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                 }
 617
 618                 // Check whether the file extension is on the unwanted list
 619                 global $wgCheckFileExtensions, $wgFileExtensions;
 620                 if ( $wgCheckFileExtensions ) {
 621                         $extensions = array_unique( $wgFileExtensions );
 622                         if ( !$this->checkFileExtension( $this->mFinalExtension, $extensions ) ) {
 623                                 $warnings['filetype-unwanted-type'] = array( $this->mFinalExtension,
 624                                         $wgLang->commaList( $extensions ), count( $extensions ) );
 625                         }
 626                 }
 627
 628                 global $wgUploadSizeWarning;
 629                 if ( $wgUploadSizeWarning && ( $this->mFileSize > $wgUploadSizeWarning ) ) {
 630                         $warnings['large-file'] = array( $wgUploadSizeWarning, $this->mFileSize );
 631                 }
 632
 633                 if ( $this->mFileSize == 0 ) {
 634                         $warnings['emptyfile'] = true;
 635                 }
 636
 637                 $exists = self::getExistsWarning( $localFile );
 638                 if ( $exists !== false ) {
 639                         $warnings['exists'] = $exists;
 640                 }
 641
 642                 // Check dupes against existing files
 643                 $hash = $this->getTempFileSha1Base36();
 644                 $dupes = RepoGroup::singleton()->findBySha1( $hash );
 645                 $title = $this->getTitle();
 646                 // Remove all matches against self
 647                 foreach ( $dupes as $key => $dupe ) {
 648                         if ( $title->equals( $dupe->getTitle() ) ) {
 649                                 unset( $dupes[$key] );
 650                         }
 651                 }
 652                 if ( $dupes ) {
 653                         $warnings['duplicate'] = $dupes;
 654                 }
 655
 656                 // Check dupes against archives
 657                 $archivedImage = new ArchivedFile( null, 0, "{$hash}.{$this->mFinalExtension}" );
 658                 if ( $archivedImage->getID() > 0 ) {
 659                         if ( $archivedImage->userCan( File::DELETED_FILE ) ) {
 660                                 $warnings['duplicate-archive'] = $archivedImage->getName();
 661                         } else {
 662                                 $warnings['duplicate-archive'] = '';
 663                         }
 664                 }
 665
 666                 wfProfileOut( __METHOD__ );
 667                 return $warnings;
 668         }
 669
 670         /**
 671          * Really perform the upload. Stores the file in the local repo, watches
 672          * if necessary and runs the UploadComplete hook.
 673          *
 674          * @param $comment
 675          * @param $pageText
 676          * @param $watch
 677          * @param $user User
 678          *
 679          * @return Status indicating the whether the upload succeeded.
 680          */
 681         public function performUpload( $comment, $pageText, $watch, $user ) {
 682                 wfProfileIn( __METHOD__ );
 683
 684                 $status = $this->getLocalFile()->upload(
 685                         $this->mTempPath,
 686                         $comment,
 687                         $pageText,
 688                         File::DELETE_SOURCE,
 689                         $this->mFileProps,
 690                         false,
 691                         $user
 692                 );
 693
 694                 if ( $status->isGood() ) {
 695                         if ( $watch ) {
 696                                 WatchAction::doWatch( $this->getLocalFile()->getTitle(), $user, WatchedItem::IGNORE_USER_RIGHTS );
 697                         }
 698                         wfRunHooks( 'UploadComplete', array( &$this ) );
 699                 }
 700
 701                 wfProfileOut( __METHOD__ );
 702                 return $status;
 703         }
 704
 705         /**
 706          * Returns the title of the file to be uploaded. Sets mTitleError in case
 707          * the name was illegal.
 708          *
 709          * @return Title The title of the file or null in case the name was illegal
 710          */
 711         public function getTitle() {
 712                 if ( $this->mTitle !== false ) {
 713                         return $this->mTitle;
 714                 }
 715                 /* Assume that if a user specified File:Something.jpg, this is an error
 716                  * and that the namespace prefix needs to be stripped of.
 717                  */
 718                 $title = Title::newFromText( $this->mDesiredDestName );
 719                 if ( $title && $title->getNamespace() == NS_FILE ) {
 720                         $this->mFilteredName = $title->getDBkey();
 721                 } else {
 722                         $this->mFilteredName = $this->mDesiredDestName;
 723                 }
 724
 725                 # oi_archive_name is max 255 bytes, which include a timestamp and an
 726                 # exclamation mark, so restrict file name to 240 bytes.
 727                 if ( strlen( $this->mFilteredName ) > 240 ) {
 728                         $this->mTitleError = self::FILENAME_TOO_LONG;
 729                         $this->mTitle = null;
 730                         return $this->mTitle;
 731                 }
 732
 733                 /**
 734                  * Chop off any directories in the given filename. Then
 735                  * filter out illegal characters, and try to make a legible name
 736                  * out of it. We'll strip some silently that Title would die on.
 737                  */
 738                 $this->mFilteredName = wfStripIllegalFilenameChars( $this->mFilteredName );
 739                 /* Normalize to title form before we do any further processing */
 740                 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 741                 if ( is_null( $nt ) ) {
 742                         $this->mTitleError = self::ILLEGAL_FILENAME;
 743                         $this->mTitle = null;
 744                         return $this->mTitle;
 745                 }
 746                 $this->mFilteredName = $nt->getDBkey();
 747
 748                 /**
 749                  * We'll want to blacklist against *any* 'extension', and use
 750                  * only the final one for the whitelist.
