includes/filerepo/backend/FileBackendStore.php

   1 <?php
   2 /**
   3  * @file
   4  * @ingroup FileBackend
   5  * @author Aaron Schulz
   6  */
   7
   8 /**
   9  * @brief Base class for all backends using particular storage medium.
  10  *
  11  * This class defines the methods as abstract that subclasses must implement.
  12  * Outside callers should *not* use functions with "Internal" in the name.
  13  *
  14  * The FileBackend operations are implemented using basic functions
  15  * such as storeInternal(), copyInternal(), deleteInternal() and the like.
  16  * This class is also responsible for path resolution and sanitization.
  17  *
  18  * @ingroup FileBackend
  19  * @since 1.19
  20  */
  21 abstract class FileBackendStore extends FileBackend {
  22         /** @var BagOStuff */
  23         protected $memCache;
  24
  25         /** @var Array Map of paths to small (RAM/disk) cache items */
  26         protected $cache = array(); // (storage path => key => value)
  27         protected $maxCacheSize = 100; // integer; max paths with entries
  28         /** @var Array Map of paths to large (RAM/disk) cache items */
  29         protected $expensiveCache = array(); // (storage path => key => value)
  30         protected $maxExpensiveCacheSize = 10; // integer; max paths with entries
  31
  32         /** @var Array Map of container names to sharding settings */
  33         protected $shardViaHashLevels = array(); // (container name => config array)
  34
  35         protected $maxFileSize = 4294967296; // integer bytes (4GiB)
  36
  37         /**
  38          * @see FileBackend::__construct()
  39          *
  40          * @param $config Array
  41          */
  42         public function __construct( array $config ) {
  43                 parent::__construct( $config );
  44                 $this->memCache = new EmptyBagOStuff(); // disabled by default
  45         }
  46
  47         /**
  48          * Get the maximum allowable file size given backend
  49          * medium restrictions and basic performance constraints.
  50          * Do not call this function from places outside FileBackend and FileOp.
  51          *
  52          * @return integer Bytes
  53          */
  54         final public function maxFileSizeInternal() {
  55                 return $this->maxFileSize;
  56         }
  57
  58         /**
  59          * Check if a file can be created at a given storage path.
  60          * FS backends should check if the parent directory exists and the file is writable.
  61          * Backends using key/value stores should check if the container exists.
  62          *
  63          * @param $storagePath string
  64          * @return bool
  65          */
  66         abstract public function isPathUsableInternal( $storagePath );
  67
  68         /**
  69          * Create a file in the backend with the given contents.
  70          * Do not call this function from places outside FileBackend and FileOp.
  71          *
  72          * $params include:
  73          *     content       : the raw file contents
  74          *     dst           : destination storage path
  75          *     overwrite     : overwrite any file that exists at the destination
  76          *
  77          * @param $params Array
  78          * @return Status
  79          */
  80         final public function createInternal( array $params ) {
  81                 wfProfileIn( __METHOD__ );
  82                 wfProfileIn( __METHOD__ . '-' . $this->name );
  83                 if ( strlen( $params['content'] ) > $this->maxFileSizeInternal() ) {
  84                         $status = Status::newFatal( 'backend-fail-maxsize',
  85                                 $params['dst'], $this->maxFileSizeInternal() );
  86                 } else {
  87                         $status = $this->doCreateInternal( $params );
  88                         $this->clearCache( array( $params['dst'] ) );
  89                 }
  90                 wfProfileOut( __METHOD__ . '-' . $this->name );
  91                 wfProfileOut( __METHOD__ );
  92                 return $status;
  93         }
  94
  95         /**
  96          * @see FileBackendStore::createInternal()
  97          */
  98         abstract protected function doCreateInternal( array $params );
  99
 100         /**
 101          * Store a file into the backend from a file on disk.
 102          * Do not call this function from places outside FileBackend and FileOp.
 103          *
 104          * $params include:
 105          *     src           : source path on disk
 106          *     dst           : destination storage path
 107          *     overwrite     : overwrite any file that exists at the destination
 108          *
 109          * @param $params Array
 110          * @return Status
 111          */
 112         final public function storeInternal( array $params ) {
 113                 wfProfileIn( __METHOD__ );
 114                 wfProfileIn( __METHOD__ . '-' . $this->name );
 115                 if ( filesize( $params['src'] ) > $this->maxFileSizeInternal() ) {
 116                         $status = Status::newFatal( 'backend-fail-store', $params['dst'] );
 117                 } else {
 118                         $status = $this->doStoreInternal( $params );
 119                         $this->clearCache( array( $params['dst'] ) );
 120                 }
 121                 wfProfileOut( __METHOD__ . '-' . $this->name );
 122                 wfProfileOut( __METHOD__ );
 123                 return $status;
 124         }
 125
 126         /**
 127          * @see FileBackendStore::storeInternal()
 128          */
 129         abstract protected function doStoreInternal( array $params );
 130
 131         /**
 132          * Copy a file from one storage path to another in the backend.
 133          * Do not call this function from places outside FileBackend and FileOp.
 134          *
 135          * $params include:
 136          *     src           : source storage path
 137          *     dst           : destination storage path
 138          *     overwrite     : overwrite any file that exists at the destination
 139          *
 140          * @param $params Array
 141          * @return Status
 142          */
 143         final public function copyInternal( array $params ) {
 144                 wfProfileIn( __METHOD__ );
 145                 wfProfileIn( __METHOD__ . '-' . $this->name );
 146                 $status = $this->doCopyInternal( $params );
 147                 $this->clearCache( array( $params['dst'] ) );
 148                 wfProfileOut( __METHOD__ . '-' . $this->name );
 149                 wfProfileOut( __METHOD__ );
 150                 return $status;
 151         }
 152
 153         /**
 154          * @see FileBackendStore::copyInternal()
 155          */
 156         abstract protected function doCopyInternal( array $params );
 157
 158         /**
 159          * Delete a file at the storage path.
 160          * Do not call this function from places outside FileBackend and FileOp.
