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(
  48                 'UTF-8',
  49                 'ISO-8859-1',
  50                 'ISO-8859-2',
  51                 'UTF-16',
  52                 'UTF-32'
  53         );
  54
  55         const SUCCESS = 0;
  56         const OK = 0;
  57         const EMPTY_FILE = 3;
  58         const MIN_LENGTH_PARTNAME = 4;
  59         const ILLEGAL_FILENAME = 5;
  60         const OVERWRITE_EXISTING_FILE = 7; # Not used anymore; handled by verifyTitlePermissions()
  61         const FILETYPE_MISSING = 8;
  62         const FILETYPE_BADTYPE = 9;
  63         const VERIFICATION_ERROR = 10;
  64
  65         # HOOK_ABORTED is the new name of UPLOAD_VERIFICATION_ERROR
  66         const UPLOAD_VERIFICATION_ERROR = 11;
  67         const HOOK_ABORTED = 11;
  68         const FILE_TOO_LARGE = 12;
  69         const WINDOWS_NONASCII_FILENAME = 13;
  70         const FILENAME_TOO_LONG = 14;
  71
  72         const SESSION_STATUS_KEY = 'wsUploadStatusData';
  73
  74         /**
  75          * @param int $error
  76          * @return string
  77          */
  78         public function getVerificationErrorCode( $error ) {
  79                 $code_to_status = array(
  80                         self::EMPTY_FILE => 'empty-file',
  81                         self::FILE_TOO_LARGE => 'file-too-large',
  82                         self::FILETYPE_MISSING => 'filetype-missing',
  83                         self::FILETYPE_BADTYPE => 'filetype-banned',
  84                         self::MIN_LENGTH_PARTNAME => 'filename-tooshort',
  85                         self::ILLEGAL_FILENAME => 'illegal-filename',
  86                         self::OVERWRITE_EXISTING_FILE => 'overwrite',
  87                         self::VERIFICATION_ERROR => 'verification-error',
  88                         self::HOOK_ABORTED => 'hookaborted',
  89                         self::WINDOWS_NONASCII_FILENAME => 'windows-nonascii-filename',
  90                         self::FILENAME_TOO_LONG => 'filename-toolong',
  91                 );
  92                 if ( isset( $code_to_status[$error] ) ) {
  93                         return $code_to_status[$error];
  94                 }
  95
  96                 return 'unknown-error';
  97         }
  98
  99         /**
 100          * Returns true if uploads are enabled.
 101          * Can be override by subclasses.
 102          * @return bool
 103          */
 104         public static function isEnabled() {
 105                 global $wgEnableUploads;
 106
 107                 if ( !$wgEnableUploads ) {
 108                         return false;
 109                 }
 110
 111                 # Check php's file_uploads setting
 112                 return wfIsHHVM() || wfIniGetBool( 'file_uploads' );
 113         }
 114
 115         /**
 116          * Returns true if the user can use this upload module or else a string
 117          * identifying the missing permission.
 118          * Can be overridden by subclasses.
 119          *
 120          * @param User $user
 121          * @return bool|string
 122          */
 123         public static function isAllowed( $user ) {
 124                 foreach ( array( 'upload', 'edit' ) as $permission ) {
 125                         if ( !$user->isAllowed( $permission ) ) {
 126                                 return $permission;
 127                         }
 128                 }
 129
 130                 return true;
 131         }
 132
 133         // Upload handlers. Should probably just be a global.
 134         private static $uploadHandlers = array( 'Stash', 'File', 'Url' );
 135
 136         /**
 137          * Create a form of UploadBase depending on wpSourceType and initializes it
 138          *
 139          * @param WebRequest $request
 140          * @param string|null $type
 141          * @return null|UploadBase
 142          */
 143         public static function createFromRequest( &$request, $type = null ) {
 144                 $type = $type ? $type : $request->getVal( 'wpSourceType', 'File' );
 145
 146                 if ( !$type ) {
 147                         return null;
 148                 }
 149
 150                 // Get the upload class
 151                 $type = ucfirst( $type );
 152
 153                 // Give hooks the chance to handle this request
 154                 $className = null;
 155                 wfRunHooks( 'UploadCreateFromRequest', array( $type, &$className ) );
 156                 if ( is_null( $className ) ) {
 157                         $className = 'UploadFrom' . $type;
 158                         wfDebug( __METHOD__ . ": class name: $className\n" );
 159                         if ( !in_array( $type, self::$uploadHandlers ) ) {
 160                                 return null;
 161                         }
 162                 }
 163
 164                 // Check whether this upload class is enabled
 165                 if ( !call_user_func( array( $className, 'isEnabled' ) ) ) {
 166                         return null;
 167                 }
 168
 169                 // Check whether the request is valid
 170                 if ( !call_user_func( array( $className, 'isValidRequest' ), $request ) ) {
 171                         return null;
 172                 }
 173
 174                 $handler = new $className;
 175
 176                 $handler->initializeFromRequest( $request );
 177
 178                 return $handler;
 179         }
 180
 181         /**
 182          * Check whether a request if valid for this handler
 183          * @param WebRequest $request
 184          * @return bool
 185          */
 186         public static function isValidRequest( $request ) {
 187                 return false;
 188         }
 189
 190         public function __construct() {
 191         }
 192
 193         /**
 194          * Returns the upload type. Should be overridden by child classes
 195          *
 196          * @since 1.18
 197          * @return string
 198          */
 199         public function getSourceType() {
 200                 return null;
 201         }
 202
 203         /**
 204          * Initialize the path information
 205          * @param string $name The desired destination name
 206          * @param string $tempPath The temporary path
 207          * @param int $fileSize The file size
 208          * @param bool $removeTempFile (false) remove the temporary file?
 209          * @throws MWException
 210          */
 211         public function initializePathInfo( $name, $tempPath, $fileSize, $removeTempFile = false ) {
 212                 $this->mDesiredDestName = $name;
 213                 if ( FileBackend::isStoragePath( $tempPath ) ) {
 214                         throw new MWException( __METHOD__ . " given storage path `$tempPath`." );
 215                 }
 216                 $this->mTempPath = $tempPath;
 217                 $this->mFileSize = $fileSize;
 218                 $this->mRemoveTempFile = $removeTempFile;
 219         }
 220
 221         /**
 222          * Initialize from a WebRequest. Override this in a subclass.
 223          *
 224          * @param WebRequest $request
 225          */
 226         abstract public function initializeFromRequest( &$request );
 227
 228         /**
 229          * Fetch the file. Usually a no-op
 230          * @return Status
 231          */
 232         public function fetchFile() {
 233                 return Status::newGood();
 234         }
 235
 236         /**
 237          * Return true if the file is empty
 238          * @return bool
 239          */
 240         public function isEmptyFile() {
 241                 return empty( $this->mFileSize );
 242         }
 243
 244         /**
 245          * Return the file size
 246          * @return int
 247          */
 248         public function getFileSize() {
 249                 return $this->mFileSize;
 250         }
 251
 252         /**
 253          * Get the base 36 SHA1 of the file
 254          * @return string
 255          */
 256         public function getTempFileSha1Base36() {
 257                 return FSFile::getSha1Base36FromPath( $this->mTempPath );
 258         }
 259
 260         /**
 261          * @param string $srcPath The source path
 262          * @return string|bool The real path if it was a virtual URL Returns false on failure
 263          */
 264         function getRealPath( $srcPath ) {
 265                 wfProfileIn( __METHOD__ );
 266                 $repo = RepoGroup::singleton()->getLocalRepo();
 267                 if ( $repo->isVirtualUrl( $srcPath ) ) {
 268                         /** @todo Just make uploads work with storage paths UploadFromStash
 269                          *  loads files via virtual URLs.
 270                          */
 271                         $tmpFile = $repo->getLocalCopy( $srcPath );
 272                         if ( $tmpFile ) {
 273                                 $tmpFile->bind( $this ); // keep alive with $this
 274                         }
 275                         $path = $tmpFile ? $tmpFile->getPath() : false;
 276                 } else {
 277                         $path = $srcPath;
 278                 }
 279                 wfProfileOut( __METHOD__ );
 280
 281                 return $path;
 282         }
 283
 284         /**
 285          * Verify whether the upload is sane.
 286          * @return mixed Const self::OK or else an array with error information
 287          */
 288         public function verifyUpload() {
 289                 wfProfileIn( __METHOD__ );
 290
 291                 /**
 292                  * If there was no filename or a zero size given, give up quick.
