includes/upload/UploadBase.php

   1 <?php
   2 /**
   3  * Base class for the backend of file upload.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup Upload
  22  */
  23
  24 /**
  25  * @defgroup Upload Upload related
  26  */
  27
  28 /**
  29  * @ingroup Upload
  30  *
  31  * UploadBase and subclasses are the backend of MediaWiki's file uploads.
  32  * The frontends are formed by ApiUpload and SpecialUpload.
  33  *
  34  * @author Brion Vibber
  35  * @author Bryan Tong Minh
  36  * @author Michael Dale
  37  */
  38 abstract class UploadBase {
  39         protected $mTempPath;
  40         protected $mDesiredDestName, $mDestName, $mRemoveTempFile, $mSourceType;
  41         protected $mTitle = false, $mTitleError = 0;
  42         protected $mFilteredName, $mFinalExtension;
  43         protected $mLocalFile, $mFileSize, $mFileProps;
  44         protected $mBlackListedExtensions;
  45         protected $mJavaDetected, $mSVGNSError;
  46
  47         protected static $safeXmlEncodings = array( 'UTF-8', 'ISO-8859-1', 'ISO-8859-2', 'UTF-16', 'UTF-32' );
  48
  49         const SUCCESS = 0;
  50         const OK = 0;
  51         const EMPTY_FILE = 3;
  52         const MIN_LENGTH_PARTNAME = 4;
  53         const ILLEGAL_FILENAME = 5;
  54         const OVERWRITE_EXISTING_FILE = 7; # Not used anymore; handled by verifyTitlePermissions()
  55         const FILETYPE_MISSING = 8;
  56         const FILETYPE_BADTYPE = 9;
  57         const VERIFICATION_ERROR = 10;
  58
  59         # HOOK_ABORTED is the new name of UPLOAD_VERIFICATION_ERROR
  60         const UPLOAD_VERIFICATION_ERROR = 11;
  61         const HOOK_ABORTED = 11;
  62         const FILE_TOO_LARGE = 12;
  63         const WINDOWS_NONASCII_FILENAME = 13;
  64         const FILENAME_TOO_LONG = 14;
  65
  66         const SESSION_STATUS_KEY = 'wsUploadStatusData';
  67
  68         /**
  69          * @param $error int
  70          * @return string
  71          */
  72         public function getVerificationErrorCode( $error ) {
  73                 $code_to_status = array(
  74                         self::EMPTY_FILE => 'empty-file',
  75                         self::FILE_TOO_LARGE => 'file-too-large',
  76                         self::FILETYPE_MISSING => 'filetype-missing',
  77                         self::FILETYPE_BADTYPE => 'filetype-banned',
  78                         self::MIN_LENGTH_PARTNAME => 'filename-tooshort',
  79                         self::ILLEGAL_FILENAME => 'illegal-filename',
  80                         self::OVERWRITE_EXISTING_FILE => 'overwrite',
  81                         self::VERIFICATION_ERROR => 'verification-error',
  82                         self::HOOK_ABORTED => 'hookaborted',
  83                         self::WINDOWS_NONASCII_FILENAME => 'windows-nonascii-filename',
  84                         self::FILENAME_TOO_LONG => 'filename-toolong',
  85                 );
  86                 if ( isset( $code_to_status[$error] ) ) {
  87                         return $code_to_status[$error];
  88                 }
  89
  90                 return 'unknown-error';
  91         }
  92
  93         /**
  94          * Returns true if uploads are enabled.
  95          * Can be override by subclasses.
  96          * @return bool
  97          */
  98         public static function isEnabled() {
  99                 global $wgEnableUploads;
 100
 101                 if ( !$wgEnableUploads ) {
 102                         return false;
 103                 }
 104
 105                 # Check php's file_uploads setting
 106                 return wfIsHHVM() || wfIniGetBool( 'file_uploads' );
 107         }
 108
 109         /**
 110          * Returns true if the user can use this upload module or else a string
 111          * identifying the missing permission.
 112          * Can be overridden by subclasses.
 113          *
 114          * @param $user User
 115          * @return bool
 116          */
 117         public static function isAllowed( $user ) {
 118                 foreach ( array( 'upload', 'edit' ) as $permission ) {
 119                         if ( !$user->isAllowed( $permission ) ) {
 120                                 return $permission;
 121                         }
 122                 }
 123                 return true;
 124         }
 125
 126         // Upload handlers. Should probably just be a global.
 127         static $uploadHandlers = array( 'Stash', 'File', 'Url' );
 128
 129         /**
 130          * Create a form of UploadBase depending on wpSourceType and initializes it
 131          *
 132          * @param $request WebRequest
 133          * @param $type
 134          * @return null
 135          */
 136         public static function createFromRequest( &$request, $type = null ) {
 137                 $type = $type ? $type : $request->getVal( 'wpSourceType', 'File' );
 138
 139                 if ( !$type ) {
 140                         return null;
 141                 }
 142
 143                 // Get the upload class
 144                 $type = ucfirst( $type );
 145
 146                 // Give hooks the chance to handle this request
 147                 $className = null;
 148                 wfRunHooks( 'UploadCreateFromRequest', array( $type, &$className ) );
 149                 if ( is_null( $className ) ) {
 150                         $className = 'UploadFrom' . $type;
 151                         wfDebug( __METHOD__ . ": class name: $className\n" );
 152                         if ( !in_array( $type, self::$uploadHandlers ) ) {
 153                                 return null;
 154                         }
 155                 }
 156
 157                 // Check whether this upload class is enabled
 158                 if ( !call_user_func( array( $className, 'isEnabled' ) ) ) {
 159                         return null;
 160                 }
 161
 162                 // Check whether the request is valid
 163                 if ( !call_user_func( array( $className, 'isValidRequest' ), $request ) ) {
 164                         return null;
 165                 }
 166
 167                 $handler = new $className;
 168
 169                 $handler->initializeFromRequest( $request );
 170                 return $handler;
 171         }
 172
 173         /**
 174          * Check whether a request if valid for this handler
 175          * @param $request
 176          * @return bool
 177          */
 178         public static function isValidRequest( $request ) {
 179                 return false;
 180         }
 181
 182         public function __construct() {}
 183
 184         /**
 185          * Returns the upload type. Should be overridden by child classes
 186          *
 187          * @since 1.18
 188          * @return string
 189          */
 190         public function getSourceType() {
 191                 return null;
 192         }
 193
 194         /**
 195          * Initialize the path information
 196          * @param string $name the desired destination name
 197          * @param string $tempPath the temporary path
 198          * @param int $fileSize the file size
 199          * @param bool $removeTempFile (false) remove the temporary file?
 200          * @throws MWException
 201          */
 202         public function initializePathInfo( $name, $tempPath, $fileSize, $removeTempFile = false ) {
 203                 $this->mDesiredDestName = $name;
 204                 if ( FileBackend::isStoragePath( $tempPath ) ) {
 205                         throw new MWException( __METHOD__ . " given storage path `$tempPath`." );
 206                 }
 207                 $this->mTempPath = $tempPath;
 208                 $this->mFileSize = $fileSize;
 209                 $this->mRemoveTempFile = $removeTempFile;
 210         }
 211
 212         /**
 213          * Initialize from a WebRequest. Override this in a subclass.
 214          */
 215         abstract public function initializeFromRequest( &$request );
 216
 217         /**
 218          * Fetch the file. Usually a no-op
 219          * @return Status
 220          */
 221         public function fetchFile() {
 222                 return Status::newGood();
 223         }
 224
 225         /**
 226          * Return true if the file is empty
 227          * @return bool
 228          */
 229         public function isEmptyFile() {
 230                 return empty( $this->mFileSize );
 231         }
 232
 233         /**
 234          * Return the file size
 235          * @return integer
 236          */
 237         public function getFileSize() {
 238                 return $this->mFileSize;
 239         }
 240
 241         /**
 242          * Get the base 36 SHA1 of the file
 243          * @return string
 244          */
 245         public function getTempFileSha1Base36() {
 246                 return FSFile::getSha1Base36FromPath( $this->mTempPath );
 247         }
 248
 249         /**
 250          * @param string $srcPath the source path
 251          * @return string|bool the real path if it was a virtual URL Returns false on failure
 252          */
 253         function getRealPath( $srcPath ) {
 254                 wfProfileIn( __METHOD__ );
 255                 $repo = RepoGroup::singleton()->getLocalRepo();
 256                 if ( $repo->isVirtualUrl( $srcPath ) ) {
 257                         // @todo just make uploads work with storage paths
 258                         // UploadFromStash loads files via virtual URLs
 259                         $tmpFile = $repo->getLocalCopy( $srcPath );
 260                         if ( $tmpFile ) {
 261                                 $tmpFile->bind( $this ); // keep alive with $this
 262                         }
 263                         $path = $tmpFile ? $tmpFile->getPath() : false;
 264                 } else {
 265                         $path = $srcPath;
 266                 }
 267                 wfProfileOut( __METHOD__ );
 268                 return $path;
 269         }
 270
 271         /**
 272          * Verify whether the upload is sane.
