includes/filerepo/backend/FileBackendStore.php

   1 <?php
   2 /**
   3  * Base class for all backends using particular storage medium.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup FileBackend
  22  * @author Aaron Schulz
  23  */
  24
  25 /**
  26  * @brief Base class for all backends using particular storage medium.
  27  *
  28  * This class defines the methods as abstract that subclasses must implement.
  29  * Outside callers should *not* use functions with "Internal" in the name.
  30  *
  31  * The FileBackend operations are implemented using basic functions
  32  * such as storeInternal(), copyInternal(), deleteInternal() and the like.
  33  * This class is also responsible for path resolution and sanitization.
  34  *
  35  * @ingroup FileBackend
  36  * @since 1.19
  37  */
  38 abstract class FileBackendStore extends FileBackend {
  39         /** @var BagOStuff */
  40         protected $memCache;
  41
  42         /** @var Array Map of paths to small (RAM/disk) cache items */
  43         protected $cache = array(); // (storage path => key => value)
  44         protected $maxCacheSize = 300; // integer; max paths with entries
  45         /** @var Array Map of paths to large (RAM/disk) cache items */
  46         protected $expensiveCache = array(); // (storage path => key => value)
  47         protected $maxExpensiveCacheSize = 5; // integer; max paths with entries
  48
  49         /** @var Array Map of container names to sharding settings */
  50         protected $shardViaHashLevels = array(); // (container name => config array)
  51
  52         protected $maxFileSize = 4294967296; // integer bytes (4GiB)
  53
  54         /**
  55          * @see FileBackend::__construct()
  56          *
  57          * @param $config Array
  58          */
  59         public function __construct( array $config ) {
  60                 parent::__construct( $config );
  61                 $this->memCache = new EmptyBagOStuff(); // disabled by default
  62         }
  63
  64         /**
  65          * Get the maximum allowable file size given backend
  66          * medium restrictions and basic performance constraints.
  67          * Do not call this function from places outside FileBackend and FileOp.
  68          *
  69          * @return integer Bytes
  70          */
  71         final public function maxFileSizeInternal() {
  72                 return $this->maxFileSize;
  73         }
  74
  75         /**
  76          * Check if a file can be created at a given storage path.
  77          * FS backends should check if the parent directory exists and the file is writable.
  78          * Backends using key/value stores should check if the container exists.
  79          *
  80          * @param $storagePath string
  81          * @return bool
  82          */
  83         abstract public function isPathUsableInternal( $storagePath );
  84
  85         /**
  86          * Create a file in the backend with the given contents.
  87          * Do not call this function from places outside FileBackend and FileOp.
  88          *
  89          * $params include:
  90          *     content       : the raw file contents
  91          *     dst           : destination storage path
  92          *     overwrite     : overwrite any file that exists at the destination
  93          *     async         : Status will be returned immediately if supported.
  94          *                     If the status is OK, then its value field will be
  95          *                     set to a FileBackendStoreOpHandle object.
  96          *
  97          * @param $params Array
  98          * @return Status
  99          */
 100         final public function createInternal( array $params ) {
 101                 wfProfileIn( __METHOD__ );
 102                 wfProfileIn( __METHOD__ . '-' . $this->name );
 103                 if ( strlen( $params['content'] ) > $this->maxFileSizeInternal() ) {
 104                         $status = Status::newFatal( 'backend-fail-maxsize',
 105                                 $params['dst'], $this->maxFileSizeInternal() );
 106                 } else {
 107                         $status = $this->doCreateInternal( $params );
 108                         $this->clearCache( array( $params['dst'] ) );
 109                         $this->deleteFileCache( $params['dst'] ); // persistent cache
 110                 }
 111                 wfProfileOut( __METHOD__ . '-' . $this->name );
 112                 wfProfileOut( __METHOD__ );
 113                 return $status;
 114         }
 115
 116         /**
 117          * @see FileBackendStore::createInternal()
 118          */
 119         abstract protected function doCreateInternal( array $params );
 120
 121         /**
 122          * Store a file into the backend from a file on disk.
 123          * Do not call this function from places outside FileBackend and FileOp.
 124          *
 125          * $params include:
 126          *     src           : source path on disk
 127          *     dst           : destination storage path
 128          *     overwrite     : overwrite any file that exists at the destination
 129          *     async         : Status will be returned immediately if supported.
 130          *                     If the status is OK, then its value field will be
 131          *                     set to a FileBackendStoreOpHandle object.
 132          *
 133          * @param $params Array
 134          * @return Status
 135          */
 136         final public function storeInternal( array $params ) {
 137                 wfProfileIn( __METHOD__ );
 138                 wfProfileIn( __METHOD__ . '-' . $this->name );
 139                 if ( filesize( $params['src'] ) > $this->maxFileSizeInternal() ) {
 140                         $status = Status::newFatal( 'backend-fail-maxsize',
 141                                 $params['dst'], $this->maxFileSizeInternal() );
 142                 } else {
 143                         $status = $this->doStoreInternal( $params );
 144                         $this->clearCache( array( $params['dst'] ) );
 145                         $this->deleteFileCache( $params['dst'] ); // persistent cache
 146                 }
 147                 wfProfileOut( __METHOD__ . '-' . $this->name );
 148                 wfProfileOut( __METHOD__ );
 149                 return $status;
 150         }
 151
 152         /**
 153          * @see FileBackendStore::storeInternal()
 154          */
 155         abstract protected function doStoreInternal( array $params );
 156
 157         /**
 158          * Copy a file from one storage path to another in the backend.
 159          * Do not call this function from places outside FileBackend and FileOp.
 160          *
 161          * $params include:
 162          *     src           : source storage path
 163          *     dst           : destination storage path
 164          *     overwrite     : overwrite any file that exists at the destination
 165          *     async         : Status will be returned immediately if supported.
 166          *                     If the status is OK, then its value field will be
 167          *                     set to a FileBackendStoreOpHandle object.
 168          *
 169          * @param $params Array
 170          * @return Status
 171          */
 172         final public function copyInternal( array $params ) {
 173                 wfProfileIn( __METHOD__ );
 174                 wfProfileIn( __METHOD__ . '-' . $this->name );
 175                 $status = $this->doCopyInternal( $params );
 176                 $this->clearCache( array( $params['dst'] ) );
 177                 $this->deleteFileCache( $params['dst'] ); // persistent cache
 178                 wfProfileOut( __METHOD__ . '-' . $this->name );
 179                 wfProfileOut( __METHOD__ );
 180                 return $status;
 181         }
 182
 183         /**
 184          * @see FileBackendStore::copyInternal()
 185          */
 186         abstract protected function doCopyInternal( array $params );
 187
 188         /**
 189          * Delete a file at the storage path.
 190          * Do not call this function from places outside FileBackend and FileOp.
 191          *
 192          * $params include:
 193          *     src                 : source storage path
 194          *     ignoreMissingSource : do nothing if the source file does not exist
 195          *     async               : Status will be returned immediately if supported.
 196          *                           If the status is OK, then its value field will be
 197          *                           set to a FileBackendStoreOpHandle object.
 198          *
 199          * @param $params Array
 200          * @return Status
 201          */
 202         final public function deleteInternal( array $params ) {
 203                 wfProfileIn( __METHOD__ );
 204                 wfProfileIn( __METHOD__ . '-' . $this->name );
 205                 $status = $this->doDeleteInternal( $params );
 206                 $this->clearCache( array( $params['src'] ) );
 207                 $this->deleteFileCache( $params['src'] ); // persistent cache
 208                 wfProfileOut( __METHOD__ . '-' . $this->name );
 209                 wfProfileOut( __METHOD__ );
 210                 return $status;
 211         }
 212
 213         /**
 214          * @see FileBackendStore::deleteInternal()
 215          */
 216         abstract protected function doDeleteInternal( array $params );
 217
 218         /**
 219          * Move a file from one storage path to another in the backend.
 220          * Do not call this function from places outside FileBackend and FileOp.
 221          *
 222          * $params include:
 223          *     src           : source storage path
 224          *     dst           : destination storage path
 225          *     overwrite     : overwrite any file that exists at the destination
 226          *     async         : Status will be returned immediately if supported.
 227          *                     If the status is OK, then its value field will be
 228          *                     set to a FileBackendStoreOpHandle object.
