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         protected $mLocalFile;
  22
  23         const SUCCESS = 0;
  24         const OK = 0;
  25         const BEFORE_PROCESSING = 1;
  26         const LARGE_FILE_SERVER = 2;
  27         const EMPTY_FILE = 3;
  28         const MIN_LENGTH_PARTNAME = 4;
  29         const ILLEGAL_FILENAME = 5;
  30         const PROTECTED_PAGE = 6;
  31         const OVERWRITE_EXISTING_FILE = 7;
  32         const FILETYPE_MISSING = 8;
  33         const FILETYPE_BADTYPE = 9;
  34         const VERIFICATION_ERROR = 10;
  35         const UPLOAD_VERIFICATION_ERROR = 11;
  36         const UPLOAD_WARNING = 12;
  37         const INTERNAL_ERROR = 13;
  38         const MIN_LENGHT_PARTNAME = 14;
  39
  40         const SESSION_VERSION = 2;
  41
  42         /**
  43          * Returns true if uploads are enabled.
  44          * Can be override by subclasses.
  45          */
  46         public static function isEnabled() {
  47                 global $wgEnableUploads;
  48                 if ( !$wgEnableUploads )
  49                         return false;
  50
  51                 # Check php's file_uploads setting
  52                 if( !wfIniGetBool( 'file_uploads' ) ) {
  53                         return false;
  54                 }
  55                 return true;
  56         }
  57
  58         /**
  59          * Returns true if the user can use this upload module or else a string
  60          * identifying the missing permission.
  61          * Can be overriden by subclasses.
  62          */
  63         public static function isAllowed( $user ) {
  64                 if( !$user->isAllowed( 'upload' ) )
  65                         return 'upload';
  66                 return true;
  67         }
  68
  69         // Upload handlers. Should probably just be a global
  70         static $uploadHandlers = array( 'Stash', 'File', 'Url' );
  71
  72         /**
  73          * Create a form of UploadBase depending on wpSourceType and initializes it
  74          */
  75         public static function createFromRequest( &$request, $type = null ) {
  76                 $type = $type ? $type : $request->getVal( 'wpSourceType', 'File' );
  77
  78                 if( !$type )
  79                         return null;
  80
  81                 // Get the upload class
  82                 $type = ucfirst( $type );
  83                 $className = 'UploadFrom' . $type;
  84                 wfDebug( __METHOD__ . ": class name: $className\n" );
  85                 if( !in_array( $type, self::$uploadHandlers ) )
  86                         return null;
  87
  88                 // Check whether this upload class is enabled
  89                 if( !call_user_func( array( $className, 'isEnabled' ) ) )
  90                         return null;
  91
  92                 // Check whether the request is valid
  93                 if( !call_user_func( array( $className, 'isValidRequest' ), $request ) )
  94                         return null;
  95
  96                 $handler = new $className;
  97
  98                 $handler->initializeFromRequest( $request );
  99                 return $handler;
 100         }
 101
 102         /**
 103          * Check whether a request if valid for this handler
 104          */
 105         public static function isValidRequest( $request ) {
 106                 return false;
 107         }
 108
 109         public function __construct() {}
 110
 111         /**
 112          * Do the real variable initialization
 113          */
 114         public function initialize( $name, $tempPath, $fileSize, $removeTempFile = false ) {
 115                 $this->mDesiredDestName = $name;
 116                 $this->mTempPath = $tempPath;
 117                 $this->mFileSize = $fileSize;
 118                 $this->mRemoveTempFile = $removeTempFile;
 119         }
 120
 121         /**
 122          * Initialize from a WebRequest. Override this in a subclass.
 123          */
 124         public abstract function initializeFromRequest( &$request );
 125
 126         /**
 127          * Fetch the file. Usually a no-op
 128          */
 129         public function fetchFile() {
 130                 return Status::newGood();
 131         }
 132
 133         /**
 134          * Return the file size
 135          */
 136         public function isEmptyFile(){
 137                 return empty( $this->mFileSize );
 138         }
 139
 140         /*
 141          * getRealPath
 142          * @param string $srcPath the source path
 143          * @returns the real path if it was a virtual url
 144          */
 145         function getRealPath( $srcPath ){
 146                 $repo = RepoGroup::singleton()->getLocalRepo();
 147                 if ( $repo->isVirtualUrl( $srcPath ) ) {
 148                         return $repo->resolveVirtualUrl( $srcPath );
 149                 }
 150                 return $srcPath;
 151         }
 152
 153         /**
 154          * Verify whether the upload is sane.
 155          * Returns self::OK or else an array with error information
 156          */
 157         public function verifyUpload() {
 158                 /**
 159                  * If there was no filename or a zero size given, give up quick.
