includes/filebackend/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         /** @var ProcessCacheLRU */
  42         protected $cheapCache; // Map of paths to small (RAM/disk) cache items
  43         /** @var ProcessCacheLRU */
  44         protected $expensiveCache; // Map of paths to large (RAM/disk) cache items
  45
  46         /** @var Array Map of container names to sharding settings */
  47         protected $shardViaHashLevels = array(); // (container name => config array)
  48
  49         protected $maxFileSize = 4294967296; // integer bytes (4GiB)
  50
  51         /**
  52          * @see FileBackend::__construct()
  53          *
  54          * @param $config Array
  55          */
  56         public function __construct( array $config ) {
  57                 parent::__construct( $config );
  58                 $this->memCache       = new EmptyBagOStuff(); // disabled by default
  59                 $this->cheapCache     = new ProcessCacheLRU( 300 );
  60                 $this->expensiveCache = new ProcessCacheLRU( 5 );
  61         }
  62
  63         /**
  64          * Get the maximum allowable file size given backend
  65          * medium restrictions and basic performance constraints.
  66          * Do not call this function from places outside FileBackend and FileOp.
  67          *
  68          * @return integer Bytes
  69          */
  70         final public function maxFileSizeInternal() {
  71                 return $this->maxFileSize;
  72         }
  73
  74         /**
  75          * Check if a file can be created at a given storage path.
  76          * FS backends should check if the parent directory exists and the file is writable.
  77          * Backends using key/value stores should check if the container exists.
  78          *
  79          * @param $storagePath string
  80          * @return bool
  81          */
  82         abstract public function isPathUsableInternal( $storagePath );
  83
  84         /**
  85          * Create a file in the backend with the given contents.
  86          * Do not call this function from places outside FileBackend and FileOp.
  87          *
  88          * $params include:
  89          *   - content       : the raw file contents
  90          *   - dst           : destination storage path
  91          *   - overwrite     : overwrite any file that exists at the destination
  92          *   - async         : Status will be returned immediately if supported.
  93          *                     If the status is OK, then its value field will be
  94          *                     set to a FileBackendStoreOpHandle object.
  95          *
  96          * @param $params Array
  97          * @return Status
  98          */
  99         final public function createInternal( array $params ) {
 100                 wfProfileIn( __METHOD__ );
 101                 wfProfileIn( __METHOD__ . '-' . $this->name );
 102                 if ( strlen( $params['content'] ) > $this->maxFileSizeInternal() ) {
 103                         $status = Status::newFatal( 'backend-fail-maxsize',
 104                                 $params['dst'], $this->maxFileSizeInternal() );
 105                 } else {
 106                         $status = $this->doCreateInternal( $params );
 107                         $this->clearCache( array( $params['dst'] ) );
 108                         $this->deleteFileCache( $params['dst'] ); // persistent cache
 109                 }
 110                 wfProfileOut( __METHOD__ . '-' . $this->name );
 111                 wfProfileOut( __METHOD__ );
 112                 return $status;
 113         }
 114
 115         /**
 116          * @see FileBackendStore::createInternal()
 117          */
 118         abstract protected function doCreateInternal( array $params );
 119
 120         /**
 121          * Store a file into the backend from a file on disk.
 122          * Do not call this function from places outside FileBackend and FileOp.
 123          *
 124          * $params include:
 125          *   - src           : source path on disk
 126          *   - dst           : destination storage path
 127          *   - overwrite     : overwrite any file that exists at the destination
 128          *   - async         : Status will be returned immediately if supported.
 129          *                     If the status is OK, then its value field will be
 130          *                     set to a FileBackendStoreOpHandle object.
 131          *
 132          * @param $params Array
 133          * @return Status
 134          */
 135         final public function storeInternal( array $params ) {
 136                 wfProfileIn( __METHOD__ );
 137                 wfProfileIn( __METHOD__ . '-' . $this->name );
 138                 if ( filesize( $params['src'] ) > $this->maxFileSizeInternal() ) {
 139                         $status = Status::newFatal( 'backend-fail-maxsize',
 140                                 $params['dst'], $this->maxFileSizeInternal() );
 141                 } else {
 142                         $status = $this->doStoreInternal( $params );
 143                         $this->clearCache( array( $params['dst'] ) );
 144                         $this->deleteFileCache( $params['dst'] ); // persistent cache
 145                 }
 146                 wfProfileOut( __METHOD__ . '-' . $this->name );
 147                 wfProfileOut( __METHOD__ );
 148                 return $status;
 149         }
 150
 151         /**
 152          * @see FileBackendStore::storeInternal()
 153          */
 154         abstract protected function doStoreInternal( array $params );
 155
 156         /**
 157          * Copy a file from one storage path to another in the backend.
 158          * Do not call this function from places outside FileBackend and FileOp.
 159          *
 160          * $params include:
 161          *   - src           : source storage path
 162          *   - dst           : destination storage path
 163          *   - overwrite     : overwrite any file that exists at the destination
 164          *   - async         : Status will be returned immediately if supported.
 165          *                     If the status is OK, then its value field will be
 166          *                     set to a FileBackendStoreOpHandle object.
 167          *
 168          * @param $params Array
 169          * @return Status
 170          */
 171         final public function copyInternal( array $params ) {
 172                 wfProfileIn( __METHOD__ );
 173                 wfProfileIn( __METHOD__ . '-' . $this->name );
 174                 $status = $this->doCopyInternal( $params );
 175                 $this->clearCache( array( $params['dst'] ) );
 176                 $this->deleteFileCache( $params['dst'] ); // persistent cache
 177                 wfProfileOut( __METHOD__ . '-' . $this->name );
 178                 wfProfileOut( __METHOD__ );
 179                 return $status;
 180         }
 181
 182         /**
 183          * @see FileBackendStore::copyInternal()
 184          */
 185         abstract protected function doCopyInternal( array $params );
 186
 187         /**
 188          * Delete a file at the storage path.
 189          * Do not call this function from places outside FileBackend and FileOp.
 190          *
 191          * $params include:
 192          *   - src                 : source storage path
 193          *   - ignoreMissingSource : do nothing if the source file does not exist
 194          *   - async               : Status will be returned immediately if supported.
 195          *                           If the status is OK, then its value field will be
 196          *                           set to a FileBackendStoreOpHandle object.
 197          *
 198          * @param $params Array
 199          * @return Status
 200          */
 201         final public function deleteInternal( array $params ) {
 202                 wfProfileIn( __METHOD__ );
 203                 wfProfileIn( __METHOD__ . '-' . $this->name );
 204                 $status = $this->doDeleteInternal( $params );
 205                 $this->clearCache( array( $params['src'] ) );
 206                 $this->deleteFileCache( $params['src'] ); // persistent cache
 207                 wfProfileOut( __METHOD__ . '-' . $this->name );
 208                 wfProfileOut( __METHOD__ );
 209                 return $status;
 210         }
 211
 212         /**
 213          * @see FileBackendStore::deleteInternal()
 214          */
 215         abstract protected function doDeleteInternal( array $params );
 216
 217         /**
 218          * Move a file from one storage path to another in the backend.
 219          * Do not call this function from places outside FileBackend and FileOp.
 220          *
 221          * $params include:
 222          *   - src           : source storage path
 223          *   - dst           : destination storage path
 224          *   - overwrite     : overwrite any file that exists at the destination
 225          *   - async         : Status will be returned immediately if supported.
 226          *                     If the status is OK, then its value field will be
 227          *                     set to a FileBackendStoreOpHandle object.
 228          *
 229          * @param $params Array
 230          * @return Status
 231          */
 232         final public function moveInternal( array $params ) {
 233                 wfProfileIn( __METHOD__ );
 234                 wfProfileIn( __METHOD__ . '-' . $this->name );
 235                 $status = $this->doMoveInternal( $params );
 236                 $this->clearCache( array( $params['src'], $params['dst'] ) );
 237                 $this->deleteFileCache( $params['src'] ); // persistent cache
 238                 $this->deleteFileCache( $params['dst'] ); // persistent cache
 239                 wfProfileOut( __METHOD__ . '-' . $this->name );
 240                 wfProfileOut( __METHOD__ );
 241                 return $status;
 242         }
 243
 244         /**
 245          * @see FileBackendStore::moveInternal()
 246          * @return Status
 247          */
 248         protected function doMoveInternal( array $params ) {
 249                 unset( $params['async'] ); // two steps, won't work here :)
 250                 // Copy source to dest
 251                 $status = $this->copyInternal( $params );
 252                 if ( $status->isOK() ) {
 253                         // Delete source (only fails due to races or medium going down)
 254                         $status->merge( $this->deleteInternal( array( 'src' => $params['src'] ) ) );
 255                         $status->setResult( true, $status->value ); // ignore delete() errors
 256                 }
 257                 return $status;
 258         }
 259
 260         /**
 261          * No-op file operation that does nothing.
