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::doPublish()
 431          * @return Status
 432          */
 433         final protected function doPublish( array $params ) {
 434                 wfProfileIn( __METHOD__ );
 435                 wfProfileIn( __METHOD__ . '-' . $this->name );
 436                 $status = Status::newGood();
 437
 438                 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
 439                 if ( $dir === null ) {
 440                         $status->fatal( 'backend-fail-invalidpath', $params['dir'] );
 441                         wfProfileOut( __METHOD__ . '-' . $this->name );
 442                         wfProfileOut( __METHOD__ );
 443                         return $status; // invalid storage path
 444                 }
 445
 446                 if ( $shard !== null ) { // confined to a single container/shard
 447                         $status->merge( $this->doPublishInternal( $fullCont, $dir, $params ) );
 448                 } else { // directory is on several shards
 449                         wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
 450                         list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
 451                         foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
 452                                 $status->merge( $this->doPublishInternal( "{$fullCont}{$suffix}", $dir, $params ) );
 453                         }
 454                 }
 455
 456                 wfProfileOut( __METHOD__ . '-' . $this->name );
 457                 wfProfileOut( __METHOD__ );
 458                 return $status;
 459         }
 460
 461         /**
 462          * @see FileBackendStore::doPublish()
 463          * @return Status
 464          */
 465         protected function doPublishInternal( $container, $dir, array $params ) {
 466                 return Status::newGood();
 467         }
 468
 469         /**
 470          * @see FileBackend::doClean()
 471          * @return Status
 472          */
 473         final protected function doClean( array $params ) {
 474                 wfProfileIn( __METHOD__ );
 475                 wfProfileIn( __METHOD__ . '-' . $this->name );
 476                 $status = Status::newGood();
 477
 478                 // Recursive: first delete all empty subdirs recursively
 479                 if ( !empty( $params['recursive'] ) && !$this->directoriesAreVirtual() ) {
 480                         $subDirsRel = $this->getTopDirectoryList( array( 'dir' => $params['dir'] ) );
 481                         if ( $subDirsRel !== null ) { // no errors
 482                                 foreach ( $subDirsRel as $subDirRel ) {
 483                                         $subDir = $params['dir'] . "/{$subDirRel}"; // full path
 484                                         $status->merge( $this->doClean( array( 'dir' => $subDir ) + $params ) );
 485                                 }
 486                         }
 487                 }
 488
 489                 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
 490                 if ( $dir === null ) {
 491                         $status->fatal( 'backend-fail-invalidpath', $params['dir'] );
 492                         wfProfileOut( __METHOD__ . '-' . $this->name );
 493                         wfProfileOut( __METHOD__ );
 494                         return $status; // invalid storage path
 495                 }
 496
 497                 // Attempt to lock this directory...
 498                 $filesLockEx = array( $params['dir'] );
 499                 $scopedLockE = $this->getScopedFileLocks( $filesLockEx, LockManager::LOCK_EX, $status );
 500                 if ( !$status->isOK() ) {
 501                         wfProfileOut( __METHOD__ . '-' . $this->name );
 502                         wfProfileOut( __METHOD__ );
 503                         return $status; // abort
 504                 }
 505
 506                 if ( $shard !== null ) { // confined to a single container/shard
 507                         $status->merge( $this->doCleanInternal( $fullCont, $dir, $params ) );
 508                         $this->deleteContainerCache( $fullCont ); // purge cache
 509                 } else { // directory is on several shards
 510                         wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
 511                         list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
 512                         foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
 513                                 $status->merge( $this->doCleanInternal( "{$fullCont}{$suffix}", $dir, $params ) );
 514                                 $this->deleteContainerCache( "{$fullCont}{$suffix}" ); // purge cache
 515                         }
 516                 }
 517
 518                 wfProfileOut( __METHOD__ . '-' . $this->name );
 519                 wfProfileOut( __METHOD__ );
 520                 return $status;
 521         }
 522
 523         /**
 524          * @see FileBackendStore::doClean()
 525          * @return Status
 526          */
 527         protected function doCleanInternal( $container, $dir, array $params ) {
 528                 return Status::newGood();
 529         }
 530
 531         /**
 532          * @see FileBackend::fileExists()
 533          * @return bool|null
 534          */
 535         final public function fileExists( array $params ) {
 536                 wfProfileIn( __METHOD__ );
 537                 wfProfileIn( __METHOD__ . '-' . $this->name );
 538                 $stat = $this->getFileStat( $params );
 539                 wfProfileOut( __METHOD__ . '-' . $this->name );
 540                 wfProfileOut( __METHOD__ );
 541                 return ( $stat === null ) ? null : (bool)$stat; // null => failure
 542         }
 543
 544         /**
 545          * @see FileBackend::getFileTimestamp()
 546          * @return bool
 547          */
 548         final public function getFileTimestamp( array $params ) {
 549                 wfProfileIn( __METHOD__ );
 550                 wfProfileIn( __METHOD__ . '-' . $this->name );
 551                 $stat = $this->getFileStat( $params );
 552                 wfProfileOut( __METHOD__ . '-' . $this->name );
 553                 wfProfileOut( __METHOD__ );
 554                 return $stat ? $stat['mtime'] : false;
 555         }
 556
 557         /**
 558          * @see FileBackend::getFileSize()
 559          * @return bool
 560          */
 561         final public function getFileSize( array $params ) {
 562                 wfProfileIn( __METHOD__ );
 563                 wfProfileIn( __METHOD__ . '-' . $this->name );
 564                 $stat = $this->getFileStat( $params );
 565                 wfProfileOut( __METHOD__ . '-' . $this->name );
 566                 wfProfileOut( __METHOD__ );
 567                 return $stat ? $stat['size'] : false;
 568         }
 569
 570         /**
 571          * @see FileBackend::getFileStat()
 572          * @return bool
 573          */
 574         final public function getFileStat( array $params ) {
 575                 $path = self::normalizeStoragePath( $params['src'] );
 576                 if ( $path === null ) {
 577                         return false; // invalid storage path
 578                 }
 579                 wfProfileIn( __METHOD__ );
 580                 wfProfileIn( __METHOD__ . '-' . $this->name );
 581                 $latest = !empty( $params['latest'] ); // use latest data?