 293                  */
 294                 if ( $this->isEmptyFile() ) {
 295                         wfProfileOut( __METHOD__ );
 296
 297                         return array( 'status' => self::EMPTY_FILE );
 298                 }
 299
 300                 /**
 301                  * Honor $wgMaxUploadSize
 302                  */
 303                 $maxSize = self::getMaxUploadSize( $this->getSourceType() );
 304                 if ( $this->mFileSize > $maxSize ) {
 305                         wfProfileOut( __METHOD__ );
 306
 307                         return array(
 308                                 'status' => self::FILE_TOO_LARGE,
 309                                 'max' => $maxSize,
 310                         );
 311                 }
 312
 313                 /**
 314                  * Look at the contents of the file; if we can recognize the
 315                  * type but it's corrupt or data of the wrong type, we should
 316                  * probably not accept it.
 317                  */
 318                 $verification = $this->verifyFile();
 319                 if ( $verification !== true ) {
 320                         wfProfileOut( __METHOD__ );
 321
 322                         return array(
 323                                 'status' => self::VERIFICATION_ERROR,
 324                                 'details' => $verification
 325                         );
 326                 }
 327
 328                 /**
 329                  * Make sure this file can be created
 330                  */
 331                 $result = $this->validateName();
 332                 if ( $result !== true ) {
 333                         wfProfileOut( __METHOD__ );
 334
 335                         return $result;
 336                 }
 337
 338                 $error = '';
 339                 if ( !wfRunHooks( 'UploadVerification',
 340                         array( $this->mDestName, $this->mTempPath, &$error ) )
 341                 ) {
 342                         wfProfileOut( __METHOD__ );
 343
 344                         return array( 'status' => self::HOOK_ABORTED, 'error' => $error );
 345                 }
 346
 347                 wfProfileOut( __METHOD__ );
 348
 349                 return array( 'status' => self::OK );
 350         }
 351
 352         /**
 353          * Verify that the name is valid and, if necessary, that we can overwrite
 354          *
 355          * @return mixed True if valid, otherwise and array with 'status'
 356          * and other keys
 357          */
 358         public function validateName() {
 359                 $nt = $this->getTitle();
 360                 if ( is_null( $nt ) ) {
 361                         $result = array( 'status' => $this->mTitleError );
 362                         if ( $this->mTitleError == self::ILLEGAL_FILENAME ) {
 363                                 $result['filtered'] = $this->mFilteredName;
 364                         }
 365                         if ( $this->mTitleError == self::FILETYPE_BADTYPE ) {
 366                                 $result['finalExt'] = $this->mFinalExtension;
 367                                 if ( count( $this->mBlackListedExtensions ) ) {
 368                                         $result['blacklistedExt'] = $this->mBlackListedExtensions;
 369                                 }
 370                         }
 371
 372                         return $result;
 373                 }
 374                 $this->mDestName = $this->getLocalFile()->getName();
 375
 376                 return true;
 377         }
 378
 379         /**
 380          * Verify the mime type.
 381          *
 382          * @note Only checks that it is not an evil mime. The "does it have
 383          *  correct extension given its mime type?" check is in verifyFile.
 384          * @param string $mime Representing the mime
 385          * @return mixed True if the file is verified, an array otherwise
 386          */
 387         protected function verifyMimeType( $mime ) {
 388                 global $wgVerifyMimeType;
 389                 wfProfileIn( __METHOD__ );
 390                 if ( $wgVerifyMimeType ) {
 391                         wfDebug( "mime: <$mime> extension: <{$this->mFinalExtension}>\n" );
 392                         global $wgMimeTypeBlacklist;
 393                         if ( $this->checkFileExtension( $mime, $wgMimeTypeBlacklist ) ) {
 394                                 wfProfileOut( __METHOD__ );
 395
 396                                 return array( 'filetype-badmime', $mime );
 397                         }
 398
 399                         # Check IE type
 400                         $fp = fopen( $this->mTempPath, 'rb' );
 401                         $chunk = fread( $fp, 256 );
 402                         fclose( $fp );
 403
 404                         $magic = MimeMagic::singleton();
 405                         $extMime = $magic->guessTypesForExtension( $this->mFinalExtension );
 406                         $ieTypes = $magic->getIEMimeTypes( $this->mTempPath, $chunk, $extMime );
 407                         foreach ( $ieTypes as $ieType ) {
 408                                 if ( $this->checkFileExtension( $ieType, $wgMimeTypeBlacklist ) ) {
 409                                         wfProfileOut( __METHOD__ );
 410
 411                                         return array( 'filetype-bad-ie-mime', $ieType );
 412                                 }
 413                         }
 414                 }
 415
 416                 wfProfileOut( __METHOD__ );
 417
 418                 return true;
 419         }
 420
 421         /**
 422          * Verifies that it's ok to include the uploaded file
 423          *
 424          * @return mixed True of the file is verified, array otherwise.
 425          */
 426         protected function verifyFile() {
 427                 global $wgVerifyMimeType;
 428                 wfProfileIn( __METHOD__ );
 429
 430                 $status = $this->verifyPartialFile();
 431                 if ( $status !== true ) {
 432                         wfProfileOut( __METHOD__ );
 433
 434                         return $status;
 435                 }
 436
 437                 $this->mFileProps = FSFile::getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
 438                 $mime = $this->mFileProps['mime'];
 439
 440                 if ( $wgVerifyMimeType ) {
 441                         # XXX: Missing extension will be caught by validateName() via getTitle()
 442                         if ( $this->mFinalExtension != '' && !$this->verifyExtension( $mime, $this->mFinalExtension ) ) {
 443                                 wfProfileOut( __METHOD__ );
 444
 445                                 return array( 'filetype-mime-mismatch', $this->mFinalExtension, $mime );
 446                         }
 447                 }
 448
 449                 $handler = MediaHandler::getHandler( $mime );
 450                 if ( $handler ) {
 451                         $handlerStatus = $handler->verifyUpload( $this->mTempPath );
 452                         if ( !$handlerStatus->isOK() ) {
 453                                 $errors = $handlerStatus->getErrorsArray();
 454                                 wfProfileOut( __METHOD__ );
 455
 456                                 return reset( $errors );
 457                         }
 458                 }
 459
 460                 wfRunHooks( 'UploadVerifyFile', array( $this, $mime, &$status ) );
 461                 if ( $status !== true ) {
 462                         wfProfileOut( __METHOD__ );
 463
 464                         return $status;
 465                 }
 466
 467                 wfDebug( __METHOD__ . ": all clear; passing.\n" );
 468                 wfProfileOut( __METHOD__ );
 469
 470                 return true;
 471         }
 472
 473         /**
 474          * A verification routine suitable for partial files
 475          *
 476          * Runs the blacklist checks, but not any checks that may
 477          * assume the entire file is present.
 478          *
 479          * @return mixed True for valid or array with error message key.
 480          */
 481         protected function verifyPartialFile() {
 482                 global $wgAllowJavaUploads, $wgDisableUploadScriptChecks;
 483                 wfProfileIn( __METHOD__ );
 484
 485                 # getTitle() sets some internal parameters like $this->mFinalExtension
 486                 $this->getTitle();
 487
 488                 $this->mFileProps = FSFile::getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
 489
 490                 # check mime type, if desired
 491                 $mime = $this->mFileProps['file-mime'];
 492                 $status = $this->verifyMimeType( $mime );
 493                 if ( $status !== true ) {
 494                         wfProfileOut( __METHOD__ );
 495
 496                         return $status;
 497                 }
 498
 499                 # check for htmlish code and javascript
 500                 if ( !$wgDisableUploadScriptChecks ) {
 501                         if ( self::detectScript( $this->mTempPath, $mime, $this->mFinalExtension ) ) {
 502                                 wfProfileOut( __METHOD__ );
 503
 504                                 return array( 'uploadscripted' );
 505                         }
 506                         if ( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
 507                                 $svgStatus = $this->detectScriptInSvg( $this->mTempPath );
 508                                 if ( $svgStatus !== false ) {
 509                                         wfProfileOut( __METHOD__ );
 510
 511                                         return $svgStatus;
 512                                 }
 513                         }
 514                 }
 515
 516                 # Check for Java applets, which if uploaded can bypass cross-site
 517                 # restrictions.
