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;
  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                                 if ( $this->detectScriptInSvg( $this->mTempPath ) ) {
 478                                         wfProfileOut( __METHOD__ );
 479                                         return array( 'uploadscripted' );
 480                                 }
 481                         }
 482                 }
 483
 484                 # Check for Java applets, which if uploaded can bypass cross-site
 485                 # restrictions.
 486                 if ( !$wgAllowJavaUploads ) {
 487                         $this->mJavaDetected = false;
 488                         $zipStatus = ZipDirectoryReader::read( $this->mTempPath,
 489                                 array( $this, 'zipEntryCallback' ) );
 490                         if ( !$zipStatus->isOK() ) {
 491                                 $errors = $zipStatus->getErrorsArray();
 492                                 $error = reset( $errors );
 493                                 if ( $error[0] !== 'zip-wrong-format' ) {
 494                                         wfProfileOut( __METHOD__ );
 495                                         return $error;
 496                                 }
 497                         }
 498                         if ( $this->mJavaDetected ) {
 499                                 wfProfileOut( __METHOD__ );
 500                                 return array( 'uploadjava' );
 501                         }
 502                 }
 503
 504                 # Scan the uploaded file for viruses
 505                 $virus = $this->detectVirus( $this->mTempPath );
 506                 if ( $virus ) {
 507                         wfProfileOut( __METHOD__ );
 508                         return array( 'uploadvirus', $virus );
 509                 }
 510
 511                 wfProfileOut( __METHOD__ );
 512                 return true;
 513         }
 514
 515         /**
 516          * Callback for ZipDirectoryReader to detect Java class files.
 517          */
 518         function zipEntryCallback( $entry ) {
 519                 $names = array( $entry['name'] );
 520
 521                 // If there is a null character, cut off the name at it, because JDK's
 522                 // ZIP_GetEntry() uses strcmp() if the name hashes match. If a file name
 523                 // were constructed which had ".class\0" followed by a string chosen to
 524                 // make the hash collide with the truncated name, that file could be
 525                 // returned in response to a request for the .class file.
 526                 $nullPos = strpos( $entry['name'], "\000" );
 527                 if ( $nullPos !== false ) {
 528                         $names[] = substr( $entry['name'], 0, $nullPos );
 529                 }
 530
 531                 // If there is a trailing slash in the file name, we have to strip it,
 532                 // because that's what ZIP_GetEntry() does.
 533                 if ( preg_grep( '!\.class/?$!', $names ) ) {
 534                         $this->mJavaDetected = true;
 535                 }
 536         }
 537
 538         /**
 539          * Alias for verifyTitlePermissions. The function was originally 'verifyPermissions'
 540          * but that suggests it's checking the user, when it's really checking the title + user combination.
 541          * @param $user User object to verify the permissions against
 542          * @return mixed An array as returned by getUserPermissionsErrors or true
 543          *               in case the user has proper permissions.
 544          */
 545         public function verifyPermissions( $user ) {
 546                 return $this->verifyTitlePermissions( $user );
 547         }
 548
 549         /**
 550          * Check whether the user can edit, upload and create the image. This
 551          * checks only against the current title; if it returns errors, it may
 552          * very well be that another title will not give errors. Therefore
 553          * isAllowed() should be called as well for generic is-user-blocked or
 554          * can-user-upload checking.
 555          *
 556          * @param $user User object to verify the permissions against
 557          * @return mixed An array as returned by getUserPermissionsErrors or true
 558          *               in case the user has proper permissions.
 559          */
 560         public function verifyTitlePermissions( $user ) {
 561                 /**
 562                  * If the image is protected, non-sysop users won't be able
 563                  * to modify it by uploading a new revision.
 564                  */
 565                 $nt = $this->getTitle();
 566                 if ( is_null( $nt ) ) {
 567                         return true;
 568                 }
 569                 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
 570                 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
 571                 if ( !$nt->exists() ) {
 572                         $permErrorsCreate = $nt->getUserPermissionsErrors( 'create', $user );
 573                 } else {
 574                         $permErrorsCreate = array();
 575                 }
 576                 if ( $permErrors || $permErrorsUpload || $permErrorsCreate ) {
 577                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
 578                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
 579                         return $permErrors;
 580                 }
 581
 582                 $overwriteError = $this->checkOverwrite( $user );
 583                 if ( $overwriteError !== true ) {
 584                         return array( $overwriteError );
 585                 }
 586
 587                 return true;
 588         }
 589
 590         /**
 591          * Check for non fatal problems with the file.
 592          *
 593          * This should not assume that mTempPath is set.
