includes/upload/UploadBase.php

   1 <?php
   2 /**
   3  * @file
   4  * @ingroup upload
   5  *
   6  * UploadBase and subclasses are the backend of MediaWiki's file uploads.
   7  * The frontends are formed by ApiUpload and SpecialUpload.
   8  *
   9  * See also includes/docs/upload.txt
  10  *
  11  * @author Brion Vibber
  12  * @author Bryan Tong Minh
  13  * @author Michael Dale
  14  */
  15
  16 abstract class UploadBase {
  17         protected $mTempPath;
  18         protected $mDesiredDestName, $mDestName, $mRemoveTempFile, $mSourceType;
  19         protected $mTitle = false, $mTitleError = 0;
  20         protected $mFilteredName, $mFinalExtension;
  21
  22         const SUCCESS = 0;
  23         const OK = 0;
  24         const BEFORE_PROCESSING = 1;
  25         const LARGE_FILE_SERVER = 2;
  26         const EMPTY_FILE = 3;
  27         const MIN_LENGTH_PARTNAME = 4;
  28         const ILLEGAL_FILENAME = 5;
  29         const PROTECTED_PAGE = 6;
  30         const OVERWRITE_EXISTING_FILE = 7;
  31         const FILETYPE_MISSING = 8;
  32         const FILETYPE_BADTYPE = 9;
  33         const VERIFICATION_ERROR = 10;
  34         const UPLOAD_VERIFICATION_ERROR = 11;
  35         const UPLOAD_WARNING = 12;
  36         const INTERNAL_ERROR = 13;
  37         const MIN_LENGHT_PARTNAME = 14;
  38
  39         const SESSION_VERSION = 2;
  40
  41         /**
  42          * Returns true if uploads are enabled.
  43          * Can be override by subclasses.
  44          */
  45         public static function isEnabled() {
  46                 global $wgEnableUploads;
  47                 if ( !$wgEnableUploads )
  48                         return false;
  49
  50                 # Check php's file_uploads setting
  51                 if( !wfIniGetBool( 'file_uploads' ) ) {
  52                         return false;
  53                 }
  54                 return true;
  55         }
  56
  57         /**
  58          * Returns true if the user can use this upload module or else a string
  59          * identifying the missing permission.
  60          * Can be overriden by subclasses.
  61          */
  62         public static function isAllowed( $user ) {
  63                 if( !$user->isAllowed( 'upload' ) )
  64                         return 'upload';
  65                 return true;
  66         }
  67
  68         // Upload handlers. Should probably just be a global
  69         static $uploadHandlers = array( 'Stash', 'File', 'Url' );
  70
  71         /**
  72          * Create a form of UploadBase depending on wpSourceType and initializes it
  73          */
  74         public static function createFromRequest( &$request, $type = null ) {
  75                 $type = $type ? $type : $request->getVal( 'wpSourceType' );
  76
  77                 if( !$type )
  78                         return null;
  79
  80                 // Get the upload class
  81                 $type = ucfirst( $type );
  82                 $className = 'UploadFrom' . $type;
  83                 wfDebug( __METHOD__ . ": class name: $className\n" );
  84                 if( !in_array( $type, self::$uploadHandlers ) )
  85                         return null;
  86
  87                 // Check whether this upload class is enabled
  88                 if( !call_user_func( array( $className, 'isEnabled' ) ) )
  89                         return null;
  90
  91                 // Check whether the request is valid
  92                 if( !call_user_func( array( $className, 'isValidRequest' ), $request ) )
  93                         return null;
  94
  95                 $handler = new $className;
  96
  97                 $handler->initializeFromRequest( $request );
  98                 return $handler;
  99         }
 100
 101         /**
 102          * Check whether a request if valid for this handler
 103          */
 104         public static function isValidRequest( $request ) {
 105                 return false;
 106         }
 107
 108         public function __construct() {}
 109
 110         /**
 111          * Do the real variable initialization
 112          */
 113         public function initialize( $name, $tempPath, $fileSize, $removeTempFile = false ) {
 114                 $this->mDesiredDestName = $name;
 115                 $this->mTempPath = $tempPath;
 116                 $this->mFileSize = $fileSize;
 117                 $this->mRemoveTempFile = $removeTempFile;
 118         }
 119
 120         /**
 121          * Initialize from a WebRequest. Override this in a subclass.
 122          */
 123         public abstract function initializeFromRequest( &$request );
 124
 125         /**
 126          * Fetch the file. Usually a no-op
 127          */
 128         public function fetchFile() {
 129                 return Status::newGood();
 130         }
 131
 132         /**
 133          * Return the file size
 134          */
 135         public function isEmptyFile(){
 136                 return empty( $this->mFileSize );
 137         }
 138
 139         /**
 140          * Verify whether the upload is sane.
 141          * Returns self::OK or else an array with error information
 142          */
 143         public function verifyUpload() {
 144                 /**
 145                  * If there was no filename or a zero size given, give up quick.