 229          *
 230          * @param $params Array
 231          * @return Status
 232          */
 233         final public function moveInternal( array $params ) {
 234                 wfProfileIn( __METHOD__ );
 235                 wfProfileIn( __METHOD__ . '-' . $this->name );
 236                 $status = $this->doMoveInternal( $params );
 237                 $this->clearCache( array( $params['src'], $params['dst'] ) );
 238                 $this->deleteFileCache( $params['src'] ); // persistent cache
 239                 $this->deleteFileCache( $params['dst'] ); // persistent cache
 240                 wfProfileOut( __METHOD__ . '-' . $this->name );
 241                 wfProfileOut( __METHOD__ );
 242                 return $status;
 243         }
 244
 245         /**
 246          * @see FileBackendStore::moveInternal()
 247          * @return Status
 248          */
 249         protected function doMoveInternal( array $params ) {
 250                 unset( $params['async'] ); // two steps, won't work here :)
 251                 // Copy source to dest
 252                 $status = $this->copyInternal( $params );
 253                 if ( $status->isOK() ) {
 254                         // Delete source (only fails due to races or medium going down)
 255                         $status->merge( $this->deleteInternal( array( 'src' => $params['src'] ) ) );
 256                         $status->setResult( true, $status->value ); // ignore delete() errors
 257                 }
 258                 return $status;
 259         }
 260
 261         /**
 262          * No-op file operation that does nothing.
 263          * Do not call this function from places outside FileBackend and FileOp.
 264          *
 265          * @param $params Array
 266          * @return Status
 267          */
 268         final public function nullInternal( array $params ) {
 269                 return Status::newGood();
 270         }
 271
 272         /**
 273          * @see FileBackend::concatenate()
 274          * @return Status
 275          */
 276         final public function concatenate( array $params ) {
 277                 wfProfileIn( __METHOD__ );
 278                 wfProfileIn( __METHOD__ . '-' . $this->name );
 279                 $status = Status::newGood();
 280
 281                 // Try to lock the source files for the scope of this function
 282                 $scopeLockS = $this->getScopedFileLocks( $params['srcs'], LockManager::LOCK_UW, $status );
 283                 if ( $status->isOK() ) {
 284                         // Actually do the concatenation
 285                         $status->merge( $this->doConcatenate( $params ) );
 286                 }
 287
 288                 wfProfileOut( __METHOD__ . '-' . $this->name );
 289                 wfProfileOut( __METHOD__ );
 290                 return $status;
 291         }
 292
 293         /**
 294          * @see FileBackendStore::concatenate()
 295          * @return Status
 296          */
 297         protected function doConcatenate( array $params ) {
 298                 $status = Status::newGood();
 299                 $tmpPath = $params['dst']; // convenience
 300
 301                 // Check that the specified temp file is valid...
 302                 wfSuppressWarnings();
 303                 $ok = ( is_file( $tmpPath ) && !filesize( $tmpPath ) );
 304                 wfRestoreWarnings();
 305                 if ( !$ok ) { // not present or not empty
 306                         $status->fatal( 'backend-fail-opentemp', $tmpPath );
 307                         return $status;
 308                 }
 309
 310                 // Build up the temp file using the source chunks (in order)...
 311                 $tmpHandle = fopen( $tmpPath, 'ab' );
 312                 if ( $tmpHandle === false ) {
 313                         $status->fatal( 'backend-fail-opentemp', $tmpPath );
 314                         return $status;
 315                 }
 316                 foreach ( $params['srcs'] as $virtualSource ) {
 317                         // Get a local FS version of the chunk
 318                         $tmpFile = $this->getLocalReference( array( 'src' => $virtualSource ) );
 319                         if ( !$tmpFile ) {
 320                                 $status->fatal( 'backend-fail-read', $virtualSource );
 321                                 return $status;
 322                         }
 323                         // Get a handle to the local FS version
 324                         $sourceHandle = fopen( $tmpFile->getPath(), 'r' );
 325                         if ( $sourceHandle === false ) {
 326                                 fclose( $tmpHandle );
 327                                 $status->fatal( 'backend-fail-read', $virtualSource );
 328                                 return $status;
 329                         }
 330                         // Append chunk to file (pass chunk size to avoid magic quotes)
 331                         if ( !stream_copy_to_stream( $sourceHandle, $tmpHandle ) ) {
 332                                 fclose( $sourceHandle );
 333                                 fclose( $tmpHandle );
 334                                 $status->fatal( 'backend-fail-writetemp', $tmpPath );
 335                                 return $status;
 336                         }
 337                         fclose( $sourceHandle );
 338                 }
 339                 if ( !fclose( $tmpHandle ) ) {
 340                         $status->fatal( 'backend-fail-closetemp', $tmpPath );
 341                         return $status;
 342                 }
 343
 344                 clearstatcache(); // temp file changed
 345
 346                 return $status;
 347         }
 348
 349         /**
 350          * @see FileBackend::doPrepare()
 351          * @return Status
 352          */
 353         final protected function doPrepare( array $params ) {
 354                 wfProfileIn( __METHOD__ );
 355                 wfProfileIn( __METHOD__ . '-' . $this->name );
 356
 357                 $status = Status::newGood();
 358                 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
 359                 if ( $dir === null ) {
 360                         $status->fatal( 'backend-fail-invalidpath', $params['dir'] );
 361                         wfProfileOut( __METHOD__ . '-' . $this->name );
 362                         wfProfileOut( __METHOD__ );
 363                         return $status; // invalid storage path
 364                 }
 365
 366                 if ( $shard !== null ) { // confined to a single container/shard
 367                         $status->merge( $this->doPrepareInternal( $fullCont, $dir, $params ) );
 368                 } else { // directory is on several shards
 369                         wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
 370                         list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
 371                         foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
 372                                 $status->merge( $this->doPrepareInternal( "{$fullCont}{$suffix}", $dir, $params ) );
 373                         }
 374                 }
 375
 376                 wfProfileOut( __METHOD__ . '-' . $this->name );
 377                 wfProfileOut( __METHOD__ );
 378                 return $status;
 379         }
 380
 381         /**
 382          * @see FileBackendStore::doPrepare()
 383          * @return Status
 384          */
 385         protected function doPrepareInternal( $container, $dir, array $params ) {
 386                 return Status::newGood();
 387         }
 388
 389         /**
 390          * @see FileBackend::doSecure()
 391          * @return Status
 392          */
 393         final protected function doSecure( array $params ) {
 394                 wfProfileIn( __METHOD__ );
 395                 wfProfileIn( __METHOD__ . '-' . $this->name );
 396                 $status = Status::newGood();
 397
 398                 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
 399                 if ( $dir === null ) {
 400                         $status->fatal( 'backend-fail-invalidpath', $params['dir'] );
 401                         wfProfileOut( __METHOD__ . '-' . $this->name );
 402                         wfProfileOut( __METHOD__ );
 403                         return $status; // invalid storage path
 404                 }
 405
 406                 if ( $shard !== null ) { // confined to a single container/shard
 407                         $status->merge( $this->doSecureInternal( $fullCont, $dir, $params ) );
 408                 } else { // directory is on several shards
 409                         wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
 410                         list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
 411                         foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
 412                                 $status->merge( $this->doSecureInternal( "{$fullCont}{$suffix}", $dir, $params ) );
 413                         }
 414                 }
 415
 416                 wfProfileOut( __METHOD__ . '-' . $this->name );
 417                 wfProfileOut( __METHOD__ );
 418                 return $status;
 419         }
 420
 421         /**
 422          * @see FileBackendStore::doSecure()
 423          * @return Status
 424          */
 425         protected function doSecureInternal( $container, $dir, array $params ) {
 426                 return Status::newGood();
 427         }
 428
 429         /**
 430          * @see FileBackend::doClean()
 431          * @return Status
 432          */
 433         final protected function doClean( array $params ) {
 434                 wfProfileIn( __METHOD__ );
 435                 wfProfileIn( __METHOD__ . '-' . $this->name );
 436                 $status = Status::newGood();
 437
 438                 // Recursive: first delete all empty subdirs recursively
 439                 if ( !empty( $params['recursive'] ) && !$this->directoriesAreVirtual() ) {
 440                         $subDirsRel = $this->getTopDirectoryList( array( 'dir' => $params['dir'] ) );
 441                         if ( $subDirsRel !== null ) { // no errors
 442                                 foreach ( $subDirsRel as $subDirRel ) {
 443                                         $subDir = $params['dir'] . "/{$subDirRel}"; // full path
 444                                         $status->merge( $this->doClean( array( 'dir' => $subDir ) + $params ) );
 445                                 }
 446                         }
 447                 }
 448
 449                 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
 450                 if ( $dir === null ) {
 451                         $status->fatal( 'backend-fail-invalidpath', $params['dir'] );
 452                         wfProfileOut( __METHOD__ . '-' . $this->name );
 453                         wfProfileOut( __METHOD__ );
 454                         return $status; // invalid storage path
 455                 }
 456
 457                 // Attempt to lock this directory...
