includes/filerepo/FileRepo.php

   1 <?php
   2
   3 /**
   4  * Base class for file repositories
   5  * Do not instantiate, use a derived class.
   6  * @ingroup FileRepo
   7  */
   8 abstract class FileRepo {
   9         const FILES_ONLY = 1;
  10         const DELETE_SOURCE = 1;
  11         const OVERWRITE = 2;
  12         const OVERWRITE_SAME = 4;
  13
  14         var $thumbScriptUrl, $transformVia404;
  15         var $descBaseUrl, $scriptDirUrl, $scriptExtension, $articleUrl;
  16         var $fetchDescription, $initialCapital;
  17         var $pathDisclosureProtection = 'paranoid';
  18         var $descriptionCacheExpiry, $hashLevels, $url, $thumbUrl;
  19
  20         /**
  21          * Factory functions for creating new files
  22          * Override these in the base class
  23          */
  24         var $fileFactory = false, $oldFileFactory = false;
  25         var $fileFactoryKey = false, $oldFileFactoryKey = false;
  26
  27         function __construct( $info ) {
  28                 // Required settings
  29                 $this->name = $info['name'];
  30
  31                 // Optional settings
  32                 $this->initialCapital = MWNamespace::isCapitalized( NS_FILE );
  33                 foreach ( array( 'descBaseUrl', 'scriptDirUrl', 'articleUrl', 'fetchDescription',
  34                         'thumbScriptUrl', 'initialCapital', 'pathDisclosureProtection',
  35                         'descriptionCacheExpiry', 'hashLevels', 'url', 'thumbUrl', 'scriptExtension' )
  36                         as $var )
  37                 {
  38                         if ( isset( $info[$var] ) ) {
  39                                 $this->$var = $info[$var];
  40                         }
  41                 }
  42                 $this->transformVia404 = !empty( $info['transformVia404'] );
  43         }
  44
  45         /**
  46          * Determine if a string is an mwrepo:// URL
  47          */
  48         static function isVirtualUrl( $url ) {
  49                 return substr( $url, 0, 9 ) == 'mwrepo://';
  50         }
  51
  52         /**
  53          * Create a new File object from the local repository
  54          *
  55          * @param $title Mixed: Title object or string
  56          * @param $time Mixed: Time at which the image was uploaded.
  57          *              If this is specified, the returned object will be an
  58          *              instance of the repository's old file class instead of a
  59          *              current file. Repositories not supporting version control
  60          *              should return false if this parameter is set.
  61          */
  62         function newFile( $title, $time = false ) {
  63                 if ( !($title instanceof Title) ) {
  64                         $title = Title::makeTitleSafe( NS_FILE, $title );
  65                         if ( !is_object( $title ) ) {
  66                                 return null;
  67                         }
  68                 }
  69                 if ( $time ) {
  70                         if ( $this->oldFileFactory ) {
  71                                 return call_user_func( $this->oldFileFactory, $title, $this, $time );
  72                         } else {
  73                                 return false;
  74                         }
  75                 } else {
  76                         return call_user_func( $this->fileFactory, $title, $this );
  77                 }
  78         }
  79
  80         /**
  81          * Find an instance of the named file created at the specified time
  82          * Returns false if the file does not exist. Repositories not supporting
  83          * version control should return false if the time is specified.
  84          *
  85          * @param $title Mixed: Title object or string
  86          * @param $options Associative array of options:
  87          *     time:           requested time for an archived image, or false for the
  88          *                     current version. An image object will be returned which was
  89          *                     created at the specified time.
  90          *
  91          *     ignoreRedirect: If true, do not follow file redirects
  92          *
  93          *     private:        If true, return restricted (deleted) files if the current
  94          *                     user is allowed to view them. Otherwise, such files will not
  95          *                     be found.