 594          *
 595          * @return Array of warnings
 596          */
 597         public function checkWarnings() {
 598                 global $wgLang;
 599                 wfProfileIn( __METHOD__ );
 600
 601                 $warnings = array();
 602
 603                 $localFile = $this->getLocalFile();
 604                 $filename = $localFile->getName();
 605
 606                 /**
 607                  * Check whether the resulting filename is different from the desired one,
 608                  * but ignore things like ucfirst() and spaces/underscore things
 609                  */
 610                 $comparableName = str_replace( ' ', '_', $this->mDesiredDestName );
 611                 $comparableName = Title::capitalize( $comparableName, NS_FILE );
 612
 613                 if ( $this->mDesiredDestName != $filename && $comparableName != $filename ) {
 614                         $warnings['badfilename'] = $filename;
 615                 }
 616
 617                 // Check whether the file extension is on the unwanted list
 618                 global $wgCheckFileExtensions, $wgFileExtensions;
 619                 if ( $wgCheckFileExtensions ) {
 620                         $extensions = array_unique( $wgFileExtensions );
 621                         if ( !$this->checkFileExtension( $this->mFinalExtension, $extensions ) ) {
 622                                 $warnings['filetype-unwanted-type'] = array( $this->mFinalExtension,
 623                                         $wgLang->commaList( $extensions ), count( $extensions ) );
 624                         }
 625                 }
 626
 627                 global $wgUploadSizeWarning;
 628                 if ( $wgUploadSizeWarning && ( $this->mFileSize > $wgUploadSizeWarning ) ) {
 629                         $warnings['large-file'] = array( $wgUploadSizeWarning, $this->mFileSize );
 630                 }
 631
 632                 if ( $this->mFileSize == 0 ) {
 633                         $warnings['emptyfile'] = true;
 634                 }
 635
 636                 $exists = self::getExistsWarning( $localFile );
 637                 if ( $exists !== false ) {
 638                         $warnings['exists'] = $exists;
 639                 }
 640
 641                 // Check dupes against existing files
 642                 $hash = $this->getTempFileSha1Base36();
 643                 $dupes = RepoGroup::singleton()->findBySha1( $hash );
 644                 $title = $this->getTitle();
 645                 // Remove all matches against self
 646                 foreach ( $dupes as $key => $dupe ) {
 647                         if ( $title->equals( $dupe->getTitle() ) ) {
 648                                 unset( $dupes[$key] );
 649                         }
 650                 }
 651                 if ( $dupes ) {
 652                         $warnings['duplicate'] = $dupes;
 653                 }
 654
 655                 // Check dupes against archives
 656                 $archivedImage = new ArchivedFile( null, 0, "{$hash}.{$this->mFinalExtension}" );
 657                 if ( $archivedImage->getID() > 0 ) {
 658                         if ( $archivedImage->userCan( File::DELETED_FILE ) ) {
 659                                 $warnings['duplicate-archive'] = $archivedImage->getName();
 660                         } else {
 661                                 $warnings['duplicate-archive'] = '';
 662                         }
 663                 }
 664
 665                 wfProfileOut( __METHOD__ );
 666                 return $warnings;
 667         }
 668
 669         /**
 670          * Really perform the upload. Stores the file in the local repo, watches
 671          * if necessary and runs the UploadComplete hook.
 672          *
 673          * @param $comment
 674          * @param $pageText
 675          * @param $watch
 676          * @param $user User
 677          *
 678          * @return Status indicating the whether the upload succeeded.
 679          */
 680         public function performUpload( $comment, $pageText, $watch, $user ) {
 681                 wfProfileIn( __METHOD__ );
 682
 683                 $status = $this->getLocalFile()->upload(
 684                         $this->mTempPath,
 685                         $comment,
 686                         $pageText,
 687                         File::DELETE_SOURCE,
 688                         $this->mFileProps,
 689                         false,
 690                         $user
 691                 );
 692
 693                 if ( $status->isGood() ) {
 694                         if ( $watch ) {
 695                                 WatchAction::doWatch( $this->getLocalFile()->getTitle(), $user, WatchedItem::IGNORE_USER_RIGHTS );
 696                         }
 697                         wfRunHooks( 'UploadComplete', array( &$this ) );
 698                 }
 699
 700                 wfProfileOut( __METHOD__ );
 701                 return $status;
 702         }
 703
 704         /**
 705          * Returns the title of the file to be uploaded. Sets mTitleError in case
 706          * the name was illegal.
 707          *
 708          * @return Title The title of the file or null in case the name was illegal
 709          */
 710         public function getTitle() {
 711                 if ( $this->mTitle !== false ) {
 712                         return $this->mTitle;
 713                 }
 714                 /* Assume that if a user specified File:Something.jpg, this is an error
 715                  * and that the namespace prefix needs to be stripped of.
 716                  */
 717                 $title = Title::newFromText( $this->mDesiredDestName );
 718                 if ( $title && $title->getNamespace() == NS_FILE ) {
 719                         $this->mFilteredName = $title->getDBkey();
 720                 } else {
 721                         $this->mFilteredName = $this->mDesiredDestName;
 722                 }
 723
 724                 # oi_archive_name is max 255 bytes, which include a timestamp and an
 725                 # exclamation mark, so restrict file name to 240 bytes.
 726                 if ( strlen( $this->mFilteredName ) > 240 ) {
 727                         $this->mTitleError = self::FILENAME_TOO_LONG;
 728                         $this->mTitle = null;
 729                         return $this->mTitle;
 730                 }
 731
 732                 /**
 733                  * Chop off any directories in the given filename. Then
 734                  * filter out illegal characters, and try to make a legible name
 735                  * out of it. We'll strip some silently that Title would die on.
 736                  */
 737                 $this->mFilteredName = wfStripIllegalFilenameChars( $this->mFilteredName );
 738                 /* Normalize to title form before we do any further processing */
 739                 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 740                 if ( is_null( $nt ) ) {
 741                         $this->mTitleError = self::ILLEGAL_FILENAME;
 742                         $this->mTitle = null;
 743                         return $this->mTitle;
 744                 }
 745                 $this->mFilteredName = $nt->getDBkey();
 746
 747                 /**
 748                  * We'll want to blacklist against *any* 'extension', and use
 749                  * only the final one for the whitelist.