 458                 $filesLockEx = array( $params['dir'] );
 459                 $scopedLockE = $this->getScopedFileLocks( $filesLockEx, LockManager::LOCK_EX, $status );
 460                 if ( !$status->isOK() ) {
 461                         wfProfileOut( __METHOD__ . '-' . $this->name );
 462                         wfProfileOut( __METHOD__ );
 463                         return $status; // abort
 464                 }
 465
 466                 if ( $shard !== null ) { // confined to a single container/shard
 467                         $status->merge( $this->doCleanInternal( $fullCont, $dir, $params ) );
 468                         $this->deleteContainerCache( $fullCont ); // purge cache
 469                 } else { // directory is on several shards
 470                         wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
 471                         list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
 472                         foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
 473                                 $status->merge( $this->doCleanInternal( "{$fullCont}{$suffix}", $dir, $params ) );
 474                                 $this->deleteContainerCache( "{$fullCont}{$suffix}" ); // purge cache
 475                         }
 476                 }
 477
 478                 wfProfileOut( __METHOD__ . '-' . $this->name );
 479                 wfProfileOut( __METHOD__ );
 480                 return $status;
 481         }
 482
 483         /**
 484          * @see FileBackendStore::doClean()
 485          * @return Status
 486          */
 487         protected function doCleanInternal( $container, $dir, array $params ) {
 488                 return Status::newGood();
 489         }
 490
 491         /**
 492          * @see FileBackend::fileExists()
 493          * @return bool|null
 494          */
 495         final public function fileExists( array $params ) {
 496                 wfProfileIn( __METHOD__ );
 497                 wfProfileIn( __METHOD__ . '-' . $this->name );
 498                 $stat = $this->getFileStat( $params );
 499                 wfProfileOut( __METHOD__ . '-' . $this->name );
 500                 wfProfileOut( __METHOD__ );
 501                 return ( $stat === null ) ? null : (bool)$stat; // null => failure
 502         }
 503
 504         /**
 505          * @see FileBackend::getFileTimestamp()
 506          * @return bool
 507          */
 508         final public function getFileTimestamp( array $params ) {
 509                 wfProfileIn( __METHOD__ );
 510                 wfProfileIn( __METHOD__ . '-' . $this->name );
 511                 $stat = $this->getFileStat( $params );
 512                 wfProfileOut( __METHOD__ . '-' . $this->name );
 513                 wfProfileOut( __METHOD__ );
 514                 return $stat ? $stat['mtime'] : false;
 515         }
 516
 517         /**
 518          * @see FileBackend::getFileSize()
 519          * @return bool
 520          */
 521         final public function getFileSize( array $params ) {
 522                 wfProfileIn( __METHOD__ );
 523                 wfProfileIn( __METHOD__ . '-' . $this->name );
 524                 $stat = $this->getFileStat( $params );
 525                 wfProfileOut( __METHOD__ . '-' . $this->name );
 526                 wfProfileOut( __METHOD__ );
 527                 return $stat ? $stat['size'] : false;
 528         }
 529
 530         /**
 531          * @see FileBackend::getFileStat()
 532          * @return bool
 533          */
 534         final public function getFileStat( array $params ) {
 535                 $path = self::normalizeStoragePath( $params['src'] );
 536                 if ( $path === null ) {
 537                         return false; // invalid storage path
 538                 }
 539                 wfProfileIn( __METHOD__ );
 540                 wfProfileIn( __METHOD__ . '-' . $this->name );
 541                 $latest = !empty( $params['latest'] ); // use latest data?
 542                 if ( !isset( $this->cache[$path]['stat'] ) ) {
 543                         $this->primeFileCache( array( $path ) ); // check persistent cache
 544                 }
 545                 if ( isset( $this->cache[$path]['stat'] ) ) {
 546                         // If we want the latest data, check that this cached
 547                         // value was in fact fetched with the latest available data.
 548                         if ( !$latest || $this->cache[$path]['stat']['latest'] ) {
 549                                 $this->pingCache( $path ); // LRU
 550                                 wfProfileOut( __METHOD__ . '-' . $this->name );
 551                                 wfProfileOut( __METHOD__ );
 552                                 return $this->cache[$path]['stat'];
 553                         }
 554                 }
 555                 wfProfileIn( __METHOD__ . '-miss' );
 556                 wfProfileIn( __METHOD__ . '-miss-' . $this->name );
 557                 $stat = $this->doGetFileStat( $params );
 558                 wfProfileOut( __METHOD__ . '-miss-' . $this->name );
 559                 wfProfileOut( __METHOD__ . '-miss' );
 560                 if ( is_array( $stat ) ) { // don't cache negatives
 561                         $stat['latest'] = $latest;
 562                         $this->trimCache(); // limit memory
 563                         $this->cache[$path]['stat'] = $stat;
 564                         $this->setFileCache( $path, $stat ); // update persistent cache
 565                         if ( isset( $stat['sha1'] ) ) { // some backends store SHA-1 as metadata
 566                                 $this->trimCache(); // limit memory
 567                                 $this->cache[$path]['sha1'] =
 568                                         array( 'hash' => $stat['sha1'], 'latest' => $latest );
 569                         }
 570                 } else {
 571                         wfDebug( __METHOD__ . ": File $path does not exist.\n" );
 572                 }
 573                 wfProfileOut( __METHOD__ . '-' . $this->name );
 574                 wfProfileOut( __METHOD__ );
 575                 return $stat;
 576         }
 577
 578         /**
 579          * @see FileBackendStore::getFileStat()
 580          */
 581         abstract protected function doGetFileStat( array $params );
 582
 583         /**
 584          * @see FileBackend::getFileContents()
 585          * @return bool|string
 586          */
 587         public function getFileContents( array $params ) {
 588                 wfProfileIn( __METHOD__ );
 589                 wfProfileIn( __METHOD__ . '-' . $this->name );
 590                 $tmpFile = $this->getLocalReference( $params );
 591                 if ( !$tmpFile ) {
 592                         wfProfileOut( __METHOD__ . '-' . $this->name );
 593                         wfProfileOut( __METHOD__ );
 594                         return false;
 595                 }
 596                 wfSuppressWarnings();
 597                 $data = file_get_contents( $tmpFile->getPath() );
 598                 wfRestoreWarnings();
 599                 wfProfileOut( __METHOD__ . '-' . $this->name );
 600                 wfProfileOut( __METHOD__ );
 601                 return $data;
 602         }
 603
 604         /**
 605          * @see FileBackend::getFileSha1Base36()
 606          * @return bool|string
 607          */
 608         final public function getFileSha1Base36( array $params ) {
 609                 $path = self::normalizeStoragePath( $params['src'] );
 610                 if ( $path === null ) {
 611                         return false; // invalid storage path
 612                 }
 613                 wfProfileIn( __METHOD__ );
 614                 wfProfileIn( __METHOD__ . '-' . $this->name );
 615                 $latest = !empty( $params['latest'] ); // use latest data?
 616                 if ( isset( $this->cache[$path]['sha1'] ) ) {
 617                         // If we want the latest data, check that this cached
 618                         // value was in fact fetched with the latest available data.
