includes/filerepo/backend/FileBackendStore.php

   1 <?php
   2 /**
   3  * Base class for all backends using particular storage medium.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup FileBackend
  22  * @author Aaron Schulz
  23  */
  24
  25 /**
  26  * @brief Base class for all backends using particular storage medium.
  27  *
  28  * This class defines the methods as abstract that subclasses must implement.
  29  * Outside callers should *not* use functions with "Internal" in the name.
  30  *
  31  * The FileBackend operations are implemented using basic functions
  32  * such as storeInternal(), copyInternal(), deleteInternal() and the like.
  33  * This class is also responsible for path resolution and sanitization.
  34  *
  35  * @ingroup FileBackend
  36  * @since 1.19
  37  */
  38 abstract class FileBackendStore extends FileBackend {
  39         /** @var BagOStuff */
  40         protected $memCache;
  41
  42         /** @var Array Map of paths to small (RAM/disk) cache items */
  43         protected $cache = array(); // (storage path => key => value)
  44         protected $maxCacheSize = 300; // integer; max paths with entries
  45         /** @var Array Map of paths to large (RAM/disk) cache items */
  46         protected $expensiveCache = array(); // (storage path => key => value)
  47         protected $maxExpensiveCacheSize = 5; // integer; max paths with entries
  48
  49         /** @var Array Map of container names to sharding settings */
  50         protected $shardViaHashLevels = array(); // (container name => config array)
  51
  52         protected $maxFileSize = 4294967296; // integer bytes (4GiB)
  53
  54         /**
  55          * @see FileBackend::__construct()
  56          *
  57          * @param $config Array
  58          */
  59         public function __construct( array $config ) {
  60                 parent::__construct( $config );
  61                 $this->memCache = new EmptyBagOStuff(); // disabled by default
  62         }
  63
  64         /**
  65          * Get the maximum allowable file size given backend
  66          * medium restrictions and basic performance constraints.
  67          * Do not call this function from places outside FileBackend and FileOp.
  68          *
  69          * @return integer Bytes
  70          */
  71         final public function maxFileSizeInternal() {
  72                 return $this->maxFileSize;
  73         }
  74
  75         /**
  76          * Check if a file can be created at a given storage path.
  77          * FS backends should check if the parent directory exists and the file is writable.
  78          * Backends using key/value stores should check if the container exists.
  79          *
  80          * @param $storagePath string
  81          * @return bool
  82          */
  83         abstract public function isPathUsableInternal( $storagePath );
  84
  85         /**
  86          * Create a file in the backend with the given contents.
  87          * Do not call this function from places outside FileBackend and FileOp.
  88          *
  89          * $params include:
  90          *     content       : the raw file contents
  91          *     dst           : destination storage path
  92          *     overwrite     : overwrite any file that exists at the destination
  93          *     async         : Status will be returned immediately if supported.
  94          *                     If the status is OK, then its value field will be
  95          *                     set to a FileBackendStoreOpHandle object.
  96          *
  97          * @param $params Array
  98          * @return Status
  99          */
 100         final public function createInternal( array $params ) {
 101                 wfProfileIn( __METHOD__ );
 102                 wfProfileIn( __METHOD__ . '-' . $this->name );
 103                 if ( strlen( $params['content'] ) > $this->maxFileSizeInternal() ) {
 104                         $status = Status::newFatal( 'backend-fail-maxsize',
 105                                 $params['dst'], $this->maxFileSizeInternal() );
 106                 } else {
 107                         $status = $this->doCreateInternal( $params );
 108                         $this->clearCache( array( $params['dst'] ) );
 109                         $this->deleteFileCache( $params['dst'] ); // persistent cache
 110                 }
 111                 wfProfileOut( __METHOD__ . '-' . $this->name );
 112                 wfProfileOut( __METHOD__ );
 113                 return $status;
 114         }
 115
 116         /**
 117          * @see FileBackendStore::createInternal()
 118          */
 119         abstract protected function doCreateInternal( array $params );
 120
 121         /**
 122          * Store a file into the backend from a file on disk.
 123          * Do not call this function from places outside FileBackend and FileOp.
 124          *
 125          * $params include:
 126          *     src           : source path on disk
 127          *     dst           : destination storage path
 128          *     overwrite     : overwrite any file that exists at the destination
 129          *     async         : Status will be returned immediately if supported.
 130          *                     If the status is OK, then its value field will be
 131          *                     set to a FileBackendStoreOpHandle object.
 132          *
 133          * @param $params Array
 134          * @return Status
 135          */
 136         final public function storeInternal( array $params ) {
 137                 wfProfileIn( __METHOD__ );
 138                 wfProfileIn( __METHOD__ . '-' . $this->name );
 139                 if ( filesize( $params['src'] ) > $this->maxFileSizeInternal() ) {
 140                         $status = Status::newFatal( 'backend-fail-maxsize',
 141                                 $params['dst'], $this->maxFileSizeInternal() );
 142                 } else {
 143                         $status = $this->doStoreInternal( $params );
 144                         $this->clearCache( array( $params['dst'] ) );
 145                         $this->deleteFileCache( $params['dst'] ); // persistent cache
 146                 }
 147                 wfProfileOut( __METHOD__ . '-' . $this->name );
 148                 wfProfileOut( __METHOD__ );
 149                 return $status;
 150         }
 151
 152         /**
 153          * @see FileBackendStore::storeInternal()
 154          */
 155         abstract protected function doStoreInternal( array $params );
 156
 157         /**
 158          * Copy a file from one storage path to another in the backend.
 159          * Do not call this function from places outside FileBackend and FileOp.
 160          *
 161          * $params include:
 162          *     src           : source storage path
 163          *     dst           : destination storage path
 164          *     overwrite     : overwrite any file that exists at the destination
 165          *     async         : Status will be returned immediately if supported.
 166          *                     If the status is OK, then its value field will be
 167          *                     set to a FileBackendStoreOpHandle object.
 168          *
 169          * @param $params Array
 170          * @return Status
 171          */
 172         final public function copyInternal( array $params ) {
 173                 wfProfileIn( __METHOD__ );
 174                 wfProfileIn( __METHOD__ . '-' . $this->name );
 175                 $status = $this->doCopyInternal( $params );
 176                 $this->clearCache( array( $params['dst'] ) );
 177                 $this->deleteFileCache( $params['dst'] ); // persistent cache
 178                 wfProfileOut( __METHOD__ . '-' . $this->name );
 179                 wfProfileOut( __METHOD__ );
 180                 return $status;
 181         }
 182
 183         /**
 184          * @see FileBackendStore::copyInternal()
 185          */
 186         abstract protected function doCopyInternal( array $params );
 187
 188         /**
 189          * Delete a file at the storage path.
 190          * Do not call this function from places outside FileBackend and FileOp.
 191          *
 192          * $params include:
 193          *     src                 : source storage path
 194          *     ignoreMissingSource : do nothing if the source file does not exist
 195          *     async               : Status will be returned immediately if supported.
 196          *                           If the status is OK, then its value field will be
 197          *                           set to a FileBackendStoreOpHandle object.
 198          *
 199          * @param $params Array
 200          * @return Status
 201          */
 202         final public function deleteInternal( array $params ) {
 203                 wfProfileIn( __METHOD__ );
 204                 wfProfileIn( __METHOD__ . '-' . $this->name );
 205                 $status = $this->doDeleteInternal( $params );
 206                 $this->clearCache( array( $params['src'] ) );
 207                 $this->deleteFileCache( $params['src'] ); // persistent cache
 208                 wfProfileOut( __METHOD__ . '-' . $this->name );
 209                 wfProfileOut( __METHOD__ );
 210                 return $status;
 211         }
 212
 213         /**
 214          * @see FileBackendStore::deleteInternal()
 215          */
 216         abstract protected function doDeleteInternal( array $params );
 217
 218         /**
 219          * Move a file from one storage path to another in the backend.
 220          * Do not call this function from places outside FileBackend and FileOp.
 221          *
 222          * $params include:
 223          *     src           : source storage path
 224          *     dst           : destination storage path
 225          *     overwrite     : overwrite any file that exists at the destination
 226          *     async         : Status will be returned immediately if supported.
 227          *                     If the status is OK, then its value field will be
 228          *                     set to a FileBackendStoreOpHandle object.