 750                  */
 751                 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName );
 752
 753                 if ( count( $ext ) ) {
 754                         $this->mFinalExtension = trim( $ext[count( $ext ) - 1] );
 755                 } else {
 756                         $this->mFinalExtension = '';
 757
 758                         # No extension, try guessing one
 759                         $magic = MimeMagic::singleton();
 760                         $mime = $magic->guessMimeType( $this->mTempPath );
 761                         if ( $mime !== 'unknown/unknown' ) {
 762                                 # Get a space separated list of extensions
 763                                 $extList = $magic->getExtensionsForType( $mime );
 764                                 if ( $extList ) {
 765                                         # Set the extension to the canonical extension
 766                                         $this->mFinalExtension = strtok( $extList, ' ' );
 767
 768                                         # Fix up the other variables
 769                                         $this->mFilteredName .= ".{$this->mFinalExtension}";
 770                                         $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 771                                         $ext = array( $this->mFinalExtension );
 772                                 }
 773                         }
 774                 }
 775
 776                 /* Don't allow users to override the blacklist (check file extension) */
 777                 global $wgCheckFileExtensions, $wgStrictFileExtensions;
 778                 global $wgFileExtensions, $wgFileBlacklist;
 779
 780                 $blackListedExtensions = $this->checkFileExtensionList( $ext, $wgFileBlacklist );
 781
 782                 if ( $this->mFinalExtension == '' ) {
 783                         $this->mTitleError = self::FILETYPE_MISSING;
 784                         $this->mTitle = null;
 785                         return $this->mTitle;
 786                 } elseif ( $blackListedExtensions ||
 787                                 ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
 788                                         !$this->checkFileExtensionList( $ext, $wgFileExtensions ) ) ) {
 789                         $this->mBlackListedExtensions = $blackListedExtensions;
 790                         $this->mTitleError = self::FILETYPE_BADTYPE;
 791                         $this->mTitle = null;
 792                         return $this->mTitle;
 793                 }
 794
 795                 // Windows may be broken with special characters, see bug XXX
 796                 if ( wfIsWindows() && !preg_match( '/^[\x0-\x7f]*$/', $nt->getText() ) ) {
 797                         $this->mTitleError = self::WINDOWS_NONASCII_FILENAME;
 798                         $this->mTitle = null;
 799                         return $this->mTitle;
 800                 }
 801
 802                 # If there was more than one "extension", reassemble the base
 803                 # filename to prevent bogus complaints about length
 804                 if ( count( $ext ) > 1 ) {
 805                         for ( $i = 0; $i < count( $ext ) - 1; $i++ ) {
 806                                 $partname .= '.' . $ext[$i];
 807                         }
 808                 }
 809
 810                 if ( strlen( $partname ) < 1 ) {
 811                         $this->mTitleError = self::MIN_LENGTH_PARTNAME;
 812                         $this->mTitle = null;
 813                         return $this->mTitle;
 814                 }
 815
 816                 $this->mTitle = $nt;
 817                 return $this->mTitle;
 818         }
 819
 820         /**
 821          * Return the local file and initializes if necessary.
 822          *
 823          * @return LocalFile|null
 824          */
 825         public function getLocalFile() {
 826                 if ( is_null( $this->mLocalFile ) ) {
 827                         $nt = $this->getTitle();
 828                         $this->mLocalFile = is_null( $nt ) ? null : wfLocalFile( $nt );
 829                 }
 830                 return $this->mLocalFile;
 831         }
 832
 833         /**
 834          * If the user does not supply all necessary information in the first upload form submission (either by accident or
 835          * by design) then we may want to stash the file temporarily, get more information, and publish the file later.
 836          *
 837          * This method will stash a file in a temporary directory for later processing, and save the necessary descriptive info
 838          * into the database.
 839          * This method returns the file object, which also has a 'fileKey' property which can be passed through a form or
 840          * API request to find this stashed file again.
 841          *
 842          * @param $user User
 843          * @return UploadStashFile stashed file
 844          */
 845         public function stashFile( User $user = null ) {
 846                 // was stashSessionFile
 847                 wfProfileIn( __METHOD__ );
 848
 849                 $stash = RepoGroup::singleton()->getLocalRepo()->getUploadStash( $user );
 850                 $file = $stash->stashFile( $this->mTempPath, $this->getSourceType() );
 851                 $this->mLocalFile = $file;
 852
 853                 wfProfileOut( __METHOD__ );
 854                 return $file;
 855         }
 856
 857         /**
 858          * Stash a file in a temporary directory, returning a key which can be used to find the file again. See stashFile().
 859          *
 860          * @return String: file key
 861          */
 862         public function stashFileGetKey() {
 863                 return $this->stashFile()->getFileKey();
 864         }
 865
 866         /**
 867          * alias for stashFileGetKey, for backwards compatibility
 868          *
 869          * @return String: file key
 870          */
 871         public function stashSession() {
 872                 return $this->stashFileGetKey();
 873         }
 874
 875         /**
 876          * If we've modified the upload file we need to manually remove it
 877          * on exit to clean up.
 878          */
 879         public function cleanupTempFile() {
 880                 if ( $this->mRemoveTempFile && $this->mTempPath && file_exists( $this->mTempPath ) ) {
 881                         wfDebug( __METHOD__ . ": Removing temporary file {$this->mTempPath}\n" );
 882                         unlink( $this->mTempPath );
 883                 }
 884         }
 885
 886         public function getTempPath() {
 887                 return $this->mTempPath;
 888         }
 889
 890         /**
 891          * Split a file into a base name and all dot-delimited 'extensions'
 892          * on the end. Some web server configurations will fall back to
 893          * earlier pseudo-'extensions' to determine type and execute
 894          * scripts, so the blacklist needs to check them all.
 895          *
 896          * @param $filename string
 897          * @return array
 898          */
 899         public static function splitExtensions( $filename ) {
 900                 $bits = explode( '.', $filename );
 901                 $basename = array_shift( $bits );
 902                 return array( $basename, $bits );
 903         }
 904
 905         /**
 906          * Perform case-insensitive match against a list of file extensions.
 907          * Returns true if the extension is in the list.
 908          *
 909          * @param $ext String
 910          * @param $list Array
 911          * @return Boolean
 912          */
 913         public static function checkFileExtension( $ext, $list ) {
 914                 return in_array( strtolower( $ext ), $list );
 915         }
 916
 917         /**
 918          * Perform case-insensitive match against a list of file extensions.
 919          * Returns an array of matching extensions.
 920          *
 921          * @param $ext Array
 922          * @param $list Array
 923          * @return Boolean
 924          */
 925         public static function checkFileExtensionList( $ext, $list ) {
 926                 return array_intersect( array_map( 'strtolower', $ext ), $list );
 927         }
 928
 929         /**
 930          * Checks if the mime type of the uploaded file matches the file extension.