  96          */
  97         function findFile( $title, $options = array() ) {
  98                 if ( !is_array( $options ) ) {
  99                         // MW 1.15 compat
 100                         $time = $options;
 101                 } else {
 102                         $time = isset( $options['time'] ) ? $options['time'] : false;
 103                 }
 104                 if ( !($title instanceof Title) ) {
 105                         $title = Title::makeTitleSafe( NS_FILE, $title );
 106                         if ( !is_object( $title ) ) {
 107                                 return false;
 108                         }
 109                 }
 110                 # First try the current version of the file to see if it precedes the timestamp
 111                 $img = $this->newFile( $title );
 112                 if ( !$img ) {
 113                         return false;
 114                 }
 115                 if ( $img->exists() && ( !$time || $img->getTimestamp() == $time ) ) {
 116                         return $img;
 117                 }
 118                 # Now try an old version of the file
 119                 if ( $time !== false ) {
 120                         $img = $this->newFile( $title, $time );
 121                         if ( $img && $img->exists() ) {
 122                                 if ( !$img->isDeleted(File::DELETED_FILE) ) {
 123                                         return $img;
 124                                 } else if ( !empty( $options['private'] )  && $img->userCan(File::DELETED_FILE) ) {
 125                                         return $img;
 126                                 }
 127                         }
 128                 }
 129
 130                 # Now try redirects
 131                 if ( !empty( $options['ignoreRedirect'] ) ) {
 132                         return false;
 133                 }
 134                 $redir = $this->checkRedirect( $title );
 135                 if( $redir && $redir->getNamespace() == NS_FILE) {
 136                         $img = $this->newFile( $redir );
 137                         if( !$img ) {
 138                                 return false;
 139                         }
 140                         if( $img->exists() ) {
 141                                 $img->redirectedFrom( $title->getDBkey() );
 142                                 return $img;
 143                         }
 144                 }
 145                 return false;
 146         }
 147
 148         /*
 149          * Find many files at once.
 150          * @param $items An array of titles, or an array of findFile() options with
 151          *    the "title" option giving the title. Example:
 152          *
 153          *     $findItem = array( 'title' => $title, 'private' => true );
 154          *     $findBatch = array( $findItem );
 155          *     $repo->findFiles( $findBatch );
 156          */
 157         function findFiles( $items ) {
 158                 $result = array();
 159                 foreach ( $items as $index => $item ) {
 160                         if ( is_array( $item ) ) {
 161                                 $title = $item['title'];
 162                                 $options = $item;
 163                                 unset( $options['title'] );
 164                         } else {
 165                                 $title = $item;
 166                                 $options = array();
 167                         }
 168                         $file = $this->findFile( $title, $options );
 169                         if ( $file )
 170                                 $result[$file->getTitle()->getDBkey()] = $file;
 171                 }
 172                 return $result;
 173         }
 174
 175         /**
 176          * Create a new File object from the local repository
 177          * @param $sha1 Mixed: SHA-1 key
 178          * @param $time Mixed: time at which the image was uploaded.
 179          *              If this is specified, the returned object will be an
 180          *              of the repository's old file class instead of a current
 181          *              file. Repositories not supporting version control should
 182          *              return false if this parameter is set.
 183          */
 184         function newFileFromKey( $sha1, $time = false ) {
 185                 if ( $time ) {
 186                         if ( $this->oldFileFactoryKey ) {
 187                                 return call_user_func( $this->oldFileFactoryKey, $sha1, $this, $time );
 188                         } else {
 189                                 return false;
 190                         }
 191                 } else {
 192                         return call_user_func( $this->fileFactoryKey, $sha1, $this );
 193                 }
 194         }
 195
 196         /**
 197          * Find an instance of the file with this key, created at the specified time
 198          * Returns false if the file does not exist. Repositories not supporting
 199          * version control should return false if the time is specified.
 200          *
 201          * @param $sha1 String
 202          * @param $options Option array, same as findFile().