 273          * @return mixed self::OK or else an array with error information
 274          */
 275         public function verifyUpload() {
 276                 wfProfileIn( __METHOD__ );
 277
 278                 /**
 279                  * If there was no filename or a zero size given, give up quick.
 280                  */
 281                 if ( $this->isEmptyFile() ) {
 282                         wfProfileOut( __METHOD__ );
 283                         return array( 'status' => self::EMPTY_FILE );
 284                 }
 285
 286                 /**
 287                  * Honor $wgMaxUploadSize
 288                  */
 289                 $maxSize = self::getMaxUploadSize( $this->getSourceType() );
 290                 if ( $this->mFileSize > $maxSize ) {
 291                         wfProfileOut( __METHOD__ );
 292                         return array(
 293                                 'status' => self::FILE_TOO_LARGE,
 294                                 'max' => $maxSize,
 295                         );
 296                 }
 297
 298                 /**
 299                  * Look at the contents of the file; if we can recognize the
 300                  * type but it's corrupt or data of the wrong type, we should
 301                  * probably not accept it.
 302                  */
 303                 $verification = $this->verifyFile();
 304                 if ( $verification !== true ) {
 305                         wfProfileOut( __METHOD__ );
 306                         return array(
 307                                 'status' => self::VERIFICATION_ERROR,
 308                                 'details' => $verification
 309                         );
 310                 }
 311
 312                 /**
 313                  * Make sure this file can be created
 314                  */
 315                 $result = $this->validateName();
 316                 if ( $result !== true ) {
 317                         wfProfileOut( __METHOD__ );
 318                         return $result;
 319                 }
 320
 321                 $error = '';
 322                 if ( !wfRunHooks( 'UploadVerification',
 323                         array( $this->mDestName, $this->mTempPath, &$error ) )
 324                 ) {
 325                         wfProfileOut( __METHOD__ );
 326                         return array( 'status' => self::HOOK_ABORTED, 'error' => $error );
 327                 }
 328
 329                 wfProfileOut( __METHOD__ );
 330                 return array( 'status' => self::OK );
 331         }
 332
 333         /**
 334          * Verify that the name is valid and, if necessary, that we can overwrite
 335          *
 336          * @return mixed true if valid, otherwise and array with 'status'
 337          * and other keys
 338          **/
 339         public function validateName() {
 340                 $nt = $this->getTitle();
 341                 if ( is_null( $nt ) ) {
 342                         $result = array( 'status' => $this->mTitleError );
 343                         if ( $this->mTitleError == self::ILLEGAL_FILENAME ) {
 344                                 $result['filtered'] = $this->mFilteredName;
 345                         }
 346                         if ( $this->mTitleError == self::FILETYPE_BADTYPE ) {
 347                                 $result['finalExt'] = $this->mFinalExtension;
 348                                 if ( count( $this->mBlackListedExtensions ) ) {
 349                                         $result['blacklistedExt'] = $this->mBlackListedExtensions;
 350                                 }
 351                         }
 352                         return $result;
 353                 }
 354                 $this->mDestName = $this->getLocalFile()->getName();
 355
 356                 return true;
 357         }
 358
 359         /**
 360          * Verify the mime type.
 361          *
 362          * @note Only checks that it is not an evil mime. The does it have
 363          *  correct extension given its mime type check is in verifyFile.
 364          * @param string $mime representing the mime
 365          * @return mixed true if the file is verified, an array otherwise
 366          */
 367         protected function verifyMimeType( $mime ) {
 368                 global $wgVerifyMimeType;
 369                 wfProfileIn( __METHOD__ );
 370                 if ( $wgVerifyMimeType ) {
 371                         wfDebug( "\n\nmime: <$mime> extension: <{$this->mFinalExtension}>\n\n" );
 372                         global $wgMimeTypeBlacklist;
 373                         if ( $this->checkFileExtension( $mime, $wgMimeTypeBlacklist ) ) {
 374                                 wfProfileOut( __METHOD__ );
 375                                 return array( 'filetype-badmime', $mime );
 376                         }
 377
 378                         # Check IE type
 379                         $fp = fopen( $this->mTempPath, 'rb' );
 380                         $chunk = fread( $fp, 256 );
 381                         fclose( $fp );
 382
 383                         $magic = MimeMagic::singleton();
 384                         $extMime = $magic->guessTypesForExtension( $this->mFinalExtension );
 385                         $ieTypes = $magic->getIEMimeTypes( $this->mTempPath, $chunk, $extMime );
 386                         foreach ( $ieTypes as $ieType ) {
 387                                 if ( $this->checkFileExtension( $ieType, $wgMimeTypeBlacklist ) ) {
 388                                         wfProfileOut( __METHOD__ );
 389                                         return array( 'filetype-bad-ie-mime', $ieType );
 390                                 }
 391                         }
 392                 }
 393
 394                 wfProfileOut( __METHOD__ );
 395                 return true;
 396         }
 397
 398         /**
 399          * Verifies that it's ok to include the uploaded file
 400          *
 401          * @return mixed true of the file is verified, array otherwise.
 402          */
 403         protected function verifyFile() {
 404                 global $wgVerifyMimeType;
 405                 wfProfileIn( __METHOD__ );
 406
 407                 $status = $this->verifyPartialFile();
 408                 if ( $status !== true ) {
 409                         wfProfileOut( __METHOD__ );
 410                         return $status;
 411                 }
 412
 413                 $this->mFileProps = FSFile::getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
 414                 $mime = $this->mFileProps['file-mime'];
 415
 416                 if ( $wgVerifyMimeType ) {
 417                         # XXX: Missing extension will be caught by validateName() via getTitle()
 418                         if ( $this->mFinalExtension != '' && !$this->verifyExtension( $mime, $this->mFinalExtension ) ) {
 419                                 wfProfileOut( __METHOD__ );
 420                                 return array( 'filetype-mime-mismatch', $this->mFinalExtension, $mime );
 421                         }
 422                 }
 423
 424                 $handler = MediaHandler::getHandler( $mime );
 425                 if ( $handler ) {
 426                         $handlerStatus = $handler->verifyUpload( $this->mTempPath );
 427                         if ( !$handlerStatus->isOK() ) {
 428                                 $errors = $handlerStatus->getErrorsArray();
 429                                 wfProfileOut( __METHOD__ );
 430                                 return reset( $errors );
 431                         }
 432                 }
 433
 434                 wfRunHooks( 'UploadVerifyFile', array( $this, $mime, &$status ) );
 435                 if ( $status !== true ) {
 436                         wfProfileOut( __METHOD__ );
 437                         return $status;
 438                 }
 439
 440                 wfDebug( __METHOD__ . ": all clear; passing.\n" );
 441                 wfProfileOut( __METHOD__ );
 442                 return true;
 443         }
 444
 445         /**
 446          * A verification routine suitable for partial files
 447          *
 448          * Runs the blacklist checks, but not any checks that may
 449          * assume the entire file is present.
 450          *
 451          * @return Mixed true for valid or array with error message key.
 452          */
 453         protected function verifyPartialFile() {
 454                 global $wgAllowJavaUploads, $wgDisableUploadScriptChecks;
 455                 wfProfileIn( __METHOD__ );
 456
 457                 # getTitle() sets some internal parameters like $this->mFinalExtension
 458                 $this->getTitle();
 459
 460                 $this->mFileProps = FSFile::getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
 461
 462                 # check mime type, if desired
 463                 $mime = $this->mFileProps['file-mime'];
 464                 $status = $this->verifyMimeType( $mime );
 465                 if ( $status !== true ) {
 466                         wfProfileOut( __METHOD__ );
 467                         return $status;
 468                 }
 469
 470                 # check for htmlish code and javascript
 471                 if ( !$wgDisableUploadScriptChecks ) {
 472                         if ( self::detectScript( $this->mTempPath, $mime, $this->mFinalExtension ) ) {
 473                                 wfProfileOut( __METHOD__ );
 474                                 return array( 'uploadscripted' );
 475                         }
 476                         if ( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
 477                                 $svgStatus = $this->detectScriptInSvg( $this->mTempPath );
 478                                 if ( $svgStatus !== false ) {
 479                                         wfProfileOut( __METHOD__ );
 480                                         return $svgStatus;
 481                                 }
 482                         }
 483                 }
 484
 485                 # Check for Java applets, which if uploaded can bypass cross-site
 486                 # restrictions.