 582                 if ( !isset( $this->cache[$path]['stat'] ) ) {
 583                         $this->primeFileCache( array( $path ) ); // check persistent cache
 584                 }
 585                 if ( isset( $this->cache[$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 || $this->cache[$path]['stat']['latest'] ) {
 589                                 $this->pingCache( $path ); // LRU
 590                                 wfProfileOut( __METHOD__ . '-' . $this->name );
 591                                 wfProfileOut( __METHOD__ );
 592                                 return $this->cache[$path]['stat'];
 593                         }
 594                 }
 595                 wfProfileIn( __METHOD__ . '-miss' );
 596                 wfProfileIn( __METHOD__ . '-miss-' . $this->name );
 597                 $stat = $this->doGetFileStat( $params );
 598                 wfProfileOut( __METHOD__ . '-miss-' . $this->name );
 599                 wfProfileOut( __METHOD__ . '-miss' );
 600                 if ( is_array( $stat ) ) { // don't cache negatives
 601                         $stat['latest'] = $latest;
 602                         $this->trimCache(); // limit memory
 603                         $this->cache[$path]['stat'] = $stat;
 604                         $this->setFileCache( $path, $stat ); // update persistent cache
 605                         if ( isset( $stat['sha1'] ) ) { // some backends store SHA-1 as metadata
 606                                 $this->trimCache(); // limit memory
 607                                 $this->cache[$path]['sha1'] =
 608                                         array( 'hash' => $stat['sha1'], 'latest' => $latest );
 609                         }
 610                 } else {
 611                         wfDebug( __METHOD__ . ": File $path does not exist.\n" );
 612                 }
 613                 wfProfileOut( __METHOD__ . '-' . $this->name );
 614                 wfProfileOut( __METHOD__ );
 615                 return $stat;
 616         }
 617
 618         /**
 619          * @see FileBackendStore::getFileStat()
 620          */
 621         abstract protected function doGetFileStat( array $params );
 622
 623         /**
 624          * @see FileBackend::getFileContents()
 625          * @return bool|string
 626          */
 627         public function getFileContents( array $params ) {
 628                 wfProfileIn( __METHOD__ );
 629                 wfProfileIn( __METHOD__ . '-' . $this->name );
 630                 $tmpFile = $this->getLocalReference( $params );
 631                 if ( !$tmpFile ) {
 632                         wfProfileOut( __METHOD__ . '-' . $this->name );
 633                         wfProfileOut( __METHOD__ );
 634                         return false;
 635                 }
 636                 wfSuppressWarnings();
 637                 $data = file_get_contents( $tmpFile->getPath() );
 638                 wfRestoreWarnings();
 639                 wfProfileOut( __METHOD__ . '-' . $this->name );
 640                 wfProfileOut( __METHOD__ );
 641                 return $data;
 642         }
 643
 644         /**
 645          * @see FileBackend::getFileSha1Base36()
 646          * @return bool|string
 647          */
 648         final public function getFileSha1Base36( array $params ) {
 649                 $path = self::normalizeStoragePath( $params['src'] );
 650                 if ( $path === null ) {
 651                         return false; // invalid storage path
 652                 }
 653                 wfProfileIn( __METHOD__ );
 654                 wfProfileIn( __METHOD__ . '-' . $this->name );
 655                 $latest = !empty( $params['latest'] ); // use latest data?
 656                 if ( isset( $this->cache[$path]['sha1'] ) ) {
 657                         // If we want the latest data, check that this cached
 658                         // value was in fact fetched with the latest available data.
 659                         if ( !$latest || $this->cache[$path]['sha1']['latest'] ) {
 660                                 $this->pingCache( $path ); // LRU
 661                                 wfProfileOut( __METHOD__ . '-' . $this->name );
 662                                 wfProfileOut( __METHOD__ );
 663                                 return $this->cache[$path]['sha1']['hash'];
 664                         }
 665                 }
 666                 wfProfileIn( __METHOD__ . '-miss' );
 667                 wfProfileIn( __METHOD__ . '-miss-' . $this->name );
 668                 $hash = $this->doGetFileSha1Base36( $params );
 669                 wfProfileOut( __METHOD__ . '-miss-' . $this->name );
 670                 wfProfileOut( __METHOD__ . '-miss' );
 671                 if ( $hash ) { // don't cache negatives
 672                         $this->trimCache(); // limit memory
 673                         $this->cache[$path]['sha1'] = array( 'hash' => $hash, 'latest' => $latest );
 674                 }
 675                 wfProfileOut( __METHOD__ . '-' . $this->name );
 676                 wfProfileOut( __METHOD__ );
 677                 return $hash;
 678         }
 679
 680         /**
 681          * @see FileBackendStore::getFileSha1Base36()
 682          * @return bool|string
 683          */
 684         protected function doGetFileSha1Base36( array $params ) {
 685                 $fsFile = $this->getLocalReference( $params );
 686                 if ( !$fsFile ) {
 687                         return false;
 688                 } else {
 689                         return $fsFile->getSha1Base36();
 690                 }
 691         }
 692
 693         /**
 694          * @see FileBackend::getFileProps()
 695          * @return Array
 696          */
 697         final public function getFileProps( array $params ) {
 698                 wfProfileIn( __METHOD__ );
 699                 wfProfileIn( __METHOD__ . '-' . $this->name );
 700                 $fsFile = $this->getLocalReference( $params );
 701                 $props = $fsFile ? $fsFile->getProps() : FSFile::placeholderProps();
 702                 wfProfileOut( __METHOD__ . '-' . $this->name );
 703                 wfProfileOut( __METHOD__ );
 704                 return $props;
 705         }
 706
 707         /**
 708          * @see FileBackend::getLocalReference()
 709          * @return TempFSFile|null
 710          */
 711         public function getLocalReference( array $params ) {
 712                 $path = self::normalizeStoragePath( $params['src'] );
 713                 if ( $path === null ) {
 714                         return null; // invalid storage path
 715                 }
 716                 wfProfileIn( __METHOD__ );
 717                 wfProfileIn( __METHOD__ . '-' . $this->name );
 718                 $latest = !empty( $params['latest'] ); // use latest data?
 719                 if ( isset( $this->expensiveCache[$path]['localRef'] ) ) {
 720                         // If we want the latest data, check that this cached
 721                         // value was in fact fetched with the latest available data.