 229          *
 230          * @param $params Array
 231          * @return Status
 232          */
 233         final public function moveInternal( array $params ) {
 234                 wfProfileIn( __METHOD__ );
 235                 wfProfileIn( __METHOD__ . '-' . $this->name );
 236                 $status = $this->doMoveInternal( $params );
 237                 $this->clearCache( array( $params['src'], $params['dst'] ) );
 238                 $this->deleteFileCache( $params['src'] ); // persistent cache
 239                 $this->deleteFileCache( $params['dst'] ); // persistent cache
 240                 wfProfileOut( __METHOD__ . '-' . $this->name );
 241                 wfProfileOut( __METHOD__ );
 242                 return $status;
 243         }
 244
 245         /**
 246          * @see FileBackendStore::moveInternal()
 247          * @return Status
 248          */
 249         protected function doMoveInternal( array $params ) {
 250                 unset( $params['async'] ); // two steps, won't work here :)
 251                 // Copy source to dest
 252                 $status = $this->copyInternal( $params );
 253                 if ( $status->isOK() ) {
 254                         // Delete source (only fails due to races or medium going down)
 255                         $status->merge( $this->deleteInternal( array( 'src' => $params['src'] ) ) );
 256                         $status->setResult( true, $status->value ); // ignore delete() errors
 257                 }
 258                 return $status;
 259         }
 260
 261         /**
 262          * No-op file operation that does nothing.
 263          * Do not call this function from places outside FileBackend and FileOp.
 264          *
 265          * @param $params Array
 266          * @return Status
 267          */
 268         final public function nullInternal( array $params ) {
 269                 return Status::newGood();
 270         }
 271
 272         /**
 273          * @see FileBackend::concatenate()
 274          * @return Status
 275          */
 276         final public function concatenate( array $params ) {
 277                 wfProfileIn( __METHOD__ );
 278                 wfProfileIn( __METHOD__ . '-' . $this->name );
 279                 $status = Status::newGood();
 280
 281                 // Try to lock the source files for the scope of this function
 282                 $scopeLockS = $this->getScopedFileLocks( $params['srcs'], LockManager::LOCK_UW, $status );
 283                 if ( $status->isOK() ) {
 284                         // Actually do the concatenation
 285                         $status->merge( $this->doConcatenate( $params ) );
 286                 }
 287
 288                 wfProfileOut( __METHOD__ . '-' . $this->name );
 289                 wfProfileOut( __METHOD__ );
 290                 return $status;
 291         }
 292
 293         /**
 294          * @see FileBackendStore::concatenate()
 295          * @return Status
 296          */
 297         protected function doConcatenate( array $params ) {
 298                 $status = Status::newGood();
 299                 $tmpPath = $params['dst']; // convenience
 300
 301                 // Check that the specified temp file is valid...
 302                 wfSuppressWarnings();
 303                 $ok = ( is_file( $tmpPath ) && !filesize( $tmpPath ) );
 304                 wfRestoreWarnings();
 305                 if ( !$ok ) { // not present or not empty
 306                         $status->fatal( 'backend-fail-opentemp', $tmpPath );
 307                         return $status;
 308                 }
 309
 310                 // Build up the temp file using the source chunks (in order)...
 311                 $tmpHandle = fopen( $tmpPath, 'ab' );
 312                 if ( $tmpHandle === false ) {
 313                         $status->fatal( 'backend-fail-opentemp', $tmpPath );
 314                         return $status;
 315                 }
 316                 foreach ( $params['srcs'] as $virtualSource ) {
 317                         // Get a local FS version of the chunk
 318                         $tmpFile = $this->getLocalReference( array( 'src' => $virtualSource ) );
 319                         if ( !$tmpFile ) {
 320                                 $status->fatal( 'backend-fail-read', $virtualSource );
 321                                 return $status;
 322                         }
 323                         // Get a handle to the local FS version
 324                         $sourceHandle = fopen( $tmpFile->getPath(), 'r' );
 325                         if ( $sourceHandle === false ) {
 326                                 fclose( $tmpHandle );
 327                                 $status->fatal( 'backend-fail-read', $virtualSource );
 328                                 return $status;
 329                         }
 330                         // Append chunk to file (pass chunk size to avoid magic quotes)
 331                         if ( !stream_copy_to_stream( $sourceHandle, $tmpHandle ) ) {
 332                                 fclose( $sourceHandle );
 333                                 fclose( $tmpHandle );
 334                                 $status->fatal( 'backend-fail-writetemp', $tmpPath );
 335                                 return $status;
 336                         }
 337                         fclose( $sourceHandle );
 338                 }
 339                 if ( !fclose( $tmpHandle ) ) {
 340                         $status->fatal( 'backend-fail-closetemp', $tmpPath );
 341                         return $status;
 342                 }
 343
 344                 clearstatcache(); // temp file changed
 345
 346                 return $status;
 347         }
 348
 349         /**
 350          * @see FileBackend::doPrepare()
 351          * @return Status
 352          */
 353         final protected function doPrepare( array $params ) {
 354                 wfProfileIn( __METHOD__ );
 355                 wfProfileIn( __METHOD__ . '-' . $this->name );
 356
 357                 $status = Status::newGood();
 358                 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
 359                 if ( $dir === null ) {
 360                         $status->fatal( 'backend-fail-invalidpath', $params['dir'] );
 361                         wfProfileOut( __METHOD__ . '-' . $this->name );
 362                         wfProfileOut( __METHOD__ );
 363                         return $status; // invalid storage path
 364                 }
 365
 366                 if ( $shard !== null ) { // confined to a single container/shard
 367                         $status->merge( $this->doPrepareInternal( $fullCont, $dir, $params ) );
 368                 } else { // directory is on several shards
 369                         wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
 370                         list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
 371                         foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
 372                                 $status->merge( $this->doPrepareInternal( "{$fullCont}{$suffix}", $dir, $params ) );
 373                         }
 374                 }
 375
 376                 wfProfileOut( __METHOD__ . '-' . $this->name );
 377                 wfProfileOut( __METHOD__ );
 378                 return $status;
 379         }
 380
 381         /**
 382          * @see FileBackendStore::doPrepare()
 383          * @return Status
 384          */
 385         protected function doPrepareInternal( $container, $dir, array $params ) {
 386                 return Status::newGood();
 387         }
 388
 389         /**
 390          * @see FileBackend::doSecure()
 391          * @return Status
 392          */
 393         final protected function doSecure( array $params ) {
 394                 wfProfileIn( __METHOD__ );
 395                 wfProfileIn( __METHOD__ . '-' . $this->name );
 396                 $status = Status::newGood();
 397
 398                 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
 399                 if ( $dir === null ) {
 400                         $status->fatal( 'backend-fail-invalidpath', $params['dir'] );
 401                         wfProfileOut( __METHOD__ . '-' . $this->name );
 402                         wfProfileOut( __METHOD__ );
 403                         return $status; // invalid storage path
 404                 }
 405
 406                 if ( $shard !== null ) { // confined to a single container/shard
 407                         $status->merge( $this->doSecureInternal( $fullCont, $dir, $params ) );
 408                 } else { // directory is on several shards
 409                         wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
 410                         list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
 411                         foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
 412                                 $status->merge( $this->doSecureInternal( "{$fullCont}{$suffix}", $dir, $params ) );
 413                         }
 414                 }
 415
 416                 wfProfileOut( __METHOD__ . '-' . $this->name );
 417                 wfProfileOut( __METHOD__ );
 418                 return $status;
 419         }
 420
 421         /**
 422          * @see FileBackendStore::doSecure()
 423          * @return Status
 424          */
 425         protected function doSecureInternal( $container, $dir, array $params ) {
 426                 return Status::newGood();
 427         }
 428
 429         /**
 430          * @see FileBackend::doClean()
 431          * @return Status
 432          */
 433         final protected function doClean( array $params ) {
 434                 wfProfileIn( __METHOD__ );
 435                 wfProfileIn( __METHOD__ . '-' . $this->name );
 436                 $status = Status::newGood();
 437
 438                 // Recursive: first delete all empty subdirs recursively
 439                 if ( !empty( $params['recursive'] ) && !$this->directoriesAreVirtual() ) {
 440                         $subDirsRel = $this->getTopDirectoryList( array( 'dir' => $params['dir'] ) );
 441                         if ( $subDirsRel !== null ) { // no errors
 442                                 foreach ( $subDirsRel as $subDirRel ) {
 443                                         $subDir = $params['dir'] . "/{$subDirRel}"; // full path
 444                                         $status->merge( $this->doClean( array( 'dir' => $subDir ) + $params ) );
 445                                 }
 446                         }
 447                 }
 448
 449                 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
 450                 if ( $dir === null ) {
 451                         $status->fatal( 'backend-fail-invalidpath', $params['dir'] );
 452                         wfProfileOut( __METHOD__ . '-' . $this->name );
 453                         wfProfileOut( __METHOD__ );
 454                         return $status; // invalid storage path
 455                 }
 456
 457                 // Attempt to lock this directory...