 487                 if ( !$wgAllowJavaUploads ) {
 488                         $this->mJavaDetected = false;
 489                         $zipStatus = ZipDirectoryReader::read( $this->mTempPath,
 490                                 array( $this, 'zipEntryCallback' ) );
 491                         if ( !$zipStatus->isOK() ) {
 492                                 $errors = $zipStatus->getErrorsArray();
 493                                 $error = reset( $errors );
 494                                 if ( $error[0] !== 'zip-wrong-format' ) {
 495                                         wfProfileOut( __METHOD__ );
 496                                         return $error;
 497                                 }
 498                         }
 499                         if ( $this->mJavaDetected ) {
 500                                 wfProfileOut( __METHOD__ );
 501                                 return array( 'uploadjava' );
 502                         }
 503                 }
 504
 505                 # Scan the uploaded file for viruses
 506                 $virus = $this->detectVirus( $this->mTempPath );
 507                 if ( $virus ) {
 508                         wfProfileOut( __METHOD__ );
 509                         return array( 'uploadvirus', $virus );
 510                 }
 511
 512                 wfProfileOut( __METHOD__ );
 513                 return true;
 514         }
 515
 516         /**
 517          * Callback for ZipDirectoryReader to detect Java class files.
 518          */
 519         function zipEntryCallback( $entry ) {
 520                 $names = array( $entry['name'] );
 521
 522                 // If there is a null character, cut off the name at it, because JDK's
 523                 // ZIP_GetEntry() uses strcmp() if the name hashes match. If a file name
 524                 // were constructed which had ".class\0" followed by a string chosen to
 525                 // make the hash collide with the truncated name, that file could be
 526                 // returned in response to a request for the .class file.
 527                 $nullPos = strpos( $entry['name'], "\000" );
 528                 if ( $nullPos !== false ) {
 529                         $names[] = substr( $entry['name'], 0, $nullPos );
 530                 }
 531
 532                 // If there is a trailing slash in the file name, we have to strip it,
 533                 // because that's what ZIP_GetEntry() does.
 534                 if ( preg_grep( '!\.class/?$!', $names ) ) {
 535                         $this->mJavaDetected = true;
 536                 }
 537         }
 538
 539         /**
 540          * Alias for verifyTitlePermissions. The function was originally 'verifyPermissions'
 541          * but that suggests it's checking the user, when it's really checking the title + user combination.
 542          * @param $user User object to verify the permissions against
 543          * @return mixed An array as returned by getUserPermissionsErrors or true
 544          *               in case the user has proper permissions.
 545          */
 546         public function verifyPermissions( $user ) {
 547                 return $this->verifyTitlePermissions( $user );
 548         }
 549
 550         /**
 551          * Check whether the user can edit, upload and create the image. This
 552          * checks only against the current title; if it returns errors, it may
 553          * very well be that another title will not give errors. Therefore
 554          * isAllowed() should be called as well for generic is-user-blocked or
 555          * can-user-upload checking.
 556          *
 557          * @param $user User object to verify the permissions against
 558          * @return mixed An array as returned by getUserPermissionsErrors or true
 559          *               in case the user has proper permissions.
 560          */
 561         public function verifyTitlePermissions( $user ) {
 562                 /**
 563                  * If the image is protected, non-sysop users won't be able
 564                  * to modify it by uploading a new revision.
 565                  */
 566                 $nt = $this->getTitle();
 567                 if ( is_null( $nt ) ) {
 568                         return true;
 569                 }
 570                 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
 571                 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
 572                 if ( !$nt->exists() ) {
 573                         $permErrorsCreate = $nt->getUserPermissionsErrors( 'create', $user );
 574                 } else {
 575                         $permErrorsCreate = array();
 576                 }
 577                 if ( $permErrors || $permErrorsUpload || $permErrorsCreate ) {
 578                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
 579                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
 580                         return $permErrors;
 581                 }
 582
 583                 $overwriteError = $this->checkOverwrite( $user );
 584                 if ( $overwriteError !== true ) {
 585                         return array( $overwriteError );
 586                 }
 587
 588                 return true;
 589         }
 590
 591         /**
 592          * Check for non fatal problems with the file.
 593          *
 594          * This should not assume that mTempPath is set.
 595          *
 596          * @return Array of warnings
 597          */
 598         public function checkWarnings() {
 599                 global $wgLang;
 600                 wfProfileIn( __METHOD__ );
 601
 602                 $warnings = array();
 603
 604                 $localFile = $this->getLocalFile();
 605                 $filename = $localFile->getName();
 606
 607                 /**
 608                  * Check whether the resulting filename is different from the desired one,
 609                  * but ignore things like ucfirst() and spaces/underscore things
 610                  */
 611                 $comparableName = str_replace( ' ', '_', $this->mDesiredDestName );
 612                 $comparableName = Title::capitalize( $comparableName, NS_FILE );
 613
 614                 if ( $this->mDesiredDestName != $filename && $comparableName != $filename ) {
 615                         $warnings['badfilename'] = $filename;
 616                         // Debugging for bug 62241
 617                         wfDebugLog( 'upload', "Filename: '$filename', mDesiredDestName: '$this->mDesiredDestName', comparableName: '$comparableName'" );
 618                 }
 619
 620                 // Check whether the file extension is on the unwanted list
 621                 global $wgCheckFileExtensions, $wgFileExtensions;
 622                 if ( $wgCheckFileExtensions ) {
 623                         $extensions = array_unique( $wgFileExtensions );
 624                         if ( !$this->checkFileExtension( $this->mFinalExtension, $extensions ) ) {
 625                                 $warnings['filetype-unwanted-type'] = array( $this->mFinalExtension,
 626                                         $wgLang->commaList( $extensions ), count( $extensions ) );
 627                         }
 628                 }
 629
 630                 global $wgUploadSizeWarning;
 631                 if ( $wgUploadSizeWarning && ( $this->mFileSize > $wgUploadSizeWarning ) ) {
 632                         $warnings['large-file'] = array( $wgUploadSizeWarning, $this->mFileSize );
 633                 }
 634
 635                 if ( $this->mFileSize == 0 ) {
 636                         $warnings['emptyfile'] = true;
 637                 }
 638
 639                 $exists = self::getExistsWarning( $localFile );
 640                 if ( $exists !== false ) {
 641                         $warnings['exists'] = $exists;
 642                 }
 643
 644                 // Check dupes against existing files
 645                 $hash = $this->getTempFileSha1Base36();
 646                 $dupes = RepoGroup::singleton()->findBySha1( $hash );
 647                 $title = $this->getTitle();
 648                 // Remove all matches against self
 649                 foreach ( $dupes as $key => $dupe ) {
 650                         if ( $title->equals( $dupe->getTitle() ) ) {
 651                                 unset( $dupes[$key] );
 652                         }
 653                 }
 654                 if ( $dupes ) {
 655                         $warnings['duplicate'] = $dupes;
 656                 }
 657
 658                 // Check dupes against archives
 659                 $archivedImage = new ArchivedFile( null, 0, "{$hash}.{$this->mFinalExtension}" );
 660                 if ( $archivedImage->getID() > 0 ) {
 661                         if ( $archivedImage->userCan( File::DELETED_FILE ) ) {
 662                                 $warnings['duplicate-archive'] = $archivedImage->getName();
 663                         } else {
 664                                 $warnings['duplicate-archive'] = '';
 665                         }
 666                 }
 667
 668                 wfProfileOut( __METHOD__ );
 669                 return $warnings;
 670         }
 671
 672         /**
 673          * Really perform the upload. Stores the file in the local repo, watches
 674          * if necessary and runs the UploadComplete hook.
 675          *
 676          * @param $comment
 677          * @param $pageText
 678          * @param $watch
 679          * @param $user User
 680          *
 681          * @return Status indicating the whether the upload succeeded.
 682          */
 683         public function performUpload( $comment, $pageText, $watch, $user ) {
 684                 wfProfileIn( __METHOD__ );
 685
 686                 $status = $this->getLocalFile()->upload(
 687                         $this->mTempPath,
 688                         $comment,
 689                         $pageText,
 690                         File::DELETE_SOURCE,
 691                         $this->mFileProps,
 692                         false,
 693                         $user
 694                 );
 695
 696                 if ( $status->isGood() ) {
 697                         if ( $watch ) {
 698                                 WatchAction::doWatch( $this->getLocalFile()->getTitle(), $user, WatchedItem::IGNORE_USER_RIGHTS );
 699                         }
 700                         wfRunHooks( 'UploadComplete', array( &$this ) );
 701                 }
 702
 703                 wfProfileOut( __METHOD__ );
 704                 return $status;
 705         }
 706
 707         /**
 708          * Returns the title of the file to be uploaded. Sets mTitleError in case
 709          * the name was illegal.