 146                  */
 147                 if( $this->isEmptyFile() )
 148                         return array( 'status' => self::EMPTY_FILE );
 149
 150                 $nt = $this->getTitle();
 151                 if( is_null( $nt ) ) {
 152                         $result = array( 'status' => $this->mTitleError );
 153                         if( $this->mTitleError == self::ILLEGAL_FILENAME )
 154                                 $result['filtered'] = $this->mFilteredName;
 155                         if ( $this->mTitleError == self::FILETYPE_BADTYPE )
 156                                 $result['finalExt'] = $this->mFinalExtension;
 157                         return $result;
 158                 }
 159                 $this->mDestName = $this->getLocalFile()->getName();
 160
 161                 /**
 162                  * In some cases we may forbid overwriting of existing files.
 163                  */
 164                 $overwrite = $this->checkOverwrite();
 165                 if( $overwrite !== true )
 166                         return array( 'status' => self::OVERWRITE_EXISTING_FILE, 'overwrite' => $overwrite );
 167
 168                 /**
 169                  * Look at the contents of the file; if we can recognize the
 170                  * type but it's corrupt or data of the wrong type, we should
 171                  * probably not accept it.
 172                  */
 173                 $verification = $this->verifyFile();
 174
 175                 if( $verification !== true ) {
 176                         if( !is_array( $verification ) )
 177                                 $verification = array( $verification );
 178                         $verification['status'] = self::VERIFICATION_ERROR;
 179                         return $verification;
 180                 }
 181
 182                 $error = '';
 183                 if( !wfRunHooks( 'UploadVerification',
 184                                 array( $this->mDestName, $this->mTempPath, &$error ) ) ) {
 185                         return array( 'status' => self::UPLOAD_VERIFICATION_ERROR, 'error' => $error );
 186                 }
 187
 188                 return self::OK;
 189         }
 190
 191         /**
 192          * Verifies that it's ok to include the uploaded file
 193          *
 194          * FIXME: this function seems to intermixes tmpfile and $this->mTempPath .. no idea why this is
 195          *
 196          * @param string $tmpfile the full path of the temporary file to verify
 197          * @return mixed true of the file is verified, a string or array otherwise.
 198          */
 199         protected function verifyFile() {
 200                 $this->mFileProps = File::getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
 201                 $this->checkMacBinary();
 202
 203                 #magically determine mime type
 204                 $magic = MimeMagic::singleton();
 205                 $mime = $magic->guessMimeType( $this->mTempFile, false );
 206
 207                 #check mime type, if desired
 208                 global $wgVerifyMimeType;
 209                 if ( $wgVerifyMimeType ) {
 210                         global $wgMimeTypeBlacklist;
 211                         if ( $this->checkFileExtension( $mime, $wgMimeTypeBlacklist ) )
 212                                 return array( 'filetype-badmime', $mime );
 213
 214                         # Check IE type
 215                         $fp = fopen( $this->mTempFile, 'rb' );
 216                         $chunk = fread( $fp, 256 );
 217                         fclose( $fp );
 218                         $extMime = $magic->guessTypesForExtension( $this->mFinalExtension );
 219                         $ieTypes = $magic->getIEMimeTypes( $tmpfile, $chunk, $extMime );
 220                         foreach ( $ieTypes as $ieType ) {
 221                                 if ( $this->checkFileExtension( $ieType, $wgMimeTypeBlacklist ) ) {
 222                                         return array( 'filetype-bad-ie-mime', $ieType );
 223                                 }
 224                         }
 225                 }
 226
 227                 #check for htmlish code and javascript
 228                 if( self::detectScript( $tmpfile, $mime, $this->mFinalExtension ) ) {
 229                         return 'uploadscripted';
 230                 }
 231                 if( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
 232                         if( self::detectScriptInSvg( $tmpfile ) ) {
 233                                 return 'uploadscripted';
 234                         }
 235                 }
 236
 237                 /**
 238                 * Scan the uploaded file for viruses
 239                 */
 240                 $virus = $this->detectVirus( $tmpfile );
 241                 if ( $virus ) {
 242                         return array( 'uploadvirus', $virus );
 243                 }
 244                 wfDebug( __METHOD__ . ": all clear; passing.\n" );
 245                 return true;
 246         }
 247
 248         /**
 249          * Check whether the user can edit, upload and create the image.
 250          *
 251          * @param User $user the user to verify the permissions against
 252          * @return mixed An array as returned by getUserPermissionsErrors or true
 253          *               in case the user has proper permissions.
 254          */
 255         public function verifyPermissions( $user ) {
 256                 /**
 257                  * If the image is protected, non-sysop users won't be able
 258                  * to modify it by uploading a new revision.