 160                  */
 161                 if( $this->isEmptyFile() )
 162                         return array( 'status' => self::EMPTY_FILE );
 163
 164                 $nt = $this->getTitle();
 165                 if( is_null( $nt ) ) {
 166                         $result = array( 'status' => $this->mTitleError );
 167                         if( $this->mTitleError == self::ILLEGAL_FILENAME )
 168                                 $result['filtered'] = $this->mFilteredName;
 169                         if ( $this->mTitleError == self::FILETYPE_BADTYPE )
 170                                 $result['finalExt'] = $this->mFinalExtension;
 171                         return $result;
 172                 }
 173                 $this->mDestName = $this->getLocalFile()->getName();
 174
 175                 /**
 176                  * In some cases we may forbid overwriting of existing files.
 177                  */
 178                 $overwrite = $this->checkOverwrite();
 179                 if( $overwrite !== true )
 180                         return array( 'status' => self::OVERWRITE_EXISTING_FILE, 'overwrite' => $overwrite );
 181
 182                 /**
 183                  * Look at the contents of the file; if we can recognize the
 184                  * type but it's corrupt or data of the wrong type, we should
 185                  * probably not accept it.
 186                  */
 187                 $verification = $this->verifyFile();
 188
 189                 if( $verification !== true ) {
 190                         if( !is_array( $verification ) )
 191                                 $verification = array( $verification );
 192                         return array( 'status' => self::VERIFICATION_ERROR,
 193                                         'details' => $verification );
 194
 195                 }
 196
 197                 $error = '';
 198                 if( !wfRunHooks( 'UploadVerification',
 199                                 array( $this->mDestName, $this->mTempPath, &$error ) ) ) {
 200                         // This status needs another name...
 201                         return array( 'status' => self::UPLOAD_VERIFICATION_ERROR, 'error' => $error );
 202                 }
 203
 204                 return array( 'status' => self::OK );
 205         }
 206
 207         /**
 208          * Verifies that it's ok to include the uploaded file
 209          *
 210          * FIXME: this function seems to intermixes tmpfile and $this->mTempPath .. no idea why this is
 211          *
 212          * @param string $tmpfile the full path of the temporary file to verify
 213          * @return mixed true of the file is verified, a string or array otherwise.
 214          */
 215         protected function verifyFile() {
 216                 $this->mFileProps = File::getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
 217                 $this->checkMacBinary();
 218
 219                 #magically determine mime type
 220                 $magic = MimeMagic::singleton();
 221                 $mime = $magic->guessMimeType( $this->mTempPath, false );
 222
 223                 #check mime type, if desired
 224                 global $wgVerifyMimeType;
 225                 if ( $wgVerifyMimeType ) {
 226                         global $wgMimeTypeBlacklist;
 227                         if ( $this->checkFileExtension( $mime, $wgMimeTypeBlacklist ) )
 228                                 return array( 'filetype-badmime', $mime );
 229
 230                         # Check IE type
 231                         $fp = fopen( $this->mTempPath, 'rb' );
 232                         $chunk = fread( $fp, 256 );
 233                         fclose( $fp );
 234                         $extMime = $magic->guessTypesForExtension( $this->mFinalExtension );
 235                         $ieTypes = $magic->getIEMimeTypes( $this->mTempPath, $chunk, $extMime );
 236                         foreach ( $ieTypes as $ieType ) {
 237                                 if ( $this->checkFileExtension( $ieType, $wgMimeTypeBlacklist ) ) {
 238                                         return array( 'filetype-bad-ie-mime', $ieType );
 239                                 }
 240                         }
 241                 }
 242
 243                 #check for htmlish code and javascript
 244                 if( self::detectScript( $this->mTempPath, $mime, $this->mFinalExtension ) ) {
 245                         return 'uploadscripted';
 246                 }
 247                 if( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
 248                         if( self::detectScriptInSvg( $this->mTempPath ) ) {
 249                                 return 'uploadscripted';
 250                         }
 251                 }
 252
 253                 /**
 254                 * Scan the uploaded file for viruses
 255                 */
 256                 $virus = $this->detectVirus( $this->mTempPath );
 257                 if ( $virus ) {
 258                         return array( 'uploadvirus', $virus );
 259                 }
 260                 wfDebug( __METHOD__ . ": all clear; passing.\n" );
 261                 return true;
 262         }
 263
 264         /**
 265          * Check whether the user can edit, upload and create the image.
 266          *
 267          * @param User $user the user to verify the permissions against
 268          * @return mixed An array as returned by getUserPermissionsErrors or true
 269          *               in case the user has proper permissions.
 270          */
 271         public function verifyPermissions( $user ) {
 272                 /**
 273                  * If the image is protected, non-sysop users won't be able
 274                  * to modify it by uploading a new revision.