 931          *
 932          * @param string $mime the mime type of the uploaded file
 933          * @param string $extension the filename extension that the file is to be served with
 934          * @return Boolean
 935          */
 936         public static function verifyExtension( $mime, $extension ) {
 937                 $magic = MimeMagic::singleton();
 938
 939                 if ( !$mime || $mime == 'unknown' || $mime == 'unknown/unknown' ) {
 940                         if ( !$magic->isRecognizableExtension( $extension ) ) {
 941                                 wfDebug( __METHOD__ . ": passing file with unknown detected mime type; " .
 942                                         "unrecognized extension '$extension', can't verify\n" );
 943                                 return true;
 944                         } else {
 945                                 wfDebug( __METHOD__ . ": rejecting file with unknown detected mime type; " .
 946                                         "recognized extension '$extension', so probably invalid file\n" );
 947                                 return false;
 948                         }
 949                 }
 950
 951                 $match = $magic->isMatchingExtension( $extension, $mime );
 952
 953                 if ( $match === null ) {
 954                         if ( $magic->getTypesForExtension( $extension ) !== null ) {
 955                                 wfDebug( __METHOD__ . ": No extension known for $mime, but we know a mime for $extension\n" );
 956                                 return false;
 957                         } else {
 958                                 wfDebug( __METHOD__ . ": no file extension known for mime type $mime, passing file\n" );
 959                                 return true;
 960                         }
 961                 } elseif ( $match === true ) {
 962                         wfDebug( __METHOD__ . ": mime type $mime matches extension $extension, passing file\n" );
 963
 964                         #TODO: if it's a bitmap, make sure PHP or ImageMagic resp. can handle it!
 965                         return true;
 966
 967                 } else {
 968                         wfDebug( __METHOD__ . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
 969                         return false;
 970                 }
 971         }
 972
 973         /**
 974          * Heuristic for detecting files that *could* contain JavaScript instructions or
 975          * things that may look like HTML to a browser and are thus
 976          * potentially harmful. The present implementation will produce false
 977          * positives in some situations.
 978          *
 979          * @param string $file pathname to the temporary upload file
 980          * @param string $mime the mime type of the file
 981          * @param string $extension the extension of the file
 982          * @return Boolean: true if the file contains something looking like embedded scripts
 983          */
 984         public static function detectScript( $file, $mime, $extension ) {
 985                 global $wgAllowTitlesInSVG;
 986                 wfProfileIn( __METHOD__ );
 987
 988                 # ugly hack: for text files, always look at the entire file.
 989                 # For binary field, just check the first K.
 990
 991                 if ( strpos( $mime, 'text/' ) === 0 ) {
 992                         $chunk = file_get_contents( $file );
 993                 } else {
 994                         $fp = fopen( $file, 'rb' );
 995                         $chunk = fread( $fp, 1024 );
 996                         fclose( $fp );
 997                 }
 998
 999                 $chunk = strtolower( $chunk );
1000
1001                 if ( !$chunk ) {
1002                         wfProfileOut( __METHOD__ );
1003                         return false;
1004                 }
1005
1006                 # decode from UTF-16 if needed (could be used for obfuscation).
1007                 if ( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
1008                         $enc = 'UTF-16BE';
1009                 } elseif ( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
1010                         $enc = 'UTF-16LE';
1011                 } else {
1012                         $enc = null;
1013                 }
1014
1015                 if ( $enc ) {
1016                         $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
1017                 }
1018
1019                 $chunk = trim( $chunk );
1020
1021                 # @todo FIXME: Convert from UTF-16 if necessary!
1022                 wfDebug( __METHOD__ . ": checking for embedded scripts and HTML stuff\n" );
1023
1024                 # check for HTML doctype
1025                 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
1026                         wfProfileOut( __METHOD__ );
1027                         return true;
1028                 }
1029
1030                 // Some browsers will interpret obscure xml encodings as UTF-8, while
1031                 // PHP/expat will interpret the given encoding in the xml declaration (bug 47304)
1032                 if ( $extension == 'svg' || strpos( $mime, 'image/svg' ) === 0 ) {
1033                         if ( self::checkXMLEncodingMissmatch( $file ) ) {
1034                                 wfProfileOut( __METHOD__ );
1035                                 return true;
1036                         }
1037                 }
1038
1039                 /**
1040                  * Internet Explorer for Windows performs some really stupid file type
1041                  * autodetection which can cause it to interpret valid image files as HTML
1042                  * and potentially execute JavaScript, creating a cross-site scripting
1043                  * attack vectors.
1044                  *
1045                  * Apple's Safari browser also performs some unsafe file type autodetection
1046                  * which can cause legitimate files to be interpreted as HTML if the
1047                  * web server is not correctly configured to send the right content-type
1048                  * (or if you're really uploading plain text and octet streams!)
1049                  *
1050                  * Returns true if IE is likely to mistake the given file for HTML.
1051                  * Also returns true if Safari would mistake the given file for HTML
1052                  * when served with a generic content-type.
1053                  */
1054                 $tags = array(
1055                         '<a href',
1056                         '<body',
1057                         '<head',
1058                         '<html',   #also in safari
1059                         '<img',
1060                         '<pre',
1061                         '<script', #also in safari
1062                         '<table'
1063                 );
1064
1065                 if ( !$wgAllowTitlesInSVG && $extension !== 'svg' && $mime !== 'image/svg' ) {
1066                         $tags[] = '<title';
1067                 }
1068
1069                 foreach ( $tags as $tag ) {
1070                         if ( false !== strpos( $chunk, $tag ) ) {
1071                                 wfDebug( __METHOD__ . ": found something that may make it be mistaken for html: $tag\n" );
1072                                 wfProfileOut( __METHOD__ );
1073                                 return true;
1074                         }
1075                 }
1076
1077                 /*
1078                  * look for JavaScript
1079                  */
1080
1081                 # resolve entity-refs to look at attributes. may be harsh on big files... cache result?