 262          * Do not call this function from places outside FileBackend and FileOp.
 263          *
 264          * @param $params Array
 265          * @return Status
 266          */
 267         final public function nullInternal( array $params ) {
 268                 return Status::newGood();
 269         }
 270
 271         /**
 272          * @see FileBackend::concatenate()
 273          * @return Status
 274          */
 275         final public function concatenate( array $params ) {
 276                 wfProfileIn( __METHOD__ );
 277                 wfProfileIn( __METHOD__ . '-' . $this->name );
 278                 $status = Status::newGood();
 279
 280                 // Try to lock the source files for the scope of this function
 281                 $scopeLockS = $this->getScopedFileLocks( $params['srcs'], LockManager::LOCK_UW, $status );
 282                 if ( $status->isOK() ) {
 283                         // Actually do the concatenation
 284                         $status->merge( $this->doConcatenate( $params ) );
 285                 }
 286
 287                 wfProfileOut( __METHOD__ . '-' . $this->name );
 288                 wfProfileOut( __METHOD__ );
 289                 return $status;
 290         }
 291
 292         /**
 293          * @see FileBackendStore::concatenate()
 294          * @return Status
 295          */
 296         protected function doConcatenate( array $params ) {
 297                 $status = Status::newGood();
 298                 $tmpPath = $params['dst']; // convenience
 299
 300                 // Check that the specified temp file is valid...
 301                 wfSuppressWarnings();
 302                 $ok = ( is_file( $tmpPath ) && !filesize( $tmpPath ) );
 303                 wfRestoreWarnings();
 304                 if ( !$ok ) { // not present or not empty
 305                         $status->fatal( 'backend-fail-opentemp', $tmpPath );
 306                         return $status;
 307                 }
 308
 309                 // Build up the temp file using the source chunks (in order)...
 310                 $tmpHandle = fopen( $tmpPath, 'ab' );
 311                 if ( $tmpHandle === false ) {
 312                         $status->fatal( 'backend-fail-opentemp', $tmpPath );
 313                         return $status;
 314                 }
 315                 foreach ( $params['srcs'] as $virtualSource ) {
 316                         // Get a local FS version of the chunk
 317                         $tmpFile = $this->getLocalReference( array( 'src' => $virtualSource ) );
 318                         if ( !$tmpFile ) {
 319                                 $status->fatal( 'backend-fail-read', $virtualSource );
 320                                 return $status;
 321                         }
 322                         // Get a handle to the local FS version
 323                         $sourceHandle = fopen( $tmpFile->getPath(), 'r' );
 324                         if ( $sourceHandle === false ) {
 325                                 fclose( $tmpHandle );
 326                                 $status->fatal( 'backend-fail-read', $virtualSource );
 327                                 return $status;
 328                         }
 329                         // Append chunk to file (pass chunk size to avoid magic quotes)
 330                         if ( !stream_copy_to_stream( $sourceHandle, $tmpHandle ) ) {
 331                                 fclose( $sourceHandle );
 332                                 fclose( $tmpHandle );
 333                                 $status->fatal( 'backend-fail-writetemp', $tmpPath );
 334                                 return $status;
 335                         }
 336                         fclose( $sourceHandle );
 337                 }
 338                 if ( !fclose( $tmpHandle ) ) {
 339                         $status->fatal( 'backend-fail-closetemp', $tmpPath );
 340                         return $status;
 341                 }
 342
 343                 clearstatcache(); // temp file changed
 344
 345                 return $status;
 346         }
 347
 348         /**
 349          * @see FileBackend::doPrepare()
 350          * @return Status
 351          */
 352         final protected function doPrepare( array $params ) {
 353                 wfProfileIn( __METHOD__ );
 354                 wfProfileIn( __METHOD__ . '-' . $this->name );
 355
 356                 $status = Status::newGood();
 357                 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
 358                 if ( $dir === null ) {
 359                         $status->fatal( 'backend-fail-invalidpath', $params['dir'] );
 360                         wfProfileOut( __METHOD__ . '-' . $this->name );
 361                         wfProfileOut( __METHOD__ );
 362                         return $status; // invalid storage path
 363                 }
 364
 365                 if ( $shard !== null ) { // confined to a single container/shard
 366                         $status->merge( $this->doPrepareInternal( $fullCont, $dir, $params ) );
 367                 } else { // directory is on several shards
 368                         wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
 369                         list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
 370                         foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
 371                                 $status->merge( $this->doPrepareInternal( "{$fullCont}{$suffix}", $dir, $params ) );
 372                         }
 373                 }
 374
 375                 wfProfileOut( __METHOD__ . '-' . $this->name );
 376                 wfProfileOut( __METHOD__ );
 377                 return $status;
 378         }
 379
 380         /**
 381          * @see FileBackendStore::doPrepare()
 382          * @return Status
 383          */
 384         protected function doPrepareInternal( $container, $dir, array $params ) {
 385                 return Status::newGood();
 386         }
 387
 388         /**
 389          * @see FileBackend::doSecure()
 390          * @return Status
 391          */
 392         final protected function doSecure( array $params ) {
 393                 wfProfileIn( __METHOD__ );
 394                 wfProfileIn( __METHOD__ . '-' . $this->name );
 395                 $status = Status::newGood();
 396
 397                 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
 398                 if ( $dir === null ) {
 399                         $status->fatal( 'backend-fail-invalidpath', $params['dir'] );
 400                         wfProfileOut( __METHOD__ . '-' . $this->name );
 401                         wfProfileOut( __METHOD__ );
 402                         return $status; // invalid storage path
 403                 }
 404
 405                 if ( $shard !== null ) { // confined to a single container/shard
 406                         $status->merge( $this->doSecureInternal( $fullCont, $dir, $params ) );
 407                 } else { // directory is on several shards
 408                         wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
 409                         list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
 410                         foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
 411                                 $status->merge( $this->doSecureInternal( "{$fullCont}{$suffix}", $dir, $params ) );
 412                         }
 413                 }
 414
 415                 wfProfileOut( __METHOD__ . '-' . $this->name );
 416                 wfProfileOut( __METHOD__ );
 417                 return $status;
 418         }
 419
 420         /**
 421          * @see FileBackendStore::doSecure()
 422          * @return Status
 423          */
 424         protected function doSecureInternal( $container, $dir, array $params ) {
 425                 return Status::newGood();
 426         }
 427
 428         /**
 429          * @see FileBackend::doPublish()
 430          * @return Status
 431          */
 432         final protected function doPublish( array $params ) {
 433                 wfProfileIn( __METHOD__ );
 434                 wfProfileIn( __METHOD__ . '-' . $this->name );
 435                 $status = Status::newGood();
 436
 437                 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
 438                 if ( $dir === null ) {
 439                         $status->fatal( 'backend-fail-invalidpath', $params['dir'] );
 440                         wfProfileOut( __METHOD__ . '-' . $this->name );
 441                         wfProfileOut( __METHOD__ );
 442                         return $status; // invalid storage path
 443                 }
 444
 445                 if ( $shard !== null ) { // confined to a single container/shard
 446                         $status->merge( $this->doPublishInternal( $fullCont, $dir, $params ) );
 447                 } else { // directory is on several shards
 448                         wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
 449                         list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
 450                         foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
 451                                 $status->merge( $this->doPublishInternal( "{$fullCont}{$suffix}", $dir, $params ) );
 452                         }
 453                 }
 454
 455                 wfProfileOut( __METHOD__ . '-' . $this->name );
 456                 wfProfileOut( __METHOD__ );
 457                 return $status;
 458         }
 459
 460         /**
 461          * @see FileBackendStore::doPublish()
 462          * @return Status
 463          */
 464         protected function doPublishInternal( $container, $dir, array $params ) {
 465                 return Status::newGood();
 466         }
 467
 468         /**
 469          * @see FileBackend::doClean()
 470          * @return Status
 471          */
 472         final protected function doClean( array $params ) {
 473                 wfProfileIn( __METHOD__ );
 474                 wfProfileIn( __METHOD__ . '-' . $this->name );
 475                 $status = Status::newGood();
 476
 477                 // Recursive: first delete all empty subdirs recursively
 478                 if ( !empty( $params['recursive'] ) && !$this->directoriesAreVirtual() ) {
 479                         $subDirsRel = $this->getTopDirectoryList( array( 'dir' => $params['dir'] ) );
 480                         if ( $subDirsRel !== null ) { // no errors
 481                                 foreach ( $subDirsRel as $subDirRel ) {
 482                                         $subDir = $params['dir'] . "/{$subDirRel}"; // full path
 483                                         $status->merge( $this->doClean( array( 'dir' => $subDir ) + $params ) );
 484                                 }
 485                         }
 486                 }
 487
 488                 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
 489                 if ( $dir === null ) {
 490                         $status->fatal( 'backend-fail-invalidpath', $params['dir'] );
 491                         wfProfileOut( __METHOD__ . '-' . $this->name );
 492                         wfProfileOut( __METHOD__ );
 493                         return $status; // invalid storage path
 494                 }
 495
 496                 // Attempt to lock this directory...