 458                 $filesLockEx = array( $params['dir'] );
 459                 $scopedLockE = $this->getScopedFileLocks( $filesLockEx, LockManager::LOCK_EX, $status );
 460                 if ( !$status->isOK() ) {
 461                         wfProfileOut( __METHOD__ . '-' . $this->name );
 462                         wfProfileOut( __METHOD__ );
 463                         return $status; // abort
 464                 }
 465
 466                 if ( $shard !== null ) { // confined to a single container/shard
 467                         $status->merge( $this->doCleanInternal( $fullCont, $dir, $params ) );
 468                         $this->deleteContainerCache( $fullCont ); // purge cache
 469                 } else { // directory is on several shards
 470                         wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
 471                         list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
 472                         foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
 473                                 $status->merge( $this->doCleanInternal( "{$fullCont}{$suffix}", $dir, $params ) );
 474                                 $this->deleteContainerCache( "{$fullCont}{$suffix}" ); // purge cache
 475                         }
 476                 }
 477
 478                 wfProfileOut( __METHOD__ . '-' . $this->name );
 479                 wfProfileOut( __METHOD__ );
 480                 return $status;
 481         }
 482
 483         /**
 484          * @see FileBackendStore::doClean()
 485          * @return Status
 486          */
 487         protected function doCleanInternal( $container, $dir, array $params ) {
 488                 return Status::newGood();
 489         }
 490
 491         /**
 492          * @see FileBackend::fileExists()
 493          * @return bool|null
 494          */
 495         final public function fileExists( array $params ) {
 496                 wfProfileIn( __METHOD__ );
 497                 wfProfileIn( __METHOD__ . '-' . $this->name );
 498                 $stat = $this->getFileStat( $params );
 499                 wfProfileOut( __METHOD__ . '-' . $this->name );
 500                 wfProfileOut( __METHOD__ );
 501                 return ( $stat === null ) ? null : (bool)$stat; // null => failure
 502         }
 503
 504         /**
 505          * @see FileBackend::getFileTimestamp()
 506          * @return bool
 507          */
 508         final public function getFileTimestamp( array $params ) {
 509                 wfProfileIn( __METHOD__ );
 510                 wfProfileIn( __METHOD__ . '-' . $this->name );
 511                 $stat = $this->getFileStat( $params );
 512                 wfProfileOut( __METHOD__ . '-' . $this->name );
 513                 wfProfileOut( __METHOD__ );
 514                 return $stat ? $stat['mtime'] : false;
 515         }
 516
 517         /**
 518          * @see FileBackend::getFileSize()
 519          * @return bool
 520          */
 521         final public function getFileSize( array $params ) {
 522                 wfProfileIn( __METHOD__ );
 523                 wfProfileIn( __METHOD__ . '-' . $this->name );
 524                 $stat = $this->getFileStat( $params );
 525                 wfProfileOut( __METHOD__ . '-' . $this->name );
 526                 wfProfileOut( __METHOD__ );
 527                 return $stat ? $stat['size'] : false;
 528         }
 529
 530         /**
 531          * @see FileBackend::getFileStat()
 532          * @return bool
 533          */
 534         final public function getFileStat( array $params ) {
 535                 wfProfileIn( __METHOD__ );
 536                 wfProfileIn( __METHOD__ . '-' . $this->name );
 537                 $path = self::normalizeStoragePath( $params['src'] );
 538                 if ( $path === null ) {
 539                         wfProfileOut( __METHOD__ . '-' . $this->name );
 540                         wfProfileOut( __METHOD__ );
 541                         return false; // invalid storage path
 542                 }
 543                 $latest = !empty( $params['latest'] ); // use latest data?
 544                 if ( !isset( $this->cache[$path]['stat'] ) ) {
 545                         $this->primeFileCache( array( $path ) ); // check persistent cache
 546                 }
 547                 if ( isset( $this->cache[$path]['stat'] ) ) {
 548                         // If we want the latest data, check that this cached
 549                         // value was in fact fetched with the latest available data.
 550                         if ( !$latest || $this->cache[$path]['stat']['latest'] ) {
 551                                 $this->pingCache( $path ); // LRU
 552                                 wfProfileOut( __METHOD__ . '-' . $this->name );
 553                                 wfProfileOut( __METHOD__ );
 554                                 return $this->cache[$path]['stat'];
 555                         }
 556                 }
 557                 wfProfileIn( __METHOD__ . '-miss' );
 558                 wfProfileIn( __METHOD__ . '-miss-' . $this->name );
 559                 $stat = $this->doGetFileStat( $params );
 560                 wfProfileOut( __METHOD__ . '-miss-' . $this->name );
 561                 wfProfileOut( __METHOD__ . '-miss' );
 562                 if ( is_array( $stat ) ) { // don't cache negatives
 563                         $stat['latest'] = $latest;
 564                         $this->trimCache(); // limit memory
 565                         $this->cache[$path]['stat'] = $stat;
 566                         $this->setFileCache( $path, $stat ); // update persistent cache
 567                 } else {
 568                         wfDebug( __METHOD__ . ": File $path does not exist.\n" );
 569                 }
 570                 wfProfileOut( __METHOD__ . '-' . $this->name );
 571                 wfProfileOut( __METHOD__ );
 572                 return $stat;
 573         }
 574
 575         /**
 576          * @see FileBackendStore::getFileStat()
 577          */
 578         abstract protected function doGetFileStat( array $params );
 579
 580         /**
 581          * @see FileBackend::getFileContents()
 582          * @return bool|string
 583          */
 584         public function getFileContents( array $params ) {
 585                 wfProfileIn( __METHOD__ );
 586                 wfProfileIn( __METHOD__ . '-' . $this->name );
 587                 $tmpFile = $this->getLocalReference( $params );
 588                 if ( !$tmpFile ) {
 589                         wfProfileOut( __METHOD__ . '-' . $this->name );
 590                         wfProfileOut( __METHOD__ );
 591                         return false;
 592                 }
 593                 wfSuppressWarnings();
 594                 $data = file_get_contents( $tmpFile->getPath() );
 595                 wfRestoreWarnings();
 596                 wfProfileOut( __METHOD__ . '-' . $this->name );
 597                 wfProfileOut( __METHOD__ );
 598                 return $data;
 599         }
 600
 601         /**
 602          * @see FileBackend::getFileSha1Base36()
 603          * @return bool|string
 604          */
 605         final public function getFileSha1Base36( array $params ) {
 606                 wfProfileIn( __METHOD__ );
 607                 wfProfileIn( __METHOD__ . '-' . $this->name );
 608                 $path = $params['src'];
 609                 if ( isset( $this->cache[$path]['sha1'] ) ) {
 610                         $this->pingCache( $path ); // LRU
 611                         wfProfileOut( __METHOD__ . '-' . $this->name );
 612                         wfProfileOut( __METHOD__ );
 613                         return $this->cache[$path]['sha1'];
 614                 }
 615                 wfProfileIn( __METHOD__ . '-miss' );
 616                 wfProfileIn( __METHOD__ . '-miss-' . $this->name );
 617                 $hash = $this->doGetFileSha1Base36( $params );
 618                 wfProfileOut( __METHOD__ . '-miss-' . $this->name );
 619                 wfProfileOut( __METHOD__ . '-miss' );
 620                 if ( $hash ) { // don't cache negatives
 621                         $this->trimCache(); // limit memory
 622                         $this->cache[$path]['sha1'] = $hash;
 623                 }
 624                 wfProfileOut( __METHOD__ . '-' . $this->name );
 625                 wfProfileOut( __METHOD__ );
 626                 return $hash;
 627         }
 628
 629         /**
 630          * @see FileBackendStore::getFileSha1Base36()
 631          * @return bool
 632          */
 633         protected function doGetFileSha1Base36( array $params ) {
 634                 $fsFile = $this->getLocalReference( $params );
 635                 if ( !$fsFile ) {
 636                         return false;
 637                 } else {
 638                         return $fsFile->getSha1Base36();
 639                 }
 640         }
 641
 642         /**
 643          * @see FileBackend::getFileProps()
 644          * @return Array
 645          */
 646         final public function getFileProps( array $params ) {
 647                 wfProfileIn( __METHOD__ );
 648                 wfProfileIn( __METHOD__ . '-' . $this->name );
 649                 $fsFile = $this->getLocalReference( $params );
 650                 $props = $fsFile ? $fsFile->getProps() : FSFile::placeholderProps();
 651                 wfProfileOut( __METHOD__ . '-' . $this->name );
 652                 wfProfileOut( __METHOD__ );
 653                 return $props;
 654         }
 655
 656         /**
 657          * @see FileBackend::getLocalReference()
 658          * @return TempFSFile|null
 659          */
 660         public function getLocalReference( array $params ) {
 661                 wfProfileIn( __METHOD__ );
 662                 wfProfileIn( __METHOD__ . '-' . $this->name );
