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