 497                 $filesLockEx = array( $params['dir'] );
 498                 $scopedLockE = $this->getScopedFileLocks( $filesLockEx, LockManager::LOCK_EX, $status );
 499                 if ( !$status->isOK() ) {
 500                         wfProfileOut( __METHOD__ . '-' . $this->name );
 501                         wfProfileOut( __METHOD__ );
 502                         return $status; // abort
 503                 }
 504
 505                 if ( $shard !== null ) { // confined to a single container/shard
 506                         $status->merge( $this->doCleanInternal( $fullCont, $dir, $params ) );
 507                         $this->deleteContainerCache( $fullCont ); // purge cache
 508                 } else { // directory is on several shards
 509                         wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
 510                         list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
 511                         foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
 512                                 $status->merge( $this->doCleanInternal( "{$fullCont}{$suffix}", $dir, $params ) );
 513                                 $this->deleteContainerCache( "{$fullCont}{$suffix}" ); // purge cache
 514                         }
 515                 }
 516
 517                 wfProfileOut( __METHOD__ . '-' . $this->name );
 518                 wfProfileOut( __METHOD__ );
 519                 return $status;
 520         }
 521
 522         /**
 523          * @see FileBackendStore::doClean()
 524          * @return Status
 525          */
 526         protected function doCleanInternal( $container, $dir, array $params ) {
 527                 return Status::newGood();
 528         }
 529
 530         /**
 531          * @see FileBackend::fileExists()
 532          * @return bool|null
 533          */
 534         final public function fileExists( array $params ) {
 535                 wfProfileIn( __METHOD__ );
 536                 wfProfileIn( __METHOD__ . '-' . $this->name );
 537                 $stat = $this->getFileStat( $params );
 538                 wfProfileOut( __METHOD__ . '-' . $this->name );
 539                 wfProfileOut( __METHOD__ );
 540                 return ( $stat === null ) ? null : (bool)$stat; // null => failure
 541         }
 542
 543         /**
 544          * @see FileBackend::getFileTimestamp()
 545          * @return bool
 546          */
 547         final public function getFileTimestamp( array $params ) {
 548                 wfProfileIn( __METHOD__ );
 549                 wfProfileIn( __METHOD__ . '-' . $this->name );
 550                 $stat = $this->getFileStat( $params );
 551                 wfProfileOut( __METHOD__ . '-' . $this->name );
 552                 wfProfileOut( __METHOD__ );
 553                 return $stat ? $stat['mtime'] : false;
 554         }
 555
 556         /**
 557          * @see FileBackend::getFileSize()
 558          * @return bool
 559          */
 560         final public function getFileSize( array $params ) {
 561                 wfProfileIn( __METHOD__ );
 562                 wfProfileIn( __METHOD__ . '-' . $this->name );
 563                 $stat = $this->getFileStat( $params );
 564                 wfProfileOut( __METHOD__ . '-' . $this->name );
 565                 wfProfileOut( __METHOD__ );
 566                 return $stat ? $stat['size'] : false;
 567         }
 568
 569         /**
 570          * @see FileBackend::getFileStat()
 571          * @return bool
 572          */
 573         final public function getFileStat( array $params ) {
 574                 $path = self::normalizeStoragePath( $params['src'] );
 575                 if ( $path === null ) {
 576                         return false; // invalid storage path
 577                 }
 578                 wfProfileIn( __METHOD__ );
 579                 wfProfileIn( __METHOD__ . '-' . $this->name );
 580                 $latest = !empty( $params['latest'] ); // use latest data?
 581                 if ( !$this->cheapCache->has( $path, 'stat' ) ) {
 582                         $this->primeFileCache( array( $path ) ); // check persistent cache
 583                 }
 584                 if ( $this->cheapCache->has( $path, 'stat' ) ) {
 585                         $stat = $this->cheapCache->get( $path, 'stat' );
 586                         // If we want the latest data, check that this cached
 587                         // value was in fact fetched with the latest available data.
 588                         if ( !$latest || $stat['latest'] ) {
 589                                 wfProfileOut( __METHOD__ . '-' . $this->name );
 590                                 wfProfileOut( __METHOD__ );
 591                                 return $stat;
 592                         }
 593                 }
 594                 wfProfileIn( __METHOD__ . '-miss' );
 595                 wfProfileIn( __METHOD__ . '-miss-' . $this->name );
 596                 $stat = $this->doGetFileStat( $params );
 597                 wfProfileOut( __METHOD__ . '-miss-' . $this->name );
 598                 wfProfileOut( __METHOD__ . '-miss' );
 599                 if ( is_array( $stat ) ) { // don't cache negatives
 600                         $stat['latest'] = $latest;
 601                         $this->cheapCache->set( $path, 'stat', $stat );
 602                         $this->setFileCache( $path, $stat ); // update persistent cache
 603                         if ( isset( $stat['sha1'] ) ) { // some backends store SHA-1 as metadata
 604                                 $this->cheapCache->set( $path, 'sha1',
 605                                         array( 'hash' => $stat['sha1'], 'latest' => $latest ) );
 606                         }
 607                 } else {
 608                         wfDebug( __METHOD__ . ": File $path does not exist.\n" );
 609                 }
 610                 wfProfileOut( __METHOD__ . '-' . $this->name );
 611                 wfProfileOut( __METHOD__ );
 612                 return $stat;
 613         }
 614
 615         /**
 616          * @see FileBackendStore::getFileStat()
 617          */
 618         abstract protected function doGetFileStat( array $params );
 619
 620         /**
 621          * @see FileBackend::getFileContents()
 622          * @return bool|string
 623          */
 624         public function getFileContents( array $params ) {
 625                 wfProfileIn( __METHOD__ );
 626                 wfProfileIn( __METHOD__ . '-' . $this->name );
 627                 $tmpFile = $this->getLocalReference( $params );
 628                 if ( !$tmpFile ) {
 629                         wfProfileOut( __METHOD__ . '-' . $this->name );
 630                         wfProfileOut( __METHOD__ );
 631                         return false;
 632                 }
 633                 wfSuppressWarnings();
 634                 $data = file_get_contents( $tmpFile->getPath() );
 635                 wfRestoreWarnings();
 636                 wfProfileOut( __METHOD__ . '-' . $this->name );
 637                 wfProfileOut( __METHOD__ );
 638                 return $data;
 639         }
 640
 641         /**
 642          * @see FileBackend::getFileSha1Base36()
 643          * @return bool|string
 644          */
 645         final public function getFileSha1Base36( array $params ) {
 646                 $path = self::normalizeStoragePath( $params['src'] );
 647                 if ( $path === null ) {
 648                         return false; // invalid storage path
 649                 }
 650                 wfProfileIn( __METHOD__ );
 651                 wfProfileIn( __METHOD__ . '-' . $this->name );
 652                 $latest = !empty( $params['latest'] ); // use latest data?
 653                 if ( $this->cheapCache->has( $path, 'sha1' ) ) {
 654                         $stat = $this->cheapCache->get( $path, 'sha1' );
 655                         // If we want the latest data, check that this cached
 656                         // value was in fact fetched with the latest available data.