 663                 $path = $params['src'];
 664                 if ( isset( $this->expensiveCache[$path]['localRef'] ) ) {
 665                         $this->pingExpensiveCache( $path );
 666                         wfProfileOut( __METHOD__ . '-' . $this->name );
 667                         wfProfileOut( __METHOD__ );
 668                         return $this->expensiveCache[$path]['localRef'];
 669                 }
 670                 $tmpFile = $this->getLocalCopy( $params );
 671                 if ( $tmpFile ) { // don't cache negatives
 672                         $this->trimExpensiveCache(); // limit memory
 673                         $this->expensiveCache[$path]['localRef'] = $tmpFile;
 674                 }
 675                 wfProfileOut( __METHOD__ . '-' . $this->name );
 676                 wfProfileOut( __METHOD__ );
 677                 return $tmpFile;
 678         }
 679
 680         /**
 681          * @see FileBackend::streamFile()
 682          * @return Status
 683          */
 684         final public function streamFile( array $params ) {
 685                 wfProfileIn( __METHOD__ );
 686                 wfProfileIn( __METHOD__ . '-' . $this->name );
 687                 $status = Status::newGood();
 688
 689                 $info = $this->getFileStat( $params );
 690                 if ( !$info ) { // let StreamFile handle the 404
 691                         $status->fatal( 'backend-fail-notexists', $params['src'] );
 692                 }
 693
 694                 // Set output buffer and HTTP headers for stream
 695                 $extraHeaders = isset( $params['headers'] ) ? $params['headers'] : array();
 696                 $res = StreamFile::prepareForStream( $params['src'], $info, $extraHeaders );
 697                 if ( $res == StreamFile::NOT_MODIFIED ) {
 698                         // do nothing; client cache is up to date
 699                 } elseif ( $res == StreamFile::READY_STREAM ) {
 700                         wfProfileIn( __METHOD__ . '-send' );
 701                         wfProfileIn( __METHOD__ . '-send-' . $this->name );
 702                         $status = $this->doStreamFile( $params );
 703                         wfProfileOut( __METHOD__ . '-send-' . $this->name );
 704                         wfProfileOut( __METHOD__ . '-send' );
 705                 } else {
 706                         $status->fatal( 'backend-fail-stream', $params['src'] );
 707                 }
 708
 709                 wfProfileOut( __METHOD__ . '-' . $this->name );
 710                 wfProfileOut( __METHOD__ );
 711                 return $status;
 712         }
 713
 714         /**
 715          * @see FileBackendStore::streamFile()
 716          * @return Status
 717          */
 718         protected function doStreamFile( array $params ) {
 719                 $status = Status::newGood();
 720
 721                 $fsFile = $this->getLocalReference( $params );
 722                 if ( !$fsFile ) {
 723                         $status->fatal( 'backend-fail-stream', $params['src'] );
 724                 } elseif ( !readfile( $fsFile->getPath() ) ) {
 725                         $status->fatal( 'backend-fail-stream', $params['src'] );
 726                 }
 727
 728                 return $status;
 729         }
 730
 731         /**
 732          * @see FileBackend::directoryExists()
 733          * @return bool|null
 734          */
 735         final public function directoryExists( array $params ) {
 736                 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
 737                 if ( $dir === null ) {
 738                         return false; // invalid storage path
 739                 }
 740                 if ( $shard !== null ) { // confined to a single container/shard
 741                         return $this->doDirectoryExists( $fullCont, $dir, $params );
 742                 } else { // directory is on several shards
 743                         wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
 744                         list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
 745                         $res = false; // response
 746                         foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
 747                                 $exists = $this->doDirectoryExists( "{$fullCont}{$suffix}", $dir, $params );
 748                                 if ( $exists ) {
 749                                         $res = true;
 750                                         break; // found one!
 751                                 } elseif ( $exists === null ) { // error?
 752                                         $res = null; // if we don't find anything, it is indeterminate
 753                                 }
 754                         }
 755                         return $res;
 756                 }
 757         }
 758
 759         /**
 760          * @see FileBackendStore::directoryExists()
 761          *
 762          * @param $container string Resolved container name
 763          * @param $dir string Resolved path relative to container
 764          * @param $params Array
 765          * @return bool|null
 766          */
 767         abstract protected function doDirectoryExists( $container, $dir, array $params );
 768
 769         /**
 770          * @see FileBackend::getDirectoryList()
 771          * @return Traversable|Array|null Returns null on failure
 772          */
 773         final public function getDirectoryList( array $params ) {
 774                 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
 775                 if ( $dir === null ) { // invalid storage path
 776                         return null;
 777                 }
 778                 if ( $shard !== null ) {
 779                         // File listing is confined to a single container/shard
 780                         return $this->getDirectoryListInternal( $fullCont, $dir, $params );
 781                 } else {
 782                         wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
 783                         // File listing spans multiple containers/shards
 784                         list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
 785                         return new FileBackendStoreShardDirIterator( $this,
 786                                 $fullCont, $dir, $this->getContainerSuffixes( $shortCont ), $params );
 787                 }
 788         }
 789
 790         /**
 791          * Do not call this function from places outside FileBackend
 792          *
 793          * @see FileBackendStore::getDirectoryList()
 794          *
 795          * @param $container string Resolved container name
 796          * @param $dir string Resolved path relative to container
 797          * @param $params Array
 798          * @return Traversable|Array|null Returns null on failure
 799          */
 800         abstract public function getDirectoryListInternal( $container, $dir, array $params );
 801
 802         /**
 803          * @see FileBackend::getFileList()
 804          * @return Traversable|Array|null Returns null on failure
 805          */
 806         final public function getFileList( array $params ) {
 807                 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
 808                 if ( $dir === null ) { // invalid storage path
 809                         return null;
 810                 }
 811                 if ( $shard !== null ) {
 812                         // File listing is confined to a single container/shard
 813                         return $this->getFileListInternal( $fullCont, $dir, $params );
 814                 } else {
 815                         wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
 816                         // File listing spans multiple containers/shards
 817                         list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
 818                         return new FileBackendStoreShardFileIterator( $this,
 819                                 $fullCont, $dir, $this->getContainerSuffixes( $shortCont ), $params );
 820                 }
 821         }
 822
 823         /**
 824          * Do not call this function from places outside FileBackend
 825          *
 826          * @see FileBackendStore::getFileList()
 827          *
 828          * @param $container string Resolved container name
 829          * @param $dir string Resolved path relative to container
 830          * @param $params Array
 831          * @return Traversable|Array|null Returns null on failure
 832          */
 833         abstract public function getFileListInternal( $container, $dir, array $params );
 834
 835         /**
 836          * Return a list of FileOp objects from a list of operations.
 837          * Do not call this function from places outside FileBackend.
 838          *
 839          * The result must have the same number of items as the input.
 840          * An exception is thrown if an unsupported operation is requested.
 841          *
 842          * @param $ops Array Same format as doOperations()
 843          * @return Array List of FileOp objects
 844          * @throws MWException
 845          */
 846         final public function getOperationsInternal( array $ops ) {
 847                 $supportedOps = array(
 848                         'store'       => 'StoreFileOp',
 849                         'copy'        => 'CopyFileOp',
 850                         'move'        => 'MoveFileOp',
 851                         'delete'      => 'DeleteFileOp',
 852                         'create'      => 'CreateFileOp',
 853                         'null'        => 'NullFileOp'
 854                 );
 855
 856                 $performOps = array(); // array of FileOp objects
 857                 // Build up ordered array of FileOps...