 161          *
 162          * $params include:
 163          *     src                 : source storage path
 164          *     ignoreMissingSource : do nothing if the source file does not exist
 165          *
 166          * @param $params Array
 167          * @return Status
 168          */
 169         final public function deleteInternal( array $params ) {
 170                 wfProfileIn( __METHOD__ );
 171                 wfProfileIn( __METHOD__ . '-' . $this->name );
 172                 $status = $this->doDeleteInternal( $params );
 173                 $this->clearCache( array( $params['src'] ) );
 174                 wfProfileOut( __METHOD__ . '-' . $this->name );
 175                 wfProfileOut( __METHOD__ );
 176                 return $status;
 177         }
 178
 179         /**
 180          * @see FileBackendStore::deleteInternal()
 181          */
 182         abstract protected function doDeleteInternal( array $params );
 183
 184         /**
 185          * Move a file from one storage path to another in the backend.
 186          * Do not call this function from places outside FileBackend and FileOp.
 187          *
 188          * $params include:
 189          *     src           : source storage path
 190          *     dst           : destination storage path
 191          *     overwrite     : overwrite any file that exists at the destination
 192          *
 193          * @param $params Array
 194          * @return Status
 195          */
 196         final public function moveInternal( array $params ) {
 197                 wfProfileIn( __METHOD__ );
 198                 wfProfileIn( __METHOD__ . '-' . $this->name );
 199                 $status = $this->doMoveInternal( $params );
 200                 $this->clearCache( array( $params['src'], $params['dst'] ) );
 201                 wfProfileOut( __METHOD__ . '-' . $this->name );
 202                 wfProfileOut( __METHOD__ );
 203                 return $status;
 204         }
 205
 206         /**
 207          * @see FileBackendStore::moveInternal()
 208          * @return Status
 209          */
 210         protected function doMoveInternal( array $params ) {
 211                 // Copy source to dest
 212                 $status = $this->copyInternal( $params );
 213                 if ( $status->isOK() ) {
 214                         // Delete source (only fails due to races or medium going down)
 215                         $status->merge( $this->deleteInternal( array( 'src' => $params['src'] ) ) );
 216                         $status->setResult( true, $status->value ); // ignore delete() errors
 217                 }
 218                 return $status;
 219         }
 220
 221         /**
 222          * @see FileBackend::concatenate()
 223          * @return Status
 224          */
 225         final public function concatenate( array $params ) {
 226                 wfProfileIn( __METHOD__ );
 227                 wfProfileIn( __METHOD__ . '-' . $this->name );
 228                 $status = Status::newGood();
 229
 230                 // Try to lock the source files for the scope of this function
 231                 $scopeLockS = $this->getScopedFileLocks( $params['srcs'], LockManager::LOCK_UW, $status );
 232                 if ( $status->isOK() ) {
 233                         // Actually do the concatenation
 234                         $status->merge( $this->doConcatenate( $params ) );
 235                 }
 236
 237                 wfProfileOut( __METHOD__ . '-' . $this->name );
 238                 wfProfileOut( __METHOD__ );
 239                 return $status;
 240         }
 241
 242         /**
 243          * @see FileBackendStore::concatenate()
 244          * @return Status
 245          */
 246         protected function doConcatenate( array $params ) {
 247                 $status = Status::newGood();
 248                 $tmpPath = $params['dst']; // convenience
 249
 250                 // Check that the specified temp file is valid...
 251                 wfSuppressWarnings();
 252                 $ok = ( is_file( $tmpPath ) && !filesize( $tmpPath ) );
 253                 wfRestoreWarnings();
 254                 if ( !$ok ) { // not present or not empty
 255                         $status->fatal( 'backend-fail-opentemp', $tmpPath );
 256                         return $status;
 257                 }
 258
 259                 // Build up the temp file using the source chunks (in order)...
 260                 $tmpHandle = fopen( $tmpPath, 'ab' );
 261                 if ( $tmpHandle === false ) {
 262                         $status->fatal( 'backend-fail-opentemp', $tmpPath );
 263                         return $status;
 264                 }
 265                 foreach ( $params['srcs'] as $virtualSource ) {
 266                         // Get a local FS version of the chunk
 267                         $tmpFile = $this->getLocalReference( array( 'src' => $virtualSource ) );
 268                         if ( !$tmpFile ) {
 269                                 $status->fatal( 'backend-fail-read', $virtualSource );
 270                                 return $status;
 271                         }
 272                         // Get a handle to the local FS version
 273                         $sourceHandle = fopen( $tmpFile->getPath(), 'r' );
 274                         if ( $sourceHandle === false ) {
 275                                 fclose( $tmpHandle );
 276                                 $status->fatal( 'backend-fail-read', $virtualSource );
 277                                 return $status;
 278                         }
 279                         // Append chunk to file (pass chunk size to avoid magic quotes)
 280                         if ( !stream_copy_to_stream( $sourceHandle, $tmpHandle ) ) {
 281                                 fclose( $sourceHandle );
 282                                 fclose( $tmpHandle );
 283                                 $status->fatal( 'backend-fail-writetemp', $tmpPath );
 284                                 return $status;
 285                         }
 286                         fclose( $sourceHandle );
 287                 }
 288                 if ( !fclose( $tmpHandle ) ) {
 289                         $status->fatal( 'backend-fail-closetemp', $tmpPath );
 290                         return $status;
 291                 }
 292
 293                 clearstatcache(); // temp file changed
 294
 295                 return $status;
 296         }
 297
 298         /**
 299          * @see FileBackend::doPrepare()
 300          * @return Status
 301          */
 302         final protected function doPrepare( array $params ) {
 303                 wfProfileIn( __METHOD__ );
 304                 wfProfileIn( __METHOD__ . '-' . $this->name );
 305
 306                 $status = Status::newGood();
 307                 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
 308                 if ( $dir === null ) {
 309                         $status->fatal( 'backend-fail-invalidpath', $params['dir'] );
 310                         wfProfileOut( __METHOD__ . '-' . $this->name );
 311                         wfProfileOut( __METHOD__ );
 312                         return $status; // invalid storage path
 313                 }
 314
 315                 if ( $shard !== null ) { // confined to a single container/shard
 316                         $status->merge( $this->doPrepareInternal( $fullCont, $dir, $params ) );
 317                 } else { // directory is on several shards
 318                         wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
 319                         list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
 320                         foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