 259                  */
 260                 $nt = $this->getTitle();
 261                 if( is_null( $nt ) )
 262                         return true;
 263                 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
 264                 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
 265                 $permErrorsCreate = ( $nt->exists() ? array() : $nt->getUserPermissionsErrors( 'create', $user ) );
 266                 if( $permErrors || $permErrorsUpload || $permErrorsCreate ) {
 267                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
 268                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
 269                         return $permErrors;
 270                 }
 271                 return true;
 272         }
 273
 274         /**
 275          * Check for non fatal problems with the file
 276          *
 277          * @return array Array of warnings
 278          */
 279         public function checkWarnings() {
 280                 $warning = array();
 281
 282                 $localFile = $this->getLocalFile();
 283                 $filename = $localFile->getName();
 284                 $n = strrpos( $filename, '.' );
 285                 $partname = $n ? substr( $filename, 0, $n ) : $filename;
 286
 287                 /*
 288                  * Check whether the resulting filename is different from the desired one,
 289                  * but ignore things like ucfirst() and spaces/underscore things
 290                  */
 291                 $comparableName = str_replace( ' ', '_', $this->mDesiredDestName );
 292                 global $wgCapitalLinks, $wgContLang;
 293                 if ( $wgCapitalLinks ) {
 294                         $comparableName = $wgContLang->ucfirst( $comparableName );
 295                 }
 296                 if( $this->mDesiredDestName != $filename && $comparableName != $filename )
 297                         $warning['badfilename'] = $filename;
 298
 299                 // Check whether the file extension is on the unwanted list
 300                 global $wgCheckFileExtensions, $wgFileExtensions;
 301                 if ( $wgCheckFileExtensions ) {
 302                         if ( !$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) )
 303                                 $warning['filetype-unwanted-type'] = $this->mFinalExtension;
 304                 }
 305
 306                 global $wgUploadSizeWarning;
 307                 if ( $wgUploadSizeWarning && ( $this->mFileSize > $wgUploadSizeWarning ) )
 308                         $warning['large-file'] = $wgUploadSizeWarning;
 309
 310                 if ( $this->mFileSize == 0 )
 311                         $warning['emptyfile'] = true;
 312
 313
 314                 $exists = self::getExistsWarning( $localFile );
 315                 if( $exists !== false )
 316                         $warning['exists'] = $exists;
 317
 318                 // Check whether this may be a thumbnail
 319                 if( $exists !== false && $exists[0] != 'thumb'
 320                                 && self::isThumbName( $filename ) ){
 321                         // Make the title
 322                         $nt = $this->getTitle();
 323                         $warning['file-thumbnail-no'] = substr( $filename, 0,
 324                                 strpos( $nt->getText() , '-' ) +1 );
 325                 }
 326
 327                 // Check dupes against existing files
 328                 $hash = File::sha1Base36( $this->mTempPath );
 329                 $dupes = RepoGroup::singleton()->findBySha1( $hash );
 330                 $title = $this->getTitle();
 331                 // Remove all matches against self
 332                 foreach ( $dupes as $key => $dupe ) {
 333                         if( $title->equals( $dupe->getTitle() ) )
 334                                 unset( $dupes[$key] );
 335                 }
 336                 if( $dupes )
 337                         $warning['duplicate'] = $dupes;
 338
 339                 // Check dupes against archives
 340                 $archivedImage = new ArchivedFile( null, 0, "{$hash}.{$this->mFinalExtension}" );
 341                 if ( $archivedImage->getID() > 0 )
 342                         $warning['duplicate-archive'] = $archivedImage->getName();
 343
 344                 $filenamePrefixBlacklist = self::getFilenamePrefixBlacklist();
 345                 foreach( $filenamePrefixBlacklist as $prefix ) {
 346                         if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
 347                                 $warning['filename-bad-prefix'] = $prefix;
 348                                 break;
 349                         }
 350                 }
 351
 352                 # If the file existed before and was deleted, warn the user of this
 353                 # Don't bother doing so if the file exists now, however
 354                 if( $localFile->wasDeleted() && !$localFile->exists() )
 355                         $warning['filewasdeleted'] = $localFile->getTitle();
 356
 357                 return $warning;
 358         }
 359
 360         /**
 361          * Really perform the upload. Stores the file in the local repo, watches
 362          * if necessary and runs the UploadComplete hook.
 363          *
 364          * @return mixed Status indicating the whether the upload succeeded.
 365          */
 366         public function performUpload( $comment, $pageText, $watch, $user ) {
 367                 wfDebug( "\n\n\performUpload: sum:" . $comment . ' c: ' . $pageText . ' w:' . $watch );
 368                 $status = $this->getLocalFile()->upload( $this->mTempPath, $comment, $pageText,
 369                         File::DELETE_SOURCE, $this->mFileProps, false, $user );
 370
 371                 if( $status->isGood() && $watch )
 372                         $user->addWatch( $this->getLocalFile()->getTitle() );
 373
 374                 if( $status->isGood() )
 375                         wfRunHooks( 'UploadComplete', array( &$this ) );
 376
 377                 return $status;
 378         }
 379
 380         /**
 381          * Returns the title of the file to be uploaded. Sets mTitleError in case
 382          * the name was illegal.
 383          *
 384          * @return Title The title of the file or null in case the name was illegal
 385          */
 386         public function getTitle() {
 387                 if ( $this->mTitle !== false )
 388                         return $this->mTitle;
 389
 390                 /**
 391                  * Chop off any directories in the given filename. Then
 392                  * filter out illegal characters, and try to make a legible name
 393                  * out of it. We'll strip some silently that Title would die on.