 518                 if ( !$wgAllowJavaUploads ) {
 519                         $this->mJavaDetected = false;
 520                         $zipStatus = ZipDirectoryReader::read( $this->mTempPath,
 521                                 array( $this, 'zipEntryCallback' ) );
 522                         if ( !$zipStatus->isOK() ) {
 523                                 $errors = $zipStatus->getErrorsArray();
 524                                 $error = reset( $errors );
 525                                 if ( $error[0] !== 'zip-wrong-format' ) {
 526                                         wfProfileOut( __METHOD__ );
 527
 528                                         return $error;
 529                                 }
 530                         }
 531                         if ( $this->mJavaDetected ) {
 532                                 wfProfileOut( __METHOD__ );
 533
 534                                 return array( 'uploadjava' );
 535                         }
 536                 }
 537
 538                 # Scan the uploaded file for viruses
 539                 $virus = $this->detectVirus( $this->mTempPath );
 540                 if ( $virus ) {
 541                         wfProfileOut( __METHOD__ );
 542
 543                         return array( 'uploadvirus', $virus );
 544                 }
 545
 546                 wfProfileOut( __METHOD__ );
 547
 548                 return true;
 549         }
 550
 551         /**
 552          * Callback for ZipDirectoryReader to detect Java class files.
 553          *
 554          * @param array $entry
 555          */
 556         function zipEntryCallback( $entry ) {
 557                 $names = array( $entry['name'] );
 558
 559                 // If there is a null character, cut off the name at it, because JDK's
 560                 // ZIP_GetEntry() uses strcmp() if the name hashes match. If a file name
 561                 // were constructed which had ".class\0" followed by a string chosen to
 562                 // make the hash collide with the truncated name, that file could be
 563                 // returned in response to a request for the .class file.
 564                 $nullPos = strpos( $entry['name'], "\000" );
 565                 if ( $nullPos !== false ) {
 566                         $names[] = substr( $entry['name'], 0, $nullPos );
 567                 }
 568
 569                 // If there is a trailing slash in the file name, we have to strip it,
 570                 // because that's what ZIP_GetEntry() does.
 571                 if ( preg_grep( '!\.class/?$!', $names ) ) {
 572                         $this->mJavaDetected = true;
 573                 }
 574         }
 575
 576         /**
 577          * Alias for verifyTitlePermissions. The function was originally
 578          * 'verifyPermissions', but that suggests it's checking the user, when it's
 579          * really checking the title + user combination.
 580          *
 581          * @param User $user User object to verify the permissions against
 582          * @return mixed An array as returned by getUserPermissionsErrors or true
 583          *   in case the user has proper permissions.
 584          */
 585         public function verifyPermissions( $user ) {
 586                 return $this->verifyTitlePermissions( $user );
 587         }
 588
 589         /**
 590          * Check whether the user can edit, upload and create the image. This
 591          * checks only against the current title; if it returns errors, it may
 592          * very well be that another title will not give errors. Therefore
 593          * isAllowed() should be called as well for generic is-user-blocked or
 594          * can-user-upload checking.
 595          *
 596          * @param User $user User object to verify the permissions against
 597          * @return mixed An array as returned by getUserPermissionsErrors or true
 598          *   in case the user has proper permissions.
 599          */
 600         public function verifyTitlePermissions( $user ) {
 601                 /**
 602                  * If the image is protected, non-sysop users won't be able
 603                  * to modify it by uploading a new revision.
 604                  */
 605                 $nt = $this->getTitle();
 606                 if ( is_null( $nt ) ) {
 607                         return true;
 608                 }
 609                 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
 610                 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
 611                 if ( !$nt->exists() ) {
 612                         $permErrorsCreate = $nt->getUserPermissionsErrors( 'create', $user );
 613                 } else {
 614                         $permErrorsCreate = array();
 615                 }
 616                 if ( $permErrors || $permErrorsUpload || $permErrorsCreate ) {
 617                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
 618                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
 619
 620                         return $permErrors;
 621                 }
 622
 623                 $overwriteError = $this->checkOverwrite( $user );
 624                 if ( $overwriteError !== true ) {
 625                         return array( $overwriteError );
 626                 }
 627
 628                 return true;
 629         }
 630
 631         /**
 632          * Check for non fatal problems with the file.
 633          *
 634          * This should not assume that mTempPath is set.
 635          *
 636          * @return array Array of warnings
 637          */
 638         public function checkWarnings() {
 639                 global $wgLang;
 640                 wfProfileIn( __METHOD__ );
 641
 642                 $warnings = array();
 643
 644                 $localFile = $this->getLocalFile();
 645                 $filename = $localFile->getName();
 646
 647                 /**
 648                  * Check whether the resulting filename is different from the desired one,
 649                  * but ignore things like ucfirst() and spaces/underscore things
 650                  */
 651                 $comparableName = str_replace( ' ', '_', $this->mDesiredDestName );
 652                 $comparableName = Title::capitalize( $comparableName, NS_FILE );
 653
 654                 if ( $this->mDesiredDestName != $filename && $comparableName != $filename ) {
 655                         $warnings['badfilename'] = $filename;
 656                         // Debugging for bug 62241
 657                         wfDebugLog( 'upload', "Filename: '$filename', mDesiredDestName: "
 658                                 . "'$this->mDesiredDestName', comparableName: '$comparableName'" );
 659                 }
 660
 661                 // Check whether the file extension is on the unwanted list
 662                 global $wgCheckFileExtensions, $wgFileExtensions;
 663                 if ( $wgCheckFileExtensions ) {
 664                         $extensions = array_unique( $wgFileExtensions );
 665                         if ( !$this->checkFileExtension( $this->mFinalExtension, $extensions ) ) {
 666                                 $warnings['filetype-unwanted-type'] = array( $this->mFinalExtension,
 667                                         $wgLang->commaList( $extensions ), count( $extensions ) );
 668                         }
 669                 }
 670
 671                 global $wgUploadSizeWarning;
 672                 if ( $wgUploadSizeWarning && ( $this->mFileSize > $wgUploadSizeWarning ) ) {
 673                         $warnings['large-file'] = array( $wgUploadSizeWarning, $this->mFileSize );
 674                 }
 675
 676                 if ( $this->mFileSize == 0 ) {
 677                         $warnings['emptyfile'] = true;
 678                 }
 679
 680                 $exists = self::getExistsWarning( $localFile );
 681                 if ( $exists !== false ) {
 682                         $warnings['exists'] = $exists;
 683                 }
 684
 685                 // Check dupes against existing files
 686                 $hash = $this->getTempFileSha1Base36();
 687                 $dupes = RepoGroup::singleton()->findBySha1( $hash );
 688                 $title = $this->getTitle();
 689                 // Remove all matches against self
 690                 foreach ( $dupes as $key => $dupe ) {
 691                         if ( $title->equals( $dupe->getTitle() ) ) {
 692                                 unset( $dupes[$key] );
 693                         }
 694                 }
 695                 if ( $dupes ) {
 696                         $warnings['duplicate'] = $dupes;
 697                 }
 698
 699                 // Check dupes against archives
 700                 $archivedImage = new ArchivedFile( null, 0, "{$hash}.{$this->mFinalExtension}" );
 701                 if ( $archivedImage->getID() > 0 ) {
 702                         if ( $archivedImage->userCan( File::DELETED_FILE ) ) {
 703                                 $warnings['duplicate-archive'] = $archivedImage->getName();
 704                         } else {
 705                                 $warnings['duplicate-archive'] = '';
 706                         }
 707                 }
 708
 709                 wfProfileOut( __METHOD__ );
 710
 711                 return $warnings;
 712         }
 713
 714         /**
 715          * Really perform the upload. Stores the file in the local repo, watches
 716          * if necessary and runs the UploadComplete hook.
 717          *
 718          * @param string $comment
 719          * @param string $pageText
 720          * @param bool $watch
 721          * @param User $user
 722          *
 723          * @return Status Indicating the whether the upload succeeded.
 724          */
 725         public function performUpload( $comment, $pageText, $watch, $user ) {
 726                 wfProfileIn( __METHOD__ );
 727
 728                 $status = $this->getLocalFile()->upload(
 729                         $this->mTempPath,
 730                         $comment,
 731                         $pageText,
 732                         File::DELETE_SOURCE,
 733                         $this->mFileProps,
 734                         false,
 735                         $user
 736                 );
 737
 738                 if ( $status->isGood() ) {
 739                         if ( $watch ) {
 740                                 WatchAction::doWatch(
 741                                         $this->getLocalFile()->getTitle(),
 742                                         $user,
 743                                         WatchedItem::IGNORE_USER_RIGHTS
 744                                 );
 745                         }
 746                         wfRunHooks( 'UploadComplete', array( &$this ) );
 747                 }
 748
 749                 wfProfileOut( __METHOD__ );
 750
 751                 return $status;
 752         }
 753
 754         /**
 755          * Returns the title of the file to be uploaded. Sets mTitleError in case
 756          * the name was illegal.