 321                                 $status->merge( $this->doPrepareInternal( "{$fullCont}{$suffix}", $dir, $params ) );
 322                         }
 323                 }
 324
 325                 wfProfileOut( __METHOD__ . '-' . $this->name );
 326                 wfProfileOut( __METHOD__ );
 327                 return $status;
 328         }
 329
 330         /**
 331          * @see FileBackendStore::doPrepare()
 332          * @return Status
 333          */
 334         protected function doPrepareInternal( $container, $dir, array $params ) {
 335                 return Status::newGood();
 336         }
 337
 338         /**
 339          * @see FileBackend::doSecure()
 340          * @return Status
 341          */
 342         final protected function doSecure( array $params ) {
 343                 wfProfileIn( __METHOD__ );
 344                 wfProfileIn( __METHOD__ . '-' . $this->name );
 345                 $status = Status::newGood();
 346
 347                 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
 348                 if ( $dir === null ) {
 349                         $status->fatal( 'backend-fail-invalidpath', $params['dir'] );
 350                         wfProfileOut( __METHOD__ . '-' . $this->name );
 351                         wfProfileOut( __METHOD__ );
 352                         return $status; // invalid storage path
 353                 }
 354
 355                 if ( $shard !== null ) { // confined to a single container/shard
 356                         $status->merge( $this->doSecureInternal( $fullCont, $dir, $params ) );
 357                 } else { // directory is on several shards
 358                         wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
 359                         list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
 360                         foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
 361                                 $status->merge( $this->doSecureInternal( "{$fullCont}{$suffix}", $dir, $params ) );
 362                         }
 363                 }
 364
 365                 wfProfileOut( __METHOD__ . '-' . $this->name );
 366                 wfProfileOut( __METHOD__ );
 367                 return $status;
 368         }
 369
 370         /**
 371          * @see FileBackendStore::doSecure()
 372          * @return Status
 373          */
 374         protected function doSecureInternal( $container, $dir, array $params ) {
 375                 return Status::newGood();
 376         }
 377
 378         /**
 379          * @see FileBackend::doClean()
 380          * @return Status
 381          */
 382         final protected function doClean( array $params ) {
 383                 wfProfileIn( __METHOD__ );
 384                 wfProfileIn( __METHOD__ . '-' . $this->name );
 385                 $status = Status::newGood();
 386
 387                 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
 388                 if ( $dir === null ) {
 389                         $status->fatal( 'backend-fail-invalidpath', $params['dir'] );
 390                         wfProfileOut( __METHOD__ . '-' . $this->name );
 391                         wfProfileOut( __METHOD__ );
 392                         return $status; // invalid storage path
 393                 }
 394
 395                 // Attempt to lock this directory...
 396                 $filesLockEx = array( $params['dir'] );
 397                 $scopedLockE = $this->getScopedFileLocks( $filesLockEx, LockManager::LOCK_EX, $status );
 398                 if ( !$status->isOK() ) {
 399                         wfProfileOut( __METHOD__ . '-' . $this->name );
 400                         wfProfileOut( __METHOD__ );
 401                         return $status; // abort
 402                 }
 403
 404                 if ( $shard !== null ) { // confined to a single container/shard
 405                         $status->merge( $this->doCleanInternal( $fullCont, $dir, $params ) );
 406                         $this->deleteContainerCache( $fullCont ); // purge cache
 407                 } else { // directory is on several shards
 408                         wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
 409                         list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
 410                         foreach ( $this->getContainerSuffixes( $shortCont ) as $suffix ) {
 411                                 $status->merge( $this->doCleanInternal( "{$fullCont}{$suffix}", $dir, $params ) );
 412                                 $this->deleteContainerCache( "{$fullCont}{$suffix}" ); // purge cache
 413                         }
 414                 }
 415
 416                 wfProfileOut( __METHOD__ . '-' . $this->name );
 417                 wfProfileOut( __METHOD__ );
 418                 return $status;
 419         }
 420
 421         /**
 422          * @see FileBackendStore::doClean()
 423          * @return Status
 424          */
 425         protected function doCleanInternal( $container, $dir, array $params ) {
 426                 return Status::newGood();
 427         }
 428
 429         /**
 430          * @see FileBackend::fileExists()
 431          * @return bool|null
 432          */
 433         final public function fileExists( array $params ) {
 434                 wfProfileIn( __METHOD__ );
 435                 wfProfileIn( __METHOD__ . '-' . $this->name );
 436                 $stat = $this->getFileStat( $params );
 437                 wfProfileOut( __METHOD__ . '-' . $this->name );
 438                 wfProfileOut( __METHOD__ );
 439                 return ( $stat === null ) ? null : (bool)$stat; // null => failure
 440         }
 441
 442         /**
 443          * @see FileBackend::getFileTimestamp()
 444          * @return bool
 445          */
 446         final public function getFileTimestamp( array $params ) {
 447                 wfProfileIn( __METHOD__ );
 448                 wfProfileIn( __METHOD__ . '-' . $this->name );
 449                 $stat = $this->getFileStat( $params );
 450                 wfProfileOut( __METHOD__ . '-' . $this->name );
 451                 wfProfileOut( __METHOD__ );
 452                 return $stat ? $stat['mtime'] : false;
 453         }
 454
 455         /**
 456          * @see FileBackend::getFileSize()
 457          * @return bool
 458          */
 459         final public function getFileSize( array $params ) {
 460                 wfProfileIn( __METHOD__ );
 461                 wfProfileIn( __METHOD__ . '-' . $this->name );
 462                 $stat = $this->getFileStat( $params );
 463                 wfProfileOut( __METHOD__ . '-' . $this->name );
 464                 wfProfileOut( __METHOD__ );
 465                 return $stat ? $stat['size'] : false;
 466         }
 467
 468         /**
 469          * @see FileBackend::getFileStat()
 470          * @return bool
 471          */
 472         final public function getFileStat( array $params ) {
 473                 wfProfileIn( __METHOD__ );
 474                 wfProfileIn( __METHOD__ . '-' . $this->name );
 475                 $path = self::normalizeStoragePath( $params['src'] );
 476                 if ( $path === null ) {
 477                         wfProfileOut( __METHOD__ . '-' . $this->name );
 478                         wfProfileOut( __METHOD__ );
 479                         return false; // invalid storage path
 480                 }
 481                 $latest = !empty( $params['latest'] );
 482                 if ( isset( $this->cache[$path]['stat'] ) ) {
 483                         // If we want the latest data, check that this cached
 484                         // value was in fact fetched with the latest available data.