 722                         if ( !$latest || $this->expensiveCache[$path]['localRef']['latest'] ) {
 723                                 $this->pingExpensiveCache( $path );
 724                                 wfProfileOut( __METHOD__ . '-' . $this->name );
 725                                 wfProfileOut( __METHOD__ );
 726                                 return $this->expensiveCache[$path]['localRef']['object'];
 727                         }
 728                 }
 729                 $tmpFile = $this->getLocalCopy( $params );
 730                 if ( $tmpFile ) { // don't cache negatives
 731                         $this->trimExpensiveCache(); // limit memory
 732                         $this->expensiveCache[$path]['localRef'] =
 733                                 array( 'object' => $tmpFile, 'latest' => $latest );
 734                 }
 735                 wfProfileOut( __METHOD__ . '-' . $this->name );
 736                 wfProfileOut( __METHOD__ );
 737                 return $tmpFile;
 738         }
 739
 740         /**
 741          * @see FileBackend::streamFile()
 742          * @return Status
 743          */
 744         final public function streamFile( array $params ) {
 745                 wfProfileIn( __METHOD__ );
 746                 wfProfileIn( __METHOD__ . '-' . $this->name );
 747                 $status = Status::newGood();
 748
 749                 $info = $this->getFileStat( $params );
 750                 if ( !$info ) { // let StreamFile handle the 404
 751                         $status->fatal( 'backend-fail-notexists', $params['src'] );
 752                 }
 753
 754                 // Set output buffer and HTTP headers for stream
 755                 $extraHeaders = isset( $params['headers'] ) ? $params['headers'] : array();
 756                 $res = StreamFile::prepareForStream( $params['src'], $info, $extraHeaders );
 757                 if ( $res == StreamFile::NOT_MODIFIED ) {
 758                         // do nothing; client cache is up to date
 759                 } elseif ( $res == StreamFile::READY_STREAM ) {
 760                         wfProfileIn( __METHOD__ . '-send' );
 761                         wfProfileIn( __METHOD__ . '-send-' . $this->name );
 762                         $status = $this->doStreamFile( $params );
 763                         wfProfileOut( __METHOD__ . '-send-' . $this->name );
 764                         wfProfileOut( __METHOD__ . '-send' );
 765                 } else {
 766                         $status->fatal( 'backend-fail-stream', $params['src'] );
 767                 }
 768
 769                 wfProfileOut( __METHOD__ . '-' . $this->name );
 770                 wfProfileOut( __METHOD__ );
 771                 return $status;
 772         }
 773
 774         /**
 775          * @see FileBackendStore::streamFile()
 776          * @return Status
 777          */
 778         protected function doStreamFile( array $params ) {
 779                 $status = Status::newGood();
 780
 781                 $fsFile = $this->getLocalReference( $params );
 782                 if ( !$fsFile ) {
 783                         $status->fatal( 'backend-fail-stream', $params['src'] );
 784                 } elseif ( !readfile( $fsFile->getPath() ) ) {
 785                         $status->fatal( 'backend-fail-stream', $params['src'] );
 786                 }
 787
 788                 return $status;
 789         }
 790
 791         /**
 792          * @see FileBackend::directoryExists()
 793          * @return bool|null
 794          */
 795         final public function directoryExists( array $params ) {
 796                 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
 797                 if ( $dir === null ) {
 798                         return false; // invalid storage path
 799                 }
 800                 if ( $shard !== null ) { // confined to a single container/shard
 801                         return $this->doDirectoryExists( $fullCont, $dir, $params );
 802                 } else { // directory is on several shards
 803                         wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
 804                         list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
 805                         $res = false; // response
 806                         foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
 807                                 $exists = $this->doDirectoryExists( "{$fullCont}{$suffix}", $dir, $params );
 808                                 if ( $exists ) {
 809                                         $res = true;
 810                                         break; // found one!
 811                                 } elseif ( $exists === null ) { // error?
 812                                         $res = null; // if we don't find anything, it is indeterminate
 813                                 }
 814                         }
 815                         return $res;
 816                 }
 817         }
 818
 819         /**
 820          * @see FileBackendStore::directoryExists()
 821          *
 822          * @param $container string Resolved container name
 823          * @param $dir string Resolved path relative to container
 824          * @param $params Array
 825          * @return bool|null
 826          */
 827         abstract protected function doDirectoryExists( $container, $dir, array $params );
 828
 829         /**
 830          * @see FileBackend::getDirectoryList()
 831          * @return Traversable|Array|null Returns null on failure
 832          */
 833         final public function getDirectoryList( array $params ) {
 834                 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
 835                 if ( $dir === null ) { // invalid storage path
 836                         return null;
 837                 }
 838                 if ( $shard !== null ) {
 839                         // File listing is confined to a single container/shard
 840                         return $this->getDirectoryListInternal( $fullCont, $dir, $params );
 841                 } else {
 842                         wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
 843                         // File listing spans multiple containers/shards
 844                         list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
 845                         return new FileBackendStoreShardDirIterator( $this,
 846                                 $fullCont, $dir, $this->getContainerSuffixes( $shortCont ), $params );
 847                 }
 848         }
 849
 850         /**
 851          * Do not call this function from places outside FileBackend
 852          *
 853          * @see FileBackendStore::getDirectoryList()
 854          *
 855          * @param $container string Resolved container name
 856          * @param $dir string Resolved path relative to container
 857          * @param $params Array
 858          * @return Traversable|Array|null Returns null on failure
 859          */
 860         abstract public function getDirectoryListInternal( $container, $dir, array $params );
 861
 862         /**
 863          * @see FileBackend::getFileList()
 864          * @return Traversable|Array|null Returns null on failure
 865          */
 866         final public function getFileList( array $params ) {
 867                 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
 868                 if ( $dir === null ) { // invalid storage path
 869                         return null;
 870                 }
 871                 if ( $shard !== null ) {
 872                         // File listing is confined to a single container/shard
 873                         return $this->getFileListInternal( $fullCont, $dir, $params );
 874                 } else {
 875                         wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
 876                         // File listing spans multiple containers/shards
 877                         list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
 878                         return new FileBackendStoreShardFileIterator( $this,
 879                                 $fullCont, $dir, $this->getContainerSuffixes( $shortCont ), $params );
 880                 }
 881         }
 882
 883         /**
 884          * Do not call this function from places outside FileBackend
 885          *
 886          * @see FileBackendStore::getFileList()
 887          *
 888          * @param $container string Resolved container name
 889          * @param $dir string Resolved path relative to container
 890          * @param $params Array
 891          * @return Traversable|Array|null Returns null on failure
 892          */
 893         abstract public function getFileListInternal( $container, $dir, array $params );
 894
 895         /**
 896          * Return a list of FileOp objects from a list of operations.
 897          * Do not call this function from places outside FileBackend.
 898          *
 899          * The result must have the same number of items as the input.
 900          * An exception is thrown if an unsupported operation is requested.
 901          *
 902          * @param $ops Array Same format as doOperations()
 903          * @return Array List of FileOp objects
 904          * @throws MWException
 905          */
 906         final public function getOperationsInternal( array $ops ) {
 907                 $supportedOps = array(
 908                         'store'       => 'StoreFileOp',
 909                         'copy'        => 'CopyFileOp',
 910                         'move'        => 'MoveFileOp',
 911                         'delete'      => 'DeleteFileOp',
 912                         'create'      => 'CreateFileOp',
 913                         'null'        => 'NullFileOp'
 914                 );
 915
 916                 $performOps = array(); // array of FileOp objects
 917                 // Build up ordered array of FileOps...
 918                 foreach ( $ops as $operation ) {
 919                         $opName = $operation['op'];
 920                         if ( isset( $supportedOps[$opName] ) ) {
 921                                 $class = $supportedOps[$opName];
 922                                 // Get params for this operation
 923                                 $params = $operation;
 924                                 // Append the FileOp class
 925                                 $performOps[] = new $class( $this, $params );
 926                         } else {
 927                                 throw new MWException( "Operation '$opName' is not supported." );
 928                         }
 929                 }
 930
 931                 return $performOps;
 932         }
 933
 934         /**
 935          * Get a list of storage paths to lock for a list of operations
 936          * Returns an array with 'sh' (shared) and 'ex' (exclusive) keys,
 937          * each corresponding to a list of storage paths to be locked.