 858                 foreach ( $ops as $operation ) {
 859                         $opName = $operation['op'];
 860                         if ( isset( $supportedOps[$opName] ) ) {
 861                                 $class = $supportedOps[$opName];
 862                                 // Get params for this operation
 863                                 $params = $operation;
 864                                 // Append the FileOp class
 865                                 $performOps[] = new $class( $this, $params );
 866                         } else {
 867                                 throw new MWException( "Operation '$opName' is not supported." );
 868                         }
 869                 }
 870
 871                 return $performOps;
 872         }
 873
 874         /**
 875          * Get a list of storage paths to lock for a list of operations
 876          * Returns an array with 'sh' (shared) and 'ex' (exclusive) keys,
 877          * each corresponding to a list of storage paths to be locked.
 878          *
 879          * @param $performOps Array List of FileOp objects
 880          * @return Array ('sh' => list of paths, 'ex' => list of paths)
 881          */
 882         final public function getPathsToLockForOpsInternal( array $performOps ) {
 883                 // Build up a list of files to lock...
 884                 $paths = array( 'sh' => array(), 'ex' => array() );
 885                 foreach ( $performOps as $fileOp ) {
 886                         $paths['sh'] = array_merge( $paths['sh'], $fileOp->storagePathsRead() );
 887                         $paths['ex'] = array_merge( $paths['ex'], $fileOp->storagePathsChanged() );
 888                 }
 889                 // Optimization: if doing an EX lock anyway, don't also set an SH one
 890                 $paths['sh'] = array_diff( $paths['sh'], $paths['ex'] );
 891                 // Get a shared lock on the parent directory of each path changed
 892                 $paths['sh'] = array_merge( $paths['sh'], array_map( 'dirname', $paths['ex'] ) );
 893
 894                 return $paths;
 895         }
 896
 897         /**
 898          * @see FileBackend::doOperationsInternal()
 899          * @return Status
 900          */
 901         protected function doOperationsInternal( array $ops, array $opts ) {
 902                 wfProfileIn( __METHOD__ );
 903                 wfProfileIn( __METHOD__ . '-' . $this->name );
 904                 $status = Status::newGood();
 905
 906                 // Build up a list of FileOps...
 907                 $performOps = $this->getOperationsInternal( $ops );
 908
 909                 // Acquire any locks as needed...
 910                 if ( empty( $opts['nonLocking'] ) ) {
 911                         // Build up a list of files to lock...
 912                         $paths = $this->getPathsToLockForOpsInternal( $performOps );
 913                         // Try to lock those files for the scope of this function...
 914                         $scopeLockS = $this->getScopedFileLocks( $paths['sh'], LockManager::LOCK_UW, $status );
 915                         $scopeLockE = $this->getScopedFileLocks( $paths['ex'], LockManager::LOCK_EX, $status );
 916                         if ( !$status->isOK() ) {
 917                                 wfProfileOut( __METHOD__ . '-' . $this->name );
 918                                 wfProfileOut( __METHOD__ );
 919                                 return $status; // abort
 920                         }
 921                 }
 922
 923                 // Clear any file cache entries (after locks acquired)
 924                 $this->clearCache();
 925
 926                 // Load from the persistent file and container caches
 927                 $this->primeFileCache( $performOps );
 928                 $this->primeContainerCache( $performOps );
 929
 930                 // Actually attempt the operation batch...
 931                 $subStatus = FileOpBatch::attempt( $performOps, $opts, $this->fileJournal );
 932
 933                 // Merge errors into status fields
 934                 $status->merge( $subStatus );
 935                 $status->success = $subStatus->success; // not done in merge()
 936
 937                 wfProfileOut( __METHOD__ . '-' . $this->name );
 938                 wfProfileOut( __METHOD__ );
 939                 return $status;
 940         }
 941
 942         /**
 943          * @see FileBackend::doQuickOperationsInternal()
 944          * @return Status
 945          * @throws MWException
 946          */
 947         final protected function doQuickOperationsInternal( array $ops ) {
 948                 wfProfileIn( __METHOD__ );
 949                 wfProfileIn( __METHOD__ . '-' . $this->name );
 950                 $status = Status::newGood();
 951
 952                 $supportedOps = array( 'create', 'store', 'copy', 'move', 'delete', 'null' );
 953                 $async = ( $this->parallelize === 'implicit' );
 954                 $maxConcurrency = $this->concurrency; // throttle
 955
 956                 $statuses = array(); // array of (index => Status)
 957                 $fileOpHandles = array(); // list of (index => handle) arrays
 958                 $curFileOpHandles = array(); // current handle batch
 959                 // Perform the sync-only ops and build up op handles for the async ops...
 960                 foreach ( $ops as $index => $params ) {
 961                         if ( !in_array( $params['op'], $supportedOps ) ) {
 962                                 wfProfileOut( __METHOD__ . '-' . $this->name );
 963                                 wfProfileOut( __METHOD__ );
 964                                 throw new MWException( "Operation '{$params['op']}' is not supported." );
 965                         }
 966                         $method = $params['op'] . 'Internal'; // e.g. "storeInternal"
 967                         $subStatus = $this->$method( array( 'async' => $async ) + $params );
 968                         if ( $subStatus->value instanceof FileBackendStoreOpHandle ) { // async
 969                                 if ( count( $curFileOpHandles ) >= $maxConcurrency ) {
 970                                         $fileOpHandles[] = $curFileOpHandles; // push this batch
 971                                         $curFileOpHandles = array();
 972                                 }
 973                                 $curFileOpHandles[$index] = $subStatus->value; // keep index
 974                         } else { // error or completed
 975                                 $statuses[$index] = $subStatus; // keep index
 976                         }
 977                 }
 978                 if ( count( $curFileOpHandles ) ) {
 979                         $fileOpHandles[] = $curFileOpHandles; // last batch
 980                 }
 981                 // Do all the async ops that can be done concurrently...
 982                 foreach ( $fileOpHandles as $fileHandleBatch ) {
 983                         $statuses = $statuses + $this->executeOpHandlesInternal( $fileHandleBatch );
 984                 }
 985                 // Marshall and merge all the responses...
 986                 foreach ( $statuses as $index => $subStatus ) {
 987                         $status->merge( $subStatus );
 988                         if ( $subStatus->isOK() ) {
 989                                 $status->success[$index] = true;
 990                                 ++$status->successCount;
 991                         } else {
 992                                 $status->success[$index] = false;
 993                                 ++$status->failCount;
 994                         }
 995                 }
 996
 997                 wfProfileOut( __METHOD__ . '-' . $this->name );
 998                 wfProfileOut( __METHOD__ );
 999                 return $status;
1000         }
1001
1002         /**
1003          * Execute a list of FileBackendStoreOpHandle handles in parallel.
1004          * The resulting Status object fields will correspond
1005          * to the order in which the handles where given.