 710          *
 711          * @return Title The title of the file or null in case the name was illegal
 712          */
 713         public function getTitle() {
 714                 if ( $this->mTitle !== false ) {
 715                         return $this->mTitle;
 716                 }
 717                 /* Assume that if a user specified File:Something.jpg, this is an error
 718                  * and that the namespace prefix needs to be stripped of.
 719                  */
 720                 $title = Title::newFromText( $this->mDesiredDestName );
 721                 if ( $title && $title->getNamespace() == NS_FILE ) {
 722                         $this->mFilteredName = $title->getDBkey();
 723                 } else {
 724                         $this->mFilteredName = $this->mDesiredDestName;
 725                 }
 726
 727                 # oi_archive_name is max 255 bytes, which include a timestamp and an
 728                 # exclamation mark, so restrict file name to 240 bytes.
 729                 if ( strlen( $this->mFilteredName ) > 240 ) {
 730                         $this->mTitleError = self::FILENAME_TOO_LONG;
 731                         $this->mTitle = null;
 732                         return $this->mTitle;
 733                 }
 734
 735                 /**
 736                  * Chop off any directories in the given filename. Then
 737                  * filter out illegal characters, and try to make a legible name
 738                  * out of it. We'll strip some silently that Title would die on.
 739                  */
 740                 $this->mFilteredName = wfStripIllegalFilenameChars( $this->mFilteredName );
 741                 /* Normalize to title form before we do any further processing */
 742                 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 743                 if ( is_null( $nt ) ) {
 744                         $this->mTitleError = self::ILLEGAL_FILENAME;
 745                         $this->mTitle = null;
 746                         return $this->mTitle;
 747                 }
 748                 $this->mFilteredName = $nt->getDBkey();
 749
 750                 /**
 751                  * We'll want to blacklist against *any* 'extension', and use
 752                  * only the final one for the whitelist.
 753                  */
 754                 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName );
 755
 756                 if ( count( $ext ) ) {
 757                         $this->mFinalExtension = trim( $ext[count( $ext ) - 1] );
 758                 } else {
 759                         $this->mFinalExtension = '';
 760
 761                         # No extension, try guessing one
 762                         $magic = MimeMagic::singleton();
 763                         $mime = $magic->guessMimeType( $this->mTempPath );
 764                         if ( $mime !== 'unknown/unknown' ) {
 765                                 # Get a space separated list of extensions
 766                                 $extList = $magic->getExtensionsForType( $mime );
 767                                 if ( $extList ) {
 768                                         # Set the extension to the canonical extension
 769                                         $this->mFinalExtension = strtok( $extList, ' ' );
 770
 771                                         # Fix up the other variables
 772                                         $this->mFilteredName .= ".{$this->mFinalExtension}";
 773                                         $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 774                                         $ext = array( $this->mFinalExtension );
 775                                 }
 776                         }
 777                 }
 778
 779                 /* Don't allow users to override the blacklist (check file extension) */
 780                 global $wgCheckFileExtensions, $wgStrictFileExtensions;
 781                 global $wgFileExtensions, $wgFileBlacklist;
 782
 783                 $blackListedExtensions = $this->checkFileExtensionList( $ext, $wgFileBlacklist );
 784
 785                 if ( $this->mFinalExtension == '' ) {
 786                         $this->mTitleError = self::FILETYPE_MISSING;
 787                         $this->mTitle = null;
 788                         return $this->mTitle;
 789                 } elseif ( $blackListedExtensions ||
 790                                 ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
 791                                         !$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) ) ) {
 792                         $this->mBlackListedExtensions = $blackListedExtensions;
 793                         $this->mTitleError = self::FILETYPE_BADTYPE;
 794                         $this->mTitle = null;
 795                         return $this->mTitle;
 796                 }
 797
 798                 // Windows may be broken with special characters, see bug XXX
 799                 if ( wfIsWindows() && !preg_match( '/^[\x0-\x7f]*$/', $nt->getText() ) ) {
 800                         $this->mTitleError = self::WINDOWS_NONASCII_FILENAME;
 801                         $this->mTitle = null;
 802                         return $this->mTitle;
 803                 }
 804
 805                 # If there was more than one "extension", reassemble the base
 806                 # filename to prevent bogus complaints about length
 807                 if ( count( $ext ) > 1 ) {
 808                         for ( $i = 0; $i < count( $ext ) - 1; $i++ ) {
 809                                 $partname .= '.' . $ext[$i];
 810                         }
 811                 }
 812
 813                 if ( strlen( $partname ) < 1 ) {
 814                         $this->mTitleError = self::MIN_LENGTH_PARTNAME;
 815                         $this->mTitle = null;
 816                         return $this->mTitle;
 817                 }
 818
 819                 $this->mTitle = $nt;
 820                 return $this->mTitle;
 821         }
 822
 823         /**
 824          * Return the local file and initializes if necessary.
 825          *
 826          * @return LocalFile|null
 827          */
 828         public function getLocalFile() {
 829                 if ( is_null( $this->mLocalFile ) ) {
 830                         $nt = $this->getTitle();
 831                         $this->mLocalFile = is_null( $nt ) ? null : wfLocalFile( $nt );
 832                 }
 833                 return $this->mLocalFile;
 834         }
 835
 836         /**
 837          * If the user does not supply all necessary information in the first upload form submission (either by accident or
 838          * by design) then we may want to stash the file temporarily, get more information, and publish the file later.
 839          *
 840          * This method will stash a file in a temporary directory for later processing, and save the necessary descriptive info
 841          * into the database.
 842          * This method returns the file object, which also has a 'fileKey' property which can be passed through a form or
 843          * API request to find this stashed file again.
 844          *
 845          * @param $user User
 846          * @return UploadStashFile stashed file
 847          */
 848         public function stashFile( User $user = null ) {
 849                 // was stashSessionFile
 850                 wfProfileIn( __METHOD__ );
 851
 852                 $stash = RepoGroup::singleton()->getLocalRepo()->getUploadStash( $user );
 853                 $file = $stash->stashFile( $this->mTempPath, $this->getSourceType() );
 854                 $this->mLocalFile = $file;
 855
 856                 wfProfileOut( __METHOD__ );
 857                 return $file;
 858         }
 859
 860         /**
 861          * Stash a file in a temporary directory, returning a key which can be used to find the file again. See stashFile().
 862          *
 863          * @return String: file key
 864          */
 865         public function stashFileGetKey() {
 866                 return $this->stashFile()->getFileKey();
 867         }
 868
 869         /**
 870          * alias for stashFileGetKey, for backwards compatibility
 871          *
 872          * @return String: file key
 873          */
 874         public function stashSession() {
 875                 return $this->stashFileGetKey();
 876         }
 877
 878         /**
 879          * If we've modified the upload file we need to manually remove it
 880          * on exit to clean up.
 881          */
 882         public function cleanupTempFile() {
 883                 if ( $this->mRemoveTempFile && $this->mTempPath && file_exists( $this->mTempPath ) ) {
 884                         wfDebug( __METHOD__ . ": Removing temporary file {$this->mTempPath}\n" );
 885                         unlink( $this->mTempPath );
 886                 }
 887         }
 888
 889         public function getTempPath() {
 890                 return $this->mTempPath;
 891         }
 892
 893         /**
 894          * Split a file into a base name and all dot-delimited 'extensions'
 895          * on the end. Some web server configurations will fall back to
 896          * earlier pseudo-'extensions' to determine type and execute
 897          * scripts, so the blacklist needs to check them all.
 898          *
 899          * @param $filename string
 900          * @return array
 901          */
 902         public static function splitExtensions( $filename ) {
 903                 $bits = explode( '.', $filename );
 904                 $basename = array_shift( $bits );
 905                 return array( $basename, $bits );
 906         }
 907
 908         /**
 909          * Perform case-insensitive match against a list of file extensions.
 910          * Returns true if the extension is in the list.
 911          *
 912          * @param $ext String
 913          * @param $list Array
 914          * @return Boolean
 915          */
 916         public static function checkFileExtension( $ext, $list ) {
 917                 return in_array( strtolower( $ext ), $list );
 918         }
 919
 920         /**
 921          * Perform case-insensitive match against a list of file extensions.
 922          * Returns an array of matching extensions.
 923          *
 924          * @param $ext Array
 925          * @param $list Array
 926          * @return Boolean
 927          */
 928         public static function checkFileExtensionList( $ext, $list ) {
 929                 return array_intersect( array_map( 'strtolower', $ext ), $list );
 930         }
 931
 932         /**
 933          * Checks if the mime type of the uploaded file matches the file extension.