 751                  */
 752                 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName );
 753
 754                 if ( count( $ext ) ) {
 755                         $this->mFinalExtension = trim( $ext[count( $ext ) - 1] );
 756                 } else {
 757                         $this->mFinalExtension = '';
 758
 759                         # No extension, try guessing one
 760                         $magic = MimeMagic::singleton();
 761                         $mime = $magic->guessMimeType( $this->mTempPath );
 762                         if ( $mime !== 'unknown/unknown' ) {
 763                                 # Get a space separated list of extensions
 764                                 $extList = $magic->getExtensionsForType( $mime );
 765                                 if ( $extList ) {
 766                                         # Set the extension to the canonical extension
 767                                         $this->mFinalExtension = strtok( $extList, ' ' );
 768
 769                                         # Fix up the other variables
 770                                         $this->mFilteredName .= ".{$this->mFinalExtension}";
 771                                         $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 772                                         $ext = array( $this->mFinalExtension );
 773                                 }
 774                         }
 775                 }
 776
 777                 /* Don't allow users to override the blacklist (check file extension) */
 778                 global $wgCheckFileExtensions, $wgStrictFileExtensions;
 779                 global $wgFileExtensions, $wgFileBlacklist;
 780
 781                 $blackListedExtensions = $this->checkFileExtensionList( $ext, $wgFileBlacklist );
 782
 783                 if ( $this->mFinalExtension == '' ) {
 784                         $this->mTitleError = self::FILETYPE_MISSING;
 785                         $this->mTitle = null;
 786                         return $this->mTitle;
 787                 } elseif ( $blackListedExtensions ||
 788                                 ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
 789                                         !$this->checkFileExtensionList( $ext, $wgFileExtensions ) ) ) {
 790                         $this->mBlackListedExtensions = $blackListedExtensions;
 791                         $this->mTitleError = self::FILETYPE_BADTYPE;
 792                         $this->mTitle = null;
 793                         return $this->mTitle;
 794                 }
 795
 796                 // Windows may be broken with special characters, see bug XXX
 797                 if ( wfIsWindows() && !preg_match( '/^[\x0-\x7f]*$/', $nt->getText() ) ) {
 798                         $this->mTitleError = self::WINDOWS_NONASCII_FILENAME;
 799                         $this->mTitle = null;
 800                         return $this->mTitle;
 801                 }
 802
 803                 # If there was more than one "extension", reassemble the base
 804                 # filename to prevent bogus complaints about length
 805                 if ( count( $ext ) > 1 ) {
 806                         for ( $i = 0; $i < count( $ext ) - 1; $i++ ) {
 807                                 $partname .= '.' . $ext[$i];
 808                         }
 809                 }
 810
 811                 if ( strlen( $partname ) < 1 ) {
 812                         $this->mTitleError = self::MIN_LENGTH_PARTNAME;
 813                         $this->mTitle = null;
 814                         return $this->mTitle;
 815                 }
 816
 817                 $this->mTitle = $nt;
 818                 return $this->mTitle;
 819         }
 820
 821         /**
 822          * Return the local file and initializes if necessary.
 823          *
 824          * @return LocalFile|null
 825          */
 826         public function getLocalFile() {
 827                 if ( is_null( $this->mLocalFile ) ) {
 828                         $nt = $this->getTitle();
 829                         $this->mLocalFile = is_null( $nt ) ? null : wfLocalFile( $nt );
 830                 }
 831                 return $this->mLocalFile;
 832         }
 833
 834         /**
 835          * If the user does not supply all necessary information in the first upload form submission (either by accident or
 836          * by design) then we may want to stash the file temporarily, get more information, and publish the file later.
 837          *
 838          * This method will stash a file in a temporary directory for later processing, and save the necessary descriptive info
 839          * into the database.
 840          * This method returns the file object, which also has a 'fileKey' property which can be passed through a form or
 841          * API request to find this stashed file again.
 842          *
 843          * @param $user User
 844          * @return UploadStashFile stashed file
 845          */
 846         public function stashFile( User $user = null ) {
 847                 // was stashSessionFile
 848                 wfProfileIn( __METHOD__ );
 849
 850                 $stash = RepoGroup::singleton()->getLocalRepo()->getUploadStash( $user );
 851                 $file = $stash->stashFile( $this->mTempPath, $this->getSourceType() );
 852                 $this->mLocalFile = $file;
 853
 854                 wfProfileOut( __METHOD__ );
 855                 return $file;
 856         }
 857
 858         /**
 859          * Stash a file in a temporary directory, returning a key which can be used to find the file again. See stashFile().
 860          *
 861          * @return String: file key
 862          */
 863         public function stashFileGetKey() {
 864                 return $this->stashFile()->getFileKey();
 865         }
 866
 867         /**
 868          * alias for stashFileGetKey, for backwards compatibility
 869          *
 870          * @return String: file key
 871          */
 872         public function stashSession() {
 873                 return $this->stashFileGetKey();
 874         }
 875
 876         /**
 877          * If we've modified the upload file we need to manually remove it
 878          * on exit to clean up.
 879          */
 880         public function cleanupTempFile() {
 881                 if ( $this->mRemoveTempFile && $this->mTempPath && file_exists( $this->mTempPath ) ) {
 882                         wfDebug( __METHOD__ . ": Removing temporary file {$this->mTempPath}\n" );
 883                         unlink( $this->mTempPath );
 884                 }
 885         }
 886
 887         public function getTempPath() {
 888                 return $this->mTempPath;
 889         }
 890
 891         /**
 892          * Split a file into a base name and all dot-delimited 'extensions'
 893          * on the end. Some web server configurations will fall back to
 894          * earlier pseudo-'extensions' to determine type and execute
 895          * scripts, so the blacklist needs to check them all.
 896          *
 897          * @param $filename string
 898          * @return array
 899          */
 900         public static function splitExtensions( $filename ) {
 901                 $bits = explode( '.', $filename );
 902                 $basename = array_shift( $bits );
 903                 return array( $basename, $bits );
 904         }
 905
 906         /**
 907          * Perform case-insensitive match against a list of file extensions.
 908          * Returns true if the extension is in the list.