 275                  */
 276                 $nt = $this->getTitle();
 277                 if( is_null( $nt ) )
 278                         return true;
 279                 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
 280                 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
 281                 $permErrorsCreate = ( $nt->exists() ? array() : $nt->getUserPermissionsErrors( 'create', $user ) );
 282                 if( $permErrors || $permErrorsUpload || $permErrorsCreate ) {
 283                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
 284                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
 285                         return $permErrors;
 286                 }
 287                 return true;
 288         }
 289
 290         /**
 291          * Check for non fatal problems with the file
 292          *
 293          * @return array Array of warnings
 294          */
 295         public function checkWarnings() {
 296                 $warnings = array();
 297
 298                 $localFile = $this->getLocalFile();
 299                 $filename = $localFile->getName();
 300                 $n = strrpos( $filename, '.' );
 301                 $partname = $n ? substr( $filename, 0, $n ) : $filename;
 302
 303                 /*
 304                  * Check whether the resulting filename is different from the desired one,
 305                  * but ignore things like ucfirst() and spaces/underscore things
 306                  */
 307                 $comparableName = str_replace( ' ', '_', $this->mDesiredDestName );
 308                 global $wgCapitalLinks, $wgContLang;
 309                 if ( $wgCapitalLinks ) {
 310                         $comparableName = $wgContLang->ucfirst( $comparableName );
 311                 }
 312                 if( $this->mDesiredDestName != $filename && $comparableName != $filename )
 313                         $warnings['badfilename'] = $filename;
 314
 315                 // Check whether the file extension is on the unwanted list
 316                 global $wgCheckFileExtensions, $wgFileExtensions;
 317                 if ( $wgCheckFileExtensions ) {
 318                         if ( !$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) )
 319                                 $warnings['filetype-unwanted-type'] = $this->mFinalExtension;
 320                 }
 321
 322                 global $wgUploadSizeWarning;
 323                 if ( $wgUploadSizeWarning && ( $this->mFileSize > $wgUploadSizeWarning ) )
 324                         $warnings['large-file'] = $wgUploadSizeWarning;
 325
 326                 if ( $this->mFileSize == 0 )
 327                         $warnings['emptyfile'] = true;
 328
 329
 330                 $exists = self::getExistsWarning( $localFile );
 331                 if( $exists !== false )
 332                         $warnings['exists'] = $exists;
 333
 334                 // Check dupes against existing files
 335                 $hash = File::sha1Base36( $this->mTempPath );
 336                 $dupes = RepoGroup::singleton()->findBySha1( $hash );
 337                 $title = $this->getTitle();
 338                 // Remove all matches against self
 339                 foreach ( $dupes as $key => $dupe ) {
 340                         if( $title->equals( $dupe->getTitle() ) )
 341                                 unset( $dupes[$key] );
 342                 }
 343                 if( $dupes )
 344                         $warnings['duplicate'] = $dupes;
 345
 346                 // Check dupes against archives
 347                 $archivedImage = new ArchivedFile( null, 0, "{$hash}.{$this->mFinalExtension}" );
 348                 if ( $archivedImage->getID() > 0 )
 349                         $warnings['duplicate-archive'] = $archivedImage->getName();
 350
 351                 return $warnings;
 352         }
 353
 354         /**
 355          * Really perform the upload. Stores the file in the local repo, watches
 356          * if necessary and runs the UploadComplete hook.
 357          *
 358          * @return mixed Status indicating the whether the upload succeeded.
 359          */
 360         public function performUpload( $comment, $pageText, $watch, $user ) {
 361                 wfDebug( "\n\n\performUpload: sum:" . $comment . ' c: ' . $pageText . ' w:' . $watch );
 362                 $status = $this->getLocalFile()->upload( $this->mTempPath, $comment, $pageText,
 363                         File::DELETE_SOURCE, $this->mFileProps, false, $user );
 364
 365                 if( $status->isGood() && $watch )
 366                         $user->addWatch( $this->getLocalFile()->getTitle() );
 367
 368                 if( $status->isGood() )
 369                         wfRunHooks( 'UploadComplete', array( &$this ) );
 370
 371                 return $status;
 372         }
 373
 374         /**
 375          * Returns the title of the file to be uploaded. Sets mTitleError in case
 376          * the name was illegal.
 377          *
 378          * @return Title The title of the file or null in case the name was illegal
 379          */
 380         public function getTitle() {
 381                 if ( $this->mTitle !== false )
 382                         return $this->mTitle;
 383
 384                 /**
 385                  * Chop off any directories in the given filename. Then
 386                  * filter out illegal characters, and try to make a legible name
 387                  * out of it. We'll strip some silently that Title would die on.
 388                  */
 389                 $basename = $this->mDesiredDestName;
 390
 391                 $this->mFilteredName = wfStripIllegalFilenameChars( $basename );
 392                 /* Normalize to title form before we do any further processing */
 393                 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 394                 if( is_null( $nt ) ) {
 395                         $this->mTitleError = self::ILLEGAL_FILENAME;
 396                         return $this->mTitle = null;
 397                 }
 398                 $this->mFilteredName = $nt->getDBkey();
 399
 400                 /**
 401                  * We'll want to blacklist against *any* 'extension', and use
 402                  * only the final one for the whitelist.