 934          *
 935          * @param string $mime the mime type of the uploaded file
 936          * @param string $extension the filename extension that the file is to be served with
 937          * @return Boolean
 938          */
 939         public static function verifyExtension( $mime, $extension ) {
 940                 $magic = MimeMagic::singleton();
 941
 942                 if ( !$mime || $mime == 'unknown' || $mime == 'unknown/unknown' ) {
 943                         if ( !$magic->isRecognizableExtension( $extension ) ) {
 944                                 wfDebug( __METHOD__ . ": passing file with unknown detected mime type; " .
 945                                         "unrecognized extension '$extension', can't verify\n" );
 946                                 return true;
 947                         } else {
 948                                 wfDebug( __METHOD__ . ": rejecting file with unknown detected mime type; " .
 949                                         "recognized extension '$extension', so probably invalid file\n" );
 950                                 return false;
 951                         }
 952                 }
 953
 954                 $match = $magic->isMatchingExtension( $extension, $mime );
 955
 956                 if ( $match === null ) {
 957                         if ( $magic->getTypesForExtension( $extension ) !== null ) {
 958                                 wfDebug( __METHOD__ . ": No extension known for $mime, but we know a mime for $extension\n" );
 959                                 return false;
 960                         } else {
 961                                 wfDebug( __METHOD__ . ": no file extension known for mime type $mime, passing file\n" );
 962                                 return true;
 963                         }
 964                 } elseif ( $match === true ) {
 965                         wfDebug( __METHOD__ . ": mime type $mime matches extension $extension, passing file\n" );
 966
 967                         #TODO: if it's a bitmap, make sure PHP or ImageMagic resp. can handle it!
 968                         return true;
 969
 970                 } else {
 971                         wfDebug( __METHOD__ . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
 972                         return false;
 973                 }
 974         }
 975
 976         /**
 977          * Heuristic for detecting files that *could* contain JavaScript instructions or
 978          * things that may look like HTML to a browser and are thus
 979          * potentially harmful. The present implementation will produce false
 980          * positives in some situations.
 981          *
 982          * @param string $file pathname to the temporary upload file
 983          * @param string $mime the mime type of the file
 984          * @param string $extension the extension of the file
 985          * @return Boolean: true if the file contains something looking like embedded scripts
 986          */
 987         public static function detectScript( $file, $mime, $extension ) {
 988                 global $wgAllowTitlesInSVG;
 989                 wfProfileIn( __METHOD__ );
 990
 991                 # ugly hack: for text files, always look at the entire file.
 992                 # For binary field, just check the first K.
 993
 994                 if ( strpos( $mime, 'text/' ) === 0 ) {
 995                         $chunk = file_get_contents( $file );
 996                 } else {
 997                         $fp = fopen( $file, 'rb' );
 998                         $chunk = fread( $fp, 1024 );
 999                         fclose( $fp );
1000                 }
1001
1002                 $chunk = strtolower( $chunk );
1003
1004                 if ( !$chunk ) {
1005                         wfProfileOut( __METHOD__ );
1006                         return false;
1007                 }
1008
1009                 # decode from UTF-16 if needed (could be used for obfuscation).
1010                 if ( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
1011                         $enc = 'UTF-16BE';
1012                 } elseif ( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
1013                         $enc = 'UTF-16LE';
1014                 } else {
1015                         $enc = null;
1016                 }
1017
1018                 if ( $enc ) {
1019                         $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
1020                 }
1021
1022                 $chunk = trim( $chunk );
1023
1024                 # @todo FIXME: Convert from UTF-16 if necessary!
1025                 wfDebug( __METHOD__ . ": checking for embedded scripts and HTML stuff\n" );
1026
1027                 # check for HTML doctype
1028                 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
1029                         wfProfileOut( __METHOD__ );
1030                         return true;
1031                 }
1032
1033                 // Some browsers will interpret obscure xml encodings as UTF-8, while
1034                 // PHP/expat will interpret the given encoding in the xml declaration (bug 47304)
1035                 if ( $extension == 'svg' || strpos( $mime, 'image/svg' ) === 0 ) {
1036                         if ( self::checkXMLEncodingMissmatch( $file ) ) {
1037                                 wfProfileOut( __METHOD__ );
1038                                 return true;
1039                         }
1040                 }
1041
1042                 /**
1043                  * Internet Explorer for Windows performs some really stupid file type
1044                  * autodetection which can cause it to interpret valid image files as HTML
1045                  * and potentially execute JavaScript, creating a cross-site scripting
1046                  * attack vectors.
1047                  *
1048                  * Apple's Safari browser also performs some unsafe file type autodetection
1049                  * which can cause legitimate files to be interpreted as HTML if the
1050                  * web server is not correctly configured to send the right content-type
1051                  * (or if you're really uploading plain text and octet streams!)
1052                  *
1053                  * Returns true if IE is likely to mistake the given file for HTML.
1054                  * Also returns true if Safari would mistake the given file for HTML
1055                  * when served with a generic content-type.
1056                  */
1057                 $tags = array(
1058                         '<a href',
1059                         '<body',
1060                         '<head',
1061                         '<html',   #also in safari
1062                         '<img',
1063                         '<pre',
1064                         '<script', #also in safari
1065                         '<table'
1066                 );
1067
1068                 if ( !$wgAllowTitlesInSVG && $extension !== 'svg' && $mime !== 'image/svg' ) {
1069                         $tags[] = '<title';
1070                 }
1071
1072                 foreach ( $tags as $tag ) {
1073                         if ( false !== strpos( $chunk, $tag ) ) {
1074                                 wfDebug( __METHOD__ . ": found something that may make it be mistaken for html: $tag\n" );
1075                                 wfProfileOut( __METHOD__ );
1076                                 return true;
1077                         }
1078                 }
1079
1080                 /*
1081                  * look for JavaScript
1082                  */
1083
1084                 # resolve entity-refs to look at attributes. may be harsh on big files... cache result?
1085                 $chunk = Sanitizer::decodeCharReferences( $chunk );
1086
1087                 # look for script-types
1088                 if ( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
1089                         wfDebug( __METHOD__ . ": found script types\n" );
1090                         wfProfileOut( __METHOD__ );
1091                         return true;
1092                 }
1093
1094                 # look for html-style script-urls
1095                 if ( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1096                         wfDebug( __METHOD__ . ": found html-style script urls\n" );
1097                         wfProfileOut( __METHOD__ );
1098                         return true;
1099                 }
1100
1101                 # look for css-style script-urls
1102                 if ( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1103                         wfDebug( __METHOD__ . ": found css-style script urls\n" );
1104                         wfProfileOut( __METHOD__ );
1105                         return true;
1106                 }
1107
1108                 wfDebug( __METHOD__ . ": no scripts found\n" );
1109                 wfProfileOut( __METHOD__ );
1110                 return false;
1111         }
1112
1113         /**
1114          * Check a whitelist of xml encodings that are known not to be interpreted differently
1115          * by the server's xml parser (expat) and some common browsers.
1116          *
1117          * @param string $file pathname to the temporary upload file
1118          * @return Boolean: true if the file contains an encoding that could be misinterpreted
1119          */
1120         public static function checkXMLEncodingMissmatch( $file ) {
1121                 global $wgSVGMetadataCutoff;
1122                 $contents = file_get_contents( $file, false, null, -1, $wgSVGMetadataCutoff );
1123                 $encodingRegex = '!encoding[ \t\n\r]*=[ \t\n\r]*[\'"](.*?)[\'"]!si';
1124
1125                 if ( preg_match( "!<\?xml\b(.*?)\?>!si", $contents, $matches ) ) {
1126                         if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1127                                 && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1128                         ) {
1129                                 wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1130                                 return true;
1131                         }
1132                 } elseif ( preg_match( "!<\?xml\b!si", $contents ) ) {
1133                         // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1134                         // bytes. There shouldn't be a legitimate reason for this to happen.
1135                         wfDebug( __METHOD__ . ": Unmatched XML declaration start\n" );
1136                         return true;
1137                 } elseif ( substr( $contents, 0, 4 ) == "\x4C\x6F\xA7\x94" ) {
1138                         // EBCDIC encoded XML
1139                         wfDebug( __METHOD__ . ": EBCDIC Encoded XML\n" );
1140                         return true;
1141                 }
1142
1143                 // It's possible the file is encoded with multi-byte encoding, so re-encode attempt to
1144                 // detect the encoding in case is specifies an encoding not whitelisted in self::$safeXmlEncodings
1145                 $attemptEncodings = array( 'UTF-16', 'UTF-16BE', 'UTF-32', 'UTF-32BE' );
1146                 foreach ( $attemptEncodings as $encoding ) {
1147                         wfSuppressWarnings();
1148                         $str = iconv( $encoding, 'UTF-8', $contents );
1149                         wfRestoreWarnings();
1150                         if ( $str != '' && preg_match( "!<\?xml\b(.*?)\?>!si", $str, $matches ) ) {
1151                                 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1152                                         && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1153                                 ) {
1154                                         wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1155                                         return true;
1156                                 }
1157                         } elseif ( $str != '' && preg_match( "!<\?xml\b!si", $str ) ) {
1158                                 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1159                                 // bytes. There shouldn't be a legitimate reason for this to happen.