 485                         if ( !$latest || $this->cache[$path]['stat']['latest'] ) {
 486                                 $this->pingCache( $path ); // LRU
 487                                 wfProfileOut( __METHOD__ . '-' . $this->name );
 488                                 wfProfileOut( __METHOD__ );
 489                                 return $this->cache[$path]['stat'];
 490                         }
 491                 }
 492                 wfProfileIn( __METHOD__ . '-miss' );
 493                 wfProfileIn( __METHOD__ . '-miss-' . $this->name );
 494                 $stat = $this->doGetFileStat( $params );
 495                 wfProfileOut( __METHOD__ . '-miss-' . $this->name );
 496                 wfProfileOut( __METHOD__ . '-miss' );
 497                 if ( is_array( $stat ) ) { // don't cache negatives
 498                         $this->trimCache(); // limit memory
 499                         $this->cache[$path]['stat'] = $stat;
 500                         $this->cache[$path]['stat']['latest'] = $latest;
 501                 }
 502                 wfProfileOut( __METHOD__ . '-' . $this->name );
 503                 wfProfileOut( __METHOD__ );
 504                 return $stat;
 505         }
 506
 507         /**
 508          * @see FileBackendStore::getFileStat()
 509          */
 510         abstract protected function doGetFileStat( array $params );
 511
 512         /**
 513          * @see FileBackend::getFileContents()
 514          * @return bool|string
 515          */
 516         public function getFileContents( array $params ) {
 517                 wfProfileIn( __METHOD__ );
 518                 wfProfileIn( __METHOD__ . '-' . $this->name );
 519                 $tmpFile = $this->getLocalReference( $params );
 520                 if ( !$tmpFile ) {
 521                         wfProfileOut( __METHOD__ . '-' . $this->name );
 522                         wfProfileOut( __METHOD__ );
 523                         return false;
 524                 }
 525                 wfSuppressWarnings();
 526                 $data = file_get_contents( $tmpFile->getPath() );
 527                 wfRestoreWarnings();
 528                 wfProfileOut( __METHOD__ . '-' . $this->name );
 529                 wfProfileOut( __METHOD__ );
 530                 return $data;
 531         }
 532
 533         /**
 534          * @see FileBackend::getFileSha1Base36()
 535          * @return bool|string
 536          */
 537         final public function getFileSha1Base36( array $params ) {
 538                 wfProfileIn( __METHOD__ );
 539                 wfProfileIn( __METHOD__ . '-' . $this->name );
 540                 $path = $params['src'];
 541                 if ( isset( $this->cache[$path]['sha1'] ) ) {
 542                         $this->pingCache( $path ); // LRU
 543                         wfProfileOut( __METHOD__ . '-' . $this->name );
 544                         wfProfileOut( __METHOD__ );
 545                         return $this->cache[$path]['sha1'];
 546                 }
 547                 wfProfileIn( __METHOD__ . '-miss' );
 548                 wfProfileIn( __METHOD__ . '-miss-' . $this->name );
 549                 $hash = $this->doGetFileSha1Base36( $params );
 550                 wfProfileOut( __METHOD__ . '-miss-' . $this->name );
 551                 wfProfileOut( __METHOD__ . '-miss' );
 552                 if ( $hash ) { // don't cache negatives
 553                         $this->trimCache(); // limit memory
 554                         $this->cache[$path]['sha1'] = $hash;
 555                 }
 556                 wfProfileOut( __METHOD__ . '-' . $this->name );
 557                 wfProfileOut( __METHOD__ );
 558                 return $hash;
 559         }
 560
 561         /**
 562          * @see FileBackendStore::getFileSha1Base36()
 563          * @return bool
 564          */
 565         protected function doGetFileSha1Base36( array $params ) {
 566                 $fsFile = $this->getLocalReference( $params );
 567                 if ( !$fsFile ) {
 568                         return false;
 569                 } else {
 570                         return $fsFile->getSha1Base36();
 571                 }
 572         }
 573
 574         /**
 575          * @see FileBackend::getFileProps()
 576          * @return Array
 577          */
 578         final public function getFileProps( array $params ) {
 579                 wfProfileIn( __METHOD__ );
 580                 wfProfileIn( __METHOD__ . '-' . $this->name );
 581                 $fsFile = $this->getLocalReference( $params );
 582                 $props = $fsFile ? $fsFile->getProps() : FSFile::placeholderProps();
 583                 wfProfileOut( __METHOD__ . '-' . $this->name );
 584                 wfProfileOut( __METHOD__ );
 585                 return $props;
 586         }
 587
 588         /**
 589          * @see FileBackend::getLocalReference()
 590          * @return TempFSFile|null
 591          */
 592         public function getLocalReference( array $params ) {
 593                 wfProfileIn( __METHOD__ );
 594                 wfProfileIn( __METHOD__ . '-' . $this->name );
 595                 $path = $params['src'];
 596                 if ( isset( $this->expensiveCache[$path]['localRef'] ) ) {
 597                         $this->pingExpensiveCache( $path );
 598                         wfProfileOut( __METHOD__ . '-' . $this->name );
 599                         wfProfileOut( __METHOD__ );
 600                         return $this->expensiveCache[$path]['localRef'];
 601                 }
 602                 $tmpFile = $this->getLocalCopy( $params );
 603                 if ( $tmpFile ) { // don't cache negatives
 604                         $this->trimExpensiveCache(); // limit memory
 605                         $this->expensiveCache[$path]['localRef'] = $tmpFile;
 606                 }
 607                 wfProfileOut( __METHOD__ . '-' . $this->name );
 608                 wfProfileOut( __METHOD__ );
 609                 return $tmpFile;
 610         }
 611
 612         /**
 613          * @see FileBackend::streamFile()
 614          * @return Status
 615          */
 616         final public function streamFile( array $params ) {
 617                 wfProfileIn( __METHOD__ );
 618                 wfProfileIn( __METHOD__ . '-' . $this->name );
 619                 $status = Status::newGood();
 620
 621                 $info = $this->getFileStat( $params );
 622                 if ( !$info ) { // let StreamFile handle the 404
 623                         $status->fatal( 'backend-fail-notexists', $params['src'] );
 624                 }
 625
 626                 // Set output buffer and HTTP headers for stream
 627                 $extraHeaders = isset( $params['headers'] ) ? $params['headers'] : array();
 628                 $res = StreamFile::prepareForStream( $params['src'], $info, $extraHeaders );
 629                 if ( $res == StreamFile::NOT_MODIFIED ) {