 619                         if ( !$latest || $this->cache[$path]['sha1']['latest'] ) {
 620                                 $this->pingCache( $path ); // LRU
 621                                 wfProfileOut( __METHOD__ . '-' . $this->name );
 622                                 wfProfileOut( __METHOD__ );
 623                                 return $this->cache[$path]['sha1']['hash'];
 624                         }
 625                 }
 626                 wfProfileIn( __METHOD__ . '-miss' );
 627                 wfProfileIn( __METHOD__ . '-miss-' . $this->name );
 628                 $hash = $this->doGetFileSha1Base36( $params );
 629                 wfProfileOut( __METHOD__ . '-miss-' . $this->name );
 630                 wfProfileOut( __METHOD__ . '-miss' );
 631                 if ( $hash ) { // don't cache negatives
 632                         $this->trimCache(); // limit memory
 633                         $this->cache[$path]['sha1'] = array( 'hash' => $hash, 'latest' => $latest );
 634                 }
 635                 wfProfileOut( __METHOD__ . '-' . $this->name );
 636                 wfProfileOut( __METHOD__ );
 637                 return $hash;
 638         }
 639
 640         /**
 641          * @see FileBackendStore::getFileSha1Base36()
 642          * @return bool|string
 643          */
 644         protected function doGetFileSha1Base36( array $params ) {
 645                 $fsFile = $this->getLocalReference( $params );
 646                 if ( !$fsFile ) {
 647                         return false;
 648                 } else {
 649                         return $fsFile->getSha1Base36();
 650                 }
 651         }
 652
 653         /**
 654          * @see FileBackend::getFileProps()
 655          * @return Array
 656          */
 657         final public function getFileProps( array $params ) {
 658                 wfProfileIn( __METHOD__ );
 659                 wfProfileIn( __METHOD__ . '-' . $this->name );
 660                 $fsFile = $this->getLocalReference( $params );
 661                 $props = $fsFile ? $fsFile->getProps() : FSFile::placeholderProps();
 662                 wfProfileOut( __METHOD__ . '-' . $this->name );
 663                 wfProfileOut( __METHOD__ );
 664                 return $props;
 665         }
 666
 667         /**
 668          * @see FileBackend::getLocalReference()
 669          * @return TempFSFile|null
 670          */
 671         public function getLocalReference( array $params ) {
 672                 $path = self::normalizeStoragePath( $params['src'] );
 673                 if ( $path === null ) {
 674                         return null; // invalid storage path
 675                 }
 676                 wfProfileIn( __METHOD__ );
 677                 wfProfileIn( __METHOD__ . '-' . $this->name );
 678                 $latest = !empty( $params['latest'] ); // use latest data?
 679                 if ( isset( $this->expensiveCache[$path]['localRef'] ) ) {
 680                         // If we want the latest data, check that this cached
 681                         // value was in fact fetched with the latest available data.
 682                         if ( !$latest || $this->expensiveCache[$path]['localRef']['latest'] ) {
 683                                 $this->pingExpensiveCache( $path );
 684                                 wfProfileOut( __METHOD__ . '-' . $this->name );
 685                                 wfProfileOut( __METHOD__ );
 686                                 return $this->expensiveCache[$path]['localRef']['object'];
 687                         }
 688                 }
 689                 $tmpFile = $this->getLocalCopy( $params );
 690                 if ( $tmpFile ) { // don't cache negatives
 691                         $this->trimExpensiveCache(); // limit memory
 692                         $this->expensiveCache[$path]['localRef'] =
 693                                 array( 'object' => $tmpFile, 'latest' => $latest );
 694                 }
 695                 wfProfileOut( __METHOD__ . '-' . $this->name );
 696                 wfProfileOut( __METHOD__ );
 697                 return $tmpFile;
 698         }
 699
 700         /**
 701          * @see FileBackend::streamFile()
 702          * @return Status
 703          */
 704         final public function streamFile( array $params ) {
 705                 wfProfileIn( __METHOD__ );
 706                 wfProfileIn( __METHOD__ . '-' . $this->name );
 707                 $status = Status::newGood();
 708
 709                 $info = $this->getFileStat( $params );
 710                 if ( !$info ) { // let StreamFile handle the 404
 711                         $status->fatal( 'backend-fail-notexists', $params['src'] );
 712                 }
 713
 714                 // Set output buffer and HTTP headers for stream
 715                 $extraHeaders = isset( $params['headers'] ) ? $params['headers'] : array();
 716                 $res = StreamFile::prepareForStream( $params['src'], $info, $extraHeaders );
 717                 if ( $res == StreamFile::NOT_MODIFIED ) {
 718                         // do nothing; client cache is up to date
 719                 } elseif ( $res == StreamFile::READY_STREAM ) {
 720                         wfProfileIn( __METHOD__ . '-send' );
 721                         wfProfileIn( __METHOD__ . '-send-' . $this->name );
 722                         $status = $this->doStreamFile( $params );
 723                         wfProfileOut( __METHOD__ . '-send-' . $this->name );
 724                         wfProfileOut( __METHOD__ . '-send' );
 725                 } else {
 726                         $status->fatal( 'backend-fail-stream', $params['src'] );
 727                 }
 728
 729                 wfProfileOut( __METHOD__ . '-' . $this->name );
 730                 wfProfileOut( __METHOD__ );
 731                 return $status;
 732         }
 733
 734         /**
 735          * @see FileBackendStore::streamFile()
 736          * @return Status
 737          */
 738         protected function doStreamFile( array $params ) {
 739                 $status = Status::newGood();
 740
 741                 $fsFile = $this->getLocalReference( $params );
 742                 if ( !$fsFile ) {
 743                         $status->fatal( 'backend-fail-stream', $params['src'] );
 744                 } elseif ( !readfile( $fsFile->getPath() ) ) {
 745                         $status->fatal( 'backend-fail-stream', $params['src'] );
 746                 }
 747
 748                 return $status;
 749         }
 750
 751         /**
 752          * @see FileBackend::directoryExists()
 753          * @return bool|null
 754          */
 755         final public function directoryExists( array $params ) {
 756                 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
 757                 if ( $dir === null ) {
 758                         return false; // invalid storage path
 759                 }
 760                 if ( $shard !== null ) { // confined to a single container/shard
 761                         return $this->doDirectoryExists( $fullCont, $dir, $params );
 762                 } else { // directory is on several shards
 763                         wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
 764                         list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
 765                         $res = false; // response
 766                         foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
 767                                 $exists = $this->doDirectoryExists( "{$fullCont}{$suffix}", $dir, $params );
 768                                 if ( $exists ) {
 769                                         $res = true;
 770                                         break; // found one!
 771                                 } elseif ( $exists === null ) { // error?
 772                                         $res = null; // if we don't find anything, it is indeterminate
 773                                 }
 774                         }
 775                         return $res;
 776                 }
 777         }
 778
 779         /**
 780          * @see FileBackendStore::directoryExists()
 781          *
 782          * @param $container string Resolved container name
 783          * @param $dir string Resolved path relative to container
 784          * @param $params Array
 785          * @return bool|null
 786          */
 787         abstract protected function doDirectoryExists( $container, $dir, array $params );
 788
 789         /**
 790          * @see FileBackend::getDirectoryList()
 791          * @return Traversable|Array|null Returns null on failure
 792          */
 793         final public function getDirectoryList( array $params ) {
 794                 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
 795                 if ( $dir === null ) { // invalid storage path
 796                         return null;
 797                 }
 798                 if ( $shard !== null ) {
 799                         // File listing is confined to a single container/shard
 800                         return $this->getDirectoryListInternal( $fullCont, $dir, $params );
 801                 } else {
 802                         wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
 803                         // File listing spans multiple containers/shards
 804                         list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
 805                         return new FileBackendStoreShardDirIterator( $this,
 806                                 $fullCont, $dir, $this->getContainerSuffixes( $shortCont ), $params );
 807                 }
 808         }
 809
 810         /**
 811          * Do not call this function from places outside FileBackend
 812          *
 813          * @see FileBackendStore::getDirectoryList()
 814          *
 815          * @param $container string Resolved container name
 816          * @param $dir string Resolved path relative to container
 817          * @param $params Array
 818          * @return Traversable|Array|null Returns null on failure
 819          */
 820         abstract public function getDirectoryListInternal( $container, $dir, array $params );
 821
 822         /**
 823          * @see FileBackend::getFileList()
 824          * @return Traversable|Array|null Returns null on failure
 825          */
 826         final public function getFileList( array $params ) {
 827                 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
 828                 if ( $dir === null ) { // invalid storage path
 829                         return null;
 830                 }
 831                 if ( $shard !== null ) {
 832                         // File listing is confined to a single container/shard
 833                         return $this->getFileListInternal( $fullCont, $dir, $params );
 834                 } else {
 835                         wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
 836                         // File listing spans multiple containers/shards
 837                         list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
 838                         return new FileBackendStoreShardFileIterator( $this,
 839                                 $fullCont, $dir, $this->getContainerSuffixes( $shortCont ), $params );
 840                 }
 841         }
 842
 843         /**
 844          * Do not call this function from places outside FileBackend
 845          *
 846          * @see FileBackendStore::getFileList()
 847          *
 848          * @param $container string Resolved container name
 849          * @param $dir string Resolved path relative to container
 850          * @param $params Array
 851          * @return Traversable|Array|null Returns null on failure
 852          */
 853         abstract public function getFileListInternal( $container, $dir, array $params );
 854
 855         /**
 856          * Return a list of FileOp objects from a list of operations.
 857          * Do not call this function from places outside FileBackend.
 858          *
 859          * The result must have the same number of items as the input.
 860          * An exception is thrown if an unsupported operation is requested.