1160                                 wfDebug( __METHOD__ . ": Unmatched XML declaration start\n" );
1161                                 return true;
1162                         }
1163                 }
1164
1165                 return false;
1166         }
1167
1168         /**
1169          * @param $filename string
1170          * @return mixed false of the file is verified (does not contain scripts), array otherwise.
1171          */
1172         protected function detectScriptInSvg( $filename ) {
1173                 $this->mSVGNSError = false;
1174                 $check = new XmlTypeCheck(
1175                         $filename,
1176                         array( $this, 'checkSvgScriptCallback' ),
1177                         true,
1178                         array( 'processing_instruction_handler' => 'UploadBase::checkSvgPICallback' )
1179                 );
1180                 if ( $check->wellFormed !== true ) {
1181                         // Invalid xml (bug 58553)
1182                         return array( 'uploadinvalidxml' );
1183                 } elseif ( $check->filterMatch ) {
1184                         if ( $this->mSVGNSError ) {
1185                                 return array( 'uploadscriptednamespace', $this->mSVGNSError );
1186                         }
1187                         return array( 'uploadscripted' );
1188                 }
1189                 return false;
1190         }
1191
1192         /**
1193          * Callback to filter SVG Processing Instructions.
1194          * @param $target string processing instruction name
1195          * @param $data string processing instruction attribute and value
1196          * @return bool (true if the filter identified something bad)
1197          */
1198         public static function checkSvgPICallback( $target, $data ) {
1199                 // Don't allow external stylesheets (bug 57550)
1200                 if ( preg_match( '/xml-stylesheet/i', $target ) ) {
1201                         return true;
1202                 }
1203                 return false;
1204         }
1205
1206         /**
1207          * @todo Replace this with a whitelist filter!
1208          * @param $element string
1209          * @param $attribs array
1210          * @return bool
1211          */
1212         public function checkSvgScriptCallback( $element, $attribs ) {
1213                 list( $namespace, $strippedElement ) = $this->splitXmlNamespace( $element );
1214
1215                 static $validNamespaces = array(
1216                         '',
1217                         'adobe:ns:meta/',
1218                         'http://creativecommons.org/ns#',
1219                         'http://inkscape.sourceforge.net/dtd/sodipodi-0.dtd',
1220                         'http://ns.adobe.com/adobeillustrator/10.0/',
1221                         'http://ns.adobe.com/adobesvgviewerextensions/3.0/',
1222                         'http://ns.adobe.com/extensibility/1.0/',
1223                         'http://ns.adobe.com/flows/1.0/',
1224                         'http://ns.adobe.com/illustrator/1.0/',
1225                         'http://ns.adobe.com/imagereplacement/1.0/',
1226                         'http://ns.adobe.com/pdf/1.3/',
1227                         'http://ns.adobe.com/photoshop/1.0/',
1228                         'http://ns.adobe.com/saveforweb/1.0/',
1229                         'http://ns.adobe.com/variables/1.0/',
1230                         'http://ns.adobe.com/xap/1.0/',
1231                         'http://ns.adobe.com/xap/1.0/g/',
1232                         'http://ns.adobe.com/xap/1.0/g/img/',
1233                         'http://ns.adobe.com/xap/1.0/mm/',
1234                         'http://ns.adobe.com/xap/1.0/rights/',
1235                         'http://ns.adobe.com/xap/1.0/stype/dimensions#',
1236                         'http://ns.adobe.com/xap/1.0/stype/font#',
1237                         'http://ns.adobe.com/xap/1.0/stype/manifestitem#',
1238                         'http://ns.adobe.com/xap/1.0/stype/resourceevent#',
1239                         'http://ns.adobe.com/xap/1.0/stype/resourceref#',
1240                         'http://ns.adobe.com/xap/1.0/t/pg/',
1241                         'http://purl.org/dc/elements/1.1/',
1242                         'http://purl.org/dc/elements/1.1',
1243                         'http://schemas.microsoft.com/visio/2003/svgextensions/',
1244                         'http://sodipodi.sourceforge.net/dtd/sodipodi-0.dtd',
1245                         'http://web.resource.org/cc/',
1246                         'http://www.freesoftware.fsf.org/bkchem/cdml',
1247                         'http://www.inkscape.org/namespaces/inkscape',
1248                         'http://www.w3.org/1999/02/22-rdf-syntax-ns#',
1249                         'http://www.w3.org/2000/svg',
1250                 );
1251
1252                 if ( !in_array( $namespace, $validNamespaces ) ) {
1253                         wfDebug( __METHOD__ . ": Non-svg namespace '$namespace' in uploaded file.\n" );
1254                         // @TODO return a status object to a closure in XmlTypeCheck, for MW1.21+
1255                         $this->mSVGNSError = $namespace;
1256                         return true;
1257                 }
1258
1259                 /*
1260                  * check for elements that can contain javascript
1261                  */
1262                 if ( $strippedElement == 'script' ) {
1263                         wfDebug( __METHOD__ . ": Found script element '$element' in uploaded file.\n" );
1264                         return true;
1265                 }
1266
1267                 # e.g., <svg xmlns="http://www.w3.org/2000/svg"> <handler xmlns:ev="http://www.w3.org/2001/xml-events" ev:event="load">alert(1)</handler> </svg>
1268                 if ( $strippedElement == 'handler' ) {
1269                         wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1270                         return true;
1271                 }
1272
1273                 # SVG reported in Feb '12 that used xml:stylesheet to generate javascript block
1274                 if ( $strippedElement == 'stylesheet' ) {
1275                         wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1276                         return true;
1277                 }
1278
1279                 # Block iframes, in case they pass the namespace check
1280                 if ( $strippedElement == 'iframe' ) {
1281                         wfDebug( __METHOD__ . ": iframe in uploaded file.\n" );
1282                         return true;
1283                 }
1284
1285                 foreach ( $attribs as $attrib => $value ) {
1286                         $stripped = $this->stripXmlNamespace( $attrib );
1287                         $value = strtolower( $value );
1288
1289                         if ( substr( $stripped, 0, 2 ) == 'on' ) {
1290                                 wfDebug( __METHOD__ . ": Found event-handler attribute '$attrib'='$value' in uploaded file.\n" );
1291                                 return true;
1292                         }
1293
1294                         # href with javascript target
1295                         if ( $stripped == 'href' && strpos( strtolower( $value ), 'javascript:' ) !== false ) {
1296                                 wfDebug( __METHOD__ . ": Found script in href attribute '$attrib'='$value' in uploaded file.\n" );
1297                                 return true;
1298                         }
1299
1300                         # href with embedded svg as target
1301                         if ( $stripped == 'href' && preg_match( '!data:[^,]*image/svg[^,]*,!sim', $value ) ) {
1302                                 wfDebug( __METHOD__ . ": Found href to embedded svg \"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1303                                 return true;
1304                         }
1305
1306                         # href with embedded (text/xml) svg as target
1307                         if ( $stripped == 'href' && preg_match( '!data:[^,]*text/xml[^,]*,!sim', $value ) ) {
1308                                 wfDebug( __METHOD__ . ": Found href to embedded svg \"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1309                                 return true;
1310                         }
1311
1312                         # use set/animate to add event-handler attribute to parent
1313                         if ( ( $strippedElement == 'set' || $strippedElement == 'animate' ) && $stripped == 'attributename' && substr( $value, 0, 2 ) == 'on' ) {
1314                                 wfDebug( __METHOD__ . ": Found svg setting event-handler attribute with \"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1315                                 return true;
1316                         }
1317
1318                         # use set to add href attribute to parent element
1319                         if ( $strippedElement == 'set' && $stripped == 'attributename' && strpos( $value, 'href' ) !== false ) {
1320                                 wfDebug( __METHOD__ . ": Found svg setting href attribute '$value' in uploaded file.\n" );
1321                                 return true;
1322                         }
1323
1324                         # use set to add a remote / data / script target to an element
1325                         if ( $strippedElement == 'set' && $stripped == 'to' && preg_match( '!(http|https|data|script):!sim', $value ) ) {
1326                                 wfDebug( __METHOD__ . ": Found svg setting attribute to '$value' in uploaded file.\n" );
1327                                 return true;
1328                         }
1329
1330                         # use handler attribute with remote / data / script
1331                         if ( $stripped == 'handler' && preg_match( '!(http|https|data|script):!sim', $value ) ) {
1332                                 wfDebug( __METHOD__ . ": Found svg setting handler with remote/data/script '$attrib'='$value' in uploaded file.\n" );
1333                                 return true;
1334                         }
1335
1336                         # use CSS styles to bring in remote code
1337                         # catch url("http:..., url('http:..., url(http:..., but not url("#..., url('#..., url(#....