 630                         // do nothing; client cache is up to date
 631                 } elseif ( $res == StreamFile::READY_STREAM ) {
 632                         wfProfileIn( __METHOD__ . '-send' );
 633                         wfProfileIn( __METHOD__ . '-send-' . $this->name );
 634                         $status = $this->doStreamFile( $params );
 635                         wfProfileOut( __METHOD__ . '-send-' . $this->name );
 636                         wfProfileOut( __METHOD__ . '-send' );
 637                 } else {
 638                         $status->fatal( 'backend-fail-stream', $params['src'] );
 639                 }
 640
 641                 wfProfileOut( __METHOD__ . '-' . $this->name );
 642                 wfProfileOut( __METHOD__ );
 643                 return $status;
 644         }
 645
 646         /**
 647          * @see FileBackendStore::streamFile()
 648          * @return Status
 649          */
 650         protected function doStreamFile( array $params ) {
 651                 $status = Status::newGood();
 652
 653                 $fsFile = $this->getLocalReference( $params );
 654                 if ( !$fsFile ) {
 655                         $status->fatal( 'backend-fail-stream', $params['src'] );
 656                 } elseif ( !readfile( $fsFile->getPath() ) ) {
 657                         $status->fatal( 'backend-fail-stream', $params['src'] );
 658                 }
 659
 660                 return $status;
 661         }
 662
 663         /**
 664          * @copydoc FileBackend::getFileList()
 665          * @return Array|null|Traversable
 666          */
 667         final public function getFileList( array $params ) {
 668                 list( $fullCont, $dir, $shard ) = $this->resolveStoragePath( $params['dir'] );
 669                 if ( $dir === null ) { // invalid storage path
 670                         return null;
 671                 }
 672                 if ( $shard !== null ) {
 673                         // File listing is confined to a single container/shard
 674                         return $this->getFileListInternal( $fullCont, $dir, $params );
 675                 } else {
 676                         wfDebug( __METHOD__ . ": iterating over all container shards.\n" );
 677                         // File listing spans multiple containers/shards
 678                         list( $b, $shortCont, $r ) = self::splitStoragePath( $params['dir'] );
 679                         return new FileBackendStoreShardListIterator( $this,
 680                                 $fullCont, $dir, $this->getContainerSuffixes( $shortCont ), $params );
 681                 }
 682         }
 683
 684         /**
 685          * Do not call this function from places outside FileBackend
 686          *
 687          * @see FileBackendStore::getFileList()
 688          *
 689          * @param $container string Resolved container name
 690          * @param $dir string Resolved path relative to container
 691          * @param $params Array
 692          * @return Traversable|Array|null
 693          */
 694         abstract public function getFileListInternal( $container, $dir, array $params );
 695
 696         /**
 697          * Get the list of supported operations and their corresponding FileOp classes.
 698          *
 699          * @return Array
 700          */
 701         protected function supportedOperations() {
 702                 return array(
 703                         'store'       => 'StoreFileOp',
 704                         'copy'        => 'CopyFileOp',
 705                         'move'        => 'MoveFileOp',
 706                         'delete'      => 'DeleteFileOp',
 707                         'create'      => 'CreateFileOp',
 708                         'null'        => 'NullFileOp'
 709                 );
 710         }
 711
 712         /**
 713          * Return a list of FileOp objects from a list of operations.
 714          * Do not call this function from places outside FileBackend.
 715          *
 716          * The result must have the same number of items as the input.
 717          * An exception is thrown if an unsupported operation is requested.
 718          *
 719          * @param $ops Array Same format as doOperations()
 720          * @return Array List of FileOp objects
 721          * @throws MWException
 722          */
 723         final public function getOperations( array $ops ) {
 724                 $supportedOps = $this->supportedOperations();
 725
 726                 $performOps = array(); // array of FileOp objects
 727                 // Build up ordered array of FileOps...
 728                 foreach ( $ops as $operation ) {
 729                         $opName = $operation['op'];
 730                         if ( isset( $supportedOps[$opName] ) ) {
 731                                 $class = $supportedOps[$opName];
 732                                 // Get params for this operation
 733                                 $params = $operation;
 734                                 // Append the FileOp class
 735                                 $performOps[] = new $class( $this, $params );
 736                         } else {
 737                                 throw new MWException( "Operation `$opName` is not supported." );
 738                         }
 739                 }
 740
 741                 return $performOps;
 742         }
 743
 744         /**
 745          * @see FileBackend::doOperationsInternal()
 746          * @return Status
 747          */
 748         protected function doOperationsInternal( array $ops, array $opts ) {
 749                 wfProfileIn( __METHOD__ );
 750                 wfProfileIn( __METHOD__ . '-' . $this->name );
 751                 $status = Status::newGood();
 752
 753                 // Build up a list of FileOps...
 754                 $performOps = $this->getOperations( $ops );
 755
 756                 // Acquire any locks as needed...
 757                 if ( empty( $opts['nonLocking'] ) ) {
 758                         // Build up a list of files to lock...
 759                         $filesLockEx = $filesLockSh = array();
 760                         foreach ( $performOps as $fileOp ) {
 761                                 $filesLockSh = array_merge( $filesLockSh, $fileOp->storagePathsRead() );
 762                                 $filesLockEx = array_merge( $filesLockEx, $fileOp->storagePathsChanged() );
 763                         }
 764                         // Optimization: if doing an EX lock anyway, don't also set an SH one
 765                         $filesLockSh = array_diff( $filesLockSh, $filesLockEx );
 766                         // Get a shared lock on the parent directory of each path changed
 767                         $filesLockSh = array_merge( $filesLockSh, array_map( 'dirname', $filesLockEx ) );
 768                         // Try to lock those files for the scope of this function...