 657                         if ( !$latest || $stat['latest'] ) {
 658                                 wfProfileOut( __METHOD__ . '-' . $this->name );
 659                                 wfProfileOut( __METHOD__ );
 660                                 return $stat['hash'];
 661                         }
 662                 }
 663                 wfProfileIn( __METHOD__ . '-miss' );
 664                 wfProfileIn( __METHOD__ . '-miss-' . $this->name );
 665                 $hash = $this->doGetFileSha1Base36( $params );
 666                 wfProfileOut( __METHOD__ . '-miss-' . $this->name );
 667                 wfProfileOut( __METHOD__ . '-miss' );
 668                 if ( $hash ) { // don't cache negatives
 669                         $this->cheapCache->set( $path, 'sha1',
 670                                 array( 'hash' => $hash, 'latest' => $latest ) );
 671                 }
 672                 wfProfileOut( __METHOD__ . '-' . $this->name );
 673                 wfProfileOut( __METHOD__ );
 674                 return $hash;
 675         }
 676
 677         /**
 678          * @see FileBackendStore::getFileSha1Base36()
 679          * @return bool|string
 680          */
 681         protected function doGetFileSha1Base36( array $params ) {
 682                 $fsFile = $this->getLocalReference( $params );
 683                 if ( !$fsFile ) {
 684                         return false;
 685                 } else {
 686                         return $fsFile->getSha1Base36();
 687                 }
 688         }
 689
 690         /**
 691          * @see FileBackend::getFileProps()
 692          * @return Array
 693          */
 694         final public function getFileProps( array $params ) {
 695                 wfProfileIn( __METHOD__ );
 696                 wfProfileIn( __METHOD__ . '-' . $this->name );
 697                 $fsFile = $this->getLocalReference( $params );
 698                 $props = $fsFile ? $fsFile->getProps() : FSFile::placeholderProps();
 699                 wfProfileOut( __METHOD__ . '-' . $this->name );
 700                 wfProfileOut( __METHOD__ );
 701                 return $props;
 702         }
 703
 704         /**
 705          * @see FileBackend::getLocalReference()
 706          * @return TempFSFile|null
 707          */
 708         public function getLocalReference( array $params ) {
 709                 $path = self::normalizeStoragePath( $params['src'] );
 710                 if ( $path === null ) {
 711                         return null; // invalid storage path
 712                 }
 713                 wfProfileIn( __METHOD__ );
 714                 wfProfileIn( __METHOD__ . '-' . $this->name );
 715                 $latest = !empty( $params['latest'] ); // use latest data?
 716                 if ( $this->expensiveCache->has( $path, 'localRef' ) ) {
 717                         $val = $this->expensiveCache->get( $path, 'localRef' );
 718                         // If we want the latest data, check that this cached
 719                         // value was in fact fetched with the latest available data.
 720                         if ( !$latest || $val['latest'] ) {
 721                                 wfProfileOut( __METHOD__ . '-' . $this->name );
 722                                 wfProfileOut( __METHOD__ );
 723                                 return $val['object'];
 724                         }
 725                 }
 726                 $tmpFile = $this->getLocalCopy( $params );
 727                 if ( $tmpFile ) { // don't cache negatives
 728                         $this->expensiveCache->set( $path, 'localRef',
 729                                 array( 'object' => $tmpFile, 'latest' => $latest ) );
 730                 }
 731                 wfProfileOut( __METHOD__ . '-' . $this->name );
 732                 wfProfileOut( __METHOD__ );
 733                 return $tmpFile;
 734         }
 735
 736         /**
 737          * @see FileBackend::streamFile()
 738          * @return Status
 739          */
 740         final public function streamFile( array $params ) {
 741                 wfProfileIn( __METHOD__ );
 742                 wfProfileIn( __METHOD__ . '-' . $this->name );
 743                 $status = Status::newGood();
 744
 745                 $info = $this->getFileStat( $params );
 746                 if ( !$info ) { // let StreamFile handle the 404
 747                         $status->fatal( 'backend-fail-notexists', $params['src'] );
 748                 }
 749
 750                 // Set output buffer and HTTP headers for stream
 751                 $extraHeaders = isset( $params['headers'] ) ? $params['headers'] : array();
 752                 $res = StreamFile::prepareForStream( $params['src'], $info, $extraHeaders );
 753                 if ( $res == StreamFile::NOT_MODIFIED ) {
 754                         // do nothing; client cache is up to date
 755                 } elseif ( $res == StreamFile::READY_STREAM ) {
 756                         wfProfileIn( __METHOD__ . '-send' );
 757                         wfProfileIn( __METHOD__ . '-send-' . $this->name );
 758                         $status = $this->doStreamFile( $params );
 759                         wfProfileOut( __METHOD__ . '-send-' . $this->name );
 760                         wfProfileOut( __METHOD__ . '-send' );
 761                 } else {
 762                         $status->fatal( 'backend-fail-stream', $params['src'] );
 763                 }
 764
 765                 wfProfileOut( __METHOD__ . '-' . $this->name );
 766                 wfProfileOut( __METHOD__ );
 767                 return $status;
 768         }
 769
 770         /**
 771          * @see FileBackendStore::streamFile()
 772          * @return Status
 773          */
 774         protected function doStreamFile( array $params ) {
 775                 $status = Status::newGood();
 776
 777                 $fsFile = $this->getLocalReference( $params );
 778                 if ( !$fsFile ) {
 779                         $status->fatal( 'backend-fail-stream', $params['src'] );
 780                 } elseif ( !readfile( $fsFile->getPath() ) ) {
 781                         $status->fatal( 'backend-fail-stream', $params['src'] );
 782                 }
 783
 784                 return $status;
 785         }
 786
 787         /**
 788          * @see FileBackend::directoryExists()
 789          * @return bool|null
 790          */
 791         final public function directoryExists( array $params ) {
 792                 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
 793                 if ( $dir === null ) {
 794                         return false; // invalid storage path
 795                 }
 796                 if ( $shard !== null ) { // confined to a single container/shard
 797                         return $this->doDirectoryExists( $fullCont, $dir, $params );
 798                 } else { // directory is on several shards
 799                         wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
 800                         list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
 801                         $res = false; // response
 802                         foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
 803                                 $exists = $this->doDirectoryExists( "{$fullCont}{$suffix}", $dir, $params );
 804                                 if ( $exists ) {
 805                                         $res = true;
 806                                         break; // found one!
 807                                 } elseif ( $exists === null ) { // error?
 808                                         $res = null; // if we don't find anything, it is indeterminate
 809                                 }
 810                         }
 811                         return $res;
 812                 }
 813         }
 814
 815         /**
 816          * @see FileBackendStore::directoryExists()
 817          *
 818          * @param $container string Resolved container name
 819          * @param $dir string Resolved path relative to container
 820          * @param $params Array
 821          * @return bool|null
 822          */
 823         abstract protected function doDirectoryExists( $container, $dir, array $params );
 824
 825         /**
 826          * @see FileBackend::getDirectoryList()
 827          * @return Traversable|Array|null Returns null on failure
 828          */
 829         final public function getDirectoryList( array $params ) {
 830                 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
 831                 if ( $dir === null ) { // invalid storage path
 832                         return null;
 833                 }
 834                 if ( $shard !== null ) {
 835                         // File listing is confined to a single container/shard
 836                         return $this->getDirectoryListInternal( $fullCont, $dir, $params );
 837                 } else {
 838                         wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
 839                         // File listing spans multiple containers/shards
 840                         list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
 841                         return new FileBackendStoreShardDirIterator( $this,
 842                                 $fullCont, $dir, $this->getContainerSuffixes( $shortCont ), $params );
 843                 }
 844         }
 845
 846         /**
 847          * Do not call this function from places outside FileBackend
 848          *
 849          * @see FileBackendStore::getDirectoryList()
 850          *
 851          * @param $container string Resolved container name
 852          * @param $dir string Resolved path relative to container
 853          * @param $params Array
 854          * @return Traversable|Array|null Returns null on failure
 855          */
 856         abstract public function getDirectoryListInternal( $container, $dir, array $params );
 857
 858         /**
 859          * @see FileBackend::getFileList()
 860          * @return Traversable|Array|null Returns null on failure
 861          */
 862         final public function getFileList( array $params ) {
 863                 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
 864                 if ( $dir === null ) { // invalid storage path
 865                         return null;
 866                 }
 867                 if ( $shard !== null ) {
 868                         // File listing is confined to a single container/shard
 869                         return $this->getFileListInternal( $fullCont, $dir, $params );
 870                 } else {
 871                         wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
 872                         // File listing spans multiple containers/shards
 873                         list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
 874                         return new FileBackendStoreShardFileIterator( $this,
 875                                 $fullCont, $dir, $this->getContainerSuffixes( $shortCont ), $params );
 876                 }
 877         }
 878
 879         /**
 880          * Do not call this function from places outside FileBackend
 881          *
 882          * @see FileBackendStore::getFileList()
 883          *
 884          * @param $container string Resolved container name
 885          * @param $dir string Resolved path relative to container
 886          * @param $params Array
 887          * @return Traversable|Array|null Returns null on failure
 888          */
 889         abstract public function getFileListInternal( $container, $dir, array $params );
 890
 891         /**
 892          * Return a list of FileOp objects from a list of operations.