 938          *
 939          * @param $performOps Array List of FileOp objects
 940          * @return Array ('sh' => list of paths, 'ex' => list of paths)
 941          */
 942         final public function getPathsToLockForOpsInternal( array $performOps ) {
 943                 // Build up a list of files to lock...
 944                 $paths = array( 'sh' => array(), 'ex' => array() );
 945                 foreach ( $performOps as $fileOp ) {
 946                         $paths['sh'] = array_merge( $paths['sh'], $fileOp->storagePathsRead() );
 947                         $paths['ex'] = array_merge( $paths['ex'], $fileOp->storagePathsChanged() );
 948                 }
 949                 // Optimization: if doing an EX lock anyway, don't also set an SH one
 950                 $paths['sh'] = array_diff( $paths['sh'], $paths['ex'] );
 951                 // Get a shared lock on the parent directory of each path changed
 952                 $paths['sh'] = array_merge( $paths['sh'], array_map( 'dirname', $paths['ex'] ) );
 953
 954                 return $paths;
 955         }
 956
 957         /**
 958          * @see FileBackend::getScopedLocksForOps()
 959          * @return Array
 960          */
 961         public function getScopedLocksForOps( array $ops, Status $status ) {
 962                 $paths = $this->getPathsToLockForOpsInternal( $this->getOperationsInternal( $ops ) );
 963                 return array(
 964                         $this->getScopedFileLocks( $paths['sh'], LockManager::LOCK_UW, $status ),
 965                         $this->getScopedFileLocks( $paths['ex'], LockManager::LOCK_EX, $status )
 966                 );
 967         }
 968
 969         /**
 970          * @see FileBackend::doOperationsInternal()
 971          * @return Status
 972          */
 973         protected function doOperationsInternal( array $ops, array $opts ) {
 974                 wfProfileIn( __METHOD__ );
 975                 wfProfileIn( __METHOD__ . '-' . $this->name );
 976                 $status = Status::newGood();
 977
 978                 // Build up a list of FileOps...
 979                 $performOps = $this->getOperationsInternal( $ops );
 980
 981                 // Acquire any locks as needed...
 982                 if ( empty( $opts['nonLocking'] ) ) {
 983                         // Build up a list of files to lock...
 984                         $paths = $this->getPathsToLockForOpsInternal( $performOps );
 985                         // Try to lock those files for the scope of this function...
 986                         $scopeLockS = $this->getScopedFileLocks( $paths['sh'], LockManager::LOCK_UW, $status );
 987                         $scopeLockE = $this->getScopedFileLocks( $paths['ex'], LockManager::LOCK_EX, $status );
 988                         if ( !$status->isOK() ) {
 989                                 wfProfileOut( __METHOD__ . '-' . $this->name );
 990                                 wfProfileOut( __METHOD__ );
 991                                 return $status; // abort
 992                         }
 993                 }
 994
 995                 // Clear any file cache entries (after locks acquired)
 996                 $this->clearCache();
 997
 998                 // Load from the persistent file and container caches
 999                 $this->primeFileCache( $performOps );
1000                 $this->primeContainerCache( $performOps );
1001
1002                 // Actually attempt the operation batch...
1003                 $subStatus = FileOpBatch::attempt( $performOps, $opts, $this->fileJournal );
1004
1005                 // Merge errors into status fields
1006                 $status->merge( $subStatus );
1007                 $status->success = $subStatus->success; // not done in merge()
1008
1009                 wfProfileOut( __METHOD__ . '-' . $this->name );
1010                 wfProfileOut( __METHOD__ );
1011                 return $status;
1012         }
1013
1014         /**
1015          * @see FileBackend::doQuickOperationsInternal()
1016          * @return Status
1017          * @throws MWException
1018          */
1019         final protected function doQuickOperationsInternal( array $ops ) {
1020                 wfProfileIn( __METHOD__ );
1021                 wfProfileIn( __METHOD__ . '-' . $this->name );
1022                 $status = Status::newGood();
1023
1024                 $supportedOps = array( 'create', 'store', 'copy', 'move', 'delete', 'null' );
1025                 $async = ( $this->parallelize === 'implicit' );
1026                 $maxConcurrency = $this->concurrency; // throttle
1027
1028                 $statuses = array(); // array of (index => Status)
1029                 $fileOpHandles = array(); // list of (index => handle) arrays
1030                 $curFileOpHandles = array(); // current handle batch
1031                 // Perform the sync-only ops and build up op handles for the async ops...
1032                 foreach ( $ops as $index => $params ) {
1033                         if ( !in_array( $params['op'], $supportedOps ) ) {
1034                                 wfProfileOut( __METHOD__ . '-' . $this->name );
1035                                 wfProfileOut( __METHOD__ );
1036                                 throw new MWException( "Operation '{$params['op']}' is not supported." );
1037                         }
1038                         $method = $params['op'] . 'Internal'; // e.g. "storeInternal"
1039                         $subStatus = $this->$method( array( 'async' => $async ) + $params );
1040                         if ( $subStatus->value instanceof FileBackendStoreOpHandle ) { // async
1041                                 if ( count( $curFileOpHandles ) >= $maxConcurrency ) {
1042                                         $fileOpHandles[] = $curFileOpHandles; // push this batch
1043                                         $curFileOpHandles = array();
1044                                 }
1045                                 $curFileOpHandles[$index] = $subStatus->value; // keep index
1046                         } else { // error or completed
1047                                 $statuses[$index] = $subStatus; // keep index
1048                         }
1049                 }
1050                 if ( count( $curFileOpHandles ) ) {
1051                         $fileOpHandles[] = $curFileOpHandles; // last batch
1052                 }
1053                 // Do all the async ops that can be done concurrently...
1054                 foreach ( $fileOpHandles as $fileHandleBatch ) {
1055                         $statuses = $statuses + $this->executeOpHandlesInternal( $fileHandleBatch );
1056                 }
1057                 // Marshall and merge all the responses...
1058                 foreach ( $statuses as $index => $subStatus ) {
1059                         $status->merge( $subStatus );
1060                         if ( $subStatus->isOK() ) {
1061                                 $status->success[$index] = true;
1062                                 ++$status->successCount;
1063                         } else {
1064                                 $status->success[$index] = false;
1065                                 ++$status->failCount;
1066                         }
1067                 }
1068
1069                 wfProfileOut( __METHOD__ . '-' . $this->name );
1070                 wfProfileOut( __METHOD__ );
1071                 return $status;
1072         }
1073
1074         /**
1075          * Execute a list of FileBackendStoreOpHandle handles in parallel.
1076          * The resulting Status object fields will correspond
1077          * to the order in which the handles where given.