 909          *
 910          * @param $ext String
 911          * @param $list Array
 912          * @return Boolean
 913          */
 914         public static function checkFileExtension( $ext, $list ) {
 915                 return in_array( strtolower( $ext ), $list );
 916         }
 917
 918         /**
 919          * Perform case-insensitive match against a list of file extensions.
 920          * Returns an array of matching extensions.
 921          *
 922          * @param $ext Array
 923          * @param $list Array
 924          * @return Boolean
 925          */
 926         public static function checkFileExtensionList( $ext, $list ) {
 927                 return array_intersect( array_map( 'strtolower', $ext ), $list );
 928         }
 929
 930         /**
 931          * Checks if the mime type of the uploaded file matches the file extension.
 932          *
 933          * @param string $mime the mime type of the uploaded file
 934          * @param string $extension the filename extension that the file is to be served with
 935          * @return Boolean
 936          */
 937         public static function verifyExtension( $mime, $extension ) {
 938                 $magic = MimeMagic::singleton();
 939
 940                 if ( !$mime || $mime == 'unknown' || $mime == 'unknown/unknown' ) {
 941                         if ( !$magic->isRecognizableExtension( $extension ) ) {
 942                                 wfDebug( __METHOD__ . ": passing file with unknown detected mime type; " .
 943                                         "unrecognized extension '$extension', can't verify\n" );
 944                                 return true;
 945                         } else {
 946                                 wfDebug( __METHOD__ . ": rejecting file with unknown detected mime type; " .
 947                                         "recognized extension '$extension', so probably invalid file\n" );
 948                                 return false;
 949                         }
 950                 }
 951
 952                 $match = $magic->isMatchingExtension( $extension, $mime );
 953
 954                 if ( $match === null ) {
 955                         if ( $magic->getTypesForExtension( $extension ) !== null ) {
 956                                 wfDebug( __METHOD__ . ": No extension known for $mime, but we know a mime for $extension\n" );
 957                                 return false;
 958                         } else {
 959                                 wfDebug( __METHOD__ . ": no file extension known for mime type $mime, passing file\n" );
 960                                 return true;
 961                         }
 962                 } elseif ( $match === true ) {
 963                         wfDebug( __METHOD__ . ": mime type $mime matches extension $extension, passing file\n" );
 964
 965                         #TODO: if it's a bitmap, make sure PHP or ImageMagic resp. can handle it!
 966                         return true;
 967
 968                 } else {
 969                         wfDebug( __METHOD__ . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
 970                         return false;
 971                 }
 972         }
 973
 974         /**
 975          * Heuristic for detecting files that *could* contain JavaScript instructions or
 976          * things that may look like HTML to a browser and are thus
 977          * potentially harmful. The present implementation will produce false
 978          * positives in some situations.
 979          *
 980          * @param string $file pathname to the temporary upload file
 981          * @param string $mime the mime type of the file
 982          * @param string $extension the extension of the file
 983          * @return Boolean: true if the file contains something looking like embedded scripts
 984          */
 985         public static function detectScript( $file, $mime, $extension ) {
 986                 global $wgAllowTitlesInSVG;
 987                 wfProfileIn( __METHOD__ );
 988
 989                 # ugly hack: for text files, always look at the entire file.
 990                 # For binary field, just check the first K.
 991
 992                 if ( strpos( $mime, 'text/' ) === 0 ) {
 993                         $chunk = file_get_contents( $file );
 994                 } else {
 995                         $fp = fopen( $file, 'rb' );
 996                         $chunk = fread( $fp, 1024 );
 997                         fclose( $fp );
 998                 }
 999
1000                 $chunk = strtolower( $chunk );
1001
1002                 if ( !$chunk ) {
1003                         wfProfileOut( __METHOD__ );
1004                         return false;
1005                 }
1006
1007                 # decode from UTF-16 if needed (could be used for obfuscation).
1008                 if ( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
1009                         $enc = 'UTF-16BE';
1010                 } elseif ( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
1011                         $enc = 'UTF-16LE';
1012                 } else {
1013                         $enc = null;
1014                 }
1015
1016                 if ( $enc ) {
1017                         $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
1018                 }
1019
1020                 $chunk = trim( $chunk );
1021
1022                 # @todo FIXME: Convert from UTF-16 if necessary!
1023                 wfDebug( __METHOD__ . ": checking for embedded scripts and HTML stuff\n" );
1024
1025                 # check for HTML doctype
1026                 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
1027                         wfProfileOut( __METHOD__ );
1028                         return true;
1029                 }
1030
1031                 // Some browsers will interpret obscure xml encodings as UTF-8, while
1032                 // PHP/expat will interpret the given encoding in the xml declaration (bug 47304)
1033                 if ( $extension == 'svg' || strpos( $mime, 'image/svg' ) === 0 ) {
1034                         if ( self::checkXMLEncodingMissmatch( $file ) ) {
1035                                 wfProfileOut( __METHOD__ );
1036                                 return true;
1037                         }
1038                 }
1039
1040                 /**
1041                  * Internet Explorer for Windows performs some really stupid file type
1042                  * autodetection which can cause it to interpret valid image files as HTML
1043                  * and potentially execute JavaScript, creating a cross-site scripting
1044                  * attack vectors.
1045                  *
1046                  * Apple's Safari browser also performs some unsafe file type autodetection
1047                  * which can cause legitimate files to be interpreted as HTML if the
1048                  * web server is not correctly configured to send the right content-type
1049                  * (or if you're really uploading plain text and octet streams!)
1050                  *
1051                  * Returns true if IE is likely to mistake the given file for HTML.
1052                  * Also returns true if Safari would mistake the given file for HTML
1053                  * when served with a generic content-type.