 861          *
 862          * @param $ops Array Same format as doOperations()
 863          * @return Array List of FileOp objects
 864          * @throws MWException
 865          */
 866         final public function getOperationsInternal( array $ops ) {
 867                 $supportedOps = array(
 868                         'store'       => 'StoreFileOp',
 869                         'copy'        => 'CopyFileOp',
 870                         'move'        => 'MoveFileOp',
 871                         'delete'      => 'DeleteFileOp',
 872                         'create'      => 'CreateFileOp',
 873                         'null'        => 'NullFileOp'
 874                 );
 875
 876                 $performOps = array(); // array of FileOp objects
 877                 // Build up ordered array of FileOps...
 878                 foreach ( $ops as $operation ) {
 879                         $opName = $operation['op'];
 880                         if ( isset( $supportedOps[$opName] ) ) {
 881                                 $class = $supportedOps[$opName];
 882                                 // Get params for this operation
 883                                 $params = $operation;
 884                                 // Append the FileOp class
 885                                 $performOps[] = new $class( $this, $params );
 886                         } else {
 887                                 throw new MWException( "Operation '$opName' is not supported." );
 888                         }
 889                 }
 890
 891                 return $performOps;
 892         }
 893
 894         /**
 895          * Get a list of storage paths to lock for a list of operations
 896          * Returns an array with 'sh' (shared) and 'ex' (exclusive) keys,
 897          * each corresponding to a list of storage paths to be locked.
 898          *
 899          * @param $performOps Array List of FileOp objects
 900          * @return Array ('sh' => list of paths, 'ex' => list of paths)
 901          */
 902         final public function getPathsToLockForOpsInternal( array $performOps ) {
 903                 // Build up a list of files to lock...
 904                 $paths = array( 'sh' => array(), 'ex' => array() );
 905                 foreach ( $performOps as $fileOp ) {
 906                         $paths['sh'] = array_merge( $paths['sh'], $fileOp->storagePathsRead() );
 907                         $paths['ex'] = array_merge( $paths['ex'], $fileOp->storagePathsChanged() );
 908                 }
 909                 // Optimization: if doing an EX lock anyway, don't also set an SH one
 910                 $paths['sh'] = array_diff( $paths['sh'], $paths['ex'] );
 911                 // Get a shared lock on the parent directory of each path changed
 912                 $paths['sh'] = array_merge( $paths['sh'], array_map( 'dirname', $paths['ex'] ) );
 913
 914                 return $paths;
 915         }
 916
 917         /**
 918          * @see FileBackend::getScopedLocksForOps()
 919          * @return Array
 920          */
 921         public function getScopedLocksForOps( array $ops, Status $status ) {
 922                 $paths = $this->getPathsToLockForOpsInternal( $this->getOperationsInternal( $ops ) );
 923                 return array(
 924                         $this->getScopedFileLocks( $paths['sh'], LockManager::LOCK_UW, $status ),
 925                         $this->getScopedFileLocks( $paths['ex'], LockManager::LOCK_EX, $status )
 926                 );
 927         }
 928
 929         /**
 930          * @see FileBackend::doOperationsInternal()
 931          * @return Status
 932          */
 933         protected function doOperationsInternal( array $ops, array $opts ) {
 934                 wfProfileIn( __METHOD__ );
 935                 wfProfileIn( __METHOD__ . '-' . $this->name );
 936                 $status = Status::newGood();
 937
 938                 // Build up a list of FileOps...
 939                 $performOps = $this->getOperationsInternal( $ops );
 940
 941                 // Acquire any locks as needed...
 942                 if ( empty( $opts['nonLocking'] ) ) {
 943                         // Build up a list of files to lock...
 944                         $paths = $this->getPathsToLockForOpsInternal( $performOps );
 945                         // Try to lock those files for the scope of this function...
 946                         $scopeLockS = $this->getScopedFileLocks( $paths['sh'], LockManager::LOCK_UW, $status );
 947                         $scopeLockE = $this->getScopedFileLocks( $paths['ex'], LockManager::LOCK_EX, $status );
 948                         if ( !$status->isOK() ) {
 949                                 wfProfileOut( __METHOD__ . '-' . $this->name );
 950                                 wfProfileOut( __METHOD__ );
 951                                 return $status; // abort
 952                         }
 953                 }
 954
 955                 // Clear any file cache entries (after locks acquired)
 956                 $this->clearCache();
 957
 958                 // Load from the persistent file and container caches
 959                 $this->primeFileCache( $performOps );
 960                 $this->primeContainerCache( $performOps );
 961
 962                 // Actually attempt the operation batch...
 963                 $subStatus = FileOpBatch::attempt( $performOps, $opts, $this->fileJournal );
 964
 965                 // Merge errors into status fields
 966                 $status->merge( $subStatus );
 967                 $status->success = $subStatus->success; // not done in merge()
 968
 969                 wfProfileOut( __METHOD__ . '-' . $this->name );
 970                 wfProfileOut( __METHOD__ );
 971                 return $status;
 972         }
 973
 974         /**
 975          * @see FileBackend::doQuickOperationsInternal()
 976          * @return Status
 977          * @throws MWException
 978          */
 979         final protected function doQuickOperationsInternal( array $ops ) {
 980                 wfProfileIn( __METHOD__ );
 981                 wfProfileIn( __METHOD__ . '-' . $this->name );
 982                 $status = Status::newGood();
 983
 984                 $supportedOps = array( 'create', 'store', 'copy', 'move', 'delete', 'null' );
 985                 $async = ( $this->parallelize === 'implicit' );
 986                 $maxConcurrency = $this->concurrency; // throttle
 987
 988                 $statuses = array(); // array of (index => Status)
 989                 $fileOpHandles = array(); // list of (index => handle) arrays
 990                 $curFileOpHandles = array(); // current handle batch
 991                 // Perform the sync-only ops and build up op handles for the async ops...
 992                 foreach ( $ops as $index => $params ) {
 993                         if ( !in_array( $params['op'], $supportedOps ) ) {
 994                                 wfProfileOut( __METHOD__ . '-' . $this->name );
 995                                 wfProfileOut( __METHOD__ );
 996                                 throw new MWException( "Operation '{$params['op']}' is not supported." );
 997                         }
 998                         $method = $params['op'] . 'Internal'; // e.g. "storeInternal"
 999                         $subStatus = $this->$method( array( 'async' => $async ) + $params );
1000                         if ( $subStatus->value instanceof FileBackendStoreOpHandle ) { // async
1001                                 if ( count( $curFileOpHandles ) >= $maxConcurrency ) {
1002                                         $fileOpHandles[] = $curFileOpHandles; // push this batch
1003                                         $curFileOpHandles = array();
1004                                 }
1005                                 $curFileOpHandles[$index] = $subStatus->value; // keep index
1006                         } else { // error or completed
1007                                 $statuses[$index] = $subStatus; // keep index
1008                         }
1009                 }
1010                 if ( count( $curFileOpHandles ) ) {
1011                         $fileOpHandles[] = $curFileOpHandles; // last batch
1012                 }
1013                 // Do all the async ops that can be done concurrently...
1014                 foreach ( $fileOpHandles as $fileHandleBatch ) {
1015                         $statuses = $statuses + $this->executeOpHandlesInternal( $fileHandleBatch );
1016                 }
1017                 // Marshall and merge all the responses...
1018                 foreach ( $statuses as $index => $subStatus ) {
1019                         $status->merge( $subStatus );
1020                         if ( $subStatus->isOK() ) {
1021                                 $status->success[$index] = true;
1022                                 ++$status->successCount;
1023                         } else {
1024                                 $status->success[$index] = false;
1025                                 ++$status->failCount;
1026                         }
1027                 }
1028
1029                 wfProfileOut( __METHOD__ . '-' . $this->name );
1030                 wfProfileOut( __METHOD__ );
1031                 return $status;
1032         }
1033
1034         /**
1035          * Execute a list of FileBackendStoreOpHandle handles in parallel.
1036          * The resulting Status object fields will correspond
1037          * to the order in which the handles where given.