1082                 $chunk = Sanitizer::decodeCharReferences( $chunk );
1083
1084                 # look for script-types
1085                 if ( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
1086                         wfDebug( __METHOD__ . ": found script types\n" );
1087                         wfProfileOut( __METHOD__ );
1088                         return true;
1089                 }
1090
1091                 # look for html-style script-urls
1092                 if ( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1093                         wfDebug( __METHOD__ . ": found html-style script urls\n" );
1094                         wfProfileOut( __METHOD__ );
1095                         return true;
1096                 }
1097
1098                 # look for css-style script-urls
1099                 if ( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1100                         wfDebug( __METHOD__ . ": found css-style script urls\n" );
1101                         wfProfileOut( __METHOD__ );
1102                         return true;
1103                 }
1104
1105                 wfDebug( __METHOD__ . ": no scripts found\n" );
1106                 wfProfileOut( __METHOD__ );
1107                 return false;
1108         }
1109
1110         /**
1111          * Check a whitelist of xml encodings that are known not to be interpreted differently
1112          * by the server's xml parser (expat) and some common browsers.
1113          *
1114          * @param string $file pathname to the temporary upload file
1115          * @return Boolean: true if the file contains an encoding that could be misinterpreted
1116          */
1117         public static function checkXMLEncodingMissmatch( $file ) {
1118                 global $wgSVGMetadataCutoff;
1119                 $contents = file_get_contents( $file, false, null, -1, $wgSVGMetadataCutoff );
1120                 $encodingRegex = '!encoding[ \t\n\r]*=[ \t\n\r]*[\'"](.*?)[\'"]!si';
1121
1122                 if ( preg_match( "!<\?xml\b(.*?)\?>!si", $contents, $matches ) ) {
1123                         if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1124                                 && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1125                         ) {
1126                                 wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1127                                 return true;
1128                         }
1129                 } elseif ( preg_match( "!<\?xml\b!si", $contents ) ) {
1130                         // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1131                         // bytes. There shouldn't be a legitimate reason for this to happen.
1132                         wfDebug( __METHOD__ . ": Unmatched XML declaration start\n" );
1133                         return true;
1134                 } elseif ( substr( $contents, 0, 4 ) == "\x4C\x6F\xA7\x94" ) {
1135                         // EBCDIC encoded XML
1136                         wfDebug( __METHOD__ . ": EBCDIC Encoded XML\n" );
1137                         return true;
1138                 }
1139
1140                 // It's possible the file is encoded with multi-byte encoding, so re-encode attempt to
1141                 // detect the encoding in case is specifies an encoding not whitelisted in self::$safeXmlEncodings
1142                 $attemptEncodings = array( 'UTF-16', 'UTF-16BE', 'UTF-32', 'UTF-32BE' );
1143                 foreach ( $attemptEncodings as $encoding ) {
1144                         wfSuppressWarnings();
1145                         $str = iconv( $encoding, 'UTF-8', $contents );
1146                         wfRestoreWarnings();
1147                         if ( $str != '' && preg_match( "!<\?xml\b(.*?)\?>!si", $str, $matches ) ) {
1148                                 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1149                                         && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1150                                 ) {
1151                                         wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1152                                         return true;
1153                                 }
1154                         } elseif ( $str != '' && preg_match( "!<\?xml\b!si", $str ) ) {
1155                                 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1156                                 // bytes. There shouldn't be a legitimate reason for this to happen.
1157                                 wfDebug( __METHOD__ . ": Unmatched XML declaration start\n" );
1158                                 return true;
1159                         }
1160                 }
1161
1162                 return false;
1163         }
1164
1165         /**
1166          * @param $filename string
1167          * @return bool
1168          */
1169         protected function detectScriptInSvg( $filename ) {
1170                 $check = new XmlTypeCheck( $filename, array( $this, 'checkSvgScriptCallback' ) );
1171                 return $check->filterMatch;
1172         }
1173
1174         /**
1175          * @todo Replace this with a whitelist filter!
1176          * @param $element string
1177          * @param $attribs array
1178          * @return bool
1179          */
1180         public function checkSvgScriptCallback( $element, $attribs ) {
1181                 $strippedElement = $this->stripXmlNamespace( $element );
1182
1183                 /*
1184                  * check for elements that can contain javascript
1185                  */
1186                 if ( $strippedElement == 'script' ) {
1187                         wfDebug( __METHOD__ . ": Found script element '$element' in uploaded file.\n" );
1188                         return true;
1189                 }
1190
1191                 # 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>
1192                 if ( $strippedElement == 'handler' ) {
1193                         wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1194                         return true;
1195                 }
1196
1197                 # SVG reported in Feb '12 that used xml:stylesheet to generate javascript block
1198                 if ( $strippedElement == 'stylesheet' ) {
1199                         wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1200                         return true;
1201                 }
1202
1203                 foreach ( $attribs as $attrib => $value ) {
1204                         $stripped = $this->stripXmlNamespace( $attrib );
1205                         $value = strtolower( $value );
1206
1207                         if ( substr( $stripped, 0, 2 ) == 'on' ) {
1208                                 wfDebug( __METHOD__ . ": Found event-handler attribute '$attrib'='$value' in uploaded file.\n" );
1209                                 return true;
1210                         }
1211
1212                         # href with javascript target
1213                         if ( $stripped == 'href' && strpos( strtolower( $value ), 'javascript:' ) !== false ) {
1214                                 wfDebug( __METHOD__ . ": Found script in href attribute '$attrib'='$value' in uploaded file.\n" );
1215                                 return true;
1216                         }
1217
1218                         # href with embedded svg as target
1219                         if ( $stripped == 'href' && preg_match( '!data:[^,]*image/svg[^,]*,!sim', $value ) ) {
1220                                 wfDebug( __METHOD__ . ": Found href to embedded svg \"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1221                                 return true;
1222                         }
1223
1224                         # href with embedded (text/xml) svg as target
1225                         if ( $stripped == 'href' && preg_match( '!data:[^,]*text/xml[^,]*,!sim', $value ) ) {
1226                                 wfDebug( __METHOD__ . ": Found href to embedded svg \"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1227                                 return true;
1228                         }
1229
1230                         # use set/animate to add event-handler attribute to parent
1231                         if ( ( $strippedElement == 'set' || $strippedElement == 'animate' ) && $stripped == 'attributename' && substr( $value, 0, 2 ) == 'on' ) {
1232                                 wfDebug( __METHOD__ . ": Found svg setting event-handler attribute with \"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1233                                 return true;
1234                         }
1235
1236                         # use set to add href attribute to parent element
1237                         if ( $strippedElement == 'set' && $stripped == 'attributename' && strpos( $value, 'href' ) !== false ) {
1238                                 wfDebug( __METHOD__ . ": Found svg setting href attribute '$value' in uploaded file.\n" );
1239                                 return true;
1240                         }
1241
1242                         # use set to add a remote / data / script target to an element
1243                         if ( $strippedElement == 'set' && $stripped == 'to' && preg_match( '!(http|https|data|script):!sim', $value ) ) {
1244                                 wfDebug( __METHOD__ . ": Found svg setting attribute to '$value' in uploaded file.\n" );
1245                                 return true;
1246                         }
1247
1248                         # use handler attribute with remote / data / script
1249                         if ( $stripped == 'handler' && preg_match( '!(http|https|data|script):!sim', $value ) ) {
1250                                 wfDebug( __METHOD__ . ": Found svg setting handler with remote/data/script '$attrib'='$value' in uploaded file.\n" );
1251                                 return true;
1252                         }
1253
1254                         # use CSS styles to bring in remote code
1255                         # catch url("http:..., url('http:..., url(http:..., but not url("#..., url('#..., url(#....