1338                         if ( $stripped == 'style' && preg_match_all( '!((?:font|clip-path|fill|filter|marker|marker-end|marker-mid|marker-start|mask|stroke)\s*:\s*url\s*\(\s*["\']?\s*[^#]+.*?\))!sim', $value, $matches ) ) {
1339                                 foreach ( $matches[1] as $match ) {
1340                                         if ( !preg_match( '!(?:font|clip-path|fill|filter|marker|marker-end|marker-mid|marker-start|mask|stroke)\s*:\s*url\s*\(\s*(#|\'#|"#)!sim', $match ) ) {
1341                                                 wfDebug( __METHOD__ . ": Found svg setting a style with remote url '$attrib'='$value' in uploaded file.\n" );
1342                                                 return true;
1343                                         }
1344                                 }
1345                         }
1346
1347                         # image filters can pull in url, which could be svg that executes scripts
1348                         if ( $strippedElement == 'image' && $stripped == 'filter' && preg_match( '!url\s*\(!sim', $value ) ) {
1349                                 wfDebug( __METHOD__ . ": Found image filter with url: \"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1350                                 return true;
1351                         }
1352
1353                 }
1354
1355                 return false; //No scripts detected
1356         }
1357
1358         /**
1359          * Divide the element name passed by the xml parser to the callback into URI and prifix.
1360          * @param $name string
1361          * @return array containing the namespace URI and prefix
1362          */
1363         private static function splitXmlNamespace( $element ) {
1364                 // 'http://www.w3.org/2000/svg:script' -> array( 'http://www.w3.org/2000/svg', 'script' )
1365                 $parts = explode( ':', strtolower( $element ) );
1366                 $name = array_pop( $parts );
1367                 $ns = implode( ':', $parts );
1368                 return array( $ns, $name );
1369         }
1370
1371         /**
1372          * @param $name string
1373          * @return string
1374          */
1375         private function stripXmlNamespace( $name ) {
1376                 // 'http://www.w3.org/2000/svg:script' -> 'script'
1377                 $parts = explode( ':', strtolower( $name ) );
1378                 return array_pop( $parts );
1379         }
1380
1381         /**
1382          * Generic wrapper function for a virus scanner program.
1383          * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
1384          * $wgAntivirusRequired may be used to deny upload if the scan fails.
1385          *
1386          * @param string $file pathname to the temporary upload file
1387          * @return mixed false if not virus is found, NULL if the scan fails or is disabled,
1388          *         or a string containing feedback from the virus scanner if a virus was found.
1389          *         If textual feedback is missing but a virus was found, this function returns true.
1390          */
1391         public static function detectVirus( $file ) {
1392                 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
1393                 wfProfileIn( __METHOD__ );
1394
1395                 if ( !$wgAntivirus ) {
1396                         wfDebug( __METHOD__ . ": virus scanner disabled\n" );
1397                         wfProfileOut( __METHOD__ );
1398                         return null;
1399                 }
1400
1401                 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
1402                         wfDebug( __METHOD__ . ": unknown virus scanner: $wgAntivirus\n" );
1403                         $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1\n</div>",
1404                                 array( 'virus-badscanner', $wgAntivirus ) );
1405                         wfProfileOut( __METHOD__ );
1406                         return wfMessage( 'virus-unknownscanner' )->text() . " $wgAntivirus";
1407                 }
1408
1409                 # look up scanner configuration
1410                 $command = $wgAntivirusSetup[$wgAntivirus]['command'];
1411                 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
1412                 $msgPattern = isset( $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ) ?
1413                         $wgAntivirusSetup[$wgAntivirus]['messagepattern'] : null;
1414
1415                 if ( strpos( $command, "%f" ) === false ) {
1416                         # simple pattern: append file to scan
1417                         $command .= " " . wfEscapeShellArg( $file );
1418                 } else {
1419                         # complex pattern: replace "%f" with file to scan
1420                         $command = str_replace( "%f", wfEscapeShellArg( $file ), $command );
1421                 }
1422
1423                 wfDebug( __METHOD__ . ": running virus scan: $command \n" );
1424
1425                 # execute virus scanner
1426                 $exitCode = false;
1427
1428                 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
1429                 #      that does not seem to be worth the pain.
1430                 #      Ask me (Duesentrieb) about it if it's ever needed.
1431                 $output = wfShellExecWithStderr( $command, $exitCode );
1432
1433                 # map exit code to AV_xxx constants.
1434                 $mappedCode = $exitCode;
1435                 if ( $exitCodeMap ) {
1436                         if ( isset( $exitCodeMap[$exitCode] ) ) {
1437                                 $mappedCode = $exitCodeMap[$exitCode];
1438                         } elseif ( isset( $exitCodeMap["*"] ) ) {
1439                                 $mappedCode = $exitCodeMap["*"];
1440                         }
1441                 }
1442
1443                 /* NB: AV_NO_VIRUS is 0 but AV_SCAN_FAILED is false,
1444                  * so we need the strict equalities === and thus can't use a switch here
1445                  */
1446                 if ( $mappedCode === AV_SCAN_FAILED ) {
1447                         # scan failed (code was mapped to false by $exitCodeMap)
1448                         wfDebug( __METHOD__ . ": failed to scan $file (code $exitCode).\n" );
1449
1450                         $output = $wgAntivirusRequired ? wfMessage( 'virus-scanfailed', array( $exitCode ) )->text() : null;
1451                 } elseif ( $mappedCode === AV_SCAN_ABORTED ) {
1452                         # scan failed because filetype is unknown (probably imune)
1453                         wfDebug( __METHOD__ . ": unsupported file type $file (code $exitCode).\n" );
1454                         $output = null;
1455                 } elseif ( $mappedCode === AV_NO_VIRUS ) {
1456                         # no virus found
1457                         wfDebug( __METHOD__ . ": file passed virus scan.\n" );
1458                         $output = false;
1459                 } else {
1460                         $output = trim( $output );
1461
1462                         if ( !$output ) {
1463                                 $output = true; #if there's no output, return true
1464                         } elseif ( $msgPattern ) {
1465                                 $groups = array();
1466                                 if ( preg_match( $msgPattern, $output, $groups ) ) {
1467                                         if ( $groups[1] ) {
1468                                                 $output = $groups[1];
1469                                         }
1470                                 }
1471                         }
1472
1473                         wfDebug( __METHOD__ . ": FOUND VIRUS! scanner feedback: $output \n" );
1474                 }
1475
1476                 wfProfileOut( __METHOD__ );
1477                 return $output;
1478         }
1479
1480         /**
1481          * Check if there's an overwrite conflict and, if so, if restrictions
1482          * forbid this user from performing the upload.
1483          *
1484          * @param $user User
1485          *
1486          * @return mixed true on success, array on failure
1487          */
1488         private function checkOverwrite( $user ) {
1489                 // First check whether the local file can be overwritten
1490                 $file = $this->getLocalFile();
1491                 if ( $file->exists() ) {
1492                         if ( !self::userCanReUpload( $user, $file ) ) {
1493                                 return array( 'fileexists-forbidden', $file->getName() );
1494                         } else {
1495                                 return true;
1496                         }
1497                 }
1498
1499                 /* Check shared conflicts: if the local file does not exist, but
1500                  * wfFindFile finds a file, it exists in a shared repository.
1501                  */
1502                 $file = wfFindFile( $this->getTitle() );
1503                 if ( $file && !$user->isAllowed( 'reupload-shared' ) ) {
1504                         return array( 'fileexists-shared-forbidden', $file->getName() );
1505                 }
1506
1507                 return true;
1508         }
1509
1510         /**
1511          * Check if a user is the last uploader
1512          *
1513          * @param $user User object
1514          * @param string $img image name
1515          * @return Boolean
1516          */
1517         public static function userCanReUpload( User $user, $img ) {
1518                 if ( $user->isAllowed( 'reupload' ) ) {
1519                         return true; // non-conditional
1520                 }
1521                 if ( !$user->isAllowed( 'reupload-own' ) ) {
1522                         return false;
1523                 }
1524                 if ( is_string( $img ) ) {
1525                         $img = wfLocalFile( $img );
1526                 }
1527                 if ( !( $img instanceof LocalFile ) ) {
1528                         return false;
1529                 }
1530
1531                 return $user->getId() == $img->getUser( 'id' );
1532         }
1533
1534         /**
1535          * Helper function that does various existence checks for a file.