 394                  */
 395                 $basename = $this->mDesiredDestName;
 396
 397                 $this->mFilteredName = wfStripIllegalFilenameChars( $basename );
 398                 /* Normalize to title form before we do any further processing */
 399                 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 400                 if( is_null( $nt ) ) {
 401                         $this->mTitleError = self::ILLEGAL_FILENAME;
 402                         return $this->mTitle = null;
 403                 }
 404                 $this->mFilteredName = $nt->getDBkey();
 405
 406                 /**
 407                  * We'll want to blacklist against *any* 'extension', and use
 408                  * only the final one for the whitelist.
 409                  */
 410                 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName );
 411
 412                 if( count( $ext ) ) {
 413                         $this->mFinalExtension = trim( $ext[count( $ext ) - 1] );
 414                 } else {
 415                         $this->mFinalExtension = '';
 416                 }
 417
 418                 /* Don't allow users to override the blacklist (check file extension) */
 419                 global $wgCheckFileExtensions, $wgStrictFileExtensions;
 420                 global $wgFileExtensions, $wgFileBlacklist;
 421                 if ( $this->mFinalExtension == '' ) {
 422                         $this->mTitleError = self::FILETYPE_MISSING;
 423                         return $this->mTitle = null;
 424                 } elseif ( $this->checkFileExtensionList( $ext, $wgFileBlacklist ) ||
 425                                 ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
 426                                         !$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) ) ) {
 427                         $this->mTitleError = self::FILETYPE_BADTYPE;
 428                         return $this->mTitle = null;
 429                 }
 430
 431                 # If there was more than one "extension", reassemble the base
 432                 # filename to prevent bogus complaints about length
 433                 if( count( $ext ) > 1 ) {
 434                         for( $i = 0; $i < count( $ext ) - 1; $i++ )
 435                                 $partname .= '.' . $ext[$i];
 436                 }
 437
 438                 if( strlen( $partname ) < 1 ) {
 439                         $this->mTitleError =  self::MIN_LENGTH_PARTNAME;
 440                         return $this->mTitle = null;
 441                 }
 442
 443                 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 444                 if( is_null( $nt ) ) {
 445                         $this->mTitleError = self::ILLEGAL_FILENAME;
 446                         return $this->mTitle = null;
 447                 }
 448                 return $this->mTitle = $nt;
 449         }
 450
 451         /**
 452          * Return the local file and initializes if necessary.
 453          */
 454         public function getLocalFile() {
 455                 if( is_null( $this->mLocalFile ) ) {
 456                         $nt = $this->getTitle();
 457                         $this->mLocalFile = is_null( $nt ) ? null : wfLocalFile( $nt );
 458                 }
 459                 return $this->mLocalFile;
 460         }
 461
 462         /**
 463          * Stash a file in a temporary directory for later processing
 464          * after the user has confirmed it.
 465          *
 466          * If the user doesn't explicitly cancel or accept, these files
 467          * can accumulate in the temp directory.
 468          *
 469          * @param string $saveName - the destination filename
 470          * @param string $tempName - the source temporary file to save
 471          * @return string - full path the stashed file, or false on failure
 472          * @access private
 473          */
 474         protected function saveTempUploadedFile( $saveName, $tempName ) {
 475                 $repo = RepoGroup::singleton()->getLocalRepo();
 476                 $status = $repo->storeTemp( $saveName, $tempName );
 477                 return $status;
 478         }
 479
 480         /**
 481          * Append a file to a stashed file.
 482          *
 483          * @param string $srcPath Path to file to append from
 484          * @param string $toAppendPath Path to file to append to
 485          * @return Status Status
 486          */
 487         public function appendToUploadFile( $srcPath, $toAppendPath ){
 488                 $repo = RepoGroup::singleton()->getLocalRepo();
 489                 $status = $repo->append( $srcPath, $toAppendPath );
 490                 return $status;
 491         }
 492
 493         /**
 494          * Stash a file in a temporary directory for later processing,
 495          * and save the necessary descriptive info into the session.
 496          * Returns a key value which will be passed through a form
 497          * to pick up the path info on a later invocation.
 498          *
 499          * @return int Session key
 500          */
 501         public function stashSession() {
 502                 $status = $this->saveTempUploadedFile( $this->mDestName, $this->mTempPath );
 503                 if( !$status->isOK() ) {
 504                         # Couldn't save the file.
 505                         return false;
 506                 }
 507                 if(!isset($_SESSION))
 508                         session_start(); // start up the session (might have been previously closed to prevent php session locking)
 509                 $key = $this->getSessionKey();
 510                 $_SESSION['wsUploadData'][$key] = array(
 511                         'mTempPath'       => $status->value,
 512                         'mFileSize'       => $this->mFileSize,
 513                         'mFileProps'      => $this->mFileProps,
 514                         'version'         => self::SESSION_VERSION,
 515                 );
 516                 return $key;
 517         }
 518
 519         /**
 520          * Generate a random session key from stash in cases where we want to start an upload without much information
 521          */
 522         protected function getSessionKey(){
 523                 $key = mt_rand( 0, 0x7fffffff );
 524                 $_SESSION['wsUploadData'][$key] = array();
 525                 return $key;
 526         }
 527
 528         /**
 529          * Remove a temporarily kept file stashed by saveTempUploadedFile().