1038          *
1039          * @param $handles Array List of FileBackendStoreOpHandle objects
1040          * @return Array Map of Status objects
1041          * @throws MWException
1042          */
1043         final public function executeOpHandlesInternal( array $fileOpHandles ) {
1044                 wfProfileIn( __METHOD__ );
1045                 wfProfileIn( __METHOD__ . '-' . $this->name );
1046                 foreach ( $fileOpHandles as $fileOpHandle ) {
1047                         if ( !( $fileOpHandle instanceof FileBackendStoreOpHandle ) ) {
1048                                 throw new MWException( "Given a non-FileBackendStoreOpHandle object." );
1049                         } elseif ( $fileOpHandle->backend->getName() !== $this->getName() ) {
1050                                 throw new MWException( "Given a FileBackendStoreOpHandle for the wrong backend." );
1051                         }
1052                 }
1053                 $res = $this->doExecuteOpHandlesInternal( $fileOpHandles );
1054                 foreach ( $fileOpHandles as $fileOpHandle ) {
1055                         $fileOpHandle->closeResources();
1056                 }
1057                 wfProfileOut( __METHOD__ . '-' . $this->name );
1058                 wfProfileOut( __METHOD__ );
1059                 return $res;
1060         }
1061
1062         /**
1063          * @see FileBackendStore::executeOpHandlesInternal()
1064          * @return Array List of corresponding Status objects
1065          */
1066         protected function doExecuteOpHandlesInternal( array $fileOpHandles ) {
1067                 foreach ( $fileOpHandles as $fileOpHandle ) { // OK if empty
1068                         throw new MWException( "This backend supports no asynchronous operations." );
1069                 }
1070                 return array();
1071         }
1072
1073         /**
1074          * @see FileBackend::clearCache()
1075          */
1076         final public function clearCache( array $paths = null ) {
1077                 if ( is_array( $paths ) ) {
1078                         $paths = array_map( 'FileBackend::normalizeStoragePath', $paths );
1079                         $paths = array_filter( $paths, 'strlen' ); // remove nulls
1080                 }
1081                 if ( $paths === null ) {
1082                         $this->cache = array();
1083                         $this->expensiveCache = array();
1084                 } else {
1085                         foreach ( $paths as $path ) {
1086                                 unset( $this->cache[$path] );
1087                                 unset( $this->expensiveCache[$path] );
1088                         }
1089                 }
1090                 $this->doClearCache( $paths );
1091         }
1092
1093         /**
1094          * Clears any additional stat caches for storage paths
1095          *
1096          * @see FileBackend::clearCache()
1097          *
1098          * @param $paths Array Storage paths (optional)
1099          * @return void
1100          */
1101         protected function doClearCache( array $paths = null ) {}
1102
1103         /**
1104          * Is this a key/value store where directories are just virtual?
1105          * Virtual directories exists in so much as files exists that are
1106          * prefixed with the directory path followed by a forward slash.
1107          *
1108          * @return bool
1109          */
1110         abstract protected function directoriesAreVirtual();
1111
1112         /**
1113          * Move a cache entry to the top (such as when accessed)
1114          *
1115          * @param $path string Storage path
1116          * @return void
1117          */
1118         protected function pingCache( $path ) {
1119                 if ( isset( $this->cache[$path] ) ) {
1120                         $tmp = $this->cache[$path];
1121                         unset( $this->cache[$path] );
1122                         $this->cache[$path] = $tmp;
1123                 }
1124         }
1125
1126         /**
1127          * Prune the inexpensive cache if it is too big to add an item
1128          *
1129          * @return void
1130          */
1131         protected function trimCache() {
1132                 if ( count( $this->cache ) >= $this->maxCacheSize ) {
1133                         reset( $this->cache );
1134                         unset( $this->cache[key( $this->cache )] );
1135                 }
1136         }
1137
1138         /**
1139          * Move a cache entry to the top (such as when accessed)
1140          *
1141          * @param $path string Storage path
1142          * @return void
1143          */
1144         protected function pingExpensiveCache( $path ) {
1145                 if ( isset( $this->expensiveCache[$path] ) ) {
1146                         $tmp = $this->expensiveCache[$path];
1147                         unset( $this->expensiveCache[$path] );
1148                         $this->expensiveCache[$path] = $tmp;
1149                 }
1150         }
1151
1152         /**
1153          * Prune the expensive cache if it is too big to add an item
1154          *
1155          * @return void
1156          */
1157         protected function trimExpensiveCache() {
1158                 if ( count( $this->expensiveCache ) >= $this->maxExpensiveCacheSize ) {
1159                         reset( $this->expensiveCache );
1160                         unset( $this->expensiveCache[key( $this->expensiveCache )] );
1161                 }
1162         }
1163
1164         /**
1165          * Check if a container name is valid.
1166          * This checks for for length and illegal characters.
1167          *
1168          * @param $container string
1169          * @return bool
1170          */
1171         final protected static function isValidContainerName( $container ) {
1172                 // This accounts for Swift and S3 restrictions while leaving room
1173                 // for things like '.xxx' (hex shard chars) or '.seg' (segments).
1174                 // This disallows directory separators or traversal characters.
1175                 // Note that matching strings URL encode to the same string;
1176                 // in Swift, the length restriction is *after* URL encoding.
1177                 return preg_match( '/^[a-z0-9][a-z0-9-_]{0,199}$/i', $container );
1178         }
1179
1180         /**
1181          * Splits a storage path into an internal container name,
1182          * an internal relative file name, and a container shard suffix.
1183          * Any shard suffix is already appended to the internal container name.
1184          * This also checks that the storage path is valid and within this backend.
1185          *
1186          * If the container is sharded but a suffix could not be determined,
1187          * this means that the path can only refer to a directory and can only
1188          * be scanned by looking in all the container shards.
1189          *
1190          * @param $storagePath string
1191          * @return Array (container, path, container suffix) or (null, null, null) if invalid
1192          */
1193         final protected function resolveStoragePath( $storagePath ) {
1194                 list( $backend, $container, $relPath ) = self::splitStoragePath( $storagePath );
1195                 if ( $backend === $this->name ) { // must be for this backend
1196                         $relPath = self::normalizeContainerPath( $relPath );
1197                         if ( $relPath !== null ) {
1198                                 // Get shard for the normalized path if this container is sharded
1199                                 $cShard = $this->getContainerShard( $container, $relPath );
1200                                 // Validate and sanitize the relative path (backend-specific)
1201                                 $relPath = $this->resolveContainerPath( $container, $relPath );
1202                                 if ( $relPath !== null ) {
1203                                         // Prepend any wiki ID prefix to the container name
1204                                         $container = $this->fullContainerName( $container );
1205                                         if ( self::isValidContainerName( $container ) ) {
1206                                                 // Validate and sanitize the container name (backend-specific)
1207                                                 $container = $this->resolveContainerName( "{$container}{$cShard}" );
1208                                                 if ( $container !== null ) {
1209                                                         return array( $container, $relPath, $cShard );
1210                                                 }
1211                                         }
1212                                 }
1213                         }
1214                 }
1215                 return array( null, null, null );
1216         }
1217
1218         /**
1219          * Like resolveStoragePath() except null values are returned if
1220          * the container is sharded and the shard could not be determined.
1221          *
1222          * @see FileBackendStore::resolveStoragePath()
1223          *
1224          * @param $storagePath string
1225          * @return Array (container, path) or (null, null) if invalid
1226          */
1227         final protected function resolveStoragePathReal( $storagePath ) {
1228                 list( $container, $relPath, $cShard ) = $this->resolveStoragePath( $storagePath );
1229                 if ( $cShard !== null ) {
1230                         return array( $container, $relPath );
1231                 }
1232                 return array( null, null );
1233         }
1234
1235         /**
1236          * Get the container name shard suffix for a given path.
1237          * Any empty suffix means the container is not sharded.
1238          *
1239          * @param $container string Container name
1240          * @param $relPath string Storage path relative to the container
1241          * @return string|null Returns null if shard could not be determined
1242          */
1243         final protected function getContainerShard( $container, $relPath ) {
1244                 list( $levels, $base, $repeat ) = $this->getContainerHashLevels( $container );
1245                 if ( $levels == 1 || $levels == 2 ) {
1246                         // Hash characters are either base 16 or 36
1247                         $char = ( $base == 36 ) ? '[0-9a-z]' : '[0-9a-f]';
1248                         // Get a regex that represents the shard portion of paths.
1249                         // The concatenation of the captures gives us the shard.
1250                         if ( $levels === 1 ) { // 16 or 36 shards per container
1251                                 $hashDirRegex = '(' . $char . ')';
1252                         } else { // 256 or 1296 shards per container
1253                                 if ( $repeat ) { // verbose hash dir format (e.g. "a/ab/abc")
1254                                         $hashDirRegex = $char . '/(' . $char . '{2})';
1255                                 } else { // short hash dir format (e.g. "a/b/c")
1256                                         $hashDirRegex = '(' . $char . ')/(' . $char . ')';
1257                                 }
1258                         }
1259                         // Allow certain directories to be above the hash dirs so as
1260                         // to work with FileRepo (e.g. "archive/a/ab" or "temp/a/ab").