 769                         $scopeLockS = $this->getScopedFileLocks( $filesLockSh, LockManager::LOCK_UW, $status );
 770                         $scopeLockE = $this->getScopedFileLocks( $filesLockEx, LockManager::LOCK_EX, $status );
 771                         if ( !$status->isOK() ) {
 772                                 wfProfileOut( __METHOD__ . '-' . $this->name );
 773                                 wfProfileOut( __METHOD__ );
 774                                 return $status; // abort
 775                         }
 776                 }
 777
 778                 // Clear any file cache entries (after locks acquired)
 779                 $this->clearCache();
 780
 781                 // Load from the persistent container cache
 782                 $this->primeContainerCache( $performOps );
 783
 784                 // Actually attempt the operation batch...
 785                 $subStatus = FileOp::attemptBatch( $performOps, $opts, $this->fileJournal );
 786
 787                 // Merge errors into status fields
 788                 $status->merge( $subStatus );
 789                 $status->success = $subStatus->success; // not done in merge()
 790
 791                 wfProfileOut( __METHOD__ . '-' . $this->name );
 792                 wfProfileOut( __METHOD__ );
 793                 return $status;
 794         }
 795
 796         /**
 797          * @see FileBackend::clearCache()
 798          */
 799         final public function clearCache( array $paths = null ) {
 800                 if ( is_array( $paths ) ) {
 801                         $paths = array_map( 'FileBackend::normalizeStoragePath', $paths );
 802                         $paths = array_filter( $paths, 'strlen' ); // remove nulls
 803                 }
 804                 if ( $paths === null ) {
 805                         $this->cache = array();
 806                         $this->expensiveCache = array();
 807                 } else {
 808                         foreach ( $paths as $path ) {
 809                                 unset( $this->cache[$path] );
 810                                 unset( $this->expensiveCache[$path] );
 811                         }
 812                 }
 813                 $this->doClearCache( $paths );
 814         }
 815
 816         /**
 817          * Clears any additional stat caches for storage paths
 818          *
 819          * @see FileBackend::clearCache()
 820          *
 821          * @param $paths Array Storage paths (optional)
 822          * @return void
 823          */
 824         protected function doClearCache( array $paths = null ) {}
 825
 826         /**
 827          * Move a cache entry to the top (such as when accessed)
 828          *
 829          * @param $path string Storage path
 830          */
 831         protected function pingCache( $path ) {
 832                 if ( isset( $this->cache[$path] ) ) {
 833                         $tmp = $this->cache[$path];
 834                         unset( $this->cache[$path] );
 835                         $this->cache[$path] = $tmp;
 836                 }
 837         }
 838
 839         /**
 840          * Prune the inexpensive cache if it is too big to add an item
 841          *
 842          * @return void
 843          */
 844         protected function trimCache() {
 845                 if ( count( $this->cache ) >= $this->maxCacheSize ) {
 846                         reset( $this->cache );
 847                         unset( $this->cache[key( $this->cache )] );
 848                 }
 849         }
 850
 851         /**
 852          * Move a cache entry to the top (such as when accessed)
 853          *
 854          * @param $path string Storage path
 855          */
 856         protected function pingExpensiveCache( $path ) {
 857                 if ( isset( $this->expensiveCache[$path] ) ) {
 858                         $tmp = $this->expensiveCache[$path];
 859                         unset( $this->expensiveCache[$path] );
 860                         $this->expensiveCache[$path] = $tmp;
 861                 }
 862         }
 863
 864         /**
 865          * Prune the expensive cache if it is too big to add an item
 866          *
 867          * @return void
 868          */
 869         protected function trimExpensiveCache() {
 870                 if ( count( $this->expensiveCache ) >= $this->maxExpensiveCacheSize ) {
 871                         reset( $this->expensiveCache );
 872                         unset( $this->expensiveCache[key( $this->expensiveCache )] );
 873                 }
 874         }
 875
 876         /**
 877          * Check if a container name is valid.
 878          * This checks for for length and illegal characters.
 879          *
 880          * @param $container string
 881          * @return bool
 882          */
 883         final protected static function isValidContainerName( $container ) {
 884                 // This accounts for Swift and S3 restrictions while leaving room
 885                 // for things like '.xxx' (hex shard chars) or '.seg' (segments).
 886                 // This disallows directory separators or traversal characters.
 887                 // Note that matching strings URL encode to the same string;
 888                 // in Swift, the length restriction is *after* URL encoding.
 889                 return preg_match( '/^[a-z0-9][a-z0-9-_]{0,199}$/i', $container );
 890         }
 891
 892         /**
 893          * Splits a storage path into an internal container name,
 894          * an internal relative file name, and a container shard suffix.
 895          * Any shard suffix is already appended to the internal container name.
 896          * This also checks that the storage path is valid and within this backend.
 897          *
 898          * If the container is sharded but a suffix could not be determined,
 899          * this means that the path can only refer to a directory and can only
 900          * be scanned by looking in all the container shards.
 901          *
 902          * @param $storagePath string
 903          * @return Array (container, path, container suffix) or (null, null, null) if invalid
 904          */
 905         final protected function resolveStoragePath( $storagePath ) {
 906                 list( $backend, $container, $relPath ) = self::splitStoragePath( $storagePath );
 907                 if ( $backend === $this->name ) { // must be for this backend
 908                         $relPath = self::normalizeContainerPath( $relPath );
 909                         if ( $relPath !== null ) {
 910                                 // Get shard for the normalized path if this container is sharded
 911                                 $cShard = $this->getContainerShard( $container, $relPath );
 912                                 // Validate and sanitize the relative path (backend-specific)
 913                                 $relPath = $this->resolveContainerPath( $container, $relPath );
 914                                 if ( $relPath !== null ) {
 915                                         // Prepend any wiki ID prefix to the container name
 916                                         $container = $this->fullContainerName( $container );
 917                                         if ( self::isValidContainerName( $container ) ) {
 918                                                 // Validate and sanitize the container name (backend-specific)
 919                                                 $container = $this->resolveContainerName( "{$container}{$cShard}" );
 920                                                 if ( $container !== null ) {
 921                                                         return array( $container, $relPath, $cShard );
 922                                                 }
 923                                         }
 924                                 }
 925                         }
 926                 }
 927                 return array( null, null, null );
 928         }
 929
 930         /**
 931          * Like resolveStoragePath() except null values are returned if
 932          * the container is sharded and the shard could not be determined.