 203          */
 204         function findFileFromKey( $sha1, $options = array() ) {
 205                 if ( !is_array( $options ) ) {
 206                         # MW 1.15 compat
 207                         $time = $options;
 208                 } else {
 209                         $time = isset( $options['time'] ) ? $options['time'] : false;
 210                 }
 211
 212                 # First try the current version of the file to see if it precedes the timestamp
 213                 $img = $this->newFileFromKey( $sha1 );
 214                 if ( !$img ) {
 215                         return false;
 216                 }
 217                 if ( $img->exists() && ( !$time || $img->getTimestamp() == $time ) ) {
 218                         return $img;
 219                 }
 220                 # Now try an old version of the file
 221                 if ( $time !== false ) {
 222                         $img = $this->newFileFromKey( $sha1, $time );
 223                         if ( $img->exists() ) {
 224                                 if ( !$img->isDeleted(File::DELETED_FILE) ) {
 225                                         return $img;
 226                                 } else if ( !empty( $options['private'] ) && $img->userCan(File::DELETED_FILE) ) {
 227                                         return $img;
 228                                 }
 229                         }
 230                 }
 231                 return false;
 232         }
 233
 234         /**
 235          * Get the URL of thumb.php
 236          */
 237         function getThumbScriptUrl() {
 238                 return $this->thumbScriptUrl;
 239         }
 240
 241         /**
 242          * Get the URL corresponding to one of the four basic zones
 243          * @param $zone String: one of: public, deleted, temp, thumb
 244          * @return String or false
 245          */
 246         function getZoneUrl( $zone ) {
 247                 return false;
 248         }
 249
 250         /**
 251          * Returns true if the repository can transform files via a 404 handler
 252          */
 253         function canTransformVia404() {
 254                 return $this->transformVia404;
 255         }
 256
 257         /**
 258          * Get the name of an image from its title object
 259          */
 260         function getNameFromTitle( $title ) {
 261                 if ( $this->initialCapital != MWNamespace::isCapitalized( NS_FILE ) ) {
 262                         global $wgContLang;
 263                         $name = $title->getUserCaseDBKey();
 264                         if ( $this->initialCapital ) {
 265                                 $name = $wgContLang->ucfirst( $name );
 266                         }
 267                 } else {
 268                         $name = $title->getDBkey();
 269                 }
 270                 return $name;
 271         }
 272
 273         static function getHashPathForLevel( $name, $levels ) {
 274                 if ( $levels == 0 ) {
 275                         return '';
 276                 } else {
 277                         $hash = md5( $name );
 278                         $path = '';
 279                         for ( $i = 1; $i <= $levels; $i++ ) {
 280                                 $path .= substr( $hash, 0, $i ) . '/';
 281                         }
 282                         return $path;
 283                 }
 284         }
 285
 286         /**
 287          * Get a relative path including trailing slash, e.g. f/fa/
 288          * If the repo is not hashed, returns an empty string
 289          */
 290         function getHashPath( $name ) {
 291                 return self::getHashPathForLevel( $name, $this->hashLevels );
 292         }
 293
 294         /**
 295          * Get the name of this repository, as specified by $info['name]' to the constructor
 296          */
 297         function getName() {
 298                 return $this->name;
 299         }
 300
 301         /**
 302          * Make an url to this repo
 303          *
 304          * @param $query mixed Query string to append
 305          * @param $entry string Entry point; defaults to index
 306          * @return string
 307          */
 308         function makeUrl( $query = '', $entry = 'index' ) {
 309                 $ext = isset( $this->scriptExtension ) ? $this->scriptExtension : '.php';
 310                 return wfAppendQuery( "{$this->scriptDirUrl}/{$entry}{$ext}?", $query );
 311         }
 312
 313         /**
 314          * Get the URL of an image description page. May return false if it is
 315          * unknown or not applicable. In general this should only be called by the
 316          * File class, since it may return invalid results for certain kinds of
 317          * repositories. Use File::getDescriptionUrl() in user code.
 318          *
 319          * In particular, it uses the article paths as specified to the repository
 320          * constructor, whereas local repositories use the local Title functions.
 321          */
 322         function getDescriptionUrl( $name ) {
 323                 $encName = wfUrlencode( $name );
 324                 if ( !is_null( $this->descBaseUrl ) ) {
 325                         # "http://example.com/wiki/Image:"
 326                         return $this->descBaseUrl . $encName;
 327                 }
 328                 if ( !is_null( $this->articleUrl ) ) {
 329                         # "http://example.com/wiki/$1"
 330                         #
 331                         # We use "Image:" as the canonical namespace for
 332                         # compatibility across all MediaWiki versions.
 333                         return str_replace( '$1',
 334                                 "Image:$encName", $this->articleUrl );
 335                 }
 336                 if ( !is_null( $this->scriptDirUrl ) ) {
 337                         # "http://example.com/w"
 338                         #
 339                         # We use "Image:" as the canonical namespace for
 340                         # compatibility across all MediaWiki versions,
 341                         # and just sort of hope index.php is right. ;)
 342                         return $this->makeUrl( "title=Image:$encName" );
 343                 }
 344                 return false;
 345         }
 346
 347         /**
 348          * Get the URL of the content-only fragment of the description page. For
 349          * MediaWiki this means action=render. This should only be called by the
 350          * repository's file class, since it may return invalid results. User code
 351          * should use File::getDescriptionText().