 403                  */
 404                 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName );
 405
 406                 if( count( $ext ) ) {
 407                         $this->mFinalExtension = trim( $ext[count( $ext ) - 1] );
 408                 } else {
 409                         $this->mFinalExtension = '';
 410                 }
 411
 412                 /* Don't allow users to override the blacklist (check file extension) */
 413                 global $wgCheckFileExtensions, $wgStrictFileExtensions;
 414                 global $wgFileExtensions, $wgFileBlacklist;
 415                 if ( $this->mFinalExtension == '' ) {
 416                         $this->mTitleError = self::FILETYPE_MISSING;
 417                         return $this->mTitle = null;
 418                 } elseif ( $this->checkFileExtensionList( $ext, $wgFileBlacklist ) ||
 419                                 ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
 420                                         !$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) ) ) {
 421                         $this->mTitleError = self::FILETYPE_BADTYPE;
 422                         return $this->mTitle = null;
 423                 }
 424
 425                 # If there was more than one "extension", reassemble the base
 426                 # filename to prevent bogus complaints about length
 427                 if( count( $ext ) > 1 ) {
 428                         for( $i = 0; $i < count( $ext ) - 1; $i++ )
 429                                 $partname .= '.' . $ext[$i];
 430                 }
 431
 432                 if( strlen( $partname ) < 1 ) {
 433                         $this->mTitleError =  self::MIN_LENGTH_PARTNAME;
 434                         return $this->mTitle = null;
 435                 }
 436
 437                 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 438                 if( is_null( $nt ) ) {
 439                         $this->mTitleError = self::ILLEGAL_FILENAME;
 440                         return $this->mTitle = null;
 441                 }
 442                 return $this->mTitle = $nt;
 443         }
 444
 445         /**
 446          * Return the local file and initializes if necessary.
 447          */
 448         public function getLocalFile() {
 449                 if( is_null( $this->mLocalFile ) ) {
 450                         $nt = $this->getTitle();
 451                         $this->mLocalFile = is_null( $nt ) ? null : wfLocalFile( $nt );
 452                 }
 453                 return $this->mLocalFile;
 454         }
 455
 456         /**
 457          * Stash a file in a temporary directory for later processing
 458          * after the user has confirmed it.
 459          *
 460          * If the user doesn't explicitly cancel or accept, these files
 461          * can accumulate in the temp directory.
 462          *
 463          * @param string $saveName - the destination filename
 464          * @param string $tempSrc - the source temporary file to save
 465          * @return string - full path the stashed file, or false on failure
 466          * @access private
 467          */
 468         protected function saveTempUploadedFile( $saveName, $tempSrc ) {
 469                 $repo = RepoGroup::singleton()->getLocalRepo();
 470                 $status = $repo->storeTemp( $saveName, $tempSrc );
 471                 return $status;
 472         }
 473
 474         /**
 475          * Append a file to a stashed file.
 476          *
 477          * @param string $srcPath Path to file to append from
 478          * @param string $toAppendPath Path to file to append to
 479          * @return Status Status
 480          */
 481         public function appendToUploadFile( $srcPath, $toAppendPath ){
 482                 $repo = RepoGroup::singleton()->getLocalRepo();
 483                 $status = $repo->append( $srcPath, $toAppendPath );
 484                 return $status;
 485         }
 486
 487         /**
 488          * Stash a file in a temporary directory for later processing,
 489          * and save the necessary descriptive info into the session.
 490          * Returns a key value which will be passed through a form
 491          * to pick up the path info on a later invocation.
 492          *
 493          * @return int Session key
 494          */
 495         public function stashSession() {
 496                 $status = $this->saveTempUploadedFile( $this->mDestName, $this->mTempPath );
 497                 if( !$status->isOK() ) {
 498                         # Couldn't save the file.
 499                         return false;
 500                 }
 501                 if(!isset($_SESSION))
 502                         session_start(); // start up the session (might have been previously closed to prevent php session locking)
 503                 $key = $this->getSessionKey();
 504                 $_SESSION['wsUploadData'][$key] = array(
 505                         'mTempPath'       => $status->value,
 506                         'mFileSize'       => $this->mFileSize,
 507                         'mFileProps'      => $this->mFileProps,
 508                         'version'         => self::SESSION_VERSION,
 509                 );
 510                 return $key;
 511         }
 512
 513         /**
 514          * Generate a random session key from stash in cases where we want to start an upload without much information
 515          */
 516         protected function getSessionKey(){
 517                 $key = mt_rand( 0, 0x7fffffff );
 518                 $_SESSION['wsUploadData'][$key] = array();
 519                 return $key;
 520         }
 521
 522         /**
 523          * Remove a temporarily kept file stashed by saveTempUploadedFile().
 524          * @return success
 525          */
 526         public function unsaveUploadedFile() {
 527                 $repo = RepoGroup::singleton()->getLocalRepo();
 528                 $success = $repo->freeTemp( $this->mTempPath );
 529                 return $success;
 530         }
 531
 532         /**
 533          * If we've modified the upload file we need to manually remove it
 534          * on exit to clean up.
 535          * @access private
 536          */
 537         public function cleanupTempFile() {
 538                 if ( $this->mRemoveTempFile && $this->mTempPath && file_exists( $this->mTempPath ) ) {
 539                         wfDebug( __METHOD__ . ": Removing temporary file {$this->mTempPath}\n" );
 540                         unlink( $this->mTempPath );
 541                 }
 542         }
 543
 544         public function getTempPath() {
 545                 return $this->mTempPath;
 546         }
 547
 548         /**
 549          * Split a file into a base name and all dot-delimited 'extensions'
 550          * on the end. Some web server configurations will fall back to
 551          * earlier pseudo-'extensions' to determine type and execute
 552          * scripts, so the blacklist needs to check them all.