 933          *
 934          * @see FileBackendStore::resolveStoragePath()
 935          *
 936          * @param $storagePath string
 937          * @return Array (container, path) or (null, null) if invalid
 938          */
 939         final protected function resolveStoragePathReal( $storagePath ) {
 940                 list( $container, $relPath, $cShard ) = $this->resolveStoragePath( $storagePath );
 941                 if ( $cShard !== null ) {
 942                         return array( $container, $relPath );
 943                 }
 944                 return array( null, null );
 945         }
 946
 947         /**
 948          * Get the container name shard suffix for a given path.
 949          * Any empty suffix means the container is not sharded.
 950          *
 951          * @param $container string Container name
 952          * @param $relStoragePath string Storage path relative to the container
 953          * @return string|null Returns null if shard could not be determined
 954          */
 955         final protected function getContainerShard( $container, $relPath ) {
 956                 list( $levels, $base, $repeat ) = $this->getContainerHashLevels( $container );
 957                 if ( $levels == 1 || $levels == 2 ) {
 958                         // Hash characters are either base 16 or 36
 959                         $char = ( $base == 36 ) ? '[0-9a-z]' : '[0-9a-f]';
 960                         // Get a regex that represents the shard portion of paths.
 961                         // The concatenation of the captures gives us the shard.
 962                         if ( $levels === 1 ) { // 16 or 36 shards per container
 963                                 $hashDirRegex = '(' . $char . ')';
 964                         } else { // 256 or 1296 shards per container
 965                                 if ( $repeat ) { // verbose hash dir format (e.g. "a/ab/abc")
 966                                         $hashDirRegex = $char . '/(' . $char . '{2})';
 967                                 } else { // short hash dir format (e.g. "a/b/c")
 968                                         $hashDirRegex = '(' . $char . ')/(' . $char . ')';
 969                                 }
 970                         }
 971                         // Allow certain directories to be above the hash dirs so as
 972                         // to work with FileRepo (e.g. "archive/a/ab" or "temp/a/ab").
 973                         // They must be 2+ chars to avoid any hash directory ambiguity.
 974                         $m = array();
 975                         if ( preg_match( "!^(?:[^/]{2,}/)*$hashDirRegex(?:/|$)!", $relPath, $m ) ) {
 976                                 return '.' . implode( '', array_slice( $m, 1 ) );
 977                         }
 978                         return null; // failed to match
 979                 }
 980                 return ''; // no sharding
 981         }
 982
 983         /**
 984          * Get the sharding config for a container.
 985          * If greater than 0, then all file storage paths within
 986          * the container are required to be hashed accordingly.
 987          *
 988          * @param $container string
 989          * @return Array (integer levels, integer base, repeat flag) or (0, 0, false)
 990          */
 991         final protected function getContainerHashLevels( $container ) {
 992                 if ( isset( $this->shardViaHashLevels[$container] ) ) {
 993                         $config = $this->shardViaHashLevels[$container];
 994                         $hashLevels = (int)$config['levels'];
 995                         if ( $hashLevels == 1 || $hashLevels == 2 ) {
 996                                 $hashBase = (int)$config['base'];
 997                                 if ( $hashBase == 16 || $hashBase == 36 ) {
 998                                         return array( $hashLevels, $hashBase, $config['repeat'] );
 999                                 }
1000                         }
1001                 }
1002                 return array( 0, 0, false ); // no sharding
1003         }
1004
1005         /**
1006          * Get a list of full container shard suffixes for a container
1007          *
1008          * @param $container string
1009          * @return Array
1010          */
1011         final protected function getContainerSuffixes( $container ) {
1012                 $shards = array();
1013                 list( $digits, $base ) = $this->getContainerHashLevels( $container );
1014                 if ( $digits > 0 ) {
1015                         $numShards = pow( $base, $digits );
1016                         for ( $index = 0; $index < $numShards; $index++ ) {
1017                                 $shards[] = '.' . wfBaseConvert( $index, 10, $base, $digits );
1018                         }
1019                 }
1020                 return $shards;
1021         }
1022
1023         /**
1024          * Get the full container name, including the wiki ID prefix
1025          *
1026          * @param $container string
1027          * @return string
1028          */
1029         final protected function fullContainerName( $container ) {
1030                 if ( $this->wikiId != '' ) {
1031                         return "{$this->wikiId}-$container";
1032                 } else {
1033                         return $container;
1034                 }
1035         }
1036
1037         /**
1038          * Resolve a container name, checking if it's allowed by the backend.
1039          * This is intended for internal use, such as encoding illegal chars.
1040          * Subclasses can override this to be more restrictive.
1041          *
1042          * @param $container string
1043          * @return string|null
1044          */
1045         protected function resolveContainerName( $container ) {
1046                 return $container;
1047         }
1048
1049         /**
1050          * Resolve a relative storage path, checking if it's allowed by the backend.
1051          * This is intended for internal use, such as encoding illegal chars or perhaps
1052          * getting absolute paths (e.g. FS based backends). Note that the relative path
1053          * may be the empty string (e.g. the path is simply to the container).
1054          *
1055          * @param $container string Container name
1056          * @param $relStoragePath string Storage path relative to the container
1057          * @return string|null Path or null if not valid
1058          */
1059         protected function resolveContainerPath( $container, $relStoragePath ) {
1060                 return $relStoragePath;
1061         }
1062
1063         /**
1064          * Get the cache key for a container
1065          *
1066          * @param $container Resolved container name
1067          * @return string
1068          */
1069         private function containerCacheKey( $container ) {
1070                 return wfMemcKey( 'backend', $this->getName(), 'container', $container );
1071         }
1072
1073         /**
1074          * Set the cached info for a container
1075          *
1076          * @param $container Resolved container name
1077          * @param $val mixed Information to cache
1078          * @return void
1079          */
1080         final protected function setContainerCache( $container, $val ) {
1081                 $this->memCache->set( $this->containerCacheKey( $container ), $val, 7*86400 );
1082         }
1083
1084         /**
1085          * Delete the cached info for a container
1086          *
1087          * @param $container Resolved container name
1088          * @return void
1089          */
1090         final protected function deleteContainerCache( $container ) {
1091                 $this->memCache->delete( $this->containerCacheKey( $container ) );
1092         }
1093
1094         /**
1095          * Do a batch lookup from cache for container stats for all containers
1096          * used in a list of container names, storage paths, or FileOp objects.
