /**
 379                  * Validate filenames and create archive directories
 380                  */
 381                 foreach ( $sourceDestPairs as $pair ) {
 382                         list( $srcRel, $archiveRel ) = $pair;
 383                         if ( !$this->validateFilename( $srcRel ) ) {
 384                                 throw new MWException( __METHOD__.':Validation error in $srcRel' );
 385                         }
 386                         if ( !$this->validateFilename( $archiveRel ) ) {
 387                                 throw new MWException( __METHOD__.':Validation error in $archiveRel' );
 388                         }
 389                         $archivePath = "{$this->deletedDir}/$archiveRel";
 390                         $archiveDir = dirname( $archivePath );
 391                         if ( !is_dir( $archiveDir ) ) {
 392                                 if ( !wfMkdirParents( $archiveDir ) ) {
 393                                         $status->fatal( 'directorycreateerror', $archiveDir );
 394                                         continue;
 395                                 }
 396                                 // Seed new directories with a blank index.html, to prevent crawling
 397                                 file_put_contents( "$archiveDir/index.html", '' );
 398                         }
 399                         // Check if the archive directory is writable
 400                         // This doesn't appear to work on NTFS
 401                         if ( !is_writable( $archiveDir ) ) {
 402                                 $status->fatal( 'filedelete-archive-read-only', $archiveDir );
 403                         }
 404                 }
 405                 if ( !$status->ok ) {
 406                         // Abort early
 407                         return $status;
 408                 }
 409
 410                 /**
 411                  * Move the files
 412                  * We're now committed to returning an OK result, which will lead to
 413                  * the files being moved in the DB also.
 414                  */
 415                 foreach ( $sourceDestPairs as $pair ) {
 416                         list( $srcRel, $archiveRel ) = $pair;
 417                         $srcPath = "{$this->directory}/$srcRel";
 418                         $archivePath = "{$this->deletedDir}/$archiveRel";
 419                         $good = true;
 420                         if ( file_exists( $archivePath ) ) {
 421                                 # A file with this content hash is already archived
 422                                 if ( !@unlink( $srcPath ) ) {
 423                                         $status->error( 'filedeleteerror', $srcPath );
 424                                         $good = false;
 425                                 }
 426                         } else{
 427                                 if ( !@rename( $srcPath, $archivePath ) ) {
 428                                         $status->error( 'filerenameerror', $srcPath, $archivePath );
 429                                         $good = false;
 430                                 } else {
 431                                         @chmod( $archivePath, 0644 );
 432                                 }
 433                         }
 434                         if ( $good ) {
 435                                 $status->successCount++;
 436                         } else {
 437                                 $status->failCount++;
 438                         }
 439                 }
 440                 return $status;
 441         }
 442
 443         /**
 444          * Get a relative path including trailing slash, e.g. f/fa/
 445          * If the repo is not hashed, returns an empty string
 446          */
 447         function getHashPath( $name ) {
 448                 return FileRepo::getHashPathForLevel( $name, $this->hashLevels );
 449         }
 450
 451         /**
 452          * Get a relative path for a deletion archive key,
 453          * e.g. s/z/a/ for sza251lrxrc1jad41h5mgilp8nysje52.jpg
 454          */
 455         function getDeletedHashPath( $key ) {
 456                 $path = '';
 457                 for ( $i = 0; $i < $this->deletedHashLevels; $i++ ) {
 458                         $path .= $key[$i] . '/';
 459                 }
 460                 return $path;
 461         }
 462
 463         /**
 464          * Call a callback function for every file in the repository.
 465          * Uses the filesystem even in child classes.
 466          */
 467         function enumFilesInFS( $callback ) {
 468                 $numDirs = 1 << ( $this->hashLevels * 4 );
 469                 for ( $flatIndex = 0; $flatIndex < $numDirs; $flatIndex++ ) {
 470                         $hexString = sprintf( "%0{$this->hashLevels}x", $flatIndex );
 471                         $path = $this->directory;
 472                         for ( $hexPos = 0; $hexPos < $this->hashLevels; $hexPos++ ) {
 473                                 $path .= '/' . substr( $hexString, 0, $hexPos + 1 );
 474                         }
 475                         if ( !file_exists( $path ) || !is_dir( $path ) ) {
 476                                 continue;
 477                         }
 478                         $dir = opendir( $path );
 479                         while ( false !== ( $name = readdir( $dir ) ) ) {
 480                                 call_user_func( $callback, $path . '/' . $name );
 481                         }
 482                 }
 483         }
 484
 485         /**
 486          * Call a callback function for every file in the repository
 487          * May use either the database or the filesystem
 488          */
 489         function enumFiles( $callback ) {
 490                 $this->enumFilesInFS( $callback );
 491         }
 492
 493         /**
 494          * Get properties of a file with a given virtual URL
 495          * The virtual URL must refer to this repo
 496          */
 497         function getFileProps( $virtualUrl ) {
 498                 $path = $this->resolveVirtualUrl( $virtualUrl );
 499                 return File::getPropsFromPath( $path );
 500         }
 501
 502         /**
 503          * Path disclosure protection functions
 504          *
 505          * Get a callback function to use for cleaning error message parameters
 506          */
 507         function getErrorCleanupFunction() {
 508                 switch ( $this->pathDisclosureProtection ) {
 509                         case 'simple':
 510                                 $callback = array( $this, 'simpleClean' );
 511                                 break;
 512                         default:
 513                                 $callback = parent::getErrorCleanupFunction();
 514                 }
 515                 return $callback;
 516         }
 517
 518         function simpleClean( $param ) {
 519                 if ( !isset( $this->simpleCleanPairs ) ) {
 520                         global $IP;
 521                         $this->simpleCleanPairs = array(
 522                                 $this->directory => 'public',
 523                                 "{$this->directory}/temp" => 'temp',
 524                                 $IP => '$IP',
 525                                 dirname( __FILE__ ) => '$IP/extensions/WebStore',
 526                         );
 527                         if ( $this->deletedDir ) {
 528                                 $this->simpleCleanPairs[$this->deletedDir] = 'deleted';
 529                         }
 530                 }
 531                 return strtr( $param, $this->simpleCleanPairs );
 532         }
 533
 534 }