 757          *
 758          * @return Title The title of the file or null in case the name was illegal
 759          */
 760         public function getTitle() {
 761                 if ( $this->mTitle !== false ) {
 762                         return $this->mTitle;
 763                 }
 764                 /* Assume that if a user specified File:Something.jpg, this is an error
 765                  * and that the namespace prefix needs to be stripped of.
 766                  */
 767                 $title = Title::newFromText( $this->mDesiredDestName );
 768                 if ( $title && $title->getNamespace() == NS_FILE ) {
 769                         $this->mFilteredName = $title->getDBkey();
 770                 } else {
 771                         $this->mFilteredName = $this->mDesiredDestName;
 772                 }
 773
 774                 # oi_archive_name is max 255 bytes, which include a timestamp and an
 775                 # exclamation mark, so restrict file name to 240 bytes.
 776                 if ( strlen( $this->mFilteredName ) > 240 ) {
 777                         $this->mTitleError = self::FILENAME_TOO_LONG;
 778                         $this->mTitle = null;
 779
 780                         return $this->mTitle;
 781                 }
 782
 783                 /**
 784                  * Chop off any directories in the given filename. Then
 785                  * filter out illegal characters, and try to make a legible name
 786                  * out of it. We'll strip some silently that Title would die on.
 787                  */
 788                 $this->mFilteredName = wfStripIllegalFilenameChars( $this->mFilteredName );
 789                 /* Normalize to title form before we do any further processing */
 790                 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 791                 if ( is_null( $nt ) ) {
 792                         $this->mTitleError = self::ILLEGAL_FILENAME;
 793                         $this->mTitle = null;
 794
 795                         return $this->mTitle;
 796                 }
 797                 $this->mFilteredName = $nt->getDBkey();
 798
 799                 /**
 800                  * We'll want to blacklist against *any* 'extension', and use
 801                  * only the final one for the whitelist.
 802                  */
 803                 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName );
 804
 805                 if ( count( $ext ) ) {
 806                         $this->mFinalExtension = trim( $ext[count( $ext ) - 1] );
 807                 } else {
 808                         $this->mFinalExtension = '';
 809
 810                         # No extension, try guessing one
 811                         $magic = MimeMagic::singleton();
 812                         $mime = $magic->guessMimeType( $this->mTempPath );
 813                         if ( $mime !== 'unknown/unknown' ) {
 814                                 # Get a space separated list of extensions
 815                                 $extList = $magic->getExtensionsForType( $mime );
 816                                 if ( $extList ) {
 817                                         # Set the extension to the canonical extension
 818                                         $this->mFinalExtension = strtok( $extList, ' ' );
 819
 820                                         # Fix up the other variables
 821                                         $this->mFilteredName .= ".{$this->mFinalExtension}";
 822                                         $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 823                                         $ext = array( $this->mFinalExtension );
 824                                 }
 825                         }
 826                 }
 827
 828                 /* Don't allow users to override the blacklist (check file extension) */
 829                 global $wgCheckFileExtensions, $wgStrictFileExtensions;
 830                 global $wgFileExtensions, $wgFileBlacklist;
 831
 832                 $blackListedExtensions = $this->checkFileExtensionList( $ext, $wgFileBlacklist );
 833
 834                 if ( $this->mFinalExtension == '' ) {
 835                         $this->mTitleError = self::FILETYPE_MISSING;
 836                         $this->mTitle = null;
 837
 838                         return $this->mTitle;
 839                 } elseif ( $blackListedExtensions ||
 840                         ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
 841                                 !$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) )
 842                 ) {
 843                         $this->mBlackListedExtensions = $blackListedExtensions;
 844                         $this->mTitleError = self::FILETYPE_BADTYPE;
 845                         $this->mTitle = null;
 846
 847                         return $this->mTitle;
 848                 }
 849
 850                 // Windows may be broken with special characters, see bug 1780
 851                 if ( !preg_match( '/^[\x0-\x7f]*$/', $nt->getText() )
 852                         && !RepoGroup::singleton()->getLocalRepo()->backendSupportsUnicodePaths()
 853                 ) {
 854                         $this->mTitleError = self::WINDOWS_NONASCII_FILENAME;
 855                         $this->mTitle = null;
 856
 857                         return $this->mTitle;
 858                 }
 859
 860                 # If there was more than one "extension", reassemble the base
 861                 # filename to prevent bogus complaints about length
 862                 if ( count( $ext ) > 1 ) {
 863                         $iterations = count( $ext ) - 1;
 864                         for ( $i = 0; $i < $iterations; $i++ ) {
 865                                 $partname .= '.' . $ext[$i];
 866                         }
 867                 }
 868
 869                 if ( strlen( $partname ) < 1 ) {
 870                         $this->mTitleError = self::MIN_LENGTH_PARTNAME;
 871                         $this->mTitle = null;
 872
 873                         return $this->mTitle;
 874                 }
 875
 876                 $this->mTitle = $nt;
 877
 878                 return $this->mTitle;
 879         }
 880
 881         /**
 882          * Return the local file and initializes if necessary.
 883          *
 884          * @return LocalFile|null
 885          */
 886         public function getLocalFile() {
 887                 if ( is_null( $this->mLocalFile ) ) {
 888                         $nt = $this->getTitle();
 889                         $this->mLocalFile = is_null( $nt ) ? null : wfLocalFile( $nt );
 890                 }
 891
 892                 return $this->mLocalFile;
 893         }
 894
 895         /**
 896          * If the user does not supply all necessary information in the first upload
 897          * form submission (either by accident or by design) then we may want to
 898          * stash the file temporarily, get more information, and publish the file
 899          * later.
 900          *
 901          * This method will stash a file in a temporary directory for later
 902          * processing, and save the necessary descriptive info into the database.
 903          * This method returns the file object, which also has a 'fileKey' property
 904          * which can be passed through a form or API request to find this stashed
 905          * file again.
 906          *
 907          * @param User $user
 908          * @return UploadStashFile Stashed file
 909          */
 910         public function stashFile( User $user = null ) {
 911                 // was stashSessionFile
 912                 wfProfileIn( __METHOD__ );
 913
 914                 $stash = RepoGroup::singleton()->getLocalRepo()->getUploadStash( $user );
 915                 $file = $stash->stashFile( $this->mTempPath, $this->getSourceType() );
 916                 $this->mLocalFile = $file;
 917
 918                 wfProfileOut( __METHOD__ );
 919
 920                 return $file;
 921         }
 922
 923         /**
 924          * Stash a file in a temporary directory, returning a key which can be used
 925          * to find the file again. See stashFile().
 926          *
 927          * @return string File key
 928          */
 929         public function stashFileGetKey() {
 930                 return $this->stashFile()->getFileKey();
 931         }
 932
 933         /**
 934          * alias for stashFileGetKey, for backwards compatibility
 935          *
 936          * @return string File key
 937          */
 938         public function stashSession() {
 939                 return $this->stashFileGetKey();
 940         }
 941
 942         /**
 943          * If we've modified the upload file we need to manually remove it
 944          * on exit to clean up.
 945          */
 946         public function cleanupTempFile() {
 947                 if ( $this->mRemoveTempFile && $this->mTempPath && file_exists( $this->mTempPath ) ) {
 948                         wfDebug( __METHOD__ . ": Removing temporary file {$this->mTempPath}\n" );
 949                         unlink( $this->mTempPath );
 950                 }
 951         }
 952
 953         public function getTempPath() {
 954                 return $this->mTempPath;
 955         }
 956
 957         /**
 958          * Split a file into a base name and all dot-delimited 'extensions'
 959          * on the end. Some web server configurations will fall back to
 960          * earlier pseudo-'extensions' to determine type and execute
 961          * scripts, so the blacklist needs to check them all.
 962          *
 963          * @param string $filename
 964          * @return array
 965          */
 966         public static function splitExtensions( $filename ) {
 967                 $bits = explode( '.', $filename );
 968                 $basename = array_shift( $bits );
 969
 970                 return array( $basename, $bits );
 971         }
 972
 973         /**
 974          * Perform case-insensitive match against a list of file extensions.