 530          * @return success
 531          */
 532         public function unsaveUploadedFile() {
 533                 $repo = RepoGroup::singleton()->getLocalRepo();
 534                 $success = $repo->freeTemp( $this->mTempPath );
 535                 return $success;
 536         }
 537
 538         /**
 539          * If we've modified the upload file we need to manually remove it
 540          * on exit to clean up.
 541          * @access private
 542          */
 543         public function cleanupTempFile() {
 544                 if ( $this->mRemoveTempFile && $this->mTempPath && file_exists( $this->mTempPath ) ) {
 545                         wfDebug( __METHOD__ . ": Removing temporary file {$this->mTempPath}\n" );
 546                         unlink( $this->mTempPath );
 547                 }
 548         }
 549
 550         public function getTempPath() {
 551                 return $this->mTempPath;
 552         }
 553
 554         /**
 555          * Split a file into a base name and all dot-delimited 'extensions'
 556          * on the end. Some web server configurations will fall back to
 557          * earlier pseudo-'extensions' to determine type and execute
 558          * scripts, so the blacklist needs to check them all.
 559          *
 560          * @return array
 561          */
 562         public static function splitExtensions( $filename ) {
 563                 $bits = explode( '.', $filename );
 564                 $basename = array_shift( $bits );
 565                 return array( $basename, $bits );
 566         }
 567
 568         /**
 569          * Perform case-insensitive match against a list of file extensions.
 570          * Returns true if the extension is in the list.
 571          *
 572          * @param string $ext
 573          * @param array $list
 574          * @return bool
 575          */
 576         public static function checkFileExtension( $ext, $list ) {
 577                 return in_array( strtolower( $ext ), $list );
 578         }
 579
 580         /**
 581          * Perform case-insensitive match against a list of file extensions.
 582          * Returns true if any of the extensions are in the list.
 583          *
 584          * @param array $ext
 585          * @param array $list
 586          * @return bool
 587          */
 588         public static function checkFileExtensionList( $ext, $list ) {
 589                 foreach( $ext as $e ) {
 590                         if( in_array( strtolower( $e ), $list ) ) {
 591                                 return true;
 592                         }
 593                 }
 594                 return false;
 595         }
 596
 597         /**
 598          * Checks if the mime type of the uploaded file matches the file extension.
 599          *
 600          * @param string $mime the mime type of the uploaded file
 601          * @param string $extension The filename extension that the file is to be served with
 602          * @return bool
 603          */
 604         public static function verifyExtension( $mime, $extension ) {
 605                 $magic = MimeMagic::singleton();
 606
 607                 if ( !$mime || $mime == 'unknown' || $mime == 'unknown/unknown' )
 608                         if ( !$magic->isRecognizableExtension( $extension ) ) {
 609                                 wfDebug( __METHOD__ . ": passing file with unknown detected mime type; " .
 610                                         "unrecognized extension '$extension', can't verify\n" );
 611                                 return true;
 612                         } else {
 613                                 wfDebug( __METHOD__ . ": rejecting file with unknown detected mime type; ".
 614                                         "recognized extension '$extension', so probably invalid file\n" );
 615                                 return false;
 616                         }
 617
 618                 $match = $magic->isMatchingExtension( $extension, $mime );
 619
 620                 if ( $match === NULL ) {
 621                         wfDebug( __METHOD__ . ": no file extension known for mime type $mime, passing file\n" );
 622                         return true;
 623                 } elseif( $match === true ) {
 624                         wfDebug( __METHOD__ . ": mime type $mime matches extension $extension, passing file\n" );
 625
 626                         #TODO: if it's a bitmap, make sure PHP or ImageMagic resp. can handle it!
 627                         return true;
 628
 629                 } else {
 630                         wfDebug( __METHOD__ . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
 631                         return false;
 632                 }
 633         }
 634
 635         /**
 636          * Heuristic for detecting files that *could* contain JavaScript instructions or
 637          * things that may look like HTML to a browser and are thus
 638          * potentially harmful. The present implementation will produce false positives in some situations.
 639          *
 640          * @param string $file Pathname to the temporary upload file
 641          * @param string $mime The mime type of the file
 642          * @param string $extension The extension of the file
 643          * @return bool true if the file contains something looking like embedded scripts
 644          */
 645         public static function detectScript( $file, $mime, $extension ) {
 646                 global $wgAllowTitlesInSVG;
 647
 648                 #ugly hack: for text files, always look at the entire file.
 649                 #For binary field, just check the first K.
 650
 651                 if( strpos( $mime,'text/' ) === 0 )
 652                         $chunk = file_get_contents( $file );
 653                 else {
 654                         $fp = fopen( $file, 'rb' );
 655                         $chunk = fread( $fp, 1024 );
 656                         fclose( $fp );
 657                 }
 658
 659                 $chunk = strtolower( $chunk );
 660
 661                 if( !$chunk )
 662                         return false;
 663
 664                 #decode from UTF-16 if needed (could be used for obfuscation).