1054                  */
1055                 $tags = array(
1056                         '<a href',
1057                         '<body',
1058                         '<head',
1059                         '<html',   #also in safari
1060                         '<img',
1061                         '<pre',
1062                         '<script', #also in safari
1063                         '<table'
1064                 );
1065
1066                 if ( !$wgAllowTitlesInSVG && $extension !== 'svg' && $mime !== 'image/svg' ) {
1067                         $tags[] = '<title';
1068                 }
1069
1070                 foreach ( $tags as $tag ) {
1071                         if ( false !== strpos( $chunk, $tag ) ) {
1072                                 wfDebug( __METHOD__ . ": found something that may make it be mistaken for html: $tag\n" );
1073                                 wfProfileOut( __METHOD__ );
1074                                 return true;
1075                         }
1076                 }
1077
1078                 /*
1079                  * look for JavaScript
1080                  */
1081
1082                 # resolve entity-refs to look at attributes. may be harsh on big files... cache result?
1083                 $chunk = Sanitizer::decodeCharReferences( $chunk );
1084
1085                 # look for script-types
1086                 if ( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
1087                         wfDebug( __METHOD__ . ": found script types\n" );
1088                         wfProfileOut( __METHOD__ );
1089                         return true;
1090                 }
1091
1092                 # look for html-style script-urls
1093                 if ( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1094                         wfDebug( __METHOD__ . ": found html-style script urls\n" );
1095                         wfProfileOut( __METHOD__ );
1096                         return true;
1097                 }
1098
1099                 # look for css-style script-urls
1100                 if ( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1101                         wfDebug( __METHOD__ . ": found css-style script urls\n" );
1102                         wfProfileOut( __METHOD__ );
1103                         return true;
1104                 }
1105
1106                 wfDebug( __METHOD__ . ": no scripts found\n" );
1107                 wfProfileOut( __METHOD__ );
1108                 return false;
1109         }
1110
1111         /**
1112          * Check a whitelist of xml encodings that are known not to be interpreted differently
1113          * by the server's xml parser (expat) and some common browsers.
1114          *
1115          * @param string $file pathname to the temporary upload file
1116          * @return Boolean: true if the file contains an encoding that could be misinterpreted
1117          */
1118         public static function checkXMLEncodingMissmatch( $file ) {
1119                 global $wgSVGMetadataCutoff;
1120                 $contents = file_get_contents( $file, false, null, -1, $wgSVGMetadataCutoff );
1121                 $encodingRegex = '!encoding[ \t\n\r]*=[ \t\n\r]*[\'"](.*?)[\'"]!si';
1122
1123                 if ( preg_match( "!<\?xml\b(.*?)\?>!si", $contents, $matches ) ) {
1124                         if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1125                                 && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1126                         ) {
1127                                 wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1128                                 return true;
1129                         }
1130                 } elseif ( preg_match( "!<\?xml\b!si", $contents ) ) {
1131                         // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1132                         // bytes. There shouldn't be a legitimate reason for this to happen.
1133                         wfDebug( __METHOD__ . ": Unmatched XML declaration start\n" );
1134                         return true;
1135                 } elseif ( substr( $contents, 0, 4 ) == "\x4C\x6F\xA7\x94" ) {
1136                         // EBCDIC encoded XML
1137                         wfDebug( __METHOD__ . ": EBCDIC Encoded XML\n" );
1138                         return true;
1139                 }
1140
1141                 // It's possible the file is encoded with multi-byte encoding, so re-encode attempt to
1142                 // detect the encoding in case is specifies an encoding not whitelisted in self::$safeXmlEncodings
1143                 $attemptEncodings = array( 'UTF-16', 'UTF-16BE', 'UTF-32', 'UTF-32BE' );
1144                 foreach ( $attemptEncodings as $encoding ) {
1145                         wfSuppressWarnings();
1146                         $str = iconv( $encoding, 'UTF-8', $contents );
1147                         wfRestoreWarnings();
1148                         if ( $str != '' && preg_match( "!<\?xml\b(.*?)\?>!si", $str, $matches ) ) {
1149                                 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1150                                         && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1151                                 ) {
1152                                         wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1153                                         return true;
1154                                 }
1155                         } elseif ( $str != '' && preg_match( "!<\?xml\b!si", $str ) ) {
1156                                 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1157                                 // bytes. There shouldn't be a legitimate reason for this to happen.
1158                                 wfDebug( __METHOD__ . ": Unmatched XML declaration start\n" );
1159                                 return true;
1160                         }
1161                 }
1162
1163                 return false;
1164         }
1165
1166         /**
1167          * @param $filename string
1168          * @return mixed false of the file is verified (does not contain scripts), array otherwise.
1169          */
1170         protected function detectScriptInSvg( $filename ) {
1171                 $this->mSVGNSError = false;
1172                 $check = new XmlTypeCheck(
1173                         $filename,
1174                         array( $this, 'checkSvgScriptCallback' ),
1175                         true,
1176                         array( 'processing_instruction_handler' => 'UploadBase::checkSvgPICallback' )
1177                 );
1178                 if ( $check->wellFormed !== true ) {
1179                         // Invalid xml (bug 58553)
1180                         return array( 'uploadinvalidxml' );
1181                 } elseif ( $check->filterMatch ) {
1182                         if ( $this->mSVGNSError ) {
1183                                 return array( 'uploadscriptednamespace', $this->mSVGNSError );
1184                         }
1185                         return array( 'uploadscripted' );
1186                 }
1187                 return false;
1188         }
1189
1190         /**
1191          * Callback to filter SVG Processing Instructions.
1192          * @param $target string processing instruction name
1193          * @param $data string processing instruction attribute and value
1194          * @return bool (true if the filter identified something bad)
1195          */
1196         public static function checkSvgPICallback( $target, $data ) {
1197                 // Don't allow external stylesheets (bug 57550)
1198                 if ( preg_match( '/xml-stylesheet/i', $target) ) {
1199                         return true;
1200                 }
1201                 return false;
1202         }
1203
1204         /**
1205          * @todo Replace this with a whitelist filter!
