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