1078          *
1079          * @param $handles Array List of FileBackendStoreOpHandle objects
1080          * @return Array Map of Status objects
1081          * @throws MWException
1082          */
1083         final public function executeOpHandlesInternal( array $fileOpHandles ) {
1084                 wfProfileIn( __METHOD__ );
1085                 wfProfileIn( __METHOD__ . '-' . $this->name );
1086                 foreach ( $fileOpHandles as $fileOpHandle ) {
1087                         if ( !( $fileOpHandle instanceof FileBackendStoreOpHandle ) ) {
1088                                 throw new MWException( "Given a non-FileBackendStoreOpHandle object." );
1089                         } elseif ( $fileOpHandle->backend->getName() !== $this->getName() ) {
1090                                 throw new MWException( "Given a FileBackendStoreOpHandle for the wrong backend." );
1091                         }
1092                 }
1093                 $res = $this->doExecuteOpHandlesInternal( $fileOpHandles );
1094                 foreach ( $fileOpHandles as $fileOpHandle ) {
1095                         $fileOpHandle->closeResources();
1096                 }
1097                 wfProfileOut( __METHOD__ . '-' . $this->name );
1098                 wfProfileOut( __METHOD__ );
1099                 return $res;
1100         }
1101
1102         /**
1103          * @see FileBackendStore::executeOpHandlesInternal()
1104          * @return Array List of corresponding Status objects
1105          */
1106         protected function doExecuteOpHandlesInternal( array $fileOpHandles ) {
1107                 foreach ( $fileOpHandles as $fileOpHandle ) { // OK if empty
1108                         throw new MWException( "This backend supports no asynchronous operations." );
1109                 }
1110                 return array();
1111         }
1112
1113         /**
1114          * @see FileBackend::clearCache()
1115          */
1116         final public function clearCache( array $paths = null ) {
1117                 if ( is_array( $paths ) ) {
1118                         $paths = array_map( 'FileBackend::normalizeStoragePath', $paths );
1119                         $paths = array_filter( $paths, 'strlen' ); // remove nulls
1120                 }
1121                 if ( $paths === null ) {
1122                         $this->cache = array();
1123                         $this->expensiveCache = array();
1124                 } else {
1125                         foreach ( $paths as $path ) {
1126                                 unset( $this->cache[$path] );
1127                                 unset( $this->expensiveCache[$path] );
1128                         }
1129                 }
1130                 $this->doClearCache( $paths );
1131         }
1132
1133         /**
1134          * Clears any additional stat caches for storage paths
1135          *
1136          * @see FileBackend::clearCache()
1137          *
1138          * @param $paths Array Storage paths (optional)
1139          * @return void
1140          */
1141         protected function doClearCache( array $paths = null ) {}
1142
1143         /**
1144          * Is this a key/value store where directories are just virtual?
1145          * Virtual directories exists in so much as files exists that are
1146          * prefixed with the directory path followed by a forward slash.
1147          *
1148          * @return bool
1149          */
1150         abstract protected function directoriesAreVirtual();
1151
1152         /**
1153          * Move a cache entry to the top (such as when accessed)
1154          *
1155          * @param $path string Storage path
1156          * @return void
1157          */
1158         protected function pingCache( $path ) {
1159                 if ( isset( $this->cache[$path] ) ) {
1160                         $tmp = $this->cache[$path];
1161                         unset( $this->cache[$path] );
1162                         $this->cache[$path] = $tmp;
1163                 }
1164         }
1165
1166         /**
1167          * Prune the inexpensive cache if it is too big to add an item
1168          *
1169          * @return void
1170          */
1171         protected function trimCache() {
1172                 if ( count( $this->cache ) >= $this->maxCacheSize ) {
1173                         reset( $this->cache );
1174                         unset( $this->cache[key( $this->cache )] );
1175                 }
1176         }
1177
1178         /**
1179          * Move a cache entry to the top (such as when accessed)
1180          *
1181          * @param $path string Storage path
1182          * @return void
1183          */
1184         protected function pingExpensiveCache( $path ) {
1185                 if ( isset( $this->expensiveCache[$path] ) ) {
1186                         $tmp = $this->expensiveCache[$path];
1187                         unset( $this->expensiveCache[$path] );
1188                         $this->expensiveCache[$path] = $tmp;
1189                 }
1190         }
1191
1192         /**
1193          * Prune the expensive cache if it is too big to add an item
1194          *
1195          * @return void
1196          */
1197         protected function trimExpensiveCache() {
1198                 if ( count( $this->expensiveCache ) >= $this->maxExpensiveCacheSize ) {
1199                         reset( $this->expensiveCache );
1200                         unset( $this->expensiveCache[key( $this->expensiveCache )] );
1201                 }
1202         }
1203
1204         /**
1205          * Check if a container name is valid.
1206          * This checks for for length and illegal characters.
1207          *
1208          * @param $container string
1209          * @return bool
1210          */
1211         final protected static function isValidContainerName( $container ) {
1212                 // This accounts for Swift and S3 restrictions while leaving room
1213                 // for things like '.xxx' (hex shard chars) or '.seg' (segments).
1214                 // This disallows directory separators or traversal characters.
1215                 // Note that matching strings URL encode to the same string;
1216                 // in Swift, the length restriction is *after* URL encoding.
1217                 return preg_match( '/^[a-z0-9][a-z0-9-_]{0,199}$/i', $container );
1218         }
1219
1220         /**
1221          * Splits a storage path into an internal container name,
1222          * an internal relative file name, and a container shard suffix.
1223          * Any shard suffix is already appended to the internal container name.
1224          * This also checks that the storage path is valid and within this backend.
1225          *
1226          * If the container is sharded but a suffix could not be determined,
1227          * this means that the path can only refer to a directory and can only
1228          * be scanned by looking in all the container shards.
1229          *
1230          * @param $storagePath string
1231          * @return Array (container, path, container suffix) or (null, null, null) if invalid
1232          */
1233         final protected function resolveStoragePath( $storagePath ) {
1234                 list( $backend, $container, $relPath ) = self::splitStoragePath( $storagePath );
1235                 if ( $backend === $this->name ) { // must be for this backend
1236                         $relPath = self::normalizeContainerPath( $relPath );
1237                         if ( $relPath !== null ) {
1238                                 // Get shard for the normalized path if this container is sharded
1239                                 $cShard = $this->getContainerShard( $container, $relPath );
1240                                 // Validate and sanitize the relative path (backend-specific)
1241                                 $relPath = $this->resolveContainerPath( $container, $relPath );
1242                                 if ( $relPath !== null ) {
1243                                         // Prepend any wiki ID prefix to the container name
1244                                         $container = $this->fullContainerName( $container );
1245                                         if ( self::isValidContainerName( $container ) ) {
1246                                                 // Validate and sanitize the container name (backend-specific)
1247                                                 $container = $this->resolveContainerName( "{$container}{$cShard}" );
1248                                                 if ( $container !== null ) {
1249                                                         return array( $container, $relPath, $cShard );
1250                                                 }
1251                                         }
1252                                 }
1253                         }
1254                 }
1255                 return array( null, null, null );
1256         }
1257
1258         /**
1259          * Like resolveStoragePath() except null values are returned if
1260          * the container is sharded and the shard could not be determined.