 975          * Returns true if the extension is in the list.
 976          *
 977          * @param string $ext
 978          * @param array $list
 979          * @return bool
 980          */
 981         public static function checkFileExtension( $ext, $list ) {
 982                 return in_array( strtolower( $ext ), $list );
 983         }
 984
 985         /**
 986          * Perform case-insensitive match against a list of file extensions.
 987          * Returns an array of matching extensions.
 988          *
 989          * @param array $ext
 990          * @param array $list
 991          * @return bool
 992          */
 993         public static function checkFileExtensionList( $ext, $list ) {
 994                 return array_intersect( array_map( 'strtolower', $ext ), $list );
 995         }
 996
 997         /**
 998          * Checks if the mime type of the uploaded file matches the file extension.
 999          *
1000          * @param string $mime The mime type of the uploaded file
1001          * @param string $extension The filename extension that the file is to be served with
1002          * @return bool
1003          */
1004         public static function verifyExtension( $mime, $extension ) {
1005                 $magic = MimeMagic::singleton();
1006
1007                 if ( !$mime || $mime == 'unknown' || $mime == 'unknown/unknown' ) {
1008                         if ( !$magic->isRecognizableExtension( $extension ) ) {
1009                                 wfDebug( __METHOD__ . ": passing file with unknown detected mime type; " .
1010                                         "unrecognized extension '$extension', can't verify\n" );
1011
1012                                 return true;
1013                         } else {
1014                                 wfDebug( __METHOD__ . ": rejecting file with unknown detected mime type; " .
1015                                         "recognized extension '$extension', so probably invalid file\n" );
1016
1017                                 return false;
1018                         }
1019                 }
1020
1021                 $match = $magic->isMatchingExtension( $extension, $mime );
1022
1023                 if ( $match === null ) {
1024                         if ( $magic->getTypesForExtension( $extension ) !== null ) {
1025                                 wfDebug( __METHOD__ . ": No extension known for $mime, but we know a mime for $extension\n" );
1026
1027                                 return false;
1028                         } else {
1029                                 wfDebug( __METHOD__ . ": no file extension known for mime type $mime, passing file\n" );
1030
1031                                 return true;
1032                         }
1033                 } elseif ( $match === true ) {
1034                         wfDebug( __METHOD__ . ": mime type $mime matches extension $extension, passing file\n" );
1035
1036                         /** @todo If it's a bitmap, make sure PHP or ImageMagick resp. can handle it! */
1037                         return true;
1038                 } else {
1039                         wfDebug( __METHOD__
1040                                 . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
1041
1042                         return false;
1043                 }
1044         }
1045
1046         /**
1047          * Heuristic for detecting files that *could* contain JavaScript instructions or
1048          * things that may look like HTML to a browser and are thus
1049          * potentially harmful. The present implementation will produce false
1050          * positives in some situations.
1051          *
1052          * @param string $file Pathname to the temporary upload file
1053          * @param string $mime The mime type of the file
1054          * @param string $extension The extension of the file
1055          * @return bool True if the file contains something looking like embedded scripts
1056          */
1057         public static function detectScript( $file, $mime, $extension ) {
1058                 global $wgAllowTitlesInSVG;
1059                 wfProfileIn( __METHOD__ );
1060
1061                 # ugly hack: for text files, always look at the entire file.
1062                 # For binary field, just check the first K.
1063
1064                 if ( strpos( $mime, 'text/' ) === 0 ) {
1065                         $chunk = file_get_contents( $file );
1066                 } else {
1067                         $fp = fopen( $file, 'rb' );
1068                         $chunk = fread( $fp, 1024 );
1069                         fclose( $fp );
1070                 }
1071
1072                 $chunk = strtolower( $chunk );
1073
1074                 if ( !$chunk ) {
1075                         wfProfileOut( __METHOD__ );
1076
1077                         return false;
1078                 }
1079
1080                 # decode from UTF-16 if needed (could be used for obfuscation).
1081                 if ( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
1082                         $enc = 'UTF-16BE';
1083                 } elseif ( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
1084                         $enc = 'UTF-16LE';
1085                 } else {
1086                         $enc = null;
1087                 }
1088
1089                 if ( $enc ) {
1090                         $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
1091                 }
1092
1093                 $chunk = trim( $chunk );
1094
1095                 /** @todo FIXME: Convert from UTF-16 if necessary! */
1096                 wfDebug( __METHOD__ . ": checking for embedded scripts and HTML stuff\n" );
1097
1098                 # check for HTML doctype
1099                 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
1100                         wfProfileOut( __METHOD__ );
1101
1102                         return true;
1103                 }
1104
1105                 // Some browsers will interpret obscure xml encodings as UTF-8, while
1106                 // PHP/expat will interpret the given encoding in the xml declaration (bug 47304)
1107                 if ( $extension == 'svg' || strpos( $mime, 'image/svg' ) === 0 ) {
1108                         if ( self::checkXMLEncodingMissmatch( $file ) ) {
1109                                 wfProfileOut( __METHOD__ );
1110
1111                                 return true;
1112                         }
1113                 }
1114
1115                 /**
1116                  * Internet Explorer for Windows performs some really stupid file type
1117                  * autodetection which can cause it to interpret valid image files as HTML
1118                  * and potentially execute JavaScript, creating a cross-site scripting
1119                  * attack vectors.
1120                  *
1121                  * Apple's Safari browser also performs some unsafe file type autodetection
1122                  * which can cause legitimate files to be interpreted as HTML if the
1123                  * web server is not correctly configured to send the right content-type
1124                  * (or if you're really uploading plain text and octet streams!)
1125                  *
1126                  * Returns true if IE is likely to mistake the given file for HTML.
1127                  * Also returns true if Safari would mistake the given file for HTML
1128                  * when served with a generic content-type.
1129                  */
1130                 $tags = array(
1131                         '<a href',
1132                         '<body',
1133                         '<head',
1134                         '<html', #also in safari
1135                         '<img',
1136                         '<pre',
1137                         '<script', #also in safari
1138                         '<table'
1139                 );
1140
1141                 if ( !$wgAllowTitlesInSVG && $extension !== 'svg' && $mime !== 'image/svg' ) {
1142                         $tags[] = '<title';
1143                 }
1144
1145                 foreach ( $tags as $tag ) {
1146                         if ( false !== strpos( $chunk, $tag ) ) {
1147                                 wfDebug( __METHOD__ . ": found something that may make it be mistaken for html: $tag\n" );
1148                                 wfProfileOut( __METHOD__ );
1149
1150                                 return true;
1151                         }
1152                 }
1153
1154                 /*
1155                  * look for JavaScript
1156                  */
1157
1158                 # resolve entity-refs to look at attributes. may be harsh on big files... cache result?
1159                 $chunk = Sanitizer::decodeCharReferences( $chunk );
1160
1161                 # look for script-types
1162                 if ( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
1163                         wfDebug( __METHOD__ . ": found script types\n" );
1164                         wfProfileOut( __METHOD__ );
1165
1166                         return true;
1167                 }
1168
1169                 # look for html-style script-urls
1170                 if ( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1171                         wfDebug( __METHOD__ . ": found html-style script urls\n" );
1172                         wfProfileOut( __METHOD__ );
1173
1174                         return true;
1175                 }
1176
1177                 # look for css-style script-urls
1178                 if ( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1179                         wfDebug( __METHOD__ . ": found css-style script urls\n" );
1180                         wfProfileOut( __METHOD__ );
1181
1182                         return true;
1183                 }
1184
1185                 wfDebug( __METHOD__ . ": no scripts found\n" );
1186                 wfProfileOut( __METHOD__ );
1187
1188                 return false;
1189         }
1190
1191         /**
1192          * Check a whitelist of xml encodings that are known not to be interpreted differently
1193          * by the server's xml parser (expat) and some common browsers.
1194          *
1195          * @param string $file Pathname to the temporary upload file
1196          * @return bool True if the file contains an encoding that could be misinterpreted
1197          */
1198         public static function checkXMLEncodingMissmatch( $file ) {
1199                 global $wgSVGMetadataCutoff;
1200                 $contents = file_get_contents( $file, false, null, -1, $wgSVGMetadataCutoff );
1201                 $encodingRegex = '!encoding[ \t\n\r]*=[ \t\n\r]*[\'"](.*?)[\'"]!si';
1202
1203                 if ( preg_match( "!<\?xml\b(.*?)\?>!si", $contents, $matches ) ) {
1204                         if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1205                                 && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1206                         ) {
1207                                 wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1208
1209                                 return true;
1210                         }
1211                 } elseif ( preg_match( "!<\?xml\b!si", $contents ) ) {
1212                         // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1213                         // bytes. There shouldn't be a legitimate reason for this to happen.