 553          *
 554          * @return array
 555          */
 556         public static function splitExtensions( $filename ) {
 557                 $bits = explode( '.', $filename );
 558                 $basename = array_shift( $bits );
 559                 return array( $basename, $bits );
 560         }
 561
 562         /**
 563          * Perform case-insensitive match against a list of file extensions.
 564          * Returns true if the extension is in the list.
 565          *
 566          * @param string $ext
 567          * @param array $list
 568          * @return bool
 569          */
 570         public static function checkFileExtension( $ext, $list ) {
 571                 return in_array( strtolower( $ext ), $list );
 572         }
 573
 574         /**
 575          * Perform case-insensitive match against a list of file extensions.
 576          * Returns true if any of the extensions are in the list.
 577          *
 578          * @param array $ext
 579          * @param array $list
 580          * @return bool
 581          */
 582         public static function checkFileExtensionList( $ext, $list ) {
 583                 foreach( $ext as $e ) {
 584                         if( in_array( strtolower( $e ), $list ) ) {
 585                                 return true;
 586                         }
 587                 }
 588                 return false;
 589         }
 590
 591         /**
 592          * Checks if the mime type of the uploaded file matches the file extension.
 593          *
 594          * @param string $mime the mime type of the uploaded file
 595          * @param string $extension The filename extension that the file is to be served with
 596          * @return bool
 597          */
 598         public static function verifyExtension( $mime, $extension ) {
 599                 $magic = MimeMagic::singleton();
 600
 601                 if ( !$mime || $mime == 'unknown' || $mime == 'unknown/unknown' )
 602                         if ( !$magic->isRecognizableExtension( $extension ) ) {
 603                                 wfDebug( __METHOD__ . ": passing file with unknown detected mime type; " .
 604                                         "unrecognized extension '$extension', can't verify\n" );
 605                                 return true;
 606                         } else {
 607                                 wfDebug( __METHOD__ . ": rejecting file with unknown detected mime type; ".
 608                                         "recognized extension '$extension', so probably invalid file\n" );
 609                                 return false;
 610                         }
 611
 612                 $match = $magic->isMatchingExtension( $extension, $mime );
 613
 614                 if ( $match === NULL ) {
 615                         wfDebug( __METHOD__ . ": no file extension known for mime type $mime, passing file\n" );
 616                         return true;
 617                 } elseif( $match === true ) {
 618                         wfDebug( __METHOD__ . ": mime type $mime matches extension $extension, passing file\n" );
 619
 620                         #TODO: if it's a bitmap, make sure PHP or ImageMagic resp. can handle it!
 621                         return true;
 622
 623                 } else {
 624                         wfDebug( __METHOD__ . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
 625                         return false;
 626                 }
 627         }
 628
 629         /**
 630          * Heuristic for detecting files that *could* contain JavaScript instructions or
 631          * things that may look like HTML to a browser and are thus
 632          * potentially harmful. The present implementation will produce false positives in some situations.
 633          *
 634          * @param string $file Pathname to the temporary upload file
 635          * @param string $mime The mime type of the file
 636          * @param string $extension The extension of the file
 637          * @return bool true if the file contains something looking like embedded scripts
 638          */
 639         public static function detectScript( $file, $mime, $extension ) {
 640                 global $wgAllowTitlesInSVG;
 641
 642                 #ugly hack: for text files, always look at the entire file.
 643                 #For binary field, just check the first K.
 644
 645                 if( strpos( $mime,'text/' ) === 0 )
 646                         $chunk = file_get_contents( $file );
 647                 else {
 648                         $fp = fopen( $file, 'rb' );
 649                         $chunk = fread( $fp, 1024 );
 650                         fclose( $fp );
 651                 }
 652
 653                 $chunk = strtolower( $chunk );
 654
 655                 if( !$chunk )
 656                         return false;
 657
 658                 #decode from UTF-16 if needed (could be used for obfuscation).
 659                 if( substr( $chunk, 0, 2 ) == "\xfe\xff" )
 660                         $enc = "UTF-16BE";
 661                 elseif( substr( $chunk, 0, 2 ) == "\xff\xfe" )
 662                         $enc = "UTF-16LE";
 663                 else
 664                         $enc = NULL;
 665
 666                 if( $enc )
 667                         $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
 668
 669                 $chunk = trim( $chunk );
 670
 671                 #FIXME: convert from UTF-16 if necessarry!
 672                 wfDebug( __METHOD__ . ": checking for embedded scripts and HTML stuff\n" );
 673
 674                 #check for HTML doctype
 675                 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) )
 676                         return true;
 677
 678                 /**
 679                 * Internet Explorer for Windows performs some really stupid file type
 680                 * autodetection which can cause it to interpret valid image files as HTML
 681                 * and potentially execute JavaScript, creating a cross-site scripting
 682                 * attack vectors.