1261          *
1262          * @see FileBackendStore::resolveStoragePath()
1263          *
1264          * @param $storagePath string
1265          * @return Array (container, path) or (null, null) if invalid
1266          */
1267         final protected function resolveStoragePathReal( $storagePath ) {
1268                 list( $container, $relPath, $cShard ) = $this->resolveStoragePath( $storagePath );
1269                 if ( $cShard !== null ) {
1270                         return array( $container, $relPath );
1271                 }
1272                 return array( null, null );
1273         }
1274
1275         /**
1276          * Get the container name shard suffix for a given path.
1277          * Any empty suffix means the container is not sharded.
1278          *
1279          * @param $container string Container name
1280          * @param $relPath string Storage path relative to the container
1281          * @return string|null Returns null if shard could not be determined
1282          */
1283         final protected function getContainerShard( $container, $relPath ) {
1284                 list( $levels, $base, $repeat ) = $this->getContainerHashLevels( $container );
1285                 if ( $levels == 1 || $levels == 2 ) {
1286                         // Hash characters are either base 16 or 36
1287                         $char = ( $base == 36 ) ? '[0-9a-z]' : '[0-9a-f]';
1288                         // Get a regex that represents the shard portion of paths.
1289                         // The concatenation of the captures gives us the shard.
1290                         if ( $levels === 1 ) { // 16 or 36 shards per container
1291                                 $hashDirRegex = '(' . $char . ')';
1292                         } else { // 256 or 1296 shards per container
1293                                 if ( $repeat ) { // verbose hash dir format (e.g. "a/ab/abc")
1294                                         $hashDirRegex = $char . '/(' . $char . '{2})';
1295                                 } else { // short hash dir format (e.g. "a/b/c")
1296                                         $hashDirRegex = '(' . $char . ')/(' . $char . ')';
1297                                 }
1298                         }
1299                         // Allow certain directories to be above the hash dirs so as
1300                         // to work with FileRepo (e.g. "archive/a/ab" or "temp/a/ab").
1301                         // They must be 2+ chars to avoid any hash directory ambiguity.
1302                         $m = array();
1303                         if ( preg_match( "!^(?:[^/]{2,}/)*$hashDirRegex(?:/|$)!", $relPath, $m ) ) {
1304                                 return '.' . implode( '', array_slice( $m, 1 ) );
1305                         }
1306                         return null; // failed to match
1307                 }
1308                 return ''; // no sharding
1309         }
1310
1311         /**
1312          * Check if a storage path maps to a single shard.
1313          * Container dirs like "a", where the container shards on "x/xy",
1314          * can reside on several shards. Such paths are tricky to handle.
1315          *
1316          * @param $storagePath string Storage path
1317          * @return bool
1318          */
1319         final public function isSingleShardPathInternal( $storagePath ) {
1320                 list( $c, $r, $shard ) = $this->resolveStoragePath( $storagePath );
1321                 return ( $shard !== null );
1322         }
1323
1324         /**
1325          * Get the sharding config for a container.
1326          * If greater than 0, then all file storage paths within
1327          * the container are required to be hashed accordingly.
1328          *
1329          * @param $container string
1330          * @return Array (integer levels, integer base, repeat flag) or (0, 0, false)
1331          */
1332         final protected function getContainerHashLevels( $container ) {
1333                 if ( isset( $this->shardViaHashLevels[$container] ) ) {
1334                         $config = $this->shardViaHashLevels[$container];
1335                         $hashLevels = (int)$config['levels'];
1336                         if ( $hashLevels == 1 || $hashLevels == 2 ) {
1337                                 $hashBase = (int)$config['base'];
1338                                 if ( $hashBase == 16 || $hashBase == 36 ) {
1339                                         return array( $hashLevels, $hashBase, $config['repeat'] );
1340                                 }
1341                         }
1342                 }
1343                 return array( 0, 0, false ); // no sharding
1344         }
1345
1346         /**
1347          * Get a list of full container shard suffixes for a container
1348          *
1349          * @param $container string
1350          * @return Array
1351          */
1352         final protected function getContainerSuffixes( $container ) {
1353                 $shards = array();
1354                 list( $digits, $base ) = $this->getContainerHashLevels( $container );
1355                 if ( $digits > 0 ) {
1356                         $numShards = pow( $base, $digits );
1357                         for ( $index = 0; $index < $numShards; $index++ ) {
1358                                 $shards[] = '.' . wfBaseConvert( $index, 10, $base, $digits );
1359                         }
1360                 }
1361                 return $shards;
1362         }
1363
1364         /**
1365          * Get the full container name, including the wiki ID prefix
1366          *
1367          * @param $container string
1368          * @return string
1369          */
1370         final protected function fullContainerName( $container ) {
1371                 if ( $this->wikiId != '' ) {
1372                         return "{$this->wikiId}-$container";
1373                 } else {
1374                         return $container;
1375                 }
1376         }
1377
1378         /**
1379          * Resolve a container name, checking if it's allowed by the backend.
1380          * This is intended for internal use, such as encoding illegal chars.
1381          * Subclasses can override this to be more restrictive.
1382          *
1383          * @param $container string
1384          * @return string|null
1385          */
1386         protected function resolveContainerName( $container ) {
1387                 return $container;
1388         }
1389
1390         /**
1391          * Resolve a relative storage path, checking if it's allowed by the backend.
1392          * This is intended for internal use, such as encoding illegal chars or perhaps
1393          * getting absolute paths (e.g. FS based backends). Note that the relative path
1394          * may be the empty string (e.g. the path is simply to the container).
1395          *
1396          * @param $container string Container name
1397          * @param $relStoragePath string Storage path relative to the container
1398          * @return string|null Path or null if not valid
1399          */
1400         protected function resolveContainerPath( $container, $relStoragePath ) {
1401                 return $relStoragePath;
1402         }
1403
1404         /**
1405          * Get the cache key for a container
1406          *
1407          * @param $container string Resolved container name
1408          * @return string
1409          */
1410         private function containerCacheKey( $container ) {
1411                 return wfMemcKey( 'backend', $this->getName(), 'container', $container );
1412         }
1413
1414         /**
1415          * Set the cached info for a container
1416          *
1417          * @param $container string Resolved container name
1418          * @param $val mixed Information to cache
1419          */
1420         final protected function setContainerCache( $container, $val ) {
1421                 $this->memCache->add( $this->containerCacheKey( $container ), $val, 14*86400 );
1422         }
1423
1424         /**
1425          * Delete the cached info for a container.