1006          *
1007          * @param $handles Array List of FileBackendStoreOpHandle objects
1008          * @return Array Map of Status objects
1009          * @throws MWException
1010          */
1011         final public function executeOpHandlesInternal( array $fileOpHandles ) {
1012                 wfProfileIn( __METHOD__ );
1013                 wfProfileIn( __METHOD__ . '-' . $this->name );
1014                 foreach ( $fileOpHandles as $fileOpHandle ) {
1015                         if ( !( $fileOpHandle instanceof FileBackendStoreOpHandle ) ) {
1016                                 throw new MWException( "Given a non-FileBackendStoreOpHandle object." );
1017                         } elseif ( $fileOpHandle->backend->getName() !== $this->getName() ) {
1018                                 throw new MWException( "Given a FileBackendStoreOpHandle for the wrong backend." );
1019                         }
1020                 }
1021                 $res = $this->doExecuteOpHandlesInternal( $fileOpHandles );
1022                 foreach ( $fileOpHandles as $fileOpHandle ) {
1023                         $fileOpHandle->closeResources();
1024                 }
1025                 wfProfileOut( __METHOD__ . '-' . $this->name );
1026                 wfProfileOut( __METHOD__ );
1027                 return $res;
1028         }
1029
1030         /**
1031          * @see FileBackendStore::executeOpHandlesInternal()
1032          * @return Array List of corresponding Status objects
1033          */
1034         protected function doExecuteOpHandlesInternal( array $fileOpHandles ) {
1035                 foreach ( $fileOpHandles as $fileOpHandle ) { // OK if empty
1036                         throw new MWException( "This backend supports no asynchronous operations." );
1037                 }
1038                 return array();
1039         }
1040
1041         /**
1042          * @see FileBackend::clearCache()
1043          */
1044         final public function clearCache( array $paths = null ) {
1045                 if ( is_array( $paths ) ) {
1046                         $paths = array_map( 'FileBackend::normalizeStoragePath', $paths );
1047                         $paths = array_filter( $paths, 'strlen' ); // remove nulls
1048                 }
1049                 if ( $paths === null ) {
1050                         $this->cache = array();
1051                         $this->expensiveCache = array();
1052                 } else {
1053                         foreach ( $paths as $path ) {
1054                                 unset( $this->cache[$path] );
1055                                 unset( $this->expensiveCache[$path] );
1056                         }
1057                 }
1058                 $this->doClearCache( $paths );
1059         }
1060
1061         /**
1062          * Clears any additional stat caches for storage paths
1063          *
1064          * @see FileBackend::clearCache()
1065          *
1066          * @param $paths Array Storage paths (optional)
1067          * @return void
1068          */
1069         protected function doClearCache( array $paths = null ) {}
1070
1071         /**
1072          * Is this a key/value store where directories are just virtual?
1073          * Virtual directories exists in so much as files exists that are
1074          * prefixed with the directory path followed by a forward slash.
1075          *
1076          * @return bool
1077          */
1078         abstract protected function directoriesAreVirtual();
1079
1080         /**
1081          * Move a cache entry to the top (such as when accessed)
1082          *
1083          * @param $path string Storage path
1084          * @return void
1085          */
1086         protected function pingCache( $path ) {
1087                 if ( isset( $this->cache[$path] ) ) {
1088                         $tmp = $this->cache[$path];
1089                         unset( $this->cache[$path] );
1090                         $this->cache[$path] = $tmp;
1091                 }
1092         }
1093
1094         /**
1095          * Prune the inexpensive cache if it is too big to add an item
1096          *
1097          * @return void
1098          */
1099         protected function trimCache() {
1100                 if ( count( $this->cache ) >= $this->maxCacheSize ) {
1101                         reset( $this->cache );
1102                         unset( $this->cache[key( $this->cache )] );
1103                 }
1104         }
1105
1106         /**
1107          * Move a cache entry to the top (such as when accessed)
1108          *
1109          * @param $path string Storage path
1110          * @return void
1111          */
1112         protected function pingExpensiveCache( $path ) {
1113                 if ( isset( $this->expensiveCache[$path] ) ) {
1114                         $tmp = $this->expensiveCache[$path];
1115                         unset( $this->expensiveCache[$path] );
1116                         $this->expensiveCache[$path] = $tmp;
1117                 }
1118         }
1119
1120         /**
1121          * Prune the expensive cache if it is too big to add an item
1122          *
1123          * @return void
1124          */
1125         protected function trimExpensiveCache() {
1126                 if ( count( $this->expensiveCache ) >= $this->maxExpensiveCacheSize ) {
1127                         reset( $this->expensiveCache );
1128                         unset( $this->expensiveCache[key( $this->expensiveCache )] );
1129                 }
1130         }
1131
1132         /**
1133          * Check if a container name is valid.
1134          * This checks for for length and illegal characters.
1135          *
1136          * @param $container string
1137          * @return bool
1138          */
1139         final protected static function isValidContainerName( $container ) {
1140                 // This accounts for Swift and S3 restrictions while leaving room
1141                 // for things like '.xxx' (hex shard chars) or '.seg' (segments).
1142                 // This disallows directory separators or traversal characters.
1143                 // Note that matching strings URL encode to the same string;
1144                 // in Swift, the length restriction is *after* URL encoding.
1145                 return preg_match( '/^[a-z0-9][a-z0-9-_]{0,199}$/i', $container );
1146         }
1147
1148         /**
1149          * Splits a storage path into an internal container name,
1150          * an internal relative file name, and a container shard suffix.
1151          * Any shard suffix is already appended to the internal container name.
1152          * This also checks that the storage path is valid and within this backend.
1153          *
1154          * If the container is sharded but a suffix could not be determined,
1155          * this means that the path can only refer to a directory and can only
1156          * be scanned by looking in all the container shards.
1157          *
1158          * @param $storagePath string
1159          * @return Array (container, path, container suffix) or (null, null, null) if invalid
1160          */
1161         final protected function resolveStoragePath( $storagePath ) {
1162                 list( $backend, $container, $relPath ) = self::splitStoragePath( $storagePath );
1163                 if ( $backend === $this->name ) { // must be for this backend
1164                         $relPath = self::normalizeContainerPath( $relPath );
1165                         if ( $relPath !== null ) {
1166                                 // Get shard for the normalized path if this container is sharded
1167                                 $cShard = $this->getContainerShard( $container, $relPath );
1168                                 // Validate and sanitize the relative path (backend-specific)
1169                                 $relPath = $this->resolveContainerPath( $container, $relPath );
1170                                 if ( $relPath !== null ) {
1171                                         // Prepend any wiki ID prefix to the container name
1172                                         $container = $this->fullContainerName( $container );
1173                                         if ( self::isValidContainerName( $container ) ) {
1174                                                 // Validate and sanitize the container name (backend-specific)
1175                                                 $container = $this->resolveContainerName( "{$container}{$cShard}" );
1176                                                 if ( $container !== null ) {
1177                                                         return array( $container, $relPath, $cShard );
1178                                                 }
1179                                         }
1180                                 }
1181                         }
1182                 }
1183                 return array( null, null, null );
1184         }
1185
1186         /**
1187          * Like resolveStoragePath() except null values are returned if
1188          * the container is sharded and the shard could not be determined.
1189          *
1190          * @see FileBackendStore::resolveStoragePath()
1191          *
1192          * @param $storagePath string
1193          * @return Array (container, path) or (null, null) if invalid
1194          */
1195         final protected function resolveStoragePathReal( $storagePath ) {
1196                 list( $container, $relPath, $cShard ) = $this->resolveStoragePath( $storagePath );
1197                 if ( $cShard !== null ) {
1198                         return array( $container, $relPath );
1199                 }
1200                 return array( null, null );
1201         }
1202
1203         /**
1204          * Get the container name shard suffix for a given path.
1205          * Any empty suffix means the container is not sharded.
1206          *
1207          * @param $container string Container name
1208          * @param $relPath string Storage path relative to the container
1209          * @return string|null Returns null if shard could not be determined
1210          */
1211         final protected function getContainerShard( $container, $relPath ) {
1212                 list( $levels, $base, $repeat ) = $this->getContainerHashLevels( $container );
1213                 if ( $levels == 1 || $levels == 2 ) {
1214                         // Hash characters are either base 16 or 36
1215                         $char = ( $base == 36 ) ? '[0-9a-z]' : '[0-9a-f]';
1216                         // Get a regex that represents the shard portion of paths.
1217                         // The concatenation of the captures gives us the shard.
1218                         if ( $levels === 1 ) { // 16 or 36 shards per container
1219                                 $hashDirRegex = '(' . $char . ')';
1220                         } else { // 256 or 1296 shards per container
1221                                 if ( $repeat ) { // verbose hash dir format (e.g. "a/ab/abc")
1222                                         $hashDirRegex = $char . '/(' . $char . '{2})';
1223                                 } else { // short hash dir format (e.g. "a/b/c")
1224                                         $hashDirRegex = '(' . $char . ')/(' . $char . ')';
1225                                 }
1226                         }
1227                         // Allow certain directories to be above the hash dirs so as
1228                         // to work with FileRepo (e.g. "archive/a/ab" or "temp/a/ab").
1229                         // They must be 2+ chars to avoid any hash directory ambiguity.
1230                         $m = array();
1231                         if ( preg_match( "!^(?:[^/]{2,}/)*$hashDirRegex(?:/|$)!", $relPath, $m ) ) {
1232                                 return '.' . implode( '', array_slice( $m, 1 ) );
1233                         }
1234                         return null; // failed to match
1235                 }
1236                 return ''; // no sharding
1237         }
1238
1239         /**
1240          * Check if a storage path maps to a single shard.