1261                         // They must be 2+ chars to avoid any hash directory ambiguity.
1262                         $m = array();
1263                         if ( preg_match( "!^(?:[^/]{2,}/)*$hashDirRegex(?:/|$)!", $relPath, $m ) ) {
1264                                 return '.' . implode( '', array_slice( $m, 1 ) );
1265                         }
1266                         return null; // failed to match
1267                 }
1268                 return ''; // no sharding
1269         }
1270
1271         /**
1272          * Check if a storage path maps to a single shard.
1273          * Container dirs like "a", where the container shards on "x/xy",
1274          * can reside on several shards. Such paths are tricky to handle.
1275          *
1276          * @param $storagePath string Storage path
1277          * @return bool
1278          */
1279         final public function isSingleShardPathInternal( $storagePath ) {
1280                 list( $c, $r, $shard ) = $this->resolveStoragePath( $storagePath );
1281                 return ( $shard !== null );
1282         }
1283
1284         /**
1285          * Get the sharding config for a container.
1286          * If greater than 0, then all file storage paths within
1287          * the container are required to be hashed accordingly.
1288          *
1289          * @param $container string
1290          * @return Array (integer levels, integer base, repeat flag) or (0, 0, false)
1291          */
1292         final protected function getContainerHashLevels( $container ) {
1293                 if ( isset( $this->shardViaHashLevels[$container] ) ) {
1294                         $config = $this->shardViaHashLevels[$container];
1295                         $hashLevels = (int)$config['levels'];
1296                         if ( $hashLevels == 1 || $hashLevels == 2 ) {
1297                                 $hashBase = (int)$config['base'];
1298                                 if ( $hashBase == 16 || $hashBase == 36 ) {
1299                                         return array( $hashLevels, $hashBase, $config['repeat'] );
1300                                 }
1301                         }
1302                 }
1303                 return array( 0, 0, false ); // no sharding
1304         }
1305
1306         /**
1307          * Get a list of full container shard suffixes for a container
1308          *
1309          * @param $container string
1310          * @return Array
1311          */
1312         final protected function getContainerSuffixes( $container ) {
1313                 $shards = array();
1314                 list( $digits, $base ) = $this->getContainerHashLevels( $container );
1315                 if ( $digits > 0 ) {
1316                         $numShards = pow( $base, $digits );
1317                         for ( $index = 0; $index < $numShards; $index++ ) {
1318                                 $shards[] = '.' . wfBaseConvert( $index, 10, $base, $digits );
1319                         }
1320                 }
1321                 return $shards;
1322         }
1323
1324         /**
1325          * Get the full container name, including the wiki ID prefix
1326          *
1327          * @param $container string
1328          * @return string
1329          */
1330         final protected function fullContainerName( $container ) {
1331                 if ( $this->wikiId != '' ) {
1332                         return "{$this->wikiId}-$container";
1333                 } else {
1334                         return $container;
1335                 }
1336         }
1337
1338         /**
1339          * Resolve a container name, checking if it's allowed by the backend.
1340          * This is intended for internal use, such as encoding illegal chars.
1341          * Subclasses can override this to be more restrictive.
1342          *
1343          * @param $container string
1344          * @return string|null
1345          */
1346         protected function resolveContainerName( $container ) {
1347                 return $container;
1348         }
1349
1350         /**
1351          * Resolve a relative storage path, checking if it's allowed by the backend.
1352          * This is intended for internal use, such as encoding illegal chars or perhaps
1353          * getting absolute paths (e.g. FS based backends). Note that the relative path
1354          * may be the empty string (e.g. the path is simply to the container).
1355          *
1356          * @param $container string Container name
1357          * @param $relStoragePath string Storage path relative to the container
1358          * @return string|null Path or null if not valid
1359          */
1360         protected function resolveContainerPath( $container, $relStoragePath ) {
1361                 return $relStoragePath;
1362         }
1363
1364         /**
1365          * Get the cache key for a container
1366          *
1367          * @param $container string Resolved container name
1368          * @return string
1369          */
1370         private function containerCacheKey( $container ) {
1371                 return wfMemcKey( 'backend', $this->getName(), 'container', $container );
1372         }
1373
1374         /**
1375          * Set the cached info for a container
1376          *
1377          * @param $container string Resolved container name
1378          * @param $val mixed Information to cache
1379          */
1380         final protected function setContainerCache( $container, $val ) {
1381                 $this->memCache->add( $this->containerCacheKey( $container ), $val, 14*86400 );
1382         }
1383
1384         /**
1385          * Delete the cached info for a container.
1386          * The cache key is salted for a while to prevent race conditions.
1387          *
1388          * @param $container string Resolved container name
1389          */
1390         final protected function deleteContainerCache( $container ) {
1391                 if ( !$this->memCache->set( $this->containerCacheKey( $container ), 'PURGED', 300 ) ) {
1392                         trigger_error( "Unable to delete stat cache for container $container." );
1393                 }
1394         }
1395
1396         /**
1397          * Do a batch lookup from cache for container stats for all containers
1398          * used in a list of container names, storage paths, or FileOp objects.
1399          *
1400          * @param $items Array
1401          * @return void
1402          */
1403         final protected function primeContainerCache( array $items ) {
1404                 wfProfileIn( __METHOD__ );
1405                 wfProfileIn( __METHOD__ . '-' . $this->name );
1406
1407                 $paths = array(); // list of storage paths
1408                 $contNames = array(); // (cache key => resolved container name)
1409                 // Get all the paths/containers from the items...
1410                 foreach ( $items as $item ) {
1411                         if ( $item instanceof FileOp ) {
1412                                 $paths = array_merge( $paths, $item->storagePathsRead() );
1413                                 $paths = array_merge( $paths, $item->storagePathsChanged() );
1414                         } elseif ( self::isStoragePath( $item ) ) {
1415                                 $paths[] = $item;
1416                         } elseif ( is_string( $item ) ) { // full container name
1417                                 $contNames[$this->containerCacheKey( $item )] = $item;
1418                         }
1419                 }
1420                 // Get all the corresponding cache keys for paths...
1421                 foreach ( $paths as $path ) {
1422                         list( $fullCont, $r, $s ) = $this->resolveStoragePath( $path );
1423                         if ( $fullCont !== null ) { // valid path for this backend
1424                                 $contNames[$this->containerCacheKey( $fullCont )] = $fullCont;
1425                         }
1426                 }
1427
1428                 $contInfo = array(); // (resolved container name => cache value)
1429                 // Get all cache entries for these container cache keys...
1430                 $values = $this->memCache->getMulti( array_keys( $contNames ) );
1431                 foreach ( $values as $cacheKey => $val ) {
1432                         $contInfo[$contNames[$cacheKey]] = $val;
1433                 }
1434
1435                 // Populate the container process cache for the backend...
1436                 $this->doPrimeContainerCache( array_filter( $contInfo, 'is_array' ) );
1437
1438                 wfProfileOut( __METHOD__ . '-' . $this->name );
1439                 wfProfileOut( __METHOD__ );
1440         }
1441
1442         /**
1443          * Fill the backend-specific process cache given an array of
1444          * resolved container names and their corresponding cached info.
1445          * Only containers that actually exist should appear in the map.
1446          *
1447          * @param $containerInfo Array Map of resolved container names to cached info
1448          * @return void
1449          */
1450         protected function doPrimeContainerCache( array $containerInfo ) {}
1451
1452         /**
1453          * Get the cache key for a file path
1454          *
1455          * @param $path string Storage path
1456          * @return string
1457          */
1458         private function fileCacheKey( $path ) {
1459                 return wfMemcKey( 'backend', $this->getName(), 'file', sha1( $path ) );
1460         }
1461
1462         /**
1463          * Set the cached stat info for a file path
1464          *
1465          * @param $path string Storage path
1466          * @param $val mixed Information to cache
1467          */
1468         final protected function setFileCache( $path, $val ) {
1469                 $this->memCache->add( $this->fileCacheKey( $path ), $val, 7*86400 );
1470         }
1471
1472         /**
1473          * Delete the cached stat info for a file path.
1474          * The cache key is salted for a while to prevent race conditions.
1475          *
1476          * @param $path string Storage path
1477          */
1478         final protected function deleteFileCache( $path ) {
1479                 if ( !$this->memCache->set( $this->fileCacheKey( $path ), 'PURGED', 300 ) ) {
1480                         trigger_error( "Unable to delete stat cache for file $path." );
1481                 }
1482         }
1483
1484         /**
1485          * Do a batch lookup from cache for file stats for all paths
1486          * used in a list of storage paths or FileOp objects.