 665                 if( substr( $chunk, 0, 2 ) == "\xfe\xff" )
 666                         $enc = "UTF-16BE";
 667                 elseif( substr( $chunk, 0, 2 ) == "\xff\xfe" )
 668                         $enc = "UTF-16LE";
 669                 else
 670                         $enc = NULL;
 671
 672                 if( $enc )
 673                         $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
 674
 675                 $chunk = trim( $chunk );
 676
 677                 #FIXME: convert from UTF-16 if necessarry!
 678                 wfDebug( __METHOD__ . ": checking for embedded scripts and HTML stuff\n" );
 679
 680                 #check for HTML doctype
 681                 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) )
 682                         return true;
 683
 684                 /**
 685                 * Internet Explorer for Windows performs some really stupid file type
 686                 * autodetection which can cause it to interpret valid image files as HTML
 687                 * and potentially execute JavaScript, creating a cross-site scripting
 688                 * attack vectors.
 689                 *
 690                 * Apple's Safari browser also performs some unsafe file type autodetection
 691                 * which can cause legitimate files to be interpreted as HTML if the
 692                 * web server is not correctly configured to send the right content-type
 693                 * (or if you're really uploading plain text and octet streams!)
 694                 *
 695                 * Returns true if IE is likely to mistake the given file for HTML.
 696                 * Also returns true if Safari would mistake the given file for HTML
 697                 * when served with a generic content-type.
 698                 */
 699                 $tags = array(
 700                         '<a',
 701                         '<body',
 702                         '<head',
 703                         '<html',   #also in safari
 704                         '<img',
 705                         '<pre',
 706                         '<script', #also in safari
 707                         '<table'
 708                 );
 709
 710                 if( !$wgAllowTitlesInSVG && $extension !== 'svg' && $mime !== 'image/svg' ) {
 711                         $tags[] = '<title';
 712                 }
 713
 714                 foreach( $tags as $tag ) {
 715                         if( false !== strpos( $chunk, $tag ) ) {
 716                                 return true;
 717                         }
 718                 }
 719
 720                 /*
 721                  * look for JavaScript
 722                  */
 723
 724                 #resolve entity-refs to look at attributes. may be harsh on big files... cache result?
 725                 $chunk = Sanitizer::decodeCharReferences( $chunk );
 726
 727                 #look for script-types
 728                 if( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) )
 729                         return true;
 730
 731                 #look for html-style script-urls
 732                 if( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) )
 733                         return true;
 734
 735                 #look for css-style script-urls
 736                 if( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) )
 737                         return true;
 738
 739                 wfDebug( __METHOD__ . ": no scripts found\n" );
 740                 return false;
 741         }
 742
 743         protected function detectScriptInSvg( $filename ) {
 744                 $check = new XmlTypeCheck( $filename, array( $this, 'checkSvgScriptCallback' ) );
 745                 return $check->filterMatch;
 746         }
 747
 748         /**
 749          * @todo Replace this with a whitelist filter!
 750          */
 751         public function checkSvgScriptCallback( $element, $attribs ) {
 752                 $stripped = $this->stripXmlNamespace( $element );
 753
 754                 if( $stripped == 'script' ) {
 755                         wfDebug( __METHOD__ . ": Found script element '$element' in uploaded file.\n" );
 756                         return true;
 757                 }
 758
 759                 foreach( $attribs as $attrib => $value ) {
 760                         $stripped = $this->stripXmlNamespace( $attrib );
 761                         if( substr( $stripped, 0, 2 ) == 'on' ) {
 762                                 wfDebug( __METHOD__ . ": Found script attribute '$attrib'='value' in uploaded file.\n" );
 763                                 return true;
 764                         }
 765                         if( $stripped == 'href' && strpos( strtolower( $value ), 'javascript:' ) !== false ) {
 766                                 wfDebug( __METHOD__ . ": Found script href attribute '$attrib'='$value' in uploaded file.\n" );
 767                                 return true;
 768                         }
 769                 }
 770         }
 771
 772         private function stripXmlNamespace( $name ) {
 773                 // 'http://www.w3.org/2000/svg:script' -> 'script'
 774                 $parts = explode( ':', strtolower( $name ) );
 775                 return array_pop( $parts );
 776         }
 777
 778         /**
 779          * Generic wrapper function for a virus scanner program.
 780          * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
 781          * $wgAntivirusRequired may be used to deny upload if the scan fails.
 782          *
 783          * @param string $file Pathname to the temporary upload file
 784          * @return mixed false if not virus is found, NULL if the scan fails or is disabled,
 785          *         or a string containing feedback from the virus scanner if a virus was found.
 786          *         If textual feedback is missing but a virus was found, this function returns true.