1426          * The cache key is salted for a while to prevent race conditions.
1427          *
1428          * @param $container string Resolved container name
1429          */
1430         final protected function deleteContainerCache( $container ) {
1431                 if ( !$this->memCache->set( $this->containerCacheKey( $container ), 'PURGED', 300 ) ) {
1432                         trigger_error( "Unable to delete stat cache for container $container." );
1433                 }
1434         }
1435
1436         /**
1437          * Do a batch lookup from cache for container stats for all containers
1438          * used in a list of container names, storage paths, or FileOp objects.
1439          *
1440          * @param $items Array
1441          * @return void
1442          */
1443         final protected function primeContainerCache( array $items ) {
1444                 wfProfileIn( __METHOD__ );
1445                 wfProfileIn( __METHOD__ . '-' . $this->name );
1446
1447                 $paths = array(); // list of storage paths
1448                 $contNames = array(); // (cache key => resolved container name)
1449                 // Get all the paths/containers from the items...
1450                 foreach ( $items as $item ) {
1451                         if ( $item instanceof FileOp ) {
1452                                 $paths = array_merge( $paths, $item->storagePathsRead() );
1453                                 $paths = array_merge( $paths, $item->storagePathsChanged() );
1454                         } elseif ( self::isStoragePath( $item ) ) {
1455                                 $paths[] = $item;
1456                         } elseif ( is_string( $item ) ) { // full container name
1457                                 $contNames[$this->containerCacheKey( $item )] = $item;
1458                         }
1459                 }
1460                 // Get all the corresponding cache keys for paths...
1461                 foreach ( $paths as $path ) {
1462                         list( $fullCont, $r, $s ) = $this->resolveStoragePath( $path );
1463                         if ( $fullCont !== null ) { // valid path for this backend
1464                                 $contNames[$this->containerCacheKey( $fullCont )] = $fullCont;
1465                         }
1466                 }
1467
1468                 $contInfo = array(); // (resolved container name => cache value)
1469                 // Get all cache entries for these container cache keys...
1470                 $values = $this->memCache->getMulti( array_keys( $contNames ) );
1471                 foreach ( $values as $cacheKey => $val ) {
1472                         $contInfo[$contNames[$cacheKey]] = $val;
1473                 }
1474
1475                 // Populate the container process cache for the backend...
1476                 $this->doPrimeContainerCache( array_filter( $contInfo, 'is_array' ) );
1477
1478                 wfProfileOut( __METHOD__ . '-' . $this->name );
1479                 wfProfileOut( __METHOD__ );
1480         }
1481
1482         /**
1483          * Fill the backend-specific process cache given an array of
1484          * resolved container names and their corresponding cached info.
1485          * Only containers that actually exist should appear in the map.
1486          *
1487          * @param $containerInfo Array Map of resolved container names to cached info
1488          * @return void
1489          */
1490         protected function doPrimeContainerCache( array $containerInfo ) {}
1491
1492         /**
1493          * Get the cache key for a file path
1494          *
1495          * @param $path string Storage path
1496          * @return string
1497          */
1498         private function fileCacheKey( $path ) {
1499                 return wfMemcKey( 'backend', $this->getName(), 'file', sha1( $path ) );
1500         }
1501
1502         /**
1503          * Set the cached stat info for a file path
1504          *
1505          * @param $path string Storage path
1506          * @param $val mixed Information to cache
1507          */
1508         final protected function setFileCache( $path, $val ) {
1509                 $this->memCache->add( $this->fileCacheKey( $path ), $val, 7*86400 );
1510         }
1511
1512         /**
1513          * Delete the cached stat info for a file path.
1514          * The cache key is salted for a while to prevent race conditions.
1515          *
1516          * @param $path string Storage path
1517          */
1518         final protected function deleteFileCache( $path ) {
1519                 if ( !$this->memCache->set( $this->fileCacheKey( $path ), 'PURGED', 300 ) ) {
1520                         trigger_error( "Unable to delete stat cache for file $path." );
1521                 }
1522         }
1523
1524         /**
1525          * Do a batch lookup from cache for file stats for all paths
1526          * used in a list of storage paths or FileOp objects.
1527          *
1528          * @param $items Array List of storage paths or FileOps
1529          * @return void
1530          */
1531         final protected function primeFileCache( array $items ) {
1532                 wfProfileIn( __METHOD__ );
1533                 wfProfileIn( __METHOD__ . '-' . $this->name );
1534
1535                 $paths = array(); // list of storage paths
1536                 $pathNames = array(); // (cache key => storage path)
1537                 // Get all the paths/containers from the items...
1538                 foreach ( $items as $item ) {
1539                         if ( $item instanceof FileOp ) {
1540                                 $paths = array_merge( $paths, $item->storagePathsRead() );
1541                                 $paths = array_merge( $paths, $item->storagePathsChanged() );
1542                         } elseif ( self::isStoragePath( $item ) ) {
1543                                 $paths[] = $item;
1544                         }
1545                 }
1546                 // Get all the corresponding cache keys for paths...
1547                 foreach ( $paths as $path ) {
1548                         list( $cont, $rel, $s ) = $this->resolveStoragePath( $path );
1549                         if ( $rel !== null ) { // valid path for this backend
1550                                 $pathNames[$this->fileCacheKey( $path )] = $path;
1551                         }
1552                 }
1553                 // Get all cache entries for these container cache keys...
1554                 $values = $this->memCache->getMulti( array_keys( $pathNames ) );
1555                 foreach ( $values as $cacheKey => $val ) {
1556                         if ( is_array( $val ) ) {
1557                                 $path = $pathNames[$cacheKey];
1558                                 $this->trimCache(); // limit memory
1559                                 $this->cache[$path]['stat'] = $val;
1560                                 if ( isset( $val['sha1'] ) ) { // some backends store SHA-1 as metadata
1561                                         $this->trimCache(); // limit memory
1562                                         $this->cache[$path]['sha1'] =
1563                                                 array( 'hash' => $val['sha1'], 'latest' => $val['latest'] );
1564                                 }
1565                         }
1566                 }
1567
1568                 wfProfileOut( __METHOD__ . '-' . $this->name );
1569                 wfProfileOut( __METHOD__ );
1570         }
1571 }
1572
1573 /**
1574  * FileBackendStore helper class for performing asynchronous file operations.
1575  *
1576  * For example, calling FileBackendStore::createInternal() with the "async"
1577  * param flag may result in a Status that contains this object as a value.
1578  * This class is largely backend-specific and is mostly just "magic" to be
1579  * passed to FileBackendStore::executeOpHandlesInternal().