1206          * @param $element string
1207          * @param $attribs array
1208          * @return bool
1209          */
1210         public function checkSvgScriptCallback( $element, $attribs ) {
1211                 list( $namespace, $strippedElement ) = $this->splitXmlNamespace( $element );
1212
1213                 static $validNamespaces = array(
1214                         '',
1215                         'adobe:ns:meta/',
1216                         'http://creativecommons.org/ns#',
1217                         'http://inkscape.sourceforge.net/dtd/sodipodi-0.dtd',
1218                         'http://ns.adobe.com/adobeillustrator/10.0/',
1219                         'http://ns.adobe.com/adobesvgviewerextensions/3.0/',
1220                         'http://ns.adobe.com/extensibility/1.0/',
1221                         'http://ns.adobe.com/flows/1.0/',
1222                         'http://ns.adobe.com/illustrator/1.0/',
1223                         'http://ns.adobe.com/imagereplacement/1.0/',
1224                         'http://ns.adobe.com/pdf/1.3/',
1225                         'http://ns.adobe.com/photoshop/1.0/',
1226                         'http://ns.adobe.com/saveforweb/1.0/',
1227                         'http://ns.adobe.com/variables/1.0/',
1228                         'http://ns.adobe.com/xap/1.0/',
1229                         'http://ns.adobe.com/xap/1.0/g/',
1230                         'http://ns.adobe.com/xap/1.0/g/img/',
1231                         'http://ns.adobe.com/xap/1.0/mm/',
1232                         'http://ns.adobe.com/xap/1.0/rights/',
1233                         'http://ns.adobe.com/xap/1.0/stype/dimensions#',
1234                         'http://ns.adobe.com/xap/1.0/stype/font#',
1235                         'http://ns.adobe.com/xap/1.0/stype/manifestitem#',
1236                         'http://ns.adobe.com/xap/1.0/stype/resourceevent#',
1237                         'http://ns.adobe.com/xap/1.0/stype/resourceref#',
1238                         'http://ns.adobe.com/xap/1.0/t/pg/',
1239                         'http://purl.org/dc/elements/1.1/',
1240                         'http://purl.org/dc/elements/1.1',
1241                         'http://schemas.microsoft.com/visio/2003/svgextensions/',
1242                         'http://sodipodi.sourceforge.net/dtd/sodipodi-0.dtd',
1243                         'http://web.resource.org/cc/',
1244                         'http://www.freesoftware.fsf.org/bkchem/cdml',
1245                         'http://www.inkscape.org/namespaces/inkscape',
1246                         'http://www.w3.org/1999/02/22-rdf-syntax-ns#',
1247                         'http://www.w3.org/2000/svg',
1248                 );
1249
1250                 if ( !in_array( $namespace, $validNamespaces ) ) {
1251                         wfDebug( __METHOD__ . ": Non-svg namespace '$namespace' in uploaded file.\n" );
1252                         // @TODO return a status object to a closure in XmlTypeCheck, for MW1.21+
1253                         $this->mSVGNSError = $namespace;
1254                         return true;
1255                 }
1256
1257                 /*
1258                  * check for elements that can contain javascript
1259                  */
1260                 if ( $strippedElement == 'script' ) {
1261                         wfDebug( __METHOD__ . ": Found script element '$element' in uploaded file.\n" );
1262                         return true;
1263                 }
1264
1265                 # 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>
1266                 if ( $strippedElement == 'handler' ) {
1267                         wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1268                         return true;
1269                 }
1270
1271                 # SVG reported in Feb '12 that used xml:stylesheet to generate javascript block
1272                 if ( $strippedElement == 'stylesheet' ) {
1273                         wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1274                         return true;
1275                 }
1276
1277                 # Block iframes, in case they pass the namespace check
1278                 if ( $strippedElement == 'iframe' ) {
1279                         wfDebug( __METHOD__ . ": iframe in uploaded file.\n" );
1280                         return true;
1281                 }
1282
1283
1284                 foreach ( $attribs as $attrib => $value ) {
1285                         $stripped = $this->stripXmlNamespace( $attrib );
1286                         $value = strtolower( $value );
1287
1288                         if ( substr( $stripped, 0, 2 ) == 'on' ) {
1289                                 wfDebug( __METHOD__ . ": Found event-handler attribute '$attrib'='$value' in uploaded file.\n" );
1290                                 return true;
1291                         }
1292
1293                         # href with javascript target
1294                         if ( $stripped == 'href' && strpos( strtolower( $value ), 'javascript:' ) !== false ) {
1295                                 wfDebug( __METHOD__ . ": Found script in href attribute '$attrib'='$value' in uploaded file.\n" );
1296                                 return true;
1297                         }
1298
1299                         # href with embedded svg as target
1300                         if ( $stripped == 'href' && preg_match( '!data:[^,]*image/svg[^,]*,!sim', $value ) ) {
1301                                 wfDebug( __METHOD__ . ": Found href to embedded svg \"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1302                                 return true;
1303                         }
1304
1305                         # href with embedded (text/xml) svg as target
1306                         if ( $stripped == 'href' && preg_match( '!data:[^,]*text/xml[^,]*,!sim', $value ) ) {
1307                                 wfDebug( __METHOD__ . ": Found href to embedded svg \"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1308                                 return true;
1309                         }
1310
1311                         # use set/animate to add event-handler attribute to parent
1312                         if ( ( $strippedElement == 'set' || $strippedElement == 'animate' ) && $stripped == 'attributename' && substr( $value, 0, 2 ) == 'on' ) {
1313                                 wfDebug( __METHOD__ . ": Found svg setting event-handler attribute with \"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1314                                 return true;
1315                         }
1316
1317                         # use set to add href attribute to parent element
1318                         if ( $strippedElement == 'set' && $stripped == 'attributename' && strpos( $value, 'href' ) !== false ) {
1319                                 wfDebug( __METHOD__ . ": Found svg setting href attribute '$value' in uploaded file.\n" );
1320                                 return true;
1321                         }
1322
1323                         # use set to add a remote / data / script target to an element
1324                         if ( $strippedElement == 'set' && $stripped == 'to' && preg_match( '!(http|https|data|script):!sim', $value ) ) {
1325                                 wfDebug( __METHOD__ . ": Found svg setting attribute to '$value' in uploaded file.\n" );
1326                                 return true;
1327                         }
1328
1329                         # use handler attribute with remote / data / script
1330                         if ( $stripped == 'handler' && preg_match( '!(http|https|data|script):!sim', $value ) ) {
1331                                 wfDebug( __METHOD__ . ": Found svg setting handler with remote/data/script '$attrib'='$value' in uploaded file.\n" );
1332                                 return true;
1333                         }
1334
1335                         # use CSS styles to bring in remote code
1336                         # catch url("http:..., url('http:..., url(http:..., but not url("#..., url('#..., url(#....