 352          * @param $name String: name of image to fetch
 353          * @param $lang String: language to fetch it in, if any.
 354          */
 355         function getDescriptionRenderUrl( $name, $lang = null ) {
 356                 $query = 'action=render';
 357                 if ( !is_null( $lang ) ) {
 358                         $query .= '&uselang=' . $lang;
 359                 }
 360                 if ( isset( $this->scriptDirUrl ) ) {
 361                         return $this->makeUrl(
 362                                 'title=' .
 363                                 wfUrlencode( 'Image:' . $name ) .
 364                                 "&$query" );
 365                 } else {
 366                         $descUrl = $this->getDescriptionUrl( $name );
 367                         if ( $descUrl ) {
 368                                 return wfAppendQuery( $descUrl, $query );
 369                         } else {
 370                                 return false;
 371                         }
 372                 }
 373         }
 374
 375         /**
 376          * Get the URL of the stylesheet to apply to description pages
 377          * @return string
 378          */
 379         function getDescriptionStylesheetUrl() {
 380                 if ( $this->scriptDirUrl ) {
 381                         return self::makeUrl( 'title=MediaWiki:Filepage.css&' .
 382                                 wfArrayToCGI( Skin::getDynamicStylesheetQuery() ) );
 383                 }
 384         }
 385
 386         /**
 387          * Store a file to a given destination.
 388          *
 389          * @param $srcPath String: source path or virtual URL
 390          * @param $dstZone String: destination zone
 391          * @param $dstRel String: destination relative path
 392          * @param $flags Integer: bitwise combination of the following flags:
 393          *     self::DELETE_SOURCE     Delete the source file after upload
 394          *     self::OVERWRITE         Overwrite an existing destination file instead of failing
 395          *     self::OVERWRITE_SAME    Overwrite the file if the destination exists and has the
 396          *                             same contents as the source
 397          * @return FileRepoStatus
 398          */
 399         function store( $srcPath, $dstZone, $dstRel, $flags = 0 ) {
 400                 $status = $this->storeBatch( array( array( $srcPath, $dstZone, $dstRel ) ), $flags );
 401                 if ( $status->successCount == 0 ) {
 402                         $status->ok = false;
 403                 }
 404                 return $status;
 405         }
 406
 407         /**
 408          * Store a batch of files
 409          *
 410          * @param $triplets Array: (src,zone,dest) triplets as per store()
 411          * @param $flags Integer: flags as per store
 412          */
 413         abstract function storeBatch( $triplets, $flags = 0 );
 414
 415         /**
 416          * Pick a random name in the temp zone and store a file to it.
 417          * Returns a FileRepoStatus object with the URL in the value.
 418          *
 419          * @param $originalName String: the base name of the file as specified
 420          *     by the user. The file extension will be maintained.
 421          * @param $srcPath String: the current location of the file.
 422          */
 423         abstract function storeTemp( $originalName, $srcPath );
 424
 425
 426         /**
 427          * Append the contents of the source path to the given file.
 428          * @param $srcPath String: location of the source file
 429          * @param $toAppendPath String: path to append to.
 430          * @param $flags Integer: bitfield, may be FileRepo::DELETE_SOURCE to indicate
 431          *        that the source file should be deleted if possible
 432          * @return mixed Status or false
 433          */
 434         abstract function append( $srcPath, $toAppendPath, $flags = 0 );
 435
 436         /**
 437          * Remove a temporary file or mark it for garbage collection
 438          * @param $virtualUrl String: the virtual URL returned by storeTemp
 439          * @return Boolean: true on success, false on failure
 440          * STUB
 441          */
 442         function freeTemp( $virtualUrl ) {
 443                 return true;
 444         }
 445
 446         /**
 447          * Copy or move a file either from the local filesystem or from an mwrepo://
 448          * virtual URL, into this repository at the specified destination location.
 449          *
 450          * Returns a FileRepoStatus object. On success, the value contains "new" or
 451          * "archived", to indicate whether the file was new with that name.