 683                 *
 684                 * Apple's Safari browser also performs some unsafe file type autodetection
 685                 * which can cause legitimate files to be interpreted as HTML if the
 686                 * web server is not correctly configured to send the right content-type
 687                 * (or if you're really uploading plain text and octet streams!)
 688                 *
 689                 * Returns true if IE is likely to mistake the given file for HTML.
 690                 * Also returns true if Safari would mistake the given file for HTML
 691                 * when served with a generic content-type.
 692                 */
 693                 $tags = array(
 694                         '<a href',
 695                         '<body',
 696                         '<head',
 697                         '<html',   #also in safari
 698                         '<img',
 699                         '<pre',
 700                         '<script', #also in safari
 701                         '<table'
 702                 );
 703
 704                 if( !$wgAllowTitlesInSVG && $extension !== 'svg' && $mime !== 'image/svg' ) {
 705                         $tags[] = '<title';
 706                 }
 707
 708                 foreach( $tags as $tag ) {
 709                         if( false !== strpos( $chunk, $tag ) ) {
 710                                 return true;
 711                         }
 712                 }
 713
 714                 /*
 715                  * look for JavaScript
 716                  */
 717
 718                 #resolve entity-refs to look at attributes. may be harsh on big files... cache result?
 719                 $chunk = Sanitizer::decodeCharReferences( $chunk );
 720
 721                 #look for script-types
 722                 if( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) )
 723                         return true;
 724
 725                 #look for html-style script-urls
 726                 if( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) )
 727                         return true;
 728
 729                 #look for css-style script-urls
 730                 if( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) )
 731                         return true;
 732
 733                 wfDebug( __METHOD__ . ": no scripts found\n" );
 734                 return false;
 735         }
 736
 737         protected function detectScriptInSvg( $filename ) {
 738                 $check = new XmlTypeCheck( $filename, array( $this, 'checkSvgScriptCallback' ) );
 739                 return $check->filterMatch;
 740         }
 741
 742         /**
 743          * @todo Replace this with a whitelist filter!
 744          */
 745         public function checkSvgScriptCallback( $element, $attribs ) {
 746                 $stripped = $this->stripXmlNamespace( $element );
 747
 748                 if( $stripped == 'script' ) {
 749                         wfDebug( __METHOD__ . ": Found script element '$element' in uploaded file.\n" );
 750                         return true;
 751                 }
 752
 753                 foreach( $attribs as $attrib => $value ) {
 754                         $stripped = $this->stripXmlNamespace( $attrib );
 755                         if( substr( $stripped, 0, 2 ) == 'on' ) {
 756                                 wfDebug( __METHOD__ . ": Found script attribute '$attrib'='value' in uploaded file.\n" );
 757                                 return true;
 758                         }
 759                         if( $stripped == 'href' && strpos( strtolower( $value ), 'javascript:' ) !== false ) {
 760                                 wfDebug( __METHOD__ . ": Found script href attribute '$attrib'='$value' in uploaded file.\n" );
 761                                 return true;
 762                         }
 763                 }
 764         }
 765
 766         private function stripXmlNamespace( $name ) {
 767                 // 'http://www.w3.org/2000/svg:script' -> 'script'
 768                 $parts = explode( ':', strtolower( $name ) );
 769                 return array_pop( $parts );
 770         }
 771
 772         /**
 773          * Generic wrapper function for a virus scanner program.
 774          * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
 775          * $wgAntivirusRequired may be used to deny upload if the scan fails.
 776          *
 777          * @param string $file Pathname to the temporary upload file
 778          * @return mixed false if not virus is found, NULL if the scan fails or is disabled,
 779          *         or a string containing feedback from the virus scanner if a virus was found.
 780          *         If textual feedback is missing but a virus was found, this function returns true.
 781          */
 782         public static function detectVirus( $file ) {
 783                 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
 784
 785                 if ( !$wgAntivirus ) {
 786                         wfDebug( __METHOD__ . ": virus scanner disabled\n" );
 787                         return NULL;
 788                 }
 789
 790                 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
 791                         wfDebug( __METHOD__ . ": unknown virus scanner: $wgAntivirus\n" );
 792                         $wgOut->wrapWikiMsg( '<div class="error">$1</div>', array( 'virus-badscanner', $wgAntivirus ) );
 793                         return wfMsg( 'virus-unknownscanner' ) . " $wgAntivirus";
 794                 }
 795
 796                 # look up scanner configuration
 797                 $command = $wgAntivirusSetup[$wgAntivirus]["command"];
 798                 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]["codemap"];
 799                 $msgPattern = isset( $wgAntivirusSetup[$wgAntivirus]["messagepattern"] ) ?
 800                         $wgAntivirusSetup[$wgAntivirus]["messagepattern"] : null;
 801
 802                 if ( strpos( $command,"%f" ) === false ) {
 803                         # simple pattern: append file to scan
 804                         $command .= " " . wfEscapeShellArg( $file );
 805                 } else {
 806                         # complex pattern: replace "%f" with file to scan
 807                         $command = str_replace( "%f", wfEscapeShellArg( $file ), $command );
 808                 }
 809
 810                 wfDebug( __METHOD__ . ": running virus scan: $command \n" );
 811
 812                 # execute virus scanner
 813                 $exitCode = false;
 814
 815                 #NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
 816                 #      that does not seem to be worth the pain.