1580  */
1581 abstract class FileBackendStoreOpHandle {
1582         /** @var Array */
1583         public $params = array(); // params to caller functions
1584         /** @var FileBackendStore */
1585         public $backend;
1586         /** @var Array */
1587         public $resourcesToClose = array();
1588
1589         public $call; // string; name that identifies the function called
1590
1591         /**
1592          * Close all open file handles
1593          *
1594          * @return void
1595          */
1596         public function closeResources() {
1597                 array_map( 'fclose', $this->resourcesToClose );
1598         }
1599 }
1600
1601 /**
1602  * FileBackendStore helper function to handle listings that span container shards.
1603  * Do not use this class from places outside of FileBackendStore.
1604  *
1605  * @ingroup FileBackend
1606  */
1607 abstract class FileBackendStoreShardListIterator implements Iterator {
1608         /** @var FileBackendStore */
1609         protected $backend;
1610         /** @var Array */
1611         protected $params;
1612         /** @var Array */
1613         protected $shardSuffixes;
1614         protected $container; // string; full container name
1615         protected $directory; // string; resolved relative path
1616
1617         /** @var Traversable */
1618         protected $iter;
1619         protected $curShard = 0; // integer
1620         protected $pos = 0; // integer
1621
1622         /** @var Array */
1623         protected $multiShardPaths = array(); // (rel path => 1)
1624
1625         /**
1626          * @param $backend FileBackendStore
1627          * @param $container string Full storage container name
1628          * @param $dir string Storage directory relative to container
1629          * @param $suffixes Array List of container shard suffixes
1630          * @param $params Array
1631          */
1632         public function __construct(
1633                 FileBackendStore $backend, $container, $dir, array $suffixes, array $params
1634         ) {
1635                 $this->backend = $backend;
1636                 $this->container = $container;
1637                 $this->directory = $dir;
1638                 $this->shardSuffixes = $suffixes;
1639                 $this->params = $params;
1640         }
1641
1642         /**
1643          * @see Iterator::key()
1644          * @return integer
1645          */
1646         public function key() {
1647                 return $this->pos;
1648         }
1649
1650         /**
1651          * @see Iterator::valid()
1652          * @return bool
1653          */
1654         public function valid() {
1655                 if ( $this->iter instanceof Iterator ) {
1656                         return $this->iter->valid();
1657                 } elseif ( is_array( $this->iter ) ) {
1658                         return ( current( $this->iter ) !== false ); // no paths can have this value
1659                 }
1660                 return false; // some failure?
1661         }
1662
1663         /**
1664          * @see Iterator::current()
1665          * @return string|bool String or false
1666          */
1667         public function current() {
1668                 return ( $this->iter instanceof Iterator )
1669                         ? $this->iter->current()
1670                         : current( $this->iter );
1671         }
1672
1673         /**
1674          * @see Iterator::next()
1675          * @return void
1676          */
1677         public function next() {
1678                 ++$this->pos;
1679                 ( $this->iter instanceof Iterator ) ? $this->iter->next() : next( $this->iter );
1680                 do {
1681                         $continue = false; // keep scanning shards?
1682                         $this->filterViaNext(); // filter out duplicates
1683                         // Find the next non-empty shard if no elements are left
1684                         if ( !$this->valid() ) {
1685                                 $this->nextShardIteratorIfNotValid();
1686                                 $continue = $this->valid(); // re-filter unless we ran out of shards
1687                         }
1688                 } while ( $continue );
1689         }
1690
1691         /**
1692          * @see Iterator::rewind()
1693          * @return void
1694          */
1695         public function rewind() {
1696                 $this->pos = 0;
1697                 $this->curShard = 0;
1698                 $this->setIteratorFromCurrentShard();
1699                 do {
1700                         $continue = false; // keep scanning shards?
1701                         $this->filterViaNext(); // filter out duplicates
1702                         // Find the next non-empty shard if no elements are left
1703                         if ( !$this->valid() ) {
1704                                 $this->nextShardIteratorIfNotValid();
1705                                 $continue = $this->valid(); // re-filter unless we ran out of shards
1706                         }
1707                 } while ( $continue );
1708         }
1709
1710         /**
1711          * Filter out duplicate items by advancing to the next ones
1712          */
1713         protected function filterViaNext() {
1714                 while ( $this->valid() ) {
1715                         $rel = $this->iter->current(); // path relative to given directory
1716                         $path = $this->params['dir'] . "/{$rel}"; // full storage path
1717                         if ( $this->backend->isSingleShardPathInternal( $path ) ) {
1718                                 break; // path is only on one shard; no issue with duplicates
1719                         } elseif ( isset( $this->multiShardPaths[$rel] ) ) {
1720                                 // Don't keep listing paths that are on multiple shards
1721                                 ( $this->iter instanceof Iterator ) ? $this->iter->next() : next( $this->iter );
1722                         } else {
1723                                 $this->multiShardPaths[$rel] = 1;
1724                                 break;
1725                         }
1726                 }
1727         }
1728
1729         /**
1730          * If the list iterator for this container shard is out of items,
1731          * then move on to the next container that has items.
1732          * If there are none, then it advances to the last container.
1733          */
1734         protected function nextShardIteratorIfNotValid() {
1735                 while ( !$this->valid() && ++$this->curShard < count( $this->shardSuffixes ) ) {
1736                         $this->setIteratorFromCurrentShard();
1737                 }
1738         }
1739
1740         /**
1741          * Set the list iterator to that of the current container shard
1742          */
1743         protected function setIteratorFromCurrentShard() {
1744                 $this->iter = $this->listFromShard(
1745                         $this->container . $this->shardSuffixes[$this->curShard],
1746                         $this->directory, $this->params );
1747                 // Start loading results so that current() works
1748                 if ( $this->iter ) {
1749                         ( $this->iter instanceof Iterator ) ? $this->iter->rewind() : reset( $this->iter );
1750                 }
1751         }
1752
1753         /**
1754          * Get the list for a given container shard
1755          *
1756          * @param $container string Resolved container name
1757          * @param $dir string Resolved path relative to container
1758          * @param $params Array
1759          * @return Traversable|Array|null
1760          */
1761         abstract protected function listFromShard( $container, $dir, array $params );
1762 }
1763
1764 /**
1765  * Iterator for listing directories
1766  */
1767 class FileBackendStoreShardDirIterator extends FileBackendStoreShardListIterator {
1768         /**
1769          * @see FileBackendStoreShardListIterator::listFromShard()
1770          * @return Array|null|Traversable
1771          */
1772         protected function listFromShard( $container, $dir, array $params ) {
1773                 return $this->backend->getDirectoryListInternal( $container, $dir, $params );
1774         }
1775 }
1776
1777 /**
1778  * Iterator for listing regular files
1779  */
1780 class FileBackendStoreShardFileIterator extends FileBackendStoreShardListIterator {
1781         /**
1782          * @see FileBackendStoreShardListIterator::listFromShard()
1783          * @return Array|null|Traversable
1784          */
1785         protected function listFromShard( $container, $dir, array $params ) {
1786                 return $this->backend->getFileListInternal( $container, $dir, $params );
1787         }
1788 }