1337                         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 ) ) {
1338                                 foreach ( $matches[1] as $match ) {
1339                                         if ( !preg_match( '!(?:font|clip-path|fill|filter|marker|marker-end|marker-mid|marker-start|mask|stroke)\s*:\s*url\s*\(\s*(#|\'#|"#)!sim', $match ) ) {
1340                                                 wfDebug( __METHOD__ . ": Found svg setting a style with remote url '$attrib'='$value' in uploaded file.\n" );
1341                                                 return true;
1342                                         }
1343                                 }
1344                         }
1345
1346                         # image filters can pull in url, which could be svg that executes scripts
1347                         if ( $strippedElement == 'image' && $stripped == 'filter' && preg_match( '!url\s*\(!sim', $value ) ) {
1348                                 wfDebug( __METHOD__ . ": Found image filter with url: \"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1349                                 return true;
1350                         }
1351
1352                 }
1353
1354                 return false; //No scripts detected
1355         }
1356
1357         /**
1358          * Divide the element name passed by the xml parser to the callback into URI and prifix.
1359          * @param $name string
1360          * @return array containing the namespace URI and prefix
1361          */
1362         private static function splitXmlNamespace( $element ) {
1363                 // 'http://www.w3.org/2000/svg:script' -> array( 'http://www.w3.org/2000/svg', 'script' )
1364                 $parts = explode( ':', strtolower( $element ) );
1365                 $name = array_pop( $parts );
1366                 $ns = implode( ':', $parts );
1367                 return array( $ns, $name );
1368         }
1369
1370         /**
1371          * @param $name string
1372          * @return string
1373          */
1374         private function stripXmlNamespace( $name ) {
1375                 // 'http://www.w3.org/2000/svg:script' -> 'script'
1376                 $parts = explode( ':', strtolower( $name ) );
1377                 return array_pop( $parts );
1378         }
1379
1380         /**
1381          * Generic wrapper function for a virus scanner program.
1382          * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
1383          * $wgAntivirusRequired may be used to deny upload if the scan fails.
1384          *
1385          * @param string $file pathname to the temporary upload file
1386          * @return mixed false if not virus is found, NULL if the scan fails or is disabled,
1387          *         or a string containing feedback from the virus scanner if a virus was found.
1388          *         If textual feedback is missing but a virus was found, this function returns true.
1389          */
1390         public static function detectVirus( $file ) {
1391                 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
1392                 wfProfileIn( __METHOD__ );
1393
1394                 if ( !$wgAntivirus ) {
1395                         wfDebug( __METHOD__ . ": virus scanner disabled\n" );
1396                         wfProfileOut( __METHOD__ );
1397                         return null;
1398                 }
1399
1400                 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
1401                         wfDebug( __METHOD__ . ": unknown virus scanner: $wgAntivirus\n" );
1402                         $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1\n</div>",
1403                                 array( 'virus-badscanner', $wgAntivirus ) );
1404                         wfProfileOut( __METHOD__ );
1405                         return wfMessage( 'virus-unknownscanner' )->text() . " $wgAntivirus";
1406                 }
1407
1408                 # look up scanner configuration
1409                 $command = $wgAntivirusSetup[$wgAntivirus]['command'];
1410                 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
1411                 $msgPattern = isset( $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ) ?
1412                         $wgAntivirusSetup[$wgAntivirus]['messagepattern'] : null;
1413
1414                 if ( strpos( $command, "%f" ) === false ) {
1415                         # simple pattern: append file to scan
1416                         $command .= " " . wfEscapeShellArg( $file );
1417                 } else {
1418                         # complex pattern: replace "%f" with file to scan
1419                         $command = str_replace( "%f", wfEscapeShellArg( $file ), $command );
1420                 }
1421
1422                 wfDebug( __METHOD__ . ": running virus scan: $command \n" );
1423
1424                 # execute virus scanner
1425                 $exitCode = false;
1426
1427                 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
1428                 #      that does not seem to be worth the pain.
1429                 #      Ask me (Duesentrieb) about it if it's ever needed.
1430                 $output = wfShellExecWithStderr( $command, $exitCode );
1431
1432                 # map exit code to AV_xxx constants.