1536          * The following checks are performed:
1537          * - The file exists
1538          * - Article with the same name as the file exists
1539          * - File exists with normalized extension
1540          * - The file looks like a thumbnail and the original exists
1541          *
1542          * @param $file File The File object to check
1543          * @return mixed False if the file does not exists, else an array
1544          */
1545         public static function getExistsWarning( $file ) {
1546                 if ( $file->exists() ) {
1547                         return array( 'warning' => 'exists', 'file' => $file );
1548                 }
1549
1550                 if ( $file->getTitle()->getArticleID() ) {
1551                         return array( 'warning' => 'page-exists', 'file' => $file );
1552                 }
1553
1554                 if ( $file->wasDeleted() && !$file->exists() ) {
1555                         return array( 'warning' => 'was-deleted', 'file' => $file );
1556                 }
1557
1558                 if ( strpos( $file->getName(), '.' ) == false ) {
1559                         $partname = $file->getName();
1560                         $extension = '';
1561                 } else {
1562                         $n = strrpos( $file->getName(), '.' );
1563                         $extension = substr( $file->getName(), $n + 1 );
1564                         $partname = substr( $file->getName(), 0, $n );
1565                 }
1566                 $normalizedExtension = File::normalizeExtension( $extension );
1567
1568                 if ( $normalizedExtension != $extension ) {
1569                         // We're not using the normalized form of the extension.
1570                         // Normal form is lowercase, using most common of alternate
1571                         // extensions (eg 'jpg' rather than 'JPEG').
1572                         //
1573                         // Check for another file using the normalized form...
1574                         $nt_lc = Title::makeTitle( NS_FILE, "{$partname}.{$normalizedExtension}" );
1575                         $file_lc = wfLocalFile( $nt_lc );
1576
1577                         if ( $file_lc->exists() ) {
1578                                 return array(
1579                                         'warning' => 'exists-normalized',
1580                                         'file' => $file,
1581                                         'normalizedFile' => $file_lc
1582                                 );
1583                         }
1584                 }
1585
1586                 // Check for files with the same name but a different extension
1587                 $similarFiles = RepoGroup::singleton()->getLocalRepo()->findFilesByPrefix(
1588                                 "{$partname}.", 1 );
1589                 if ( count( $similarFiles ) ) {
1590                         return array(
1591                                 'warning' => 'exists-normalized',
1592                                 'file' => $file,
1593                                 'normalizedFile' => $similarFiles[0],
1594                         );
1595                 }
1596
1597                 if ( self::isThumbName( $file->getName() ) ) {
1598                         # Check for filenames like 50px- or 180px-, these are mostly thumbnails
1599                         $nt_thb = Title::newFromText( substr( $partname, strpos( $partname, '-' ) + 1 ) . '.' . $extension, NS_FILE );
1600                         $file_thb = wfLocalFile( $nt_thb );
1601                         if ( $file_thb->exists() ) {
1602                                 return array(
1603                                         'warning' => 'thumb',
1604                                         'file' => $file,
1605                                         'thumbFile' => $file_thb
1606                                 );
1607                         } else {
1608                                 // File does not exist, but we just don't like the name
1609                                 return array(
1610                                         'warning' => 'thumb-name',
1611                                         'file' => $file,
1612                                         'thumbFile' => $file_thb
1613                                 );
1614                         }
1615                 }
1616
1617                 foreach ( self::getFilenamePrefixBlacklist() as $prefix ) {
1618                         if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
1619                                 return array(
1620                                         'warning' => 'bad-prefix',
1621                                         'file' => $file,
1622                                         'prefix' => $prefix
1623                                 );
1624                         }
1625                 }
1626
1627                 return false;
1628         }
1629
1630         /**
1631          * Helper function that checks whether the filename looks like a thumbnail
1632          * @param $filename string
1633          * @return bool
1634          */
1635         public static function isThumbName( $filename ) {
1636                 $n = strrpos( $filename, '.' );
1637                 $partname = $n ? substr( $filename, 0, $n ) : $filename;
1638                 return (
1639                                         substr( $partname, 3, 3 ) == 'px-' ||
1640                                         substr( $partname, 2, 3 ) == 'px-'
1641                                 ) &&
1642                                 preg_match( "/[0-9]{2}/", substr( $partname, 0, 2 ) );
1643         }
1644
1645         /**
1646          * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
1647          *
1648          * @return array list of prefixes
1649          */
1650         public static function getFilenamePrefixBlacklist() {
1651                 $blacklist = array();
1652                 $message = wfMessage( 'filename-prefix-blacklist' )->inContentLanguage();
1653                 if ( !$message->isDisabled() ) {
1654                         $lines = explode( "\n", $message->plain() );
1655                         foreach ( $lines as $line ) {
1656                                 // Remove comment lines
1657                                 $comment = substr( trim( $line ), 0, 1 );
1658                                 if ( $comment == '#' || $comment == '' ) {
1659                                         continue;
1660                                 }
1661                                 // Remove additional comments after a prefix
1662                                 $comment = strpos( $line, '#' );
1663                                 if ( $comment > 0 ) {
1664                                         $line = substr( $line, 0, $comment - 1 );
1665                                 }
1666                                 $blacklist[] = trim( $line );
1667                         }
1668                 }
1669                 return $blacklist;
1670         }
1671
1672         /**
1673          * Gets image info about the file just uploaded.
1674          *
1675          * Also has the effect of setting metadata to be an 'indexed tag name' in returned API result if
1676          * 'metadata' was requested. Oddly, we have to pass the "result" object down just so it can do that
1677          * with the appropriate format, presumably.
1678          *
1679          * @param $result ApiResult:
1680          * @return Array: image info
1681          */
1682         public function getImageInfo( $result ) {
1683                 $file = $this->getLocalFile();
1684                 // TODO This cries out for refactoring. We really want to say $file->getAllInfo(); here.
1685                 // Perhaps "info" methods should be moved into files, and the API should just wrap them in queries.
1686                 if ( $file instanceof UploadStashFile ) {
1687                         $imParam = ApiQueryStashImageInfo::getPropertyNames();
1688                         $info = ApiQueryStashImageInfo::getInfo( $file, array_flip( $imParam ), $result );
1689                 } else {
1690                         $imParam = ApiQueryImageInfo::getPropertyNames();
1691                         $info = ApiQueryImageInfo::getInfo( $file, array_flip( $imParam ), $result );
1692                 }
1693                 return $info;
1694         }
1695
1696         /**
1697          * @param $error array
1698          * @return Status
1699          */
1700         public function convertVerifyErrorToStatus( $error ) {
1701                 $code = $error['status'];
1702                 unset( $code['status'] );
1703                 return Status::newFatal( $this->getVerificationErrorCode( $code ), $error );
1704         }
1705
1706         /**
1707          * @param $forType null|string
1708          * @return int
1709          */
1710         public static function getMaxUploadSize( $forType = null ) {
1711                 global $wgMaxUploadSize;
1712
1713                 if ( is_array( $wgMaxUploadSize ) ) {
1714                         if ( !is_null( $forType ) && isset( $wgMaxUploadSize[$forType] ) ) {
1715                                 return $wgMaxUploadSize[$forType];
1716                         } else {
1717                                 return $wgMaxUploadSize['*'];
1718                         }
1719                 } else {
1720                         return intval( $wgMaxUploadSize );
1721                 }
1722         }
1723
1724         /**
1725          * Get the current status of a chunked upload (used for polling).
1726          * The status will be read from the *current* user session.
1727          * @param $statusKey string
1728          * @return Array|bool
1729          */
1730         public static function getSessionStatus( $statusKey ) {
1731                 return isset( $_SESSION[self::SESSION_STATUS_KEY][$statusKey] )
1732                         ? $_SESSION[self::SESSION_STATUS_KEY][$statusKey]
1733                         : false;
1734         }
1735
1736         /**
1737          * Set the current status of a chunked upload (used for polling).
1738          * The status will be stored in the *current* user session.
1739          * @param $statusKey string
1740          * @param $value array|false
1741          * @return void
1742          */
1743         public static function setSessionStatus( $statusKey, $value ) {
1744                 if ( $value === false ) {
1745                         unset( $_SESSION[self::SESSION_STATUS_KEY][$statusKey] );
1746                 } else {
1747                         $_SESSION[self::SESSION_STATUS_KEY][$statusKey] = $value;
1748                 }
1749         }
1750 }