1097          *
1098          * @param $items Array List of storage paths or FileOps
1099          * @return void
1100          */
1101         final protected function primeContainerCache( array $items ) {
1102                 $paths = array(); // list of storage paths
1103                 $contNames = array(); // (cache key => resolved container name)
1104                 // Get all the paths/containers from the items...
1105                 foreach ( $items as $item ) {
1106                         if ( $item instanceof FileOp ) {
1107                                 $paths = array_merge( $paths, $item->storagePathsRead() );
1108                                 $paths = array_merge( $paths, $item->storagePathsChanged() );
1109                         } elseif ( self::isStoragePath( $item ) ) {
1110                                 $paths[] = $item;
1111                         } elseif ( is_string( $item ) ) { // full container name
1112                                 $contNames[$this->containerCacheKey( $item )] = $item;
1113                         }
1114                 }
1115                 // Get all the corresponding cache keys for paths...
1116                 foreach ( $paths as $path ) {
1117                         list( $fullCont, $r, $s ) = $this->resolveStoragePath( $path );
1118                         if ( $fullCont !== null ) { // valid path for this backend
1119                                 $contNames[$this->containerCacheKey( $fullCont )] = $fullCont;
1120                         }
1121                 }
1122
1123                 $contInfo = array(); // (resolved container name => cache value)
1124                 // Get all cache entries for these container cache keys...
1125                 $values = $this->memCache->getBatch( array_keys( $contNames ) );
1126                 foreach ( $values as $cacheKey => $val ) {
1127                         $contInfo[$contNames[$cacheKey]] = $val;
1128                 }
1129
1130                 // Populate the container process cache for the backend...
1131                 $this->doPrimeContainerCache( array_filter( $contInfo, 'is_array' ) );
1132         }
1133
1134         /**
1135          * Fill the backend-specific process cache given an array of
1136          * resolved container names and their corresponding cached info.
1137          * Only containers that actually exist should appear in the map.
1138          *
1139          * @param $containerInfo Array Map of resolved container names to cached info
1140          * @return void
1141          */
1142         protected function doPrimeContainerCache( array $containerInfo ) {}
1143 }
1144
1145 /**
1146  * FileBackendStore helper function to handle file listings that span container shards.
1147  * Do not use this class from places outside of FileBackendStore.
1148  *
1149  * @ingroup FileBackend
1150  */
1151 class FileBackendStoreShardListIterator implements Iterator {
1152         /* @var FileBackendStore */
1153         protected $backend;
1154         /* @var Array */
1155         protected $params;
1156         /* @var Array */
1157         protected $shardSuffixes;
1158         protected $container; // string
1159         protected $directory; // string
1160
1161         /* @var Traversable */
1162         protected $iter;
1163         protected $curShard = 0; // integer
1164         protected $pos = 0; // integer
1165
1166         /**
1167          * @param $backend FileBackendStore
1168          * @param $container string Full storage container name
1169          * @param $dir string Storage directory relative to container
1170          * @param $suffixes Array List of container shard suffixes
1171          * @param $params Array
1172          */
1173         public function __construct(
1174                 FileBackendStore $backend, $container, $dir, array $suffixes, array $params
1175         ) {
1176                 $this->backend = $backend;
1177                 $this->container = $container;
1178                 $this->directory = $dir;
1179                 $this->shardSuffixes = $suffixes;
1180                 $this->params = $params;
1181         }
1182
1183         /**
1184          * @see Iterator::current()
1185          * @return string|bool String or false
1186          */
1187         public function current() {
1188                 if ( is_array( $this->iter ) ) {
1189                         return current( $this->iter );
1190                 } else {
1191                         return $this->iter->current();
1192                 }
1193         }
1194
1195         /**
1196          * @see Iterator::key()
1197          * @return integer
1198          */
1199         public function key() {
1200                 return $this->pos;
1201         }
1202
1203         /**
1204          * @see Iterator::next()
1205          * @return void
1206          */
1207         public function next() {
1208                 ++$this->pos;
1209                 if ( is_array( $this->iter ) ) {
1210                         next( $this->iter );
1211                 } else {
1212                         $this->iter->next();
1213                 }
1214                 // Find the next non-empty shard if no elements are left
1215                 $this->nextShardIteratorIfNotValid();
1216         }
1217
1218         /**
1219          * @see Iterator::rewind()
1220          * @return void
1221          */
1222         public function rewind() {
1223                 $this->pos = 0;
1224                 $this->curShard = 0;
1225                 $this->setIteratorFromCurrentShard();
1226                 // Find the next non-empty shard if this one has no elements
1227                 $this->nextShardIteratorIfNotValid();
1228         }
1229
1230         /**
1231          * @see Iterator::valid()
1232          * @return bool
1233          */
1234         public function valid() {
1235                 if ( $this->iter == null ) {
1236                         return false; // some failure?
1237                 } elseif ( is_array( $this->iter ) ) {
1238                         return ( current( $this->iter ) !== false ); // no paths can have this value
1239                 } else {
1240                         return $this->iter->valid();
1241                 }
1242         }
1243
1244         /**
1245          * If the list iterator for this container shard is out of items,
1246          * then move on to the next container that has items.
1247          * If there are none, then it advances to the last container.
1248          */
1249         protected function nextShardIteratorIfNotValid() {
1250                 while ( !$this->valid() ) {
1251                         if ( ++$this->curShard >= count( $this->shardSuffixes ) ) {
1252                                 break; // no more container shards
1253                         }
1254                         $this->setIteratorFromCurrentShard();
1255                 }
1256         }
1257
1258         /**
1259          * Set the list iterator to that of the current container shard
1260          */
1261         protected function setIteratorFromCurrentShard() {
1262                 $suffix = $this->shardSuffixes[$this->curShard];
1263                 $this->iter = $this->backend->getFileListInternal(
1264                         "{$this->container}{$suffix}", $this->directory, $this->params );
1265         }
1266 }