 787          */
 788         public static function detectVirus( $file ) {
 789                 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
 790
 791                 if ( !$wgAntivirus ) {
 792                         wfDebug( __METHOD__ . ": virus scanner disabled\n" );
 793                         return NULL;
 794                 }
 795
 796                 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
 797                         wfDebug( __METHOD__ . ": unknown virus scanner: $wgAntivirus\n" );
 798                         $wgOut->wrapWikiMsg( '<div class="error">$1</div>', array( 'virus-badscanner', $wgAntivirus ) );
 799                         return wfMsg( 'virus-unknownscanner' ) . " $wgAntivirus";
 800                 }
 801
 802                 # look up scanner configuration
 803                 $command = $wgAntivirusSetup[$wgAntivirus]["command"];
 804                 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]["codemap"];
 805                 $msgPattern = isset( $wgAntivirusSetup[$wgAntivirus]["messagepattern"] ) ?
 806                         $wgAntivirusSetup[$wgAntivirus]["messagepattern"] : null;
 807
 808                 if ( strpos( $command,"%f" ) === false ) {
 809                         # simple pattern: append file to scan
 810                         $command .= " " . wfEscapeShellArg( $file );
 811                 } else {
 812                         # complex pattern: replace "%f" with file to scan
 813                         $command = str_replace( "%f", wfEscapeShellArg( $file ), $command );
 814                 }
 815
 816                 wfDebug( __METHOD__ . ": running virus scan: $command \n" );
 817
 818                 # execute virus scanner
 819                 $exitCode = false;
 820
 821                 #NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
 822                 #      that does not seem to be worth the pain.
 823                 #      Ask me (Duesentrieb) about it if it's ever needed.
 824                 $output = array();
 825                 if ( wfIsWindows() ) {
 826                         exec( "$command", $output, $exitCode );
 827                 } else {
 828                         exec( "$command 2>&1", $output, $exitCode );
 829                 }
 830
 831                 # map exit code to AV_xxx constants.
 832                 $mappedCode = $exitCode;
 833                 if ( $exitCodeMap ) {
 834                         if ( isset( $exitCodeMap[$exitCode] ) ) {
 835                                 $mappedCode = $exitCodeMap[$exitCode];
 836                         } elseif ( isset( $exitCodeMap["*"] ) ) {
 837                                 $mappedCode = $exitCodeMap["*"];
 838                         }
 839                 }
 840
 841                 if ( $mappedCode === AV_SCAN_FAILED ) {
 842                         # scan failed (code was mapped to false by $exitCodeMap)
 843                         wfDebug( __METHOD__ . ": failed to scan $file (code $exitCode).\n" );
 844
 845                         if ( $wgAntivirusRequired ) {
 846                                 return wfMsg( 'virus-scanfailed', array( $exitCode ) );
 847                         } else {
 848                                 return NULL;
 849                         }
 850                 } else if ( $mappedCode === AV_SCAN_ABORTED ) {
 851                         # scan failed because filetype is unknown (probably imune)
 852                         wfDebug( __METHOD__ . ": unsupported file type $file (code $exitCode).\n" );
 853                         return NULL;
 854                 } else if ( $mappedCode === AV_NO_VIRUS ) {
 855                         # no virus found
 856                         wfDebug( __METHOD__ . ": file passed virus scan.\n" );
 857                         return false;
 858                 } else {
 859                         $output = join( "\n", $output );
 860                         $output = trim( $output );
 861
 862                         if ( !$output ) {
 863                                 $output = true; #if there's no output, return true
 864                         } elseif ( $msgPattern ) {
 865                                 $groups = array();
 866                                 if ( preg_match( $msgPattern, $output, $groups ) ) {
 867                                         if ( $groups[1] ) {
 868                                                 $output = $groups[1];
 869                                         }
 870                                 }
 871                         }
 872
 873                         wfDebug( __METHOD__ . ": FOUND VIRUS! scanner feedback: $output \n" );
 874                         return $output;
 875                 }
 876         }
 877
 878         /**
 879          * Check if the temporary file is MacBinary-encoded, as some uploads
 880          * from Internet Explorer on Mac OS Classic and Mac OS X will be.
 881          * If so, the data fork will be extracted to a second temporary file,
 882          * which will then be checked for validity and either kept or discarded.
 883          *
 884          * @access private
 885          */
 886         private function checkMacBinary() {
 887                 $macbin = new MacBinary( $this->mTempPath );
 888                 if( $macbin->isValid() ) {
 889                         $dataFile = tempnam( wfTempDir(), 'WikiMacBinary' );
 890                         $dataHandle = fopen( $dataFile, 'wb' );
 891
 892                         wfDebug( __METHOD__ . ": Extracting MacBinary data fork to $dataFile\n" );
 893                         $macbin->extractData( $dataHandle );
 894
 895                         $this->mTempPath = $dataFile;
 896                         $this->mFileSize = $macbin->dataForkLength();
 897
 898                         // We'll have to manually remove the new file if it's not kept.
 899                         $this->mRemoveTempFile = true;
 900                 }
 901                 $macbin->close();
 902         }
 903
 904         /**
 905          * Check if there's an overwrite conflict and, if so, if restrictions
 906          * forbid this user from performing the upload.