1241          * Container dirs like "a", where the container shards on "x/xy",
1242          * can reside on several shards. Such paths are tricky to handle.
1243          *
1244          * @param $storagePath string Storage path
1245          * @return bool
1246          */
1247         final public function isSingleShardPathInternal( $storagePath ) {
1248                 list( $c, $r, $shard ) = $this->resolveStoragePath( $storagePath );
1249                 return ( $shard !== null );
1250         }
1251
1252         /**
1253          * Get the sharding config for a container.
1254          * If greater than 0, then all file storage paths within
1255          * the container are required to be hashed accordingly.
1256          *
1257          * @param $container string
1258          * @return Array (integer levels, integer base, repeat flag) or (0, 0, false)
1259          */
1260         final protected function getContainerHashLevels( $container ) {
1261                 if ( isset( $this->shardViaHashLevels[$container] ) ) {
1262                         $config = $this->shardViaHashLevels[$container];
1263                         $hashLevels = (int)$config['levels'];
1264                         if ( $hashLevels == 1 || $hashLevels == 2 ) {
1265                                 $hashBase = (int)$config['base'];
1266                                 if ( $hashBase == 16 || $hashBase == 36 ) {
1267                                         return array( $hashLevels, $hashBase, $config['repeat'] );
1268                                 }
1269                         }
1270                 }
1271                 return array( 0, 0, false ); // no sharding
1272         }
1273
1274         /**
1275          * Get a list of full container shard suffixes for a container
1276          *
1277          * @param $container string
1278          * @return Array
1279          */
1280         final protected function getContainerSuffixes( $container ) {
1281                 $shards = array();
1282                 list( $digits, $base ) = $this->getContainerHashLevels( $container );
1283                 if ( $digits > 0 ) {
1284                         $numShards = pow( $base, $digits );
1285                         for ( $index = 0; $index < $numShards; $index++ ) {
1286                                 $shards[] = '.' . wfBaseConvert( $index, 10, $base, $digits );
1287                         }
1288                 }
1289                 return $shards;
1290         }
1291
1292         /**
1293          * Get the full container name, including the wiki ID prefix
1294          *
1295          * @param $container string
1296          * @return string
1297          */
1298         final protected function fullContainerName( $container ) {
1299                 if ( $this->wikiId != '' ) {
1300                         return "{$this->wikiId}-$container";
1301                 } else {
1302                         return $container;
1303                 }
1304         }
1305
1306         /**
1307          * Resolve a container name, checking if it's allowed by the backend.
1308          * This is intended for internal use, such as encoding illegal chars.
1309          * Subclasses can override this to be more restrictive.
1310          *
1311          * @param $container string
1312          * @return string|null
1313          */
1314         protected function resolveContainerName( $container ) {
1315                 return $container;
1316         }
1317
1318         /**
1319          * Resolve a relative storage path, checking if it's allowed by the backend.
1320          * This is intended for internal use, such as encoding illegal chars or perhaps
1321          * getting absolute paths (e.g. FS based backends). Note that the relative path
1322          * may be the empty string (e.g. the path is simply to the container).
1323          *
1324          * @param $container string Container name
1325          * @param $relStoragePath string Storage path relative to the container
1326          * @return string|null Path or null if not valid
1327          */
1328         protected function resolveContainerPath( $container, $relStoragePath ) {
1329                 return $relStoragePath;
1330         }
1331
1332         /**
1333          * Get the cache key for a container
1334          *
1335          * @param $container string Resolved container name
1336          * @return string
1337          */
1338         private function containerCacheKey( $container ) {
1339                 return wfMemcKey( 'backend', $this->getName(), 'container', $container );
1340         }
1341
1342         /**
1343          * Set the cached info for a container
1344          *
1345          * @param $container string Resolved container name
1346          * @param $val mixed Information to cache
1347          */
1348         final protected function setContainerCache( $container, $val ) {
1349                 $this->memCache->set( $this->containerCacheKey( $container ), $val, 14*86400 );
1350         }
1351
1352         /**
1353          * Delete the cached info for a container
1354          *
1355          * @param $container string Resolved container name
1356          */
1357         final protected function deleteContainerCache( $container ) {
1358                 if ( !$this->memCache->delete( $this->containerCacheKey( $container ) ) ) {
1359                         trigger_error( "Unable to delete stat cache for container $container." );
1360                 }
1361         }
1362
1363         /**
1364          * Do a batch lookup from cache for container stats for all containers
1365          * used in a list of container names, storage paths, or FileOp objects.
1366          *
1367          * @param $items Array
1368          * @return void
1369          */
1370         final protected function primeContainerCache( array $items ) {
1371                 wfProfileIn( __METHOD__ );
1372                 wfProfileIn( __METHOD__ . '-' . $this->name );
1373
1374                 $paths = array(); // list of storage paths
1375                 $contNames = array(); // (cache key => resolved container name)
1376                 // Get all the paths/containers from the items...
1377                 foreach ( $items as $item ) {
1378                         if ( $item instanceof FileOp ) {
1379                                 $paths = array_merge( $paths, $item->storagePathsRead() );
1380                                 $paths = array_merge( $paths, $item->storagePathsChanged() );
1381                         } elseif ( self::isStoragePath( $item ) ) {
1382                                 $paths[] = $item;
1383                         } elseif ( is_string( $item ) ) { // full container name
1384                                 $contNames[$this->containerCacheKey( $item )] = $item;
1385                         }
1386                 }
1387                 // Get all the corresponding cache keys for paths...
1388                 foreach ( $paths as $path ) {
1389                         list( $fullCont, $r, $s ) = $this->resolveStoragePath( $path );
1390                         if ( $fullCont !== null ) { // valid path for this backend
1391                                 $contNames[$this->containerCacheKey( $fullCont )] = $fullCont;
1392                         }
1393                 }
1394
1395                 $contInfo = array(); // (resolved container name => cache value)
1396                 // Get all cache entries for these container cache keys...
1397                 $values = $this->memCache->getMulti( array_keys( $contNames ) );
1398                 foreach ( $values as $cacheKey => $val ) {
1399                         $contInfo[$contNames[$cacheKey]] = $val;
1400                 }
1401
1402                 // Populate the container process cache for the backend...
1403                 $this->doPrimeContainerCache( array_filter( $contInfo, 'is_array' ) );
1404
1405                 wfProfileOut( __METHOD__ . '-' . $this->name );
1406                 wfProfileOut( __METHOD__ );
1407         }
1408
1409         /**
1410          * Fill the backend-specific process cache given an array of
1411          * resolved container names and their corresponding cached info.
1412          * Only containers that actually exist should appear in the map.
1413          *
1414          * @param $containerInfo Array Map of resolved container names to cached info
1415          * @return void
1416          */
1417         protected function doPrimeContainerCache( array $containerInfo ) {}
1418
1419         /**
1420          * Get the cache key for a file path
1421          *
1422          * @param $path string Storage path
1423          * @return string
1424          */
1425         private function fileCacheKey( $path ) {
1426                 return wfMemcKey( 'backend', $this->getName(), 'file', sha1( $path ) );
1427         }
1428
1429         /**
1430          * Set the cached stat info for a file path
1431          *
1432          * @param $path string Storage path
1433          * @param $val mixed Information to cache
1434          */
1435         final protected function setFileCache( $path, $val ) {
1436                 $this->memCache->set( $this->fileCacheKey( $path ), $val, 7*86400 );
1437         }
1438
1439         /**
1440          * Delete the cached stat info for a file path
1441          *
1442          * @param $path string Storage path
1443          */
1444         final protected function deleteFileCache( $path ) {
1445                 if ( !$this->memCache->delete( $this->fileCacheKey( $path ) ) ) {
1446                         trigger_error( "Unable to delete stat cache for file $path." );
1447                 }
1448         }
1449
1450         /**
1451          * Do a batch lookup from cache for file stats for all paths
1452          * used in a list of storage paths or FileOp objects.
1453          *
1454          * @param $items Array List of storage paths or FileOps
1455          * @return void
1456          */
1457         final protected function primeFileCache( array $items ) {
1458                 wfProfileIn( __METHOD__ );
1459                 wfProfileIn( __METHOD__ . '-' . $this->name );
1460
1461                 $paths = array(); // list of storage paths
1462                 $pathNames = array(); // (cache key => storage path)
1463                 // Get all the paths/containers from the items...