1214                         wfDebug( __METHOD__ . ": Unmatched XML declaration start\n" );
1215
1216                         return true;
1217                 } elseif ( substr( $contents, 0, 4 ) == "\x4C\x6F\xA7\x94" ) {
1218                         // EBCDIC encoded XML
1219                         wfDebug( __METHOD__ . ": EBCDIC Encoded XML\n" );
1220
1221                         return true;
1222                 }
1223
1224                 // It's possible the file is encoded with multi-byte encoding, so re-encode attempt to
1225                 // detect the encoding in case is specifies an encoding not whitelisted in self::$safeXmlEncodings
1226                 $attemptEncodings = array( 'UTF-16', 'UTF-16BE', 'UTF-32', 'UTF-32BE' );
1227                 foreach ( $attemptEncodings as $encoding ) {
1228                         wfSuppressWarnings();
1229                         $str = iconv( $encoding, 'UTF-8', $contents );
1230                         wfRestoreWarnings();
1231                         if ( $str != '' && preg_match( "!<\?xml\b(.*?)\?>!si", $str, $matches ) ) {
1232                                 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1233                                         && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1234                                 ) {
1235                                         wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1236
1237                                         return true;
1238                                 }
1239                         } elseif ( $str != '' && preg_match( "!<\?xml\b!si", $str ) ) {
1240                                 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1241                                 // bytes. There shouldn't be a legitimate reason for this to happen.
1242                                 wfDebug( __METHOD__ . ": Unmatched XML declaration start\n" );
1243
1244                                 return true;
1245                         }
1246                 }
1247
1248                 return false;
1249         }
1250
1251         /**
1252          * @param string $filename
1253          * @return mixed False of the file is verified (does not contain scripts), array otherwise.
1254          */
1255         protected function detectScriptInSvg( $filename ) {
1256                 $this->mSVGNSError = false;
1257                 $check = new XmlTypeCheck(
1258                         $filename,
1259                         array( $this, 'checkSvgScriptCallback' ),
1260                         true,
1261                         array( 'processing_instruction_handler' => 'UploadBase::checkSvgPICallback' )
1262                 );
1263                 if ( $check->wellFormed !== true ) {
1264                         // Invalid xml (bug 58553)
1265                         return array( 'uploadinvalidxml' );
1266                 } elseif ( $check->filterMatch ) {
1267                         if ( $this->mSVGNSError ) {
1268                                 return array( 'uploadscriptednamespace', $this->mSVGNSError );
1269                         }
1270
1271                         return array( 'uploadscripted' );
1272                 }
1273
1274                 return false;
1275         }
1276
1277         /**
1278          * Callback to filter SVG Processing Instructions.
1279          * @param string $target Processing instruction name
1280          * @param string $data Processing instruction attribute and value
1281          * @return bool (true if the filter identified something bad)
1282          */
1283         public static function checkSvgPICallback( $target, $data ) {
1284                 // Don't allow external stylesheets (bug 57550)
1285                 if ( preg_match( '/xml-stylesheet/i', $target ) ) {
1286                         return true;
1287                 }
1288
1289                 return false;
1290         }
1291
1292         /**
1293          * @todo Replace this with a whitelist filter!
1294          * @param string $element
1295          * @param array $attribs
1296          * @return bool
1297          */
1298         public function checkSvgScriptCallback( $element, $attribs ) {
1299                 list( $namespace, $strippedElement ) = $this->splitXmlNamespace( $element );
1300
1301                 // We specifically don't include:
1302                 // http://www.w3.org/1999/xhtml (bug 60771)
1303                 static $validNamespaces = array(
1304                         '',
1305                         'adobe:ns:meta/',
1306                         'http://creativecommons.org/ns#',
1307                         'http://inkscape.sourceforge.net/dtd/sodipodi-0.dtd',
1308                         'http://ns.adobe.com/adobeillustrator/10.0/',
1309                         'http://ns.adobe.com/adobesvgviewerextensions/3.0/',
1310                         'http://ns.adobe.com/extensibility/1.0/',
1311                         'http://ns.adobe.com/flows/1.0/',
1312                         'http://ns.adobe.com/illustrator/1.0/',
1313                         'http://ns.adobe.com/imagereplacement/1.0/',
1314                         'http://ns.adobe.com/pdf/1.3/',
1315                         'http://ns.adobe.com/photoshop/1.0/',
1316                         'http://ns.adobe.com/saveforweb/1.0/',
1317                         'http://ns.adobe.com/variables/1.0/',
1318                         'http://ns.adobe.com/xap/1.0/',
1319                         'http://ns.adobe.com/xap/1.0/g/',
1320                         'http://ns.adobe.com/xap/1.0/g/img/',
1321                         'http://ns.adobe.com/xap/1.0/mm/',
1322                         'http://ns.adobe.com/xap/1.0/rights/',
1323                         'http://ns.adobe.com/xap/1.0/stype/dimensions#',
1324                         'http://ns.adobe.com/xap/1.0/stype/font#',
1325                         'http://ns.adobe.com/xap/1.0/stype/manifestitem#',
1326                         'http://ns.adobe.com/xap/1.0/stype/resourceevent#',
1327                         'http://ns.adobe.com/xap/1.0/stype/resourceref#',
1328                         'http://ns.adobe.com/xap/1.0/t/pg/',
1329                         'http://purl.org/dc/elements/1.1/',
1330                         'http://purl.org/dc/elements/1.1',
1331                         'http://schemas.microsoft.com/visio/2003/svgextensions/',
1332                         'http://sodipodi.sourceforge.net/dtd/sodipodi-0.dtd',
1333                         'http://taptrix.com/inkpad/svg_extensions',
1334                         'http://web.resource.org/cc/',
1335                         'http://www.freesoftware.fsf.org/bkchem/cdml',
1336                         'http://www.inkscape.org/namespaces/inkscape',
1337                         'http://www.opengis.net/gml',
1338                         'http://www.w3.org/1999/02/22-rdf-syntax-ns#',
1339                         'http://www.w3.org/2000/svg',
1340                         'http://www.w3.org/tr/rec-rdf-syntax/',
1341                 );
1342
1343                 if ( !in_array( $namespace, $validNamespaces ) ) {
1344                         wfDebug( __METHOD__ . ": Non-svg namespace '$namespace' in uploaded file.\n" );
1345                         /** @todo Return a status object to a closure in XmlTypeCheck, for MW1.21+ */
1346                         $this->mSVGNSError = $namespace;
1347
1348                         return true;
1349                 }
1350
1351                 /*
1352                  * check for elements that can contain javascript
1353                  */
1354                 if ( $strippedElement == 'script' ) {
1355                         wfDebug( __METHOD__ . ": Found script element '$element' in uploaded file.\n" );
1356
1357                         return true;
1358                 }
1359
1360                 # e.g., <svg xmlns="http://www.w3.org/2000/svg">
1361                 #  <handler xmlns:ev="http://www.w3.org/2001/xml-events" ev:event="load">alert(1)</handler> </svg>
1362                 if ( $strippedElement == 'handler' ) {
1363                         wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1364
1365                         return true;
1366                 }
1367
1368                 # SVG reported in Feb '12 that used xml:stylesheet to generate javascript block
1369                 if ( $strippedElement == 'stylesheet' ) {
1370                         wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1371
1372                         return true;
1373                 }
1374
1375                 # Block iframes, in case they pass the namespace check
1376                 if ( $strippedElement == 'iframe' ) {
1377                         wfDebug( __METHOD__ . ": iframe in uploaded file.\n" );
1378
1379                         return true;
1380                 }
1381
1382                 foreach ( $attribs as $attrib => $value ) {
1383                         $stripped = $this->stripXmlNamespace( $attrib );
1384                         $value = strtolower( $value );
1385
1386                         if ( substr( $stripped, 0, 2 ) == 'on' ) {
1387                                 wfDebug( __METHOD__
1388                                         . ": Found event-handler attribute '$attrib'='$value' in uploaded file.\n" );
1389
1390                                 return true;
1391                         }
1392
1393                         # href with non-local target (don't allow http://, javascript:, etc)
1394                         if ( $stripped == 'href'
1395                                 && strpos( $value, 'data:' ) !== 0
1396                                 && strpos( $value, '#' ) !== 0
1397                         ) {
1398                                 if ( !( $strippedElement === 'a'