1256                         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 ) ) {
1257                                 foreach ( $matches[1] as $match ) {
1258                                         if ( !preg_match( '!(?:font|clip-path|fill|filter|marker|marker-end|marker-mid|marker-start|mask|stroke)\s*:\s*url\s*\(\s*(#|\'#|"#)!sim', $match ) ) {
1259                                                 wfDebug( __METHOD__ . ": Found svg setting a style with remote url '$attrib'='$value' in uploaded file.\n" );
1260                                                 return true;
1261                                         }
1262                                 }
1263                         }
1264
1265                         # image filters can pull in url, which could be svg that executes scripts
1266                         if ( $strippedElement == 'image' && $stripped == 'filter' && preg_match( '!url\s*\(!sim', $value ) ) {
1267                                 wfDebug( __METHOD__ . ": Found image filter with url: \"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1268                                 return true;
1269                         }
1270
1271                 }
1272
1273                 return false; //No scripts detected
1274         }
1275
1276         /**
1277          * @param $name string
1278          * @return string
1279          */
1280         private function stripXmlNamespace( $name ) {
1281                 // 'http://www.w3.org/2000/svg:script' -> 'script'
1282                 $parts = explode( ':', strtolower( $name ) );
1283                 return array_pop( $parts );
1284         }
1285
1286         /**
1287          * Generic wrapper function for a virus scanner program.
1288          * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
1289          * $wgAntivirusRequired may be used to deny upload if the scan fails.
1290          *
1291          * @param string $file pathname to the temporary upload file
1292          * @return mixed false if not virus is found, NULL if the scan fails or is disabled,
1293          *         or a string containing feedback from the virus scanner if a virus was found.
1294          *         If textual feedback is missing but a virus was found, this function returns true.
1295          */
1296         public static function detectVirus( $file ) {
1297                 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
1298                 wfProfileIn( __METHOD__ );
1299
1300                 if ( !$wgAntivirus ) {
1301                         wfDebug( __METHOD__ . ": virus scanner disabled\n" );
1302                         wfProfileOut( __METHOD__ );
1303                         return null;
1304                 }
1305
1306                 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
1307                         wfDebug( __METHOD__ . ": unknown virus scanner: $wgAntivirus\n" );
1308                         $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1\n</div>",
1309                                 array( 'virus-badscanner', $wgAntivirus ) );
1310                         wfProfileOut( __METHOD__ );
1311                         return wfMessage( 'virus-unknownscanner' )->text() . " $wgAntivirus";
1312                 }
1313
1314                 # look up scanner configuration
1315                 $command = $wgAntivirusSetup[$wgAntivirus]['command'];
1316                 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
1317                 $msgPattern = isset( $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ) ?
1318                         $wgAntivirusSetup[$wgAntivirus]['messagepattern'] : null;
1319
1320                 if ( strpos( $command, "%f" ) === false ) {
1321                         # simple pattern: append file to scan
1322                         $command .= " " . wfEscapeShellArg( $file );
1323                 } else {
1324                         # complex pattern: replace "%f" with file to scan
1325                         $command = str_replace( "%f", wfEscapeShellArg( $file ), $command );
1326                 }
1327
1328                 wfDebug( __METHOD__ . ": running virus scan: $command \n" );
1329
1330                 # execute virus scanner
1331                 $exitCode = false;
1332
1333                 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
1334                 #      that does not seem to be worth the pain.
1335                 #      Ask me (Duesentrieb) about it if it's ever needed.
1336                 $output = wfShellExecWithStderr( $command, $exitCode );
1337
1338                 # map exit code to AV_xxx constants.