1433                 $mappedCode = $exitCode;
1434                 if ( $exitCodeMap ) {
1435                         if ( isset( $exitCodeMap[$exitCode] ) ) {
1436                                 $mappedCode = $exitCodeMap[$exitCode];
1437                         } elseif ( isset( $exitCodeMap["*"] ) ) {
1438                                 $mappedCode = $exitCodeMap["*"];
1439                         }
1440                 }
1441
1442                 /* NB: AV_NO_VIRUS is 0 but AV_SCAN_FAILED is false,
1443                  * so we need the strict equalities === and thus can't use a switch here
1444                  */
1445                 if ( $mappedCode === AV_SCAN_FAILED ) {
1446                         # scan failed (code was mapped to false by $exitCodeMap)
1447                         wfDebug( __METHOD__ . ": failed to scan $file (code $exitCode).\n" );
1448
1449                         $output = $wgAntivirusRequired ? wfMessage( 'virus-scanfailed', array( $exitCode ) )->text() : null;
1450                 } elseif ( $mappedCode === AV_SCAN_ABORTED ) {
1451                         # scan failed because filetype is unknown (probably imune)
1452                         wfDebug( __METHOD__ . ": unsupported file type $file (code $exitCode).\n" );
1453                         $output = null;
1454                 } elseif ( $mappedCode === AV_NO_VIRUS ) {
1455                         # no virus found
1456                         wfDebug( __METHOD__ . ": file passed virus scan.\n" );
1457                         $output = false;
1458                 } else {
1459                         $output = trim( $output );
1460
1461                         if ( !$output ) {
1462                                 $output = true; #if there's no output, return true
1463                         } elseif ( $msgPattern ) {
1464                                 $groups = array();
1465                                 if ( preg_match( $msgPattern, $output, $groups ) ) {
1466                                         if ( $groups[1] ) {
1467                                                 $output = $groups[1];
1468                                         }
1469                                 }
1470                         }
1471
1472                         wfDebug( __METHOD__ . ": FOUND VIRUS! scanner feedback: $output \n" );
1473                 }
1474
1475                 wfProfileOut( __METHOD__ );
1476                 return $output;
1477         }
1478
1479         /**
1480          * Check if there's an overwrite conflict and, if so, if restrictions
1481          * forbid this user from performing the upload.
1482          *
1483          * @param $user User
1484          *
1485          * @return mixed true on success, array on failure
1486          */
1487         private function checkOverwrite( $user ) {
1488                 // First check whether the local file can be overwritten
1489                 $file = $this->getLocalFile();
1490                 if ( $file->exists() ) {
1491                         if ( !self::userCanReUpload( $user, $file ) ) {
1492                                 return array( 'fileexists-forbidden', $file->getName() );
1493                         } else {
1494                                 return true;
1495                         }
1496                 }
1497
1498                 /* Check shared conflicts: if the local file does not exist, but
1499                  * wfFindFile finds a file, it exists in a shared repository.
1500                  */
1501                 $file = wfFindFile( $this->getTitle() );
1502                 if ( $file && !$user->isAllowed( 'reupload-shared' ) ) {
1503                         return array( 'fileexists-shared-forbidden', $file->getName() );
1504                 }
1505
1506                 return true;
1507         }
1508
1509         /**
1510          * Check if a user is the last uploader
1511          *
1512          * @param $user User object
1513          * @param string $img image name
1514          * @return Boolean
1515          */
1516         public static function userCanReUpload( User $user, $img ) {
1517                 if ( $user->isAllowed( 'reupload' ) ) {
1518                         return true; // non-conditional
1519                 }
1520                 if ( !$user->isAllowed( 'reupload-own' ) ) {
1521                         return false;
1522                 }
1523                 if ( is_string( $img ) ) {
1524                         $img = wfLocalFile( $img );
1525                 }
1526                 if ( !( $img instanceof LocalFile ) ) {
1527                         return false;
1528                 }
1529
1530                 return $user->getId() == $img->getUser( 'id' );
1531         }
1532
1533         /**
1534          * Helper function that does various existence checks for a file.
1535          * The following checks are performed:
1536          * - The file exists
1537          * - Article with the same name as the file exists
1538          * - File exists with normalized extension
1539          * - The file looks like a thumbnail and the original exists
1540          *
1541          * @param $file File The File object to check
1542          * @return mixed False if the file does not exists, else an array
1543          */
1544         public static function getExistsWarning( $file ) {
1545                 if ( $file->exists() ) {
1546                         return array( 'warning' => 'exists', 'file' => $file );
1547                 }
1548
1549                 if ( $file->getTitle()->getArticleID() ) {
1550                         return array( 'warning' => 'page-exists', 'file' => $file );
1551                 }
1552
1553                 if ( $file->wasDeleted() && !$file->exists() ) {
1554                         return array( 'warning' => 'was-deleted', 'file' => $file );
1555                 }
1556
1557                 if ( strpos( $file->getName(), '.' ) == false ) {
1558                         $partname = $file->getName();
1559                         $extension = '';
1560                 } else {
1561                         $n = strrpos( $file->getName(), '.' );
1562                         $extension = substr( $file->getName(), $n + 1 );
1563                         $partname = substr( $file->getName(), 0, $n );
1564                 }
1565                 $normalizedExtension = File::normalizeExtension( $extension );
1566
1567                 if ( $normalizedExtension != $extension ) {
1568                         // We're not using the normalized form of the extension.
1569                         // Normal form is lowercase, using most common of alternate
1570                         // extensions (eg 'jpg' rather than 'JPEG').
1571                         //
1572                         // Check for another file using the normalized form...