 893          * Do not call this function from places outside FileBackend.
 894          *
 895          * The result must have the same number of items as the input.
 896          * An exception is thrown if an unsupported operation is requested.
 897          *
 898          * @param $ops Array Same format as doOperations()
 899          * @return Array List of FileOp objects
 900          * @throws MWException
 901          */
 902         final public function getOperationsInternal( array $ops ) {
 903                 $supportedOps = array(
 904                         'store'       => 'StoreFileOp',
 905                         'copy'        => 'CopyFileOp',
 906                         'move'        => 'MoveFileOp',
 907                         'delete'      => 'DeleteFileOp',
 908                         'create'      => 'CreateFileOp',
 909                         'null'        => 'NullFileOp'
 910                 );
 911
 912                 $performOps = array(); // array of FileOp objects
 913                 // Build up ordered array of FileOps...
 914                 foreach ( $ops as $operation ) {
 915                         $opName = $operation['op'];
 916                         if ( isset( $supportedOps[$opName] ) ) {
 917                                 $class = $supportedOps[$opName];
 918                                 // Get params for this operation
 919                                 $params = $operation;
 920                                 // Append the FileOp class
 921                                 $performOps[] = new $class( $this, $params );
 922                         } else {
 923                                 throw new MWException( "Operation '$opName' is not supported." );
 924                         }
 925                 }
 926
 927                 return $performOps;
 928         }
 929
 930         /**
 931          * Get a list of storage paths to lock for a list of operations
 932          * Returns an array with 'sh' (shared) and 'ex' (exclusive) keys,
 933          * each corresponding to a list of storage paths to be locked.
 934          *
 935          * @param $performOps Array List of FileOp objects
 936          * @return Array ('sh' => list of paths, 'ex' => list of paths)
 937          */
 938         final public function getPathsToLockForOpsInternal( array $performOps ) {
 939                 // Build up a list of files to lock...
 940                 $paths = array( 'sh' => array(), 'ex' => array() );
 941                 foreach ( $performOps as $fileOp ) {
 942                         $paths['sh'] = array_merge( $paths['sh'], $fileOp->storagePathsRead() );
 943                         $paths['ex'] = array_merge( $paths['ex'], $fileOp->storagePathsChanged() );
 944                 }
 945                 // Optimization: if doing an EX lock anyway, don't also set an SH one
 946                 $paths['sh'] = array_diff( $paths['sh'], $paths['ex'] );
 947                 // Get a shared lock on the parent directory of each path changed
 948                 $paths['sh'] = array_merge( $paths['sh'], array_map( 'dirname', $paths['ex'] ) );
 949
 950                 return $paths;
 951         }
 952
 953         /**
 954          * @see FileBackend::getScopedLocksForOps()
 955          * @return Array
 956          */
 957         public function getScopedLocksForOps( array $ops, Status $status ) {
 958                 $paths = $this->getPathsToLockForOpsInternal( $this->getOperationsInternal( $ops ) );
 959                 return array(
 960                         $this->getScopedFileLocks( $paths['sh'], LockManager::LOCK_UW, $status ),
 961                         $this->getScopedFileLocks( $paths['ex'], LockManager::LOCK_EX, $status )
 962                 );
 963         }
 964
 965         /**
 966          * @see FileBackend::doOperationsInternal()
 967          * @return Status
 968          */
 969         final protected function doOperationsInternal( array $ops, array $opts ) {
 970                 wfProfileIn( __METHOD__ );
 971                 wfProfileIn( __METHOD__ . '-' . $this->name );
 972                 $status = Status::newGood();
 973
 974                 // Build up a list of FileOps...
 975                 $performOps = $this->getOperationsInternal( $ops );
 976
 977                 // Acquire any locks as needed...
 978                 if ( empty( $opts['nonLocking'] ) ) {
 979                         // Build up a list of files to lock...
 980                         $paths = $this->getPathsToLockForOpsInternal( $performOps );
 981                         // Try to lock those files for the scope of this function...
 982                         $scopeLockS = $this->getScopedFileLocks( $paths['sh'], LockManager::LOCK_UW, $status );
 983                         $scopeLockE = $this->getScopedFileLocks( $paths['ex'], LockManager::LOCK_EX, $status );
 984                         if ( !$status->isOK() ) {
 985                                 wfProfileOut( __METHOD__ . '-' . $this->name );
 986                                 wfProfileOut( __METHOD__ );
 987                                 return $status; // abort
 988                         }
 989                 }
 990
 991                 // Clear any file cache entries (after locks acquired)
 992                 $this->clearCache();
 993
 994                 // Load from the persistent file and container caches
 995                 $this->primeFileCache( $performOps );
 996                 $this->primeContainerCache( $performOps );
 997
 998                 // Actually attempt the operation batch...
 999                 $subStatus = FileOpBatch::attempt( $performOps, $opts, $this->fileJournal );
1000
1001                 // Merge errors into status fields
1002                 $status->merge( $subStatus );
1003                 $status->success = $subStatus->success; // not done in merge()
1004
1005                 wfProfileOut( __METHOD__ . '-' . $this->name );
1006                 wfProfileOut( __METHOD__ );
1007                 return $status;
1008         }
1009
1010         /**
1011          * @see FileBackend::doQuickOperationsInternal()
1012          * @return Status
1013          * @throws MWException
1014          */
1015         final protected function doQuickOperationsInternal( array $ops ) {
1016                 wfProfileIn( __METHOD__ );
1017                 wfProfileIn( __METHOD__ . '-' . $this->name );
1018                 $status = Status::newGood();
1019
1020                 $supportedOps = array( 'create', 'store', 'copy', 'move', 'delete', 'null' );
1021                 $async = ( $this->parallelize === 'implicit' );
1022                 $maxConcurrency = $this->concurrency; // throttle
1023
1024                 $statuses = array(); // array of (index => Status)
1025                 $fileOpHandles = array(); // list of (index => handle) arrays
1026                 $curFileOpHandles = array(); // current handle batch
1027                 // Perform the sync-only ops and build up op handles for the async ops...
1028                 foreach ( $ops as $index => $params ) {
1029                         if ( !in_array( $params['op'], $supportedOps ) ) {
1030                                 wfProfileOut( __METHOD__ . '-' . $this->name );
1031                                 wfProfileOut( __METHOD__ );
1032                                 throw new MWException( "Operation '{$params['op']}' is not supported." );
1033                         }
1034                         $method = $params['op'] . 'Internal'; // e.g. "storeInternal"
1035                         $subStatus = $this->$method( array( 'async' => $async ) + $params );
1036                         if ( $subStatus->value instanceof FileBackendStoreOpHandle ) { // async
1037                                 if ( count( $curFileOpHandles ) >= $maxConcurrency ) {
1038                                         $fileOpHandles[] = $curFileOpHandles; // push this batch
1039                                         $curFileOpHandles = array();
1040                                 }
1041                                 $curFileOpHandles[$index] = $subStatus->value; // keep index
1042                         } else { // error or completed
1043                                 $statuses[$index] = $subStatus; // keep index
1044                         }
1045                 }
1046                 if ( count( $curFileOpHandles ) ) {
1047                         $fileOpHandles[] = $curFileOpHandles; // last batch
1048                 }
1049                 // Do all the async ops that can be done concurrently...
1050                 foreach ( $fileOpHandles as $fileHandleBatch ) {
1051                         $statuses = $statuses + $this->executeOpHandlesInternal( $fileHandleBatch );
1052                 }
1053                 // Marshall and merge all the responses...
1054                 foreach ( $statuses as $index => $subStatus ) {
1055                         $status->merge( $subStatus );
1056                         if ( $subStatus->isOK() ) {
1057                                 $status->success[$index] = true;
1058                                 ++$status->successCount;
1059                         } else {
1060                                 $status->success[$index] = false;
1061                                 ++$status->failCount;
1062                         }
1063                 }
1064
1065                 wfProfileOut( __METHOD__ . '-' . $this->name );
1066                 wfProfileOut( __METHOD__ );
1067                 return $status;
1068         }
1069
1070         /**
1071          * Execute a list of FileBackendStoreOpHandle handles in parallel.
1072          * The resulting Status object fields will correspond
1073          * to the order in which the handles where given.