 452          *
 453          * @param $srcPath String: the source path or URL
 454          * @param $dstRel String: the destination relative path
 455          * @param $archiveRel String: rhe relative path where the existing file is to
 456          *        be archived, if there is one. Relative to the public zone root.
 457          * @param $flags Integer: bitfield, may be FileRepo::DELETE_SOURCE to indicate
 458          *        that the source file should be deleted if possible
 459          */
 460         function publish( $srcPath, $dstRel, $archiveRel, $flags = 0 ) {
 461                 $status = $this->publishBatch( array( array( $srcPath, $dstRel, $archiveRel ) ), $flags );
 462                 if ( $status->successCount == 0 ) {
 463                         $status->ok = false;
 464                 }
 465                 if ( isset( $status->value[0] ) ) {
 466                         $status->value = $status->value[0];
 467                 } else {
 468                         $status->value = false;
 469                 }
 470                 return $status;
 471         }
 472
 473         /**
 474          * Publish a batch of files
 475          * @param $triplets Array: (source,dest,archive) triplets as per publish()
 476          * @param $flags Integer: bitfield, may be FileRepo::DELETE_SOURCE to indicate
 477          *        that the source files should be deleted if possible
 478          */
 479         abstract function publishBatch( $triplets, $flags = 0 );
 480
 481         function fileExists( $file, $flags = 0 ) {
 482                 $result = $this->fileExistsBatch( array( $file ), $flags );
 483                 return $result[0];
 484         }
 485
 486         /**
 487          * Checks existence of an array of files.
 488          *
 489          * @param $files Array: URLs (or paths) of files to check
 490          * @param $flags Integer: bitwise combination of the following flags:
 491          *     self::FILES_ONLY     Mark file as existing only if it is a file (not directory)
 492          * @return Either array of files and existence flags, or false
 493          */
 494         abstract function fileExistsBatch( $files, $flags = 0 );
 495
 496         /**
 497          * Move a group of files to the deletion archive.
 498          *
 499          * If no valid deletion archive is configured, this may either delete the
 500          * file or throw an exception, depending on the preference of the repository.
 501          *
 502          * The overwrite policy is determined by the repository -- currently FSRepo
 503          * assumes a naming scheme in the deleted zone based on content hash, as
 504          * opposed to the public zone which is assumed to be unique.
 505          *
 506          * @param $sourceDestPairs Array of source/destination pairs. Each element
 507          *        is a two-element array containing the source file path relative to the
 508          *        public root in the first element, and the archive file path relative
 509          *        to the deleted zone root in the second element.
 510          * @return FileRepoStatus
 511          */
 512         abstract function deleteBatch( $sourceDestPairs );
 513
 514         /**
 515          * Move a file to the deletion archive.
 516          * If no valid deletion archive exists, this may either delete the file
 517          * or throw an exception, depending on the preference of the repository
 518          * @param $srcRel Mixed: relative path for the file to be deleted
 519          * @param $archiveRel Mixed: relative path for the archive location.
 520          *        Relative to a private archive directory.
 521          * @return FileRepoStatus object
 522          */
 523         function delete( $srcRel, $archiveRel ) {
 524                 return $this->deleteBatch( array( array( $srcRel, $archiveRel ) ) );
 525         }
 526
 527         /**
 528          * Get properties of a file with a given virtual URL
 529          * The virtual URL must refer to this repo
 530          * Properties should ultimately be obtained via File::getPropsFromPath()
 531          */
 532         abstract function getFileProps( $virtualUrl );
 533
 534         /**
 535          * Call a callback function for every file in the repository
 536          * May use either the database or the filesystem
 537          * STUB
 538          */
 539         function enumFiles( $callback ) {
 540                 throw new MWException( 'enumFiles is not supported by ' . get_class( $this ) );
 541         }
 542
 543         /**
 544          * Determine if a relative path is valid, i.e. not blank or involving directory traveral
 545          */
 546         function validateFilename( $filename ) {
 547                 if ( strval( $filename ) == '' ) {
 548                         return false;
 549                 }
 550                 if ( wfIsWindows() ) {
 551                         $filename = strtr( $filename, '\\', '/' );
 552                 }
 553                 /**
 554                  * Use the same traversal protection as Title::secureAndSplit()
 555                  */