1464                 foreach ( $items as $item ) {
1465                         if ( $item instanceof FileOp ) {
1466                                 $paths = array_merge( $paths, $item->storagePathsRead() );
1467                                 $paths = array_merge( $paths, $item->storagePathsChanged() );
1468                         } elseif ( self::isStoragePath( $item ) ) {
1469                                 $paths[] = $item;
1470                         }
1471                 }
1472                 // Get all the corresponding cache keys for paths...
1473                 foreach ( $paths as $path ) {
1474                         list( $cont, $rel, $s ) = $this->resolveStoragePath( $path );
1475                         if ( $rel !== null ) { // valid path for this backend
1476                                 $pathNames[$this->fileCacheKey( $path )] = $path;
1477                         }
1478                 }
1479                 // Get all cache entries for these container cache keys...
1480                 $values = $this->memCache->getMulti( array_keys( $pathNames ) );
1481                 foreach ( $values as $cacheKey => $val ) {
1482                         if ( is_array( $val ) ) {
1483                                 $this->trimCache(); // limit memory
1484                                 $this->cache[$pathNames[$cacheKey]]['stat'] = $val;
1485                         }
1486                 }
1487
1488                 wfProfileOut( __METHOD__ . '-' . $this->name );
1489                 wfProfileOut( __METHOD__ );
1490         }
1491 }
1492
1493 /**
1494  * FileBackendStore helper class for performing asynchronous file operations.
1495  *
1496  * For example, calling FileBackendStore::createInternal() with the "async"
1497  * param flag may result in a Status that contains this object as a value.
1498  * This class is largely backend-specific and is mostly just "magic" to be
1499  * passed to FileBackendStore::executeOpHandlesInternal().
1500  */
1501 abstract class FileBackendStoreOpHandle {
1502         /** @var Array */
1503         public $params = array(); // params to caller functions
1504         /** @var FileBackendStore */
1505         public $backend;
1506         /** @var Array */
1507         public $resourcesToClose = array();
1508
1509         public $call; // string; name that identifies the function called
1510
1511         /**
1512          * Close all open file handles
1513          *
1514          * @return void
1515          */
1516         public function closeResources() {
1517                 array_map( 'fclose', $this->resourcesToClose );
1518         }
1519 }
1520
1521 /**
1522  * FileBackendStore helper function to handle listings that span container shards.
1523  * Do not use this class from places outside of FileBackendStore.
1524  *
1525  * @ingroup FileBackend
1526  */
1527 abstract class FileBackendStoreShardListIterator implements Iterator {
1528         /** @var FileBackendStore */
1529         protected $backend;
1530         /** @var Array */
1531         protected $params;
1532         /** @var Array */
1533         protected $shardSuffixes;
1534         protected $container; // string; full container name
1535         protected $directory; // string; resolved relative path
1536
1537         /** @var Traversable */
1538         protected $iter;
1539         protected $curShard = 0; // integer
1540         protected $pos = 0; // integer
1541
1542         /** @var Array */
1543         protected $multiShardPaths = array(); // (rel path => 1)
1544
1545         /**
1546          * @param $backend FileBackendStore
1547          * @param $container string Full storage container name
1548          * @param $dir string Storage directory relative to container
1549          * @param $suffixes Array List of container shard suffixes
1550          * @param $params Array
1551          */
1552         public function __construct(
1553                 FileBackendStore $backend, $container, $dir, array $suffixes, array $params
1554         ) {
1555                 $this->backend = $backend;
1556                 $this->container = $container;
1557                 $this->directory = $dir;
1558                 $this->shardSuffixes = $suffixes;
1559                 $this->params = $params;
1560         }
1561
1562         /**
1563          * @see Iterator::current()
1564          * @return string|bool String or false
1565          */
1566         public function current() {
1567                 if ( is_array( $this->iter ) ) {
1568                         return current( $this->iter );
1569                 } else {
1570                         return $this->iter->current();
1571                 }
1572         }
1573
1574         /**
1575          * @see Iterator::key()
1576          * @return integer
1577          */
1578         public function key() {
1579                 return $this->pos;
1580         }
1581
1582         /**
1583          * @see Iterator::next()
1584          * @return void
1585          */
1586         public function next() {
1587                 ++$this->pos;
1588                 if ( is_array( $this->iter ) ) {
1589                         next( $this->iter );
1590                 } else {
1591                         $this->iter->next();
1592                 }
1593                 // Filter out items that we already listed
1594                 $this->filterViaNext();
1595                 // Find the next non-empty shard if no elements are left
1596                 $this->nextShardIteratorIfNotValid();
1597         }
1598
1599         /**
1600          * @see Iterator::rewind()
1601          * @return void
1602          */
1603         public function rewind() {
1604                 $this->pos = 0;
1605                 $this->curShard = 0;
1606                 $this->setIteratorFromCurrentShard();
1607                 // Filter out items that we already listed
1608                 $this->filterViaNext();
1609                 // Find the next non-empty shard if this one has no elements
1610                 $this->nextShardIteratorIfNotValid();
1611         }
1612
1613         /**
1614          * @see Iterator::valid()
1615          * @return bool
1616          */
1617         public function valid() {
1618                 if ( $this->iter === null ) {
1619                         return false; // some failure?
1620                 } elseif ( is_array( $this->iter ) ) {
1621                         return ( current( $this->iter ) !== false ); // no paths can have this value
1622                 } else {
1623                         return $this->iter->valid();
1624                 }
1625         }
1626
1627         /**
1628          * Filter out duplicate items by advancing to the next ones
1629          */
1630         protected function filterViaNext() {
1631                 while ( $this->iter->valid() ) {
1632                         $rel = $this->iter->current(); // path relative to given directory
1633                         $path = $this->params['dir'] . "/{$rel}"; // full storage path
1634                         if ( !$this->backend->isSingleShardPathInternal( $path ) ) {
1635                                 // Don't keep listing paths that are on multiple shards
1636                                 if ( isset( $this->multiShardPaths[$rel] ) ) {
1637                                         $this->iter->next(); // we already listed this path
1638                                 } else {
1639                                         $this->multiShardPaths[$rel] = 1;
1640                                         break;
1641                                 }
1642                         }
1643                 }
1644         }
1645
1646         /**
1647          * If the list iterator for this container shard is out of items,
1648          * then move on to the next container that has items.
1649          * If there are none, then it advances to the last container.
1650          */
1651         protected function nextShardIteratorIfNotValid() {
1652                 while ( !$this->valid() ) {
1653                         if ( ++$this->curShard >= count( $this->shardSuffixes ) ) {
1654                                 break; // no more container shards
1655                         }
1656                         $this->setIteratorFromCurrentShard();
1657                 }
1658         }
1659
1660         /**
1661          * Set the list iterator to that of the current container shard
1662          */
1663         protected function setIteratorFromCurrentShard() {
1664                 $suffix = $this->shardSuffixes[$this->curShard];
1665                 $this->iter = $this->listFromShard(
1666                         "{$this->container}{$suffix}", $this->directory, $this->params );
1667         }
1668
1669         /**
1670          * Get the list for a given container shard
1671          *
1672          * @param $container string Resolved container name
1673          * @param $dir string Resolved path relative to container
1674          * @param $params Array
1675          * @return Traversable|Array|null
1676          */
1677         abstract protected function listFromShard( $container, $dir, array $params );
1678 }
1679
1680 /**
1681  * Iterator for listing directories
1682  */
1683 class FileBackendStoreShardDirIterator extends FileBackendStoreShardListIterator {
1684         /**
1685          * @param string $container
1686          * @param string $dir
1687          * @param array $params
1688          * @return Array|null|Traversable
1689          */
1690         protected function listFromShard( $container, $dir, array $params ) {
1691                 return $this->backend->getDirectoryListInternal( $container, $dir, $params );
1692         }
1693 }
1694
1695 /**
1696  * Iterator for listing regular files
1697  */
1698 class FileBackendStoreShardFileIterator extends FileBackendStoreShardListIterator {
1699         /**
1700          * @param string $container
1701          * @param string $dir
1702          * @param array $params
1703          * @return Array|null|Traversable
1704          */
1705         protected function listFromShard( $container, $dir, array $params ) {
1706                 return $this->backend->getFileListInternal( $container, $dir, $params );
1707         }
1708 }