1074          *
1075          * @param $handles Array List of FileBackendStoreOpHandle objects
1076          * @return Array Map of Status objects
1077          * @throws MWException
1078          */
1079         final public function executeOpHandlesInternal( array $fileOpHandles ) {
1080                 wfProfileIn( __METHOD__ );
1081                 wfProfileIn( __METHOD__ . '-' . $this->name );
1082                 foreach ( $fileOpHandles as $fileOpHandle ) {
1083                         if ( !( $fileOpHandle instanceof FileBackendStoreOpHandle ) ) {
1084                                 throw new MWException( "Given a non-FileBackendStoreOpHandle object." );
1085                         } elseif ( $fileOpHandle->backend->getName() !== $this->getName() ) {
1086                                 throw new MWException( "Given a FileBackendStoreOpHandle for the wrong backend." );
1087                         }
1088                 }
1089                 $res = $this->doExecuteOpHandlesInternal( $fileOpHandles );
1090                 foreach ( $fileOpHandles as $fileOpHandle ) {
1091                         $fileOpHandle->closeResources();
1092                 }
1093                 wfProfileOut( __METHOD__ . '-' . $this->name );
1094                 wfProfileOut( __METHOD__ );
1095                 return $res;
1096         }
1097
1098         /**
1099          * @see FileBackendStore::executeOpHandlesInternal()
1100          * @return Array List of corresponding Status objects
1101          */
1102         protected function doExecuteOpHandlesInternal( array $fileOpHandles ) {
1103                 foreach ( $fileOpHandles as $fileOpHandle ) { // OK if empty
1104                         throw new MWException( "This backend supports no asynchronous operations." );
1105                 }
1106                 return array();
1107         }
1108
1109         /**
1110          * @see FileBackend::preloadCache()
1111          */
1112         final public function preloadCache( array $paths ) {
1113                 $fullConts = array(); // full container names
1114                 foreach ( $paths as $path ) {
1115                         list( $fullCont, $r, $s ) = $this->resolveStoragePath( $path );
1116                         $fullConts[] = $fullCont;
1117                 }
1118                 // Load from the persistent file and container caches
1119                 $this->primeContainerCache( $fullConts );
1120                 $this->primeFileCache( $paths );
1121         }
1122
1123         /**
1124          * @see FileBackend::clearCache()
1125          */
1126         final public function clearCache( array $paths = null ) {
1127                 if ( is_array( $paths ) ) {
1128                         $paths = array_map( 'FileBackend::normalizeStoragePath', $paths );
1129                         $paths = array_filter( $paths, 'strlen' ); // remove nulls
1130                 }
1131                 if ( $paths === null ) {
1132                         $this->cheapCache->clear();
1133                         $this->expensiveCache->clear();
1134                 } else {
1135                         foreach ( $paths as $path ) {
1136                                 $this->cheapCache->clear( $path );
1137                                 $this->expensiveCache->clear( $path );
1138                         }
1139                 }
1140                 $this->doClearCache( $paths );
1141         }
1142
1143         /**
1144          * Clears any additional stat caches for storage paths
1145          *
1146          * @see FileBackend::clearCache()
1147          *
1148          * @param $paths Array Storage paths (optional)
1149          * @return void
1150          */
1151         protected function doClearCache( array $paths = null ) {}
1152
1153         /**
1154          * Is this a key/value store where directories are just virtual?
1155          * Virtual directories exists in so much as files exists that are
1156          * prefixed with the directory path followed by a forward slash.
1157          *
1158          * @return bool
1159          */
1160         abstract protected function directoriesAreVirtual();
1161
1162         /**
1163          * Check if a container name is valid.
1164          * This checks for for length and illegal characters.
1165          *
1166          * @param $container string
1167          * @return bool
1168          */
1169         final protected static function isValidContainerName( $container ) {
1170                 // This accounts for Swift and S3 restrictions while leaving room
1171                 // for things like '.xxx' (hex shard chars) or '.seg' (segments).
1172                 // This disallows directory separators or traversal characters.
1173                 // Note that matching strings URL encode to the same string;
1174                 // in Swift, the length restriction is *after* URL encoding.
1175                 return preg_match( '/^[a-z0-9][a-z0-9-_]{0,199}$/i', $container );
1176         }
1177
1178         /**
1179          * Splits a storage path into an internal container name,
1180          * an internal relative file name, and a container shard suffix.
1181          * Any shard suffix is already appended to the internal container name.
1182          * This also checks that the storage path is valid and within this backend.
1183          *
1184          * If the container is sharded but a suffix could not be determined,
1185          * this means that the path can only refer to a directory and can only
1186          * be scanned by looking in all the container shards.
1187          *
1188          * @param $storagePath string
1189          * @return Array (container, path, container suffix) or (null, null, null) if invalid
1190          */
1191         final protected function resolveStoragePath( $storagePath ) {
1192                 list( $backend, $container, $relPath ) = self::splitStoragePath( $storagePath );
1193                 if ( $backend === $this->name ) { // must be for this backend
1194                         $relPath = self::normalizeContainerPath( $relPath );
1195                         if ( $relPath !== null ) {
1196                                 // Get shard for the normalized path if this container is sharded
1197                                 $cShard = $this->getContainerShard( $container, $relPath );
1198                                 // Validate and sanitize the relative path (backend-specific)
1199                                 $relPath = $this->resolveContainerPath( $container, $relPath );
1200                                 if ( $relPath !== null ) {
1201                                         // Prepend any wiki ID prefix to the container name
1202                                         $container = $this->fullContainerName( $container );
1203                                         if ( self::isValidContainerName( $container ) ) {
1204                                                 // Validate and sanitize the container name (backend-specific)
1205                                                 $container = $this->resolveContainerName( "{$container}{$cShard}" );
1206                                                 if ( $container !== null ) {
1207                                                         return array( $container, $relPath, $cShard );
1208                                                 }
1209                                         }
1210                                 }
1211                         }
1212                 }
1213                 return array( null, null, null );
1214         }
1215
1216         /**
1217          * Like resolveStoragePath() except null values are returned if
1218          * the container is sharded and the shard could not be determined.
1219          *
1220          * @see FileBackendStore::resolveStoragePath()
1221          *
1222          * @param $storagePath string
1223          * @return Array (container, path) or (null, null) if invalid
1224          */
1225         final protected function resolveStoragePathReal( $storagePath ) {
1226                 list( $container, $relPath, $cShard ) = $this->resolveStoragePath( $storagePath );
1227                 if ( $cShard !== null ) {
1228                         return array( $container, $relPath );
1229                 }
1230                 return array( null, null );
1231         }
1232
1233         /**
1234          * Get the container name shard suffix for a given path.
1235          * Any empty suffix means the container is not sharded.
1236          *
1237          * @param $container string Container name
1238          * @param $relPath string Storage path relative to the container
1239          * @return string|null Returns null if shard could not be determined
1240          */
1241         final protected function getContainerShard( $container, $relPath ) {
1242                 list( $levels, $base, $repeat ) = $this->getContainerHashLevels( $container );
1243                 if ( $levels == 1 || $levels == 2 ) {
1244                         // Hash characters are either base 16 or 36
1245                         $char = ( $base == 36 ) ? '[0-9a-z]' : '[0-9a-f]';
1246                         // Get a regex that represents the shard portion of paths.
1247                         // The concatenation of the captures gives us the shard.
1248                         if ( $levels === 1 ) { // 16 or 36 shards per container
1249                                 $hashDirRegex = '(' . $char . ')';
1250                         } else { // 256 or 1296 shards per container
1251                                 if ( $repeat ) { // verbose hash dir format (e.g. "a/ab/abc")
1252                                         $hashDirRegex = $char . '/(' . $char . '{2})';
1253                                 } else { // short hash dir format (e.g. "a/b/c")
1254                                         $hashDirRegex = '(' . $char . ')/(' . $char . ')';
1255                                 }
1256                         }
1257                         // Allow certain directories to be above the hash dirs so as
1258                         // to work with FileRepo (e.g. "archive/a/ab" or "temp/a/ab").
1259                         // They must be 2+ chars to avoid any hash directory ambiguity.
1260                         $m = array();
1261                         if ( preg_match( "!^(?:[^/]{2,}/)*$hashDirRegex(?:/|$)!", $relPath, $m ) ) {
1262                                 return '.' . implode( '', array_slice( $m, 1 ) );
1263                         }
1264                         return null; // failed to match
1265                 }
1266                 return ''; // no sharding
1267         }
1268
1269         /**
1270          * Check if a storage path maps to a single shard.
1271          * Container dirs like "a", where the container shards on "x/xy",
1272          * can reside on several shards. Such paths are tricky to handle.
1273          *
1274          * @param $storagePath string Storage path
1275          * @return bool
1276          */
1277         final public function isSingleShardPathInternal( $storagePath ) {
1278                 list( $c, $r, $shard ) = $this->resolveStoragePath( $storagePath );
1279                 return ( $shard !== null );
1280         }
1281
1282         /**
1283          * Get the sharding config for a container.