 817                 #      Ask me (Duesentrieb) about it if it's ever needed.
 818                 $output = array();
 819                 if ( wfIsWindows() ) {
 820                         exec( "$command", $output, $exitCode );
 821                 } else {
 822                         exec( "$command 2>&1", $output, $exitCode );
 823                 }
 824
 825                 # map exit code to AV_xxx constants.
 826                 $mappedCode = $exitCode;
 827                 if ( $exitCodeMap ) {
 828                         if ( isset( $exitCodeMap[$exitCode] ) ) {
 829                                 $mappedCode = $exitCodeMap[$exitCode];
 830                         } elseif ( isset( $exitCodeMap["*"] ) ) {
 831                                 $mappedCode = $exitCodeMap["*"];
 832                         }
 833                 }
 834
 835                 if ( $mappedCode === AV_SCAN_FAILED ) {
 836                         # scan failed (code was mapped to false by $exitCodeMap)
 837                         wfDebug( __METHOD__ . ": failed to scan $file (code $exitCode).\n" );
 838
 839                         if ( $wgAntivirusRequired ) {
 840                                 return wfMsg( 'virus-scanfailed', array( $exitCode ) );
 841                         } else {
 842                                 return NULL;
 843                         }
 844                 } else if ( $mappedCode === AV_SCAN_ABORTED ) {
 845                         # scan failed because filetype is unknown (probably imune)
 846                         wfDebug( __METHOD__ . ": unsupported file type $file (code $exitCode).\n" );
 847                         return NULL;
 848                 } else if ( $mappedCode === AV_NO_VIRUS ) {
 849                         # no virus found
 850                         wfDebug( __METHOD__ . ": file passed virus scan.\n" );
 851                         return false;
 852                 } else {
 853                         $output = join( "\n", $output );
 854                         $output = trim( $output );
 855
 856                         if ( !$output ) {
 857                                 $output = true; #if there's no output, return true
 858                         } elseif ( $msgPattern ) {
 859                                 $groups = array();
 860                                 if ( preg_match( $msgPattern, $output, $groups ) ) {
 861                                         if ( $groups[1] ) {
 862                                                 $output = $groups[1];
 863                                         }
 864                                 }
 865                         }
 866
 867                         wfDebug( __METHOD__ . ": FOUND VIRUS! scanner feedback: $output \n" );
 868                         return $output;
 869                 }
 870         }
 871
 872         /**
 873          * Check if the temporary file is MacBinary-encoded, as some uploads
 874          * from Internet Explorer on Mac OS Classic and Mac OS X will be.
 875          * If so, the data fork will be extracted to a second temporary file,
 876          * which will then be checked for validity and either kept or discarded.
 877          *
 878          * @access private
 879          */
 880         private function checkMacBinary() {
 881                 $macbin = new MacBinary( $this->mTempPath );
 882                 if( $macbin->isValid() ) {
 883                         $dataFile = tempnam( wfTempDir(), 'WikiMacBinary' );
 884                         $dataHandle = fopen( $dataFile, 'wb' );
 885
 886                         wfDebug( __METHOD__ . ": Extracting MacBinary data fork to $dataFile\n" );
 887                         $macbin->extractData( $dataHandle );
 888
 889                         $this->mTempPath = $dataFile;
 890                         $this->mFileSize = $macbin->dataForkLength();
 891
 892                         // We'll have to manually remove the new file if it's not kept.
 893                         $this->mRemoveTempFile = true;
 894                 }
 895                 $macbin->close();
 896         }
 897
 898         /**
 899          * Check if there's an overwrite conflict and, if so, if restrictions
 900          * forbid this user from performing the upload.
 901          *
 902          * @return mixed true on success, error string on failure
 903          * @access private
 904          */
 905         private function checkOverwrite() {
 906                 global $wgUser;
 907                 // First check whether the local file can be overwritten
 908                 $file = $this->getLocalFile();
 909                 if( $file->exists() ) {
 910                         if( !self::userCanReUpload( $wgUser, $file ) )
 911                                 return 'fileexists-forbidden';
 912                         else
 913                                 return true;
 914                 }
 915
 916                 /* Check shared conflicts: if the local file does not exist, but
 917                  * wfFindFile finds a file, it exists in a shared repository.
 918                  */
 919                 $file = wfFindFile( $this->getTitle() );
 920                 if ( $file && !$wgUser->isAllowed( 'reupload-shared' ) )
 921                         return 'fileexists-shared-forbidden';
 922
 923                 return true;
 924         }
 925
 926         /**
 927          * Check if a user is the last uploader
 928          *
 929          * @param User $user
 930          * @param string $img, image name
 931          * @return bool
 932          */
 933         public static function userCanReUpload( User $user, $img ) {
 934                 if( $user->isAllowed( 'reupload' ) )
 935                         return true; // non-conditional
 936                 if( !$user->isAllowed( 'reupload-own' ) )
 937                         return false;
 938                 if( is_string( $img ) )
 939                         $img = wfLocalFile( $img );
 940                 if ( !( $img instanceof LocalFile ) )
 941                         return false;
 942
 943                 return $user->getId() == $img->getUser( 'id' );
 944         }
 945
 946         /**
 947          * Helper function that does various existence checks for a file.