 907          *
 908          * @return mixed true on success, error string on failure
 909          * @access private
 910          */
 911         private function checkOverwrite() {
 912                 global $wgUser;
 913                 // First check whether the local file can be overwritten
 914                 $file = $this->getLocalFile();
 915                 if( $file->exists() )
 916                         if( !self::userCanReUpload( $wgUser, $file ) )
 917                                 return 'fileexists-forbidden';
 918
 919                 // Check shared conflicts
 920                 $file = wfFindFile( $file->getName() );
 921                 if ( $file && ( !$wgUser->isAllowed( 'reupload' ) ||
 922                                 !$wgUser->isAllowed( 'reupload-shared' ) ) )
 923                         return 'fileexists-shared-forbidden';
 924
 925                 return true;
 926         }
 927
 928         /**
 929          * Check if a user is the last uploader
 930          *
 931          * @param User $user
 932          * @param string $img, image name
 933          * @return bool
 934          */
 935         public static function userCanReUpload( User $user, $img ) {
 936                 if( $user->isAllowed( 'reupload' ) )
 937                         return true; // non-conditional
 938                 if( !$user->isAllowed( 'reupload-own' ) )
 939                         return false;
 940                 if( is_string( $img ) )
 941                         $img = wfLocalFile( $img );
 942                 if ( !( $img instanceof LocalFile ) )
 943                         return false;
 944
 945                 return $user->getId() == $img->getUser( 'id' );
 946         }
 947
 948         /**
 949          * Helper function that does various existence checks for a file.
 950          * The following checks are performed:
 951          * - The file exists
 952          * - Article with the same name as the file exists
 953          * - File exists with normalized extension
 954          * - The file looks like a thumbnail and the original exists
 955          *
 956          * @param File $file The file to check
 957          * @return mixed False if the file does not exists, else an array
 958          */
 959         public static function getExistsWarning( $file ) {
 960                 if( $file->exists() )
 961                         return array( 'exists', $file );
 962
 963                 if( $file->getTitle()->getArticleID() )
 964                         return array( 'page-exists', $file );
 965
 966                 if( strpos( $file->getName(), '.' ) == false ) {
 967                         $partname = $file->getName();
 968                         $rawExtension = '';
 969                 } else {
 970                         $n = strrpos( $file->getName(), '.' );
 971                         $rawExtension = substr( $file->getName(), $n + 1 );
 972                         $partname = substr( $file->getName(), 0, $n );
 973                 }
 974
 975                 if ( $rawExtension != $file->getExtension() ) {
 976                         // We're not using the normalized form of the extension.
 977                         // Normal form is lowercase, using most common of alternate
 978                         // extensions (eg 'jpg' rather than 'JPEG').
 979                         //
 980                         // Check for another file using the normalized form...
 981                         $nt_lc = Title::makeTitle( NS_FILE, $partname . '.' . $file->getExtension() );
 982                         $file_lc = wfLocalFile( $nt_lc );
 983
 984                         if( $file_lc->exists() )
 985                                 return array( 'exists-normalized', $file_lc );
 986                 }
 987
 988                 if ( self::isThumbName( $file->getName() ) ) {
 989                         # Check for filenames like 50px- or 180px-, these are mostly thumbnails
 990                         $nt_thb = Title::newFromText( substr( $partname , strpos( $partname , '-' ) +1 ) . '.' . $rawExtension );
 991                         $file_thb = wfLocalFile( $nt_thb );
 992                         if( $file_thb->exists() )
 993                                 return array( 'thumb', $file_thb );
 994                 }
 995
 996                 return false;
 997         }
 998
 999         /**
1000          * Helper function that checks whether the filename looks like a thumbnail
1001          */
1002         public static function isThumbName( $filename ) {
1003                 $n = strrpos( $filename, '.' );
1004                 $partname = $n ? substr( $filename, 0, $n ) : $filename;
1005                 return (
1006                                         substr( $partname , 3, 3 ) == 'px-' ||
1007                                         substr( $partname , 2, 3 ) == 'px-'
1008                                 ) &&
1009                                 preg_match( "/[0-9]{2}/" , substr( $partname , 0, 2 ) );
1010         }
1011
1012         /**
1013          * Get a list of blacklisted filename prefixes from [[MediaWiki:filename-prefix-blacklist]]
1014          *
1015          * @return array list of prefixes
1016          */
1017         public static function getFilenamePrefixBlacklist() {
1018                 $blacklist = array();
1019                 $message = wfMsgForContent( 'filename-prefix-blacklist' );
1020                 if( $message && !( wfEmptyMsg( 'filename-prefix-blacklist', $message ) || $message == '-' ) ) {
1021                         $lines = explode( "\n", $message );
1022                         foreach( $lines as $line ) {
1023                                 // Remove comment lines
1024                                 $comment = substr( trim( $line ), 0, 1 );
1025                                 if ( $comment == '#' || $comment == '' ) {
1026                                         continue;
1027                                 }
1028                                 // Remove additional comments after a prefix
1029                                 $comment = strpos( $line, '#' );
1030                                 if ( $comment > 0 ) {
1031                                         $line = substr( $line, 0, $comment-1 );
1032                                 }
1033                                 $blacklist[] = trim( $line );
1034                         }
1035                 }
1036                 return $blacklist;
1037         }
1038
1039 }