1339                 $mappedCode = $exitCode;
1340                 if ( $exitCodeMap ) {
1341                         if ( isset( $exitCodeMap[$exitCode] ) ) {
1342                                 $mappedCode = $exitCodeMap[$exitCode];
1343                         } elseif ( isset( $exitCodeMap["*"] ) ) {
1344                                 $mappedCode = $exitCodeMap["*"];
1345                         }
1346                 }
1347
1348                 /* NB: AV_NO_VIRUS is 0 but AV_SCAN_FAILED is false,
1349                  * so we need the strict equalities === and thus can't use a switch here
1350                  */
1351                 if ( $mappedCode === AV_SCAN_FAILED ) {
1352                         # scan failed (code was mapped to false by $exitCodeMap)
1353                         wfDebug( __METHOD__ . ": failed to scan $file (code $exitCode).\n" );
1354
1355                         $output = $wgAntivirusRequired ? wfMessage( 'virus-scanfailed', array( $exitCode ) )->text() : null;
1356                 } elseif ( $mappedCode === AV_SCAN_ABORTED ) {
1357                         # scan failed because filetype is unknown (probably imune)
1358                         wfDebug( __METHOD__ . ": unsupported file type $file (code $exitCode).\n" );
1359                         $output = null;
1360                 } elseif ( $mappedCode === AV_NO_VIRUS ) {
1361                         # no virus found
1362                         wfDebug( __METHOD__ . ": file passed virus scan.\n" );
1363                         $output = false;
1364                 } else {
1365                         $output = trim( $output );
1366
1367                         if ( !$output ) {
1368                                 $output = true; #if there's no output, return true
1369                         } elseif ( $msgPattern ) {
1370                                 $groups = array();
1371                                 if ( preg_match( $msgPattern, $output, $groups ) ) {
1372                                         if ( $groups[1] ) {
1373                                                 $output = $groups[1];
1374                                         }
1375                                 }
1376                         }
1377
1378                         wfDebug( __METHOD__ . ": FOUND VIRUS! scanner feedback: $output \n" );
1379                 }
1380
1381                 wfProfileOut( __METHOD__ );
1382                 return $output;
1383         }
1384
1385         /**
1386          * Check if there's an overwrite conflict and, if so, if restrictions
1387          * forbid this user from performing the upload.
1388          *
1389          * @param $user User
1390          *
1391          * @return mixed true on success, array on failure
1392          */
1393         private function checkOverwrite( $user ) {
1394                 // First check whether the local file can be overwritten
1395                 $file = $this->getLocalFile();
1396                 if ( $file->exists() ) {
1397                         if ( !self::userCanReUpload( $user, $file ) ) {
1398                                 return array( 'fileexists-forbidden', $file->getName() );
1399                         } else {
1400                                 return true;
1401                         }
1402                 }
1403
1404                 /* Check shared conflicts: if the local file does not exist, but
1405                  * wfFindFile finds a file, it exists in a shared repository.
1406                  */
1407                 $file = wfFindFile( $this->getTitle() );
1408                 if ( $file && !$user->isAllowed( 'reupload-shared' ) ) {
1409                         return array( 'fileexists-shared-forbidden', $file->getName() );
1410                 }
1411
1412                 return true;
1413         }
1414
1415         /**
1416          * Check if a user is the last uploader
1417          *
1418          * @param $user User object
1419          * @param string $img image name
1420          * @return Boolean
1421          */
1422         public static function userCanReUpload( User $user, $img ) {
1423                 if ( $user->isAllowed( 'reupload' ) ) {
1424                         return true; // non-conditional
1425                 }
1426                 if ( !$user->isAllowed( 'reupload-own' ) ) {
1427                         return false;
1428                 }
1429                 if ( is_string( $img ) ) {
1430                         $img = wfLocalFile( $img );
1431                 }
1432                 if ( !( $img instanceof LocalFile ) ) {
1433                         return false;
1434                 }
1435
1436                 return $user->getId() == $img->getUser( 'id' );
1437         }
1438
1439         /**
1440          * Helper function that does various existence checks for a file.
1441          * The following checks are performed:
1442          * - The file exists
1443          * - Article with the same name as the file exists
1444          * - File exists with normalized extension
1445          * - The file looks like a thumbnail and the original exists
1446          *
1447          * @param $file File The File object to check
1448          * @return mixed False if the file does not exists, else an array
1449          */
1450         public static function getExistsWarning( $file ) {
1451                 if ( $file->exists() ) {
1452                         return array( 'warning' => 'exists', 'file' => $file );
1453                 }
1454
1455                 if ( $file->getTitle()->getArticleID() ) {
1456                         return array( 'warning' => 'page-exists', 'file' => $file );
1457                 }
1458
1459                 if ( $file->wasDeleted() && !$file->exists() ) {
1460                         return array( 'warning' => 'was-deleted', 'file' => $file );
1461                 }
1462
1463                 if ( strpos( $file->getName(), '.' ) == false ) {
1464                         $partname = $file->getName();
1465                         $extension = '';
1466                 } else {
1467                         $n = strrpos( $file->getName(), '.' );
1468                         $extension = substr( $file->getName(), $n + 1 );
1469                         $partname = substr( $file->getName(), 0, $n );
1470                 }
1471                 $normalizedExtension = File::normalizeExtension( $extension );
1472
1473                 if ( $normalizedExtension != $extension ) {
1474                         // We're not using the normalized form of the extension.
1475                         // Normal form is lowercase, using most common of alternate
1476                         // extensions (eg 'jpg' rather than 'JPEG').
1477                         //
1478                         // Check for another file using the normalized form...