 948          * The following checks are performed:
 949          * - The file exists
 950          * - Article with the same name as the file exists
 951          * - File exists with normalized extension
 952          * - The file looks like a thumbnail and the original exists
 953          *
 954          * @param File $file The file to check
 955          * @return mixed False if the file does not exists, else an array
 956          */
 957         public static function getExistsWarning( $file ) {
 958                 if( $file->exists() )
 959                         return array( 'warning' => 'exists', 'file' => $file );
 960
 961                 if( $file->getTitle()->getArticleID() )
 962                         return array( 'warning' => 'page-exists', 'file' => $file );
 963
 964                 if ( $file->wasDeleted() && !$file->exists() )
 965                         return array( 'warning' => 'was-deleted', 'file' => $file );
 966
 967                 if( strpos( $file->getName(), '.' ) == false ) {
 968                         $partname = $file->getName();
 969                         $extension = '';
 970                 } else {
 971                         $n = strrpos( $file->getName(), '.' );
 972                         $extension = substr( $file->getName(), $n + 1 );
 973                         $partname = substr( $file->getName(), 0, $n );
 974                 }
 975                 $normalizedExtension = File::normalizeExtension( $extension );
 976
 977                 if ( $normalizedExtension != $extension ) {
 978                         // We're not using the normalized form of the extension.
 979                         // Normal form is lowercase, using most common of alternate
 980                         // extensions (eg 'jpg' rather than 'JPEG').
 981                         //
 982                         // Check for another file using the normalized form...
 983                         $nt_lc = Title::makeTitle( NS_FILE, "{$partname}.{$normalizedExtension}" );
 984                         $file_lc = wfLocalFile( $nt_lc );
 985
 986                         if( $file_lc->exists() )
 987                                 return array( 'warning' => 'exists-normalized', 'file' => $file, 'normalizedFile' => $file_lc );
 988                 }
 989
 990                 if ( self::isThumbName( $file->getName() ) ) {
 991                         # Check for filenames like 50px- or 180px-, these are mostly thumbnails
 992                         $nt_thb = Title::newFromText( substr( $partname , strpos( $partname , '-' ) +1 ) . '.' . $extension, NS_FILE );
 993                         $file_thb = wfLocalFile( $nt_thb );
 994                         if( $file_thb->exists() )
 995                                 return array( 'warning' => 'thumb', 'file' => $file, 'thumbFile' => $file_thb );
 996                         else
 997                                 // File does not exist, but we just don't like the name
 998                                 return array( 'warning' => 'thumb-name', 'file' => $file, 'thumbFile' => $file_thb );
 999                 }
1000
1001
1002                 foreach( self::getFilenamePrefixBlacklist() as $prefix ) {
1003                         if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix )
1004                                 return array( 'warning' => 'bad-prefix', 'file' => $file, 'prefix' => $prefix );
1005                 }
1006
1007
1008
1009                 return false;
1010         }
1011
1012         /**
1013          * Helper function that checks whether the filename looks like a thumbnail
1014          */
1015         public static function isThumbName( $filename ) {
1016                 $n = strrpos( $filename, '.' );
1017                 $partname = $n ? substr( $filename, 0, $n ) : $filename;
1018                 return (
1019                                         substr( $partname , 3, 3 ) == 'px-' ||
1020                                         substr( $partname , 2, 3 ) == 'px-'
1021                                 ) &&
1022                                 preg_match( "/[0-9]{2}/" , substr( $partname , 0, 2 ) );
1023         }
1024
1025         /**
1026          * Get a list of blacklisted filename prefixes from [[MediaWiki:filename-prefix-blacklist]]
1027          *
1028          * @return array list of prefixes
1029          */
1030         public static function getFilenamePrefixBlacklist() {
1031                 $blacklist = array();
1032                 $message = wfMsgForContent( 'filename-prefix-blacklist' );
1033                 if( $message && !( wfEmptyMsg( 'filename-prefix-blacklist', $message ) || $message == '-' ) ) {
1034                         $lines = explode( "\n", $message );
1035                         foreach( $lines as $line ) {
1036                                 // Remove comment lines
1037                                 $comment = substr( trim( $line ), 0, 1 );
1038                                 if ( $comment == '#' || $comment == '' ) {
1039                                         continue;
1040                                 }
1041                                 // Remove additional comments after a prefix
1042                                 $comment = strpos( $line, '#' );
1043                                 if ( $comment > 0 ) {
1044                                         $line = substr( $line, 0, $comment-1 );
1045                                 }
1046                                 $blacklist[] = trim( $line );
1047                         }
1048                 }
1049                 return $blacklist;
1050         }
1051
1052 }