1399                                         && preg_match( '!^https?://!im', $value ) )
1400                                 ) {
1401                                         wfDebug( __METHOD__ . ": Found href attribute <$strippedElement "
1402                                                 . "'$attrib'='$value' in uploaded file.\n" );
1403
1404                                         return true;
1405                                 }
1406                         }
1407
1408                         # href with embedded svg as target
1409                         if ( $stripped == 'href' && preg_match( '!data:[^,]*image/svg[^,]*,!sim', $value ) ) {
1410                                 wfDebug( __METHOD__ . ": Found href to embedded svg "
1411                                         . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1412
1413                                 return true;
1414                         }
1415
1416                         # href with embedded (text/xml) svg as target
1417                         if ( $stripped == 'href' && preg_match( '!data:[^,]*text/xml[^,]*,!sim', $value ) ) {
1418                                 wfDebug( __METHOD__ . ": Found href to embedded svg "
1419                                         . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1420
1421                                 return true;
1422                         }
1423
1424                         # use set/animate to add event-handler attribute to parent
1425                         if ( ( $strippedElement == 'set' || $strippedElement == 'animate' )
1426                                 && $stripped == 'attributename'
1427                                 && substr( $value, 0, 2 ) == 'on'
1428                         ) {
1429                                 wfDebug( __METHOD__ . ": Found svg setting event-handler attribute with "
1430                                         . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1431
1432                                 return true;
1433                         }
1434
1435                         # use set to add href attribute to parent element
1436                         if ( $strippedElement == 'set'
1437                                 && $stripped == 'attributename'
1438                                 && strpos( $value, 'href' ) !== false
1439                         ) {
1440                                 wfDebug( __METHOD__ . ": Found svg setting href attribute '$value' in uploaded file.\n" );
1441
1442                                 return true;
1443                         }
1444
1445                         # use set to add a remote / data / script target to an element
1446                         if ( $strippedElement == 'set'
1447                                 && $stripped == 'to'
1448                                 && preg_match( '!(http|https|data|script):!sim', $value )
1449                         ) {
1450                                 wfDebug( __METHOD__ . ": Found svg setting attribute to '$value' in uploaded file.\n" );
1451
1452                                 return true;
1453                         }
1454
1455                         # use handler attribute with remote / data / script
1456                         if ( $stripped == 'handler' && preg_match( '!(http|https|data|script):!sim', $value ) ) {
1457                                 wfDebug( __METHOD__ . ": Found svg setting handler with remote/data/script "
1458                                         . "'$attrib'='$value' in uploaded file.\n" );
1459
1460                                 return true;
1461                         }
1462
1463                         # use CSS styles to bring in remote code
1464                         # catch url("http:..., url('http:..., url(http:..., but not url("#..., url('#..., url(#....
1465                         $tagsList = "font|clip-path|fill|filter|marker|marker-end|marker-mid|marker-start|mask|stroke";
1466                         if ( $stripped == 'style'
1467                                 && preg_match_all(
1468                                         '!((?:' . $tagsList . ')\s*:\s*url\s*\(\s*["\']?\s*[^#]+.*?\))!sim',
1469                                         $value,
1470                                         $matches
1471                                 )
1472                         ) {
1473                                 foreach ( $matches[1] as $match ) {
1474                                         if ( !preg_match( '!(?:' . $tagsList . ')\s*:\s*url\s*\(\s*(#|\'#|"#)!sim', $match ) ) {
1475                                                 wfDebug( __METHOD__ . ": Found svg setting a style with "
1476                                                         . "remote url '$attrib'='$value' in uploaded file.\n" );
1477
1478                                                 return true;
1479                                         }
1480                                 }
1481                         }
1482
1483                         # image filters can pull in url, which could be svg that executes scripts
1484                         if ( $strippedElement == 'image'
1485                                 && $stripped == 'filter'
1486                                 && preg_match( '!url\s*\(!sim', $value )
1487                         ) {
1488                                 wfDebug( __METHOD__ . ": Found image filter with url: "
1489                                         . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1490
1491                                 return true;
1492                         }
1493                 }
1494
1495                 return false; //No scripts detected
1496         }
1497
1498         /**
1499          * Divide the element name passed by the xml parser to the callback into URI and prifix.
1500          * @param string $name
1501          * @return array Containing the namespace URI and prefix
1502          */
1503         private static function splitXmlNamespace( $element ) {
1504                 // 'http://www.w3.org/2000/svg:script' -> array( 'http://www.w3.org/2000/svg', 'script' )
1505                 $parts = explode( ':', strtolower( $element ) );
1506                 $name = array_pop( $parts );
1507                 $ns = implode( ':', $parts );
1508
1509                 return array( $ns, $name );
1510         }
1511
1512         /**
1513          * @param string $name
1514          * @return string
1515          */
1516         private function stripXmlNamespace( $name ) {
1517                 // 'http://www.w3.org/2000/svg:script' -> 'script'
1518                 $parts = explode( ':', strtolower( $name ) );
1519
1520                 return array_pop( $parts );
1521         }
1522
1523         /**
1524          * Generic wrapper function for a virus scanner program.
1525          * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
1526          * $wgAntivirusRequired may be used to deny upload if the scan fails.
1527          *
1528          * @param string $file Pathname to the temporary upload file
1529          * @return mixed False if not virus is found, null if the scan fails or is disabled,
1530          *   or a string containing feedback from the virus scanner if a virus was found.
1531          *   If textual feedback is missing but a virus was found, this function returns true.
1532          */
1533         public static function detectVirus( $file ) {
1534                 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
1535                 wfProfileIn( __METHOD__ );
1536
1537                 if ( !$wgAntivirus ) {
1538                         wfDebug( __METHOD__ . ": virus scanner disabled\n" );
1539                         wfProfileOut( __METHOD__ );
1540
1541                         return null;
1542                 }
1543
1544                 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
1545                         wfDebug( __METHOD__ . ": unknown virus scanner: $wgAntivirus\n" );
1546                         $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1\n</div>",
1547                                 array( 'virus-badscanner', $wgAntivirus ) );
1548                         wfProfileOut( __METHOD__ );
1549
1550                         return wfMessage( 'virus-unknownscanner' )->text() . " $wgAntivirus";
1551                 }
1552
1553                 # look up scanner configuration
1554                 $command = $wgAntivirusSetup[$wgAntivirus]['command'];
1555                 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
1556                 $msgPattern = isset( $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ) ?
1557                         $wgAntivirusSetup[$wgAntivirus]['messagepattern'] : null;
1558
1559                 if ( strpos( $command, "%f" ) === false ) {
1560                         # simple pattern: append file to scan
1561                         $command .= " " . wfEscapeShellArg( $file );
1562                 } else {
1563                         # complex pattern: replace "%f" with file to scan
1564                         $command = str_replace( "%f", wfEscapeShellArg( $file ), $command );
1565                 }
1566
1567                 wfDebug( __METHOD__ . ": running virus scan: $command \n" );
1568
1569                 # execute virus scanner
1570                 $exitCode = false;
1571
1572                 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
1573                 #      that does not seem to be worth the pain.
1574                 #      Ask me (Duesentrieb) about it if it's ever needed.
1575                 $output = wfShellExecWithStderr( $command, $exitCode );
1576
1577                 # map exit code to AV_xxx constants.
1578                 $mappedCode = $exitCode;
1579                 if ( $exitCodeMap ) {
1580                         if ( isset( $exitCodeMap[$exitCode] ) ) {
1581                                 $mappedCode = $exitCodeMap[$exitCode];
1582                         } elseif ( isset( $exitCodeMap["*"] ) ) {
1583                                 $mappedCode = $exitCodeMap["*"];
1584                         }
1585                 }
1586
1587                 /* NB: AV_NO_VIRUS is 0 but AV_SCAN_FAILED is false,
1588                  * so we need the strict equalities === and thus can't use a switch here
1589                  */
1590                 if ( $mappedCode === AV_SCAN_FAILED ) {
1591                         # scan failed (code was mapped to false by $exitCodeMap)
1592                         wfDebug( __METHOD__ . ": failed to scan $file (code $exitCode).\n" );
1593
1594                         $output = $wgAntivirusRequired
1595                                 ? wfMessage( 'virus-scanfailed', array( $exitCode ) )->text()
1596                                 : null;
1597                 } elseif ( $mappedCode === AV_SCAN_ABORTED ) {
1598                         # scan failed because filetype is unknown (probably imune)
1599                         wfDebug( __METHOD__ . ": unsupported file type $file (code $exitCode).\n" );
1600                         $output = null;
1601                 } elseif ( $mappedCode === AV_NO_VIRUS ) {
1602                         # no virus found
1603                         wfDebug( __METHOD__ . ": file passed virus scan.\n" );
1604                         $output = false;
1605                 } else {
1606                         $output = trim( $output );
1607
1608                         if ( !$output ) {
1609                                 $output = true; #if there's no output, return true
1610                         } elseif ( $msgPattern ) {
1611                                 $groups = array();
1612                                 if ( preg_match( $msgPattern, $output, $groups ) ) {
1613                                         if ( $groups[1] ) {
1614                                                 $output = $groups[1];
1615                                         }
1616                                 }
1617                         }
1618
1619                         wfDebug( __METHOD__ . ": FOUND VIRUS! scanner feedback: $output \n" );
1620                 }
1621
1622                 wfProfileOut( __METHOD__ );
1623
1624                 return $output;
1625         }
1626
1627         /**
1628          * Check if there's an overwrite conflict and, if so, if restrictions
1629          * forbid this user from performing the upload.
1630          *
1631          * @param User $user
1632          *
1633          * @return mixed True on success, array on failure
1634          */
1635         private function checkOverwrite( $user ) {
1636                 // First check whether the local file can be overwritten
1637                 $file = $this->getLocalFile();
1638                 if ( $file->exists() ) {
1639                         if ( !self::userCanReUpload( $user, $file ) ) {
1640                                 return array( 'fileexists-forbidden', $file->getName() );
1641                         } else {
1642                                 return true;
1643                         }
1644                 }
1645
1646                 /* Check shared conflicts: if the local file does not exist, but
1647                  * wfFindFile finds a file, it exists in a shared repository.
1648                  */
1649                 $file = wfFindFile( $this->getTitle() );
1650                 if ( $file && !$user->isAllowed( 'reupload-shared' ) ) {
1651                         return array( 'fileexists-shared-forbidden', $file->getName() );
1652                 }
1653
1654                 return true;
1655         }
1656
1657         /**
1658          * Check if a user is the last uploader
1659          *
1660          * @param User $user
1661          * @param string $img Image name
1662          * @return bool
1663          */
1664         public static function userCanReUpload( User $user, $img ) {
1665                 if ( $user->isAllowed( 'reupload' ) ) {
1666                         return true; // non-conditional
1667                 }
1668                 if ( !$user->isAllowed( 'reupload-own' ) ) {
1669                         return false;
1670                 }
1671                 if ( is_string( $img ) ) {
1672                         $img = wfLocalFile( $img );
1673                 }
1674                 if ( !( $img instanceof LocalFile ) ) {
1675                         return false;
1676                 }
1677
1678                 return $user->getId() == $img->getUser( 'id' );
1679         }
1680
1681         /**
1682          * Helper function that does various existence checks for a file.
1683          * The following checks are performed:
1684          * - The file exists
1685          * - Article with the same name as the file exists
1686          * - File exists with normalized extension
1687          * - The file looks like a thumbnail and the original exists
1688          *
1689          * @param File $file The File object to check
1690          * @return mixed False if the file does not exists, else an array
1691          */
1692         public static function getExistsWarning( $file ) {
1693                 if ( $file->exists() ) {
1694                         return array( 'warning' => 'exists', 'file' => $file );
1695                 }
1696
1697                 if ( $file->getTitle()->getArticleID() ) {
1698                         return array( 'warning' => 'page-exists', 'file' => $file );
1699                 }
1700
1701                 if ( $file->wasDeleted() && !$file->exists() ) {
1702                         return array( 'warning' => 'was-deleted', 'file' => $file );
1703                 }
1704
1705                 if ( strpos( $file->getName(), '.' ) == false ) {
1706                         $partname = $file->getName();
1707                         $extension = '';
1708                 } else {
1709                         $n = strrpos( $file->getName(), '.' );
1710                         $extension = substr( $file->getName(), $n + 1 );
1711                         $partname = substr( $file->getName(), 0, $n );
1712                 }
1713                 $normalizedExtension = File::normalizeExtension( $extension );
1714
1715                 if ( $normalizedExtension != $extension ) {
1716                         // We're not using the normalized form of the extension.
1717                         // Normal form is lowercase, using most common of alternate
1718                         // extensions (eg 'jpg' rather than 'JPEG').
1719                         //
1720                         // Check for another file using the normalized form...
1721                         $nt_lc = Title::makeTitle( NS_FILE, "{$partname}.{$normalizedExtension}" );
1722                         $file_lc = wfLocalFile( $nt_lc );
1723
1724                         if ( $file_lc->exists() ) {
1725                                 return array(
1726                                         'warning' => 'exists-normalized',
1727                                         'file' => $file,
1728                                         'normalizedFile' => $file_lc
1729                                 );
1730                         }
1731                 }
1732
1733                 // Check for files with the same name but a different extension
1734                 $similarFiles = RepoGroup::singleton()->getLocalRepo()->findFilesByPrefix(
1735                         "{$partname}.", 1 );
1736                 if ( count( $similarFiles ) ) {
1737                         return array(
1738                                 'warning' => 'exists-normalized',
1739                                 'file' => $file,
1740                                 'normalizedFile' => $similarFiles[0],
1741                         );
1742                 }
1743
1744                 if ( self::isThumbName( $file->getName() ) ) {
1745                         # Check for filenames like 50px- or 180px-, these are mostly thumbnails
1746                         $nt_thb = Title::newFromText(
1747                                 substr( $partname, strpos( $partname, '-' ) + 1 ) . '.' . $extension,
1748                                 NS_FILE
1749                         );
1750                         $file_thb = wfLocalFile( $nt_thb );
1751                         if ( $file_thb->exists() ) {
1752                                 return array(
1753                                         'warning' => 'thumb',
1754                                         'file' => $file,
1755                                         'thumbFile' => $file_thb
1756                                 );
1757                         } else {
1758                                 // File does not exist, but we just don't like the name
1759                                 return array(
1760                                         'warning' => 'thumb-name',
1761                                         'file' => $file,
1762                                         'thumbFile' => $file_thb
1763                                 );
1764                         }
1765                 }
1766
1767                 foreach ( self::getFilenamePrefixBlacklist() as $prefix ) {
1768                         if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
1769                                 return array(
1770                                         'warning' => 'bad-prefix',
1771                                         'file' => $file,
1772                                         'prefix' => $prefix
1773                                 );
1774                         }
1775                 }
1776
1777                 return false;
1778         }
1779
1780         /**
1781          * Helper function that checks whether the filename looks like a thumbnail
1782          * @param string $filename
1783          * @return bool
1784          */
1785         public static function isThumbName( $filename ) {
1786                 $n = strrpos( $filename, '.' );
1787                 $partname = $n ? substr( $filename, 0, $n ) : $filename;
1788
1789                 return (
1790                         substr( $partname, 3, 3 ) == 'px-' ||
1791                         substr( $partname, 2, 3 ) == 'px-'
1792                 ) &&
1793                 preg_match( "/[0-9]{2}/", substr( $partname, 0, 2 ) );
1794         }
1795
1796         /**
1797          * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
1798          *
1799          * @return array List of prefixes
1800          */
1801         public static function getFilenamePrefixBlacklist() {
1802                 $blacklist = array();
1803                 $message = wfMessage( 'filename-prefix-blacklist' )->inContentLanguage();
1804                 if ( !$message->isDisabled() ) {
1805                         $lines = explode( "\n", $message->plain() );
1806                         foreach ( $lines as $line ) {
1807                                 // Remove comment lines
1808                                 $comment = substr( trim( $line ), 0, 1 );
1809                                 if ( $comment == '#' || $comment == '' ) {
1810                                         continue;
1811                                 }
1812                                 // Remove additional comments after a prefix
1813                                 $comment = strpos( $line, '#' );
1814                                 if ( $comment > 0 ) {
1815                                         $line = substr( $line, 0, $comment - 1 );
1816                                 