1479                         $nt_lc = Title::makeTitle( NS_FILE, "{$partname}.{$normalizedExtension}" );
1480                         $file_lc = wfLocalFile( $nt_lc );
1481
1482                         if ( $file_lc->exists() ) {
1483                                 return array(
1484                                         'warning' => 'exists-normalized',
1485                                         'file' => $file,
1486                                         'normalizedFile' => $file_lc
1487                                 );
1488                         }
1489                 }
1490
1491                 // Check for files with the same name but a different extension
1492                 $similarFiles = RepoGroup::singleton()->getLocalRepo()->findFilesByPrefix(
1493                                 "{$partname}.", 1 );
1494                 if ( count( $similarFiles ) ) {
1495                         return array(
1496                                 'warning' => 'exists-normalized',
1497                                 'file' => $file,
1498                                 'normalizedFile' => $similarFiles[0],
1499                         );
1500                 }
1501
1502                 if ( self::isThumbName( $file->getName() ) ) {
1503                         # Check for filenames like 50px- or 180px-, these are mostly thumbnails
1504                         $nt_thb = Title::newFromText( substr( $partname, strpos( $partname, '-' ) + 1 ) . '.' . $extension, NS_FILE );
1505                         $file_thb = wfLocalFile( $nt_thb );
1506                         if ( $file_thb->exists() ) {
1507                                 return array(
1508                                         'warning' => 'thumb',
1509                                         'file' => $file,
1510                                         'thumbFile' => $file_thb
1511                                 );
1512                         } else {
1513                                 // File does not exist, but we just don't like the name
1514                                 return array(
1515                                         'warning' => 'thumb-name',
1516                                         'file' => $file,
1517                                         'thumbFile' => $file_thb
1518                                 );
1519                         }
1520                 }
1521
1522                 foreach ( self::getFilenamePrefixBlacklist() as $prefix ) {
1523                         if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
1524                                 return array(
1525                                         'warning' => 'bad-prefix',
1526                                         'file' => $file,
1527                                         'prefix' => $prefix
1528                                 );
1529                         }
1530                 }
1531
1532                 return false;
1533         }
1534
1535         /**
1536          * Helper function that checks whether the filename looks like a thumbnail
1537          * @param $filename string
1538          * @return bool
1539          */
1540         public static function isThumbName( $filename ) {
1541                 $n = strrpos( $filename, '.' );
1542                 $partname = $n ? substr( $filename, 0, $n ) : $filename;
1543                 return (
1544                                         substr( $partname, 3, 3 ) == 'px-' ||
1545                                         substr( $partname, 2, 3 ) == 'px-'
1546                                 ) &&
1547                                 preg_match( "/[0-9]{2}/", substr( $partname, 0, 2 ) );
1548         }
1549
1550         /**
1551          * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
1552          *
1553          * @return array list of prefixes
1554          */
1555         public static function getFilenamePrefixBlacklist() {
1556                 $blacklist = array();
1557                 $message = wfMessage( 'filename-prefix-blacklist' )->inContentLanguage();
1558                 if ( !$message->isDisabled() ) {
1559                         $lines = explode( "\n", $message->plain() );
1560                         foreach ( $lines as $line ) {
1561                                 // Remove comment lines
1562                                 $comment = substr( trim( $line ), 0, 1 );
1563                                 if ( $comment == '#' || $comment == '' ) {
1564                                         continue;
1565                                 }
1566                                 // Remove additional comments after a prefix
1567                                 $comment = strpos( $line, '#' );
1568                                 if ( $comment > 0 ) {
1569                                         $line = substr( $line, 0, $comment - 1 );
1570                                 }
1571                                 $blacklist[] = trim( $line );
1572                         }
1573                 }
1574                 return $blacklist;
1575         }
1576
1577         /**
1578          * Gets image info about the file just uploaded.
1579          *
1580          * Also has the effect of setting metadata to be an 'indexed tag name' in returned API result if
1581          * 'metadata' was requested. Oddly, we have to pass the "result" object down just so it can do that
1582          * with the appropriate format, presumably.
1583          *
1584          * @param $result ApiResult:
1585          * @return Array: image info
1586          */
1587         public function getImageInfo( $result ) {
1588                 $file = $this->getLocalFile();
1589                 // TODO This cries out for refactoring. We really want to say $file->getAllInfo(); here.
1590                 // Perhaps "info" methods should be moved into files, and the API should just wrap them in queries.
1591                 if ( $file instanceof UploadStashFile ) {
1592                         $imParam = ApiQueryStashImageInfo::getPropertyNames();
1593                         $info = ApiQueryStashImageInfo::getInfo( $file, array_flip( $imParam ), $result );
1594                 } else {
1595                         $imParam = ApiQueryImageInfo::getPropertyNames();
1596                         $info = ApiQueryImageInfo::getInfo( $file, array_flip( $imParam ), $result );
1597                 }
1598                 return $info;
1599         }
1600
1601         /**
1602          * @param $error array
1603          * @return Status
1604          */
1605         public function convertVerifyErrorToStatus( $error ) {
1606                 $code = $error['status'];
1607                 unset( $code['status'] );
1608                 return Status::newFatal( $this->getVerificationErrorCode( $code ), $error );
1609         }
1610
1611         /**
1612          * @param $forType null|string
1613          * @return int
1614          */
1615         public static function getMaxUploadSize( $forType = null ) {
1616                 global $wgMaxUploadSize;
1617
1618                 if ( is_array( $wgMaxUploadSize ) ) {
1619                         if ( !is_null( $forType ) && isset( $wgMaxUploadSize[$forType] ) ) {
1620                                 return $wgMaxUploadSize[$forType];
1621                         } else {
1622                                 return $wgMaxUploadSize['*'];
1623                         }
1624                 } else {
1625                         return intval( $wgMaxUploadSize );
1626                 }
1627         }
1628
1629         /**
1630          * Get the current status of a chunked upload (used for polling).
1631          * The status will be read from the *current* user session.
1632          * @param $statusKey string
1633          * @return Array|bool
1634          */
1635         public static function getSessionStatus( $statusKey ) {
1636                 return isset( $_SESSION[self::SESSION_STATUS_KEY][$statusKey] )
1637                         ? $_SESSION[self::SESSION_STATUS_KEY][$statusKey]
1638                         : false;
1639         }
1640
1641         /**
1642          * Set the current status of a chunked upload (used for polling).
1643          * The status will be stored in the *current* user session.
1644          * @param $statusKey string
1645          * @param $value array|false
1646          * @return void
1647          */
1648         public static function setSessionStatus( $statusKey, $value ) {
1649                 if ( $value === false ) {
1650                         unset( $_SESSION[self::SESSION_STATUS_KEY][$statusKey] );
1651                 } else {
1652                         $_SESSION[self::SESSION_STATUS_KEY][$statusKey] = $value;
1653                 }
1654         }
1655 }