1573                         $nt_lc = Title::makeTitle( NS_FILE, "{$partname}.{$normalizedExtension}" );
1574                         $file_lc = wfLocalFile( $nt_lc );
1575
1576                         if ( $file_lc->exists() ) {
1577                                 return array(
1578                                         'warning' => 'exists-normalized',
1579                                         'file' => $file,
1580                                         'normalizedFile' => $file_lc
1581                                 );
1582                         }
1583                 }
1584
1585                 // Check for files with the same name but a different extension
1586                 $similarFiles = RepoGroup::singleton()->getLocalRepo()->findFilesByPrefix(
1587                                 "{$partname}.", 1 );
1588                 if ( count( $similarFiles ) ) {
1589                         return array(
1590                                 'warning' => 'exists-normalized',
1591                                 'file' => $file,
1592                                 'normalizedFile' => $similarFiles[0],
1593                         );
1594                 }
1595
1596                 if ( self::isThumbName( $file->getName() ) ) {
1597                         # Check for filenames like 50px- or 180px-, these are mostly thumbnails
1598                         $nt_thb = Title::newFromText( substr( $partname, strpos( $partname, '-' ) + 1 ) . '.' . $extension, NS_FILE );
1599                         $file_thb = wfLocalFile( $nt_thb );
1600                         if ( $file_thb->exists() ) {
1601                                 return array(
1602                                         'warning' => 'thumb',
1603                                         'file' => $file,
1604                                         'thumbFile' => $file_thb
1605                                 );
1606                         } else {
1607                                 // File does not exist, but we just don't like the name
1608                                 return array(
1609                                         'warning' => 'thumb-name',
1610                                         'file' => $file,
1611                                         'thumbFile' => $file_thb
1612                                 );
1613                         }
1614                 }
1615
1616                 foreach ( self::getFilenamePrefixBlacklist() as $prefix ) {
1617                         if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
1618                                 return array(
1619                                         'warning' => 'bad-prefix',
1620                                         'file' => $file,
1621                                         'prefix' => $prefix
1622                                 );
1623                         }
1624                 }
1625
1626                 return false;
1627         }
1628
1629         /**
1630          * Helper function that checks whether the filename looks like a thumbnail
1631          * @param $filename string
1632          * @return bool
1633          */
1634         public static function isThumbName( $filename ) {
1635                 $n = strrpos( $filename, '.' );
1636                 $partname = $n ? substr( $filename, 0, $n ) : $filename;
1637                 return (
1638                                         substr( $partname, 3, 3 ) == 'px-' ||
1639                                         substr( $partname, 2, 3 ) == 'px-'
1640                                 ) &&
1641                                 preg_match( "/[0-9]{2}/", substr( $partname, 0, 2 ) );
1642         }
1643
1644         /**
1645          * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
1646          *
1647          * @return array list of prefixes
1648          */
1649         public static function getFilenamePrefixBlacklist() {
1650                 $blacklist = array();
1651                 $message = wfMessage( 'filename-prefix-blacklist' )->inContentLanguage();
1652                 if ( !$message->isDisabled() ) {
1653                         $lines = explode( "\n", $message->plain() );
1654                         foreach ( $lines as $line ) {
1655                                 // Remove comment lines
1656                                 $comment = substr( trim( $line ), 0, 1 );
1657                                 if ( $comment == '#' || $comment == '' ) {
1658                                         continue;
1659                                 }
1660                                 // Remove additional comments after a prefix
1661                                 $comment = strpos( $line, '#' );
1662                                 if ( $comment > 0 ) {
1663                                         $line = substr( $line, 0, $comment - 1 );
1664                                 }
1665                                 $blacklist[] = trim( $line );
1666                         }
1667                 }
1668                 return $blacklist;
1669         }
1670
1671         /**
1672          * Gets image info about the file just uploaded.
1673          *
1674          * Also has the effect of setting metadata to be an 'indexed tag name' in returned API result if
1675          * 'metadata' was requested. Oddly, we have to pass the "result" object down just so it can do that
1676          * with the appropriate format, presumably.
1677          *
1678          * @param $result ApiResult:
1679          * @return Array: image info
1680          */
1681         public function getImageInfo( $result ) {
1682                 $file = $this->getLocalFile();
1683                 // TODO This cries out for refactoring. We really want to say $file->getAllInfo(); here.
1684                 // Perhaps "info" methods should be moved into files, and the API should just wrap them in queries.
1685                 if ( $file instanceof UploadStashFile ) {
1686                         $imParam = ApiQueryStashImageInfo::getPropertyNames();
1687                         $info = ApiQueryStashImageInfo::getInfo( $file, array_flip( $imParam ), $result );
1688                 } else {
1689                         $imParam = ApiQueryImageInfo::getPropertyNames();
1690                         $info = ApiQueryImageInfo::getInfo( $file, array_flip( $imParam ), $result );
1691                 }
1692                 return $info;
1693         }
1694
1695         /**
1696          * @param $error array
1697          * @return Status
1698          */
1699         public function convertVerifyErrorToStatus( $error ) {
1700                 $code = $error['status'];
1701                 unset( $code['status'] );
1702                 return Status::newFatal( $this->getVerificationErrorCode( $code ), $error );
1703         }
1704
1705         /**
1706          * @param $forType null|string
1707          * @return int
1708          */
1709         public static function getMaxUploadSize( $forType = null ) {
1710                 global $wgMaxUploadSize;
1711
1712                 if ( is_array( $wgMaxUploadSize ) ) {
1713                         if ( !is_null( $forType ) && isset( $wgMaxUploadSize[$forType] ) ) {
1714                                 return $wgMaxUploadSize[$forType];
1715                         } else {
1716                                 return $wgMaxUploadSize['*'];
1717                         }
1718                 } else {
1719                         return intval( $wgMaxUploadSize );
1720                 }
1721         }
1722
1723         /**
1724          * Get the current status of a chunked upload (used for polling).
1725          * The status will be read from the *current* user session.
1726          * @param $statusKey string
1727          * @return Array|bool
1728          */
1729         public static function getSessionStatus( $statusKey ) {
1730                 return isset( $_SESSION[self::SESSION_STATUS_KEY][$statusKey] )
1731                         ? $_SESSION[self::SESSION_STATUS_KEY][$statusKey]
1732                         : false;
1733         }
1734
1735         /**
1736          * Set the current status of a chunked upload (used for polling).
1737          * The status will be stored in the *current* user session.
1738          * @param $statusKey string
1739          * @param $value array|false
1740          * @return void
1741          */
1742         public static function setSessionStatus( $statusKey, $value ) {
1743                 if ( $value === false ) {
1744                         unset( $_SESSION[self::SESSION_STATUS_KEY][$statusKey] );
1745                 } else {
1746                         $_SESSION[self::SESSION_STATUS_KEY][$statusKey] = $value;
1747                 }
1748         }
1749 }