1284          * If greater than 0, then all file storage paths within
1285          * the container are required to be hashed accordingly.
1286          *
1287          * @param $container string
1288          * @return Array (integer levels, integer base, repeat flag) or (0, 0, false)
1289          */
1290         final protected function getContainerHashLevels( $container ) {
1291                 if ( isset( $this->shardViaHashLevels[$container] ) ) {
1292                         $config = $this->shardViaHashLevels[$container];
1293                         $hashLevels = (int)$config['levels'];
1294                         if ( $hashLevels == 1 || $hashLevels == 2 ) {
1295                                 $hashBase = (int)$config['base'];
1296                                 if ( $hashBase == 16 || $hashBase == 36 ) {
1297                                         return array( $hashLevels, $hashBase, $config['repeat'] );
1298                                 }
1299                         }
1300                 }
1301                 return array( 0, 0, false ); // no sharding
1302         }
1303
1304         /**
1305          * Get a list of full container shard suffixes for a container
1306          *
1307          * @param $container string
1308          * @return Array
1309          */
1310         final protected function getContainerSuffixes( $container ) {
1311                 $shards = array();
1312                 list( $digits, $base ) = $this->getContainerHashLevels( $container );
1313                 if ( $digits > 0 ) {
1314                         $numShards = pow( $base, $digits );
1315                         for ( $index = 0; $index < $numShards; $index++ ) {
1316                                 $shards[] = '.' . wfBaseConvert( $index, 10, $base, $digits );
1317                         }
1318                 }
1319                 return $shards;
1320         }
1321
1322         /**
1323          * Get the full container name, including the wiki ID prefix
1324          *
1325          * @param $container string
1326          * @return string
1327          */
1328         final protected function fullContainerName( $container ) {
1329                 if ( $this->wikiId != '' ) {
1330                         return "{$this->wikiId}-$container";
1331                 } else {
1332                         return $container;
1333                 }
1334         }
1335
1336         /**
1337          * Resolve a container name, checking if it's allowed by the backend.
1338          * This is intended for internal use, such as encoding illegal chars.
1339          * Subclasses can override this to be more restrictive.
1340          *
1341          * @param $container string
1342          * @return string|null
1343          */
1344         protected function resolveContainerName( $container ) {
1345                 return $container;
1346         }
1347
1348         /**
1349          * Resolve a relative storage path, checking if it's allowed by the backend.
1350          * This is intended for internal use, such as encoding illegal chars or perhaps
1351          * getting absolute paths (e.g. FS based backends). Note that the relative path
1352          * may be the empty string (e.g. the path is simply to the container).
1353          *
1354          * @param $container string Container name
1355          * @param $relStoragePath string Storage path relative to the container
1356          * @return string|null Path or null if not valid
1357          */
1358         protected function resolveContainerPath( $container, $relStoragePath ) {
1359                 return $relStoragePath;
1360         }
1361
1362         /**
1363          * Get the cache key for a container
1364          *
1365          * @param $container string Resolved container name
1366          * @return string
1367          */
1368         private function containerCacheKey( $container ) {
1369                 return wfMemcKey( 'backend', $this->getName(), 'container', $container );
1370         }
1371
1372         /**
1373          * Set the cached info for a container
1374          *
1375          * @param $container string Resolved container name
1376          * @param $val mixed Information to cache
1377          */
1378         final protected function setContainerCache( $container, $val ) {
1379                 $this->memCache->add( $this->containerCacheKey( $container ), $val, 14*86400 );
1380         }
1381
1382         /**
1383          * Delete the cached info for a container.
1384          * The cache key is salted for a while to prevent race conditions.
1385          *
1386          * @param $container string Resolved container name
1387          */
1388         final protected function deleteContainerCache( $container ) {
1389                 if ( !$this->memCache->set( $this->containerCacheKey( $container ), 'PURGED', 300 ) ) {
1390                         trigger_error( "Unable to delete stat cache for container $container." );
1391                 }
1392         }
1393
1394         /**
1395          * Do a batch lookup from cache for container stats for all containers
1396          * used in a list of container names, storage paths, or FileOp objects.
1397          *
1398          * @param $items Array
1399          * @return void
1400          */
1401         final protected function primeContainerCache( array $items ) {
1402                 wfProfileIn( __METHOD__ );
1403                 wfProfileIn( __METHOD__ . '-' . $this->name );
1404
1405                 $paths = array(); // list of storage paths
1406                 $contNames = array(); // (cache key => resolved container name)
1407                 // Get all the paths/containers from the items...
1408                 foreach ( $items as $item ) {
1409                         if ( $item instanceof FileOp ) {
1410                                 $paths = array_merge( $paths, $item->storagePathsRead() );
1411                                 $paths = array_merge( $paths, $item->storagePathsChanged() );
1412                         } elseif ( self::isStoragePath( $item ) ) {
1413                                 $paths[] = $item;
1414                         } elseif ( is_string( $item ) ) { // full container name
1415                                 $contNames[$this->containerCacheKey( $item )] = $item;
1416                         }
1417                 }
1418                 // Get all the corresponding cache keys for paths...
1419                 foreach ( $paths as $path ) {
1420                         list( $fullCont, $r, $s ) = $this->resolveStoragePath( $path );
1421                         if ( $fullCont !== null ) { // valid path for this backend
1422                                 $contNames[$this->containerCacheKey( $fullCont )] = $fullCont;
1423                         }
1424                 }
1425
1426                 $contInfo = array(); // (resolved container name => cache value)
1427                 // Get all cache entries for these container cache keys...
1428                 $values = $this->memCache->getMulti( array_keys( $contNames ) );
1429                 foreach ( $values as $cacheKey => $val ) {
1430                         $contInfo[$contNames[$cacheKey]] = $val;
1431                 }
1432
1433                 // Populate the container process cache for the backend...
1434                 $this->doPrimeContainerCache( array_filter( $contInfo, 'is_array' ) );
1435
1436                 wfProfileOut( __METHOD__ . '-' . $this->name );
1437                 wfProfileOut( __METHOD__ );
1438         }
1439
1440         /**
1441          * Fill the backend-specific process cache given an array of
1442          * resolved container names and their corresponding cached info.
1443          * Only containers that actually exist should appear in the map.
1444          *
1445          * @param $containerInfo Array Map of resolved container names to cached info
1446          * @return void
1447          */
1448         protected function doPrimeContainerCache( array $containerInfo ) {}
1449
1450         /**
1451          * Get the cache key for a file path
1452          *
1453          * @param $path string Storage path
1454          * @return string
1455          */
1456         private function fileCacheKey( $path ) {
1457                 return wfMemcKey( 'backend', $this->getName(), 'file', sha1( $path ) );
1458         }
1459
1460         /**
1461          * Set the cached stat info for a file path
1462          *
1463          * @param $path string Storage path
1464          * @param $val mixed Information to cache
1465          */
1466         final protected function setFileCache( $path, $val ) {
1467                 $this->memCache->add( $this->fileCacheKey( $path ), $val, 7*86400 );
1468         }
1469
1470         /**
1471          * Delete the cached stat info for a file path.
1472          * The cache key is salted for a while to prevent race conditions.
1473          *
1474          * @param $path string Storage path
1475          */
1476         final protected function deleteFileCache( $path ) {
1477                 if ( !$this->memCache->set( $this->fileCacheKey( $path ), 'PURGED', 300 ) ) {
1478                         trigger_error( "Unable to delete stat cache for file $path." );
1479                 }
1480         }
1481
1482         /**
1483          * Do a batch lookup from cache for file stats for all paths
1484          * used in a list of storage paths or FileOp objects.
1485          *
1486          * @param $items Array List of storage paths or FileOps
1487          * @return void
1488          */
1489         final protected function primeFileCache( array $items ) {
1490                 wfProfileIn( __METHOD__ );
1491                 wfProfileIn( __METHOD__ . '-' . $this->name );
1492
1493                 $paths = array(); // list of storage paths
1494                 $pathNames = array(); // (cache key => storage path)
1495                 // Get all the paths/containers from the items...
1496                 foreach ( $items as $item ) {
1497                         if ( $item instanceof FileOp ) {
1498                                 $paths = array_merge( $paths, $item->storagePathsRead() );
1499                                 $paths = array_merge( $paths, $item->storagePathsChanged() );
1500                         } elseif ( self::isStoragePath( $item ) ) {
1501                                 $paths[] = $item;
1502                         }
1503                 }
1504                 // Get all the corresponding cache keys for paths...