1487          *
1488          * @param $items Array List of storage paths or FileOps
1489          * @return void
1490          */
1491         final protected function primeFileCache( array $items ) {
1492                 wfProfileIn( __METHOD__ );
1493                 wfProfileIn( __METHOD__ . '-' . $this->name );
1494
1495                 $paths = array(); // list of storage paths
1496                 $pathNames = array(); // (cache key => storage path)
1497                 // Get all the paths/containers from the items...
1498                 foreach ( $items as $item ) {
1499                         if ( $item instanceof FileOp ) {
1500                                 $paths = array_merge( $paths, $item->storagePathsRead() );
1501                                 $paths = array_merge( $paths, $item->storagePathsChanged() );
1502                         } elseif ( self::isStoragePath( $item ) ) {
1503                                 $paths[] = $item;
1504                         }
1505                 }
1506                 // Get all the corresponding cache keys for paths...
1507                 foreach ( $paths as $path ) {
1508                         list( $cont, $rel, $s ) = $this->resolveStoragePath( $path );
1509                         if ( $rel !== null ) { // valid path for this backend
1510                                 $pathNames[$this->fileCacheKey( $path )] = $path;
1511                         }
1512                 }
1513                 // Get all cache entries for these container cache keys...
1514                 $values = $this->memCache->getMulti( array_keys( $pathNames ) );
1515                 foreach ( $values as $cacheKey => $val ) {
1516                         if ( is_array( $val ) ) {
1517                                 $path = $pathNames[$cacheKey];
1518                                 $this->trimCache(); // limit memory
1519                                 $this->cache[$path]['stat'] = $val;
1520                                 if ( isset( $val['sha1'] ) ) { // some backends store SHA-1 as metadata
1521                                         $this->trimCache(); // limit memory
1522                                         $this->cache[$path]['sha1'] =
1523                                                 array( 'hash' => $val['sha1'], 'latest' => $val['latest'] );
1524                                 }
1525                         }
1526                 }
1527
1528                 wfProfileOut( __METHOD__ . '-' . $this->name );
1529                 wfProfileOut( __METHOD__ );
1530         }
1531 }
1532
1533 /**
1534  * FileBackendStore helper class for performing asynchronous file operations.
1535  *
1536  * For example, calling FileBackendStore::createInternal() with the "async"
1537  * param flag may result in a Status that contains this object as a value.
1538  * This class is largely backend-specific and is mostly just "magic" to be
1539  * passed to FileBackendStore::executeOpHandlesInternal().
1540  */
1541 abstract class FileBackendStoreOpHandle {
1542         /** @var Array */
1543         public $params = array(); // params to caller functions
1544         /** @var FileBackendStore */
1545         public $backend;
1546         /** @var Array */
1547         public $resourcesToClose = array();
1548
1549         public $call; // string; name that identifies the function called
1550
1551         /**
1552          * Close all open file handles
1553          *
1554          * @return void
1555          */
1556         public function closeResources() {
1557                 array_map( 'fclose', $this->resourcesToClose );
1558         }
1559 }
1560
1561 /**
1562  * FileBackendStore helper function to handle listings that span container shards.
1563  * Do not use this class from places outside of FileBackendStore.
1564  *
1565  * @ingroup FileBackend
1566  */
1567 abstract class FileBackendStoreShardListIterator implements Iterator {
1568         /** @var FileBackendStore */
1569         protected $backend;
1570         /** @var Array */
1571         protected $params;
1572         /** @var Array */
1573         protected $shardSuffixes;
1574         protected $container; // string; full container name
1575         protected $directory; // string; resolved relative path
1576
1577         /** @var Traversable */
1578         protected $iter;
1579         protected $curShard = 0; // integer
1580         protected $pos = 0; // integer
1581
1582         /** @var Array */
1583         protected $multiShardPaths = array(); // (rel path => 1)
1584
1585         /**
1586          * @param $backend FileBackendStore
1587          * @param $container string Full storage container name
1588          * @param $dir string Storage directory relative to container
1589          * @param $suffixes Array List of container shard suffixes
1590          * @param $params Array
1591          */
1592         public function __construct(
1593                 FileBackendStore $backend, $container, $dir, array $suffixes, array $params
1594         ) {
1595                 $this->backend = $backend;
1596                 $this->container = $container;
1597                 $this->directory = $dir;
1598                 $this->shardSuffixes = $suffixes;
1599                 $this->params = $params;
1600         }
1601
1602         /**
1603          * @see Iterator::current()
1604          * @return string|bool String or false
1605          */
1606         public function current() {
1607                 if ( is_array( $this->iter ) ) {
1608                         return current( $this->iter );
1609                 } else {
1610                         return $this->iter->current();
1611                 }
1612         }
1613
1614         /**
1615          * @see Iterator::key()
1616          * @return integer
1617          */
1618         public function key() {
1619                 return $this->pos;
1620         }
1621
1622         /**
1623          * @see Iterator::next()
1624          * @return void
1625          */
1626         public function next() {
1627                 ++$this->pos;
1628                 if ( is_array( $this->iter ) ) {
1629                         next( $this->iter );
1630                 } else {
1631                         $this->iter->next();
1632                 }
1633                 // Filter out items that we already listed
1634                 $this->filterViaNext();
1635                 // Find the next non-empty shard if no elements are left
1636                 $this->nextShardIteratorIfNotValid();
1637         }
1638
1639         /**
1640          * @see Iterator::rewind()
1641          * @return void
1642          */
1643         public function rewind() {
1644                 $this->pos = 0;
1645                 $this->curShard = 0;
1646                 $this->setIteratorFromCurrentShard();
1647                 // Filter out items that we already listed
1648                 $this->filterViaNext();
1649                 // Find the next non-empty shard if this one has no elements
1650                 $this->nextShardIteratorIfNotValid();
1651         }
1652
1653         /**
1654          * @see Iterator::valid()
1655          * @return bool
1656          */
1657         public function valid() {
1658                 if ( $this->iter === null ) {
1659                         return false; // some failure?
1660                 } elseif ( is_array( $this->iter ) ) {
1661                         return ( current( $this->iter ) !== false ); // no paths can have this value
1662                 } else {
1663                         return $this->iter->valid();
1664                 }
1665         }
1666
1667         /**
1668          * Filter out duplicate items by advancing to the next ones
1669          */
1670         protected function filterViaNext() {
1671                 while ( $this->iter->valid() ) {
1672                         $rel = $this->iter->current(); // path relative to given directory
1673                         $path = $this->params['dir'] . "/{$rel}"; // full storage path
1674                         if ( !$this->backend->isSingleShardPathInternal( $path ) ) {
1675                                 // Don't keep listing paths that are on multiple shards
1676                                 if ( isset( $this->multiShardPaths[$rel] ) ) {
1677                                         $this->iter->next(); // we already listed this path
1678                                 } else {
1679                                         $this->multiShardPaths[$rel] = 1;
1680                                         break;
1681                                 }
1682                         }
1683                 }
1684         }
1685
1686         /**
1687          * If the list iterator for this container shard is out of items,
1688          * then move on to the next container that has items.
1689          * If there are none, then it advances to the last container.
1690          */
1691         protected function nextShardIteratorIfNotValid() {
1692                 while ( !$this->valid() ) {
1693                         if ( ++$this->curShard >= count( $this->shardSuffixes ) ) {
1694                                 break; // no more container shards
1695                         }
1696                         $this->setIteratorFromCurrentShard();
1697                 }
1698         }
1699
1700         /**
1701          * Set the list iterator to that of the current container shard
1702          */
1703         protected function setIteratorFromCurrentShard() {
1704                 $suffix = $this->shardSuffixes[$this->curShard];
1705                 $this->iter = $this->listFromShard(
1706                         "{$this->container}{$suffix}", $this->directory, $this->params );
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          * @param string $container
1726          * @param string $dir
1727          * @param array $params
1728          * @return Array|null|Traversable
1729          */
1730         protected function listFromShard( $container, $dir, array $params ) {
1731                 return $this->backend->getDirectoryListInternal( $container, $dir, $params );
1732         }
1733 }
1734
1735 /**
1736  * Iterator for listing regular files
1737  */
1738 class FileBackendStoreShardFileIterator extends FileBackendStoreShardListIterator {
1739         /**
1740          * @param string $container
1741          * @param string $dir
1742          * @param array $params
1743          * @return Array|null|Traversable
1744          */
1745         protected function listFromShard( $container, $dir, array $params ) {
1746                 return $this->backend->getFileListInternal( $container, $dir, $params );
1747         }
1748 }