 556                 if ( strpos( $filename, '.' ) !== false &&
 557                      ( $filename === '.' || $filename === '..' ||
 558                        strpos( $filename, './' ) === 0  ||
 559                        strpos( $filename, '../' ) === 0 ||
 560                        strpos( $filename, '/./' ) !== false ||
 561                        strpos( $filename, '/../' ) !== false ) )
 562                 {
 563                         return false;
 564                 } else {
 565                         return true;
 566                 }
 567         }
 568
 569         /**#@+
 570          * Path disclosure protection functions
 571          */
 572         function paranoidClean( $param ) { return '[hidden]'; }
 573         function passThrough( $param ) { return $param; }
 574
 575         /**
 576          * Get a callback function to use for cleaning error message parameters
 577          */
 578         function getErrorCleanupFunction() {
 579                 switch ( $this->pathDisclosureProtection ) {
 580                         case 'none':
 581                                 $callback = array( $this, 'passThrough' );
 582                                 break;
 583                         default: // 'paranoid'
 584                                 $callback = array( $this, 'paranoidClean' );
 585                 }
 586                 return $callback;
 587         }
 588         /**#@-*/
 589
 590         /**
 591          * Create a new fatal error
 592          */
 593         function newFatal( $message /*, parameters...*/ ) {
 594                 $params = func_get_args();
 595                 array_unshift( $params, $this );
 596                 return call_user_func_array( array( 'FileRepoStatus', 'newFatal' ), $params );
 597         }
 598
 599         /**
 600          * Create a new good result
 601          */
 602         function newGood( $value = null ) {
 603                 return FileRepoStatus::newGood( $this, $value );
 604         }
 605
 606         /**
 607          * Delete files in the deleted directory if they are not referenced in the filearchive table
 608          * STUB
 609          */
 610         function cleanupDeletedBatch( $storageKeys ) {}
 611
 612         /**
 613          * Checks if there is a redirect named as $title. If there is, return the
 614          * title object. If not, return false.
 615          * STUB
 616          *
 617          * @param $title Title of image
 618          */
 619         function checkRedirect( $title ) {
 620                 return false;
 621         }
 622
 623         /**
 624          * Invalidates image redirect cache related to that image
 625          * Doesn't do anything for repositories that don't support image redirects.
 626          *
 627          * STUB
 628          * @param $title Title of image
 629          */
 630         function invalidateImageRedirect( $title ) {}
 631
 632         /**
 633          * Get an array or iterator of file objects for files that have a given
 634          * SHA-1 content hash.
 635          *
 636          * STUB
 637          */
 638         function findBySha1( $hash ) {
 639                 return array();
 640         }
 641
 642         /**
 643          * Get the human-readable name of the repo.
 644          * @return string
 645          */
 646         public function getDisplayName() {
 647                 // We don't name our own repo, return nothing
 648                 if ( $this->isLocal() ) {
 649                         return null;
 650                 }
 651                 // 'shared-repo-name-wikimediacommons' is used when $wgUseInstantCommons = true
 652                 $repoName = wfMsg( 'shared-repo-name-' . $this->name );
 653                 if ( !wfEmptyMsg( 'shared-repo-name-' . $this->name, $repoName ) ) {
 654                         return $repoName;
 655                 }
 656                 return wfMsg( 'shared-repo' );
 657         }
 658
 659         /**
 660          * Returns true if this the local file repository.
 661          *
 662          * @return bool
 663          */
 664         function isLocal() {
 665                 return $this->getName() == 'local';
 666         }
 667
 668
 669         /**
 670          * Get a key on the primary cache for this repository.
 671          * Returns false if the repository's cache is not accessible at this site.
 672          * The parameters are the parts of the key, as for wfMemcKey().
 673          *
 674          * STUB
 675          */
 676         function getSharedCacheKey( /*...*/ ) {
 677                 return false;
 678         }
 679
 680         /**
 681          * Get a key for this repo in the local cache domain. These cache keys are
 682          * not shared with remote instances of the repo.
 683          * The parameters are the parts of the key, as for wfMemcKey().
 684          */
 685         function getLocalCacheKey( /*...*/ ) {
 686                 $args = func_get_args();
 687                 array_unshift( $args, 'filerepo', $this->getName() );
 688                 return call_user_func_array( 'wfMemcKey', $args );
 689         }
 690 }