1505                 foreach ( $paths as $path ) {
1506                         list( $cont, $rel, $s ) = $this->resolveStoragePath( $path );
1507                         if ( $rel !== null ) { // valid path for this backend
1508                                 $pathNames[$this->fileCacheKey( $path )] = $path;
1509                         }
1510                 }
1511                 // Get all cache entries for these container cache keys...
1512                 $values = $this->memCache->getMulti( array_keys( $pathNames ) );
1513                 foreach ( $values as $cacheKey => $val ) {
1514                         if ( is_array( $val ) ) {
1515                                 $path = $pathNames[$cacheKey];
1516                                 $this->cheapCache->set( $path, 'stat', $val );
1517                                 if ( isset( $val['sha1'] ) ) { // some backends store SHA-1 as metadata
1518                                         $this->cheapCache->set( $path, 'sha1',
1519                                                 array( 'hash' => $val['sha1'], 'latest' => $val['latest'] ) );
1520                                 }
1521                         }
1522                 }
1523
1524                 wfProfileOut( __METHOD__ . '-' . $this->name );
1525                 wfProfileOut( __METHOD__ );
1526         }
1527 }
1528
1529 /**
1530  * FileBackendStore helper class for performing asynchronous file operations.
1531  *
1532  * For example, calling FileBackendStore::createInternal() with the "async"
1533  * param flag may result in a Status that contains this object as a value.
1534  * This class is largely backend-specific and is mostly just "magic" to be
1535  * passed to FileBackendStore::executeOpHandlesInternal().
1536  */
1537 abstract class FileBackendStoreOpHandle {
1538         /** @var Array */
1539         public $params = array(); // params to caller functions
1540         /** @var FileBackendStore */
1541         public $backend;
1542         /** @var Array */
1543         public $resourcesToClose = array();
1544
1545         public $call; // string; name that identifies the function called
1546
1547         /**
1548          * Close all open file handles
1549          *
1550          * @return void
1551          */
1552         public function closeResources() {
1553                 array_map( 'fclose', $this->resourcesToClose );
1554         }
1555 }
1556
1557 /**
1558  * FileBackendStore helper function to handle listings that span container shards.
1559  * Do not use this class from places outside of FileBackendStore.
1560  *
1561  * @ingroup FileBackend
1562  */
1563 abstract class FileBackendStoreShardListIterator implements Iterator {
1564         /** @var FileBackendStore */
1565         protected $backend;
1566         /** @var Array */
1567         protected $params;
1568         /** @var Array */
1569         protected $shardSuffixes;
1570         protected $container; // string; full container name
1571         protected $directory; // string; resolved relative path
1572
1573         /** @var Traversable */
1574         protected $iter;
1575         protected $curShard = 0; // integer
1576         protected $pos = 0; // integer
1577
1578         /** @var Array */
1579         protected $multiShardPaths = array(); // (rel path => 1)
1580
1581         /**
1582          * @param $backend FileBackendStore
1583          * @param $container string Full storage container name
1584          * @param $dir string Storage directory relative to container
1585          * @param $suffixes Array List of container shard suffixes
1586          * @param $params Array
1587          */
1588         public function __construct(
1589                 FileBackendStore $backend, $container, $dir, array $suffixes, array $params
1590         ) {
1591                 $this->backend = $backend;
1592                 $this->container = $container;
1593                 $this->directory = $dir;
1594                 $this->shardSuffixes = $suffixes;
1595                 $this->params = $params;
1596         }
1597
1598         /**
1599          * @see Iterator::key()
1600          * @return integer
1601          */
1602         public function key() {
1603                 return $this->pos;
1604         }
1605
1606         /**
1607          * @see Iterator::valid()
1608          * @return bool
1609          */
1610         public function valid() {
1611                 if ( $this->iter instanceof Iterator ) {
1612                         return $this->iter->valid();
1613                 } elseif ( is_array( $this->iter ) ) {
1614                         return ( current( $this->iter ) !== false ); // no paths can have this value
1615                 }
1616                 return false; // some failure?
1617         }
1618
1619         /**
1620          * @see Iterator::current()
1621          * @return string|bool String or false
1622          */
1623         public function current() {
1624                 return ( $this->iter instanceof Iterator )
1625                         ? $this->iter->current()
1626                         : current( $this->iter );
1627         }
1628
1629         /**
1630          * @see Iterator::next()
1631          * @return void
1632          */
1633         public function next() {
1634                 ++$this->pos;
1635                 ( $this->iter instanceof Iterator ) ? $this->iter->next() : next( $this->iter );
1636                 do {
1637                         $continue = false; // keep scanning shards?
1638                         $this->filterViaNext(); // filter out duplicates
1639                         // Find the next non-empty shard if no elements are left
1640                         if ( !$this->valid() ) {
1641                                 $this->nextShardIteratorIfNotValid();
1642                                 $continue = $this->valid(); // re-filter unless we ran out of shards
1643                         }
1644                 } while ( $continue );
1645         }
1646
1647         /**
1648          * @see Iterator::rewind()
1649          * @return void
1650          */
1651         public function rewind() {
1652                 $this->pos = 0;
1653                 $this->curShard = 0;
1654                 $this->setIteratorFromCurrentShard();
1655                 do {
1656                         $continue = false; // keep scanning shards?
1657                         $this->filterViaNext(); // filter out duplicates
1658                         // Find the next non-empty shard if no elements are left
1659                         if ( !$this->valid() ) {
1660                                 $this->nextShardIteratorIfNotValid();
1661                                 $continue = $this->valid(); // re-filter unless we ran out of shards
1662                         }
1663                 } while ( $continue );
1664         }
1665
1666         /**
1667          * Filter out duplicate items by advancing to the next ones
1668          */
1669         protected function filterViaNext() {
1670                 while ( $this->valid() ) {
1671                         $rel = $this->iter->current(); // path relative to given directory
1672                         $path = $this->params['dir'] . "/{$rel}"; // full storage path
1673                         if ( $this->backend->isSingleShardPathInternal( $path ) ) {
1674                                 break; // path is only on one shard; no issue with duplicates
1675                         } elseif ( isset( $this->multiShardPaths[$rel] ) ) {
1676                                 // Don't keep listing paths that are on multiple shards
1677                                 ( $this->iter instanceof Iterator ) ? $this->iter->next() : next( $this->iter );
1678                         } else {
1679                                 $this->multiShardPaths[$rel] = 1;
1680                                 break;
1681                         }
1682                 }
1683         }
1684
1685         /**
1686          * If the list iterator for this container shard is out of items,
1687          * then move on to the next container that has items.
1688          * If there are none, then it advances to the last container.
1689          */
1690         protected function nextShardIteratorIfNotValid() {
1691                 while ( !$this->valid() && ++$this->curShard < count( $this->shardSuffixes ) ) {
1692                         $this->setIteratorFromCurrentShard();
1693                 }
1694         }
1695
1696         /**
1697          * Set the list iterator to that of the current container shard
1698          */
1699         protected function setIteratorFromCurrentShard() {
1700                 $this->iter = $this->listFromShard(
1701                         $this->container . $this->shardSuffixes[$this->curShard],
1702                         $this->directory, $this->params );
1703                 // Start loading results so that current() works
1704                 if ( $this->iter ) {
1705                         ( $this->iter instanceof Iterator ) ? $this->iter->rewind() : reset( $this->iter );
1706                 }
1707         }
1708
1709         /**
1710          * Get the list for a given container shard
1711          *
1712          * @param $container string Resolved container name
1713          * @param $dir string Resolved path relative to container
1714          * @param $params Array
1715          * @return Traversable|Array|null
1716          */
1717         abstract protected function listFromShard( $container, $dir, array $params );
1718 }
1719
1720 /**
1721  * Iterator for listing directories
1722  */
1723 class FileBackendStoreShardDirIterator extends FileBackendStoreShardListIterator {
1724         /**
1725          * @see FileBackendStoreShardListIterator::listFromShard()
1726          * @return Array|null|Traversable
1727          */
1728         protected function listFromShard( $container, $dir, array $params ) {
1729                 return $this->backend->getDirectoryListInternal( $container, $dir, $params );
1730         }
1731 }
1732
1733 /**
1734  * Iterator for listing regular files
1735  */
1736 class FileBackendStoreShardFileIterator extends FileBackendStoreShardListIterator {
1737         /**
1738          * @see FileBackendStoreShardListIterator::listFromShard()
1739          * @return Array|null|Traversable
1740          */
1741         protected function listFromShard( $container, $dir, array $params ) {
1742                 return $this->backend->getFileListInternal( $container, $dir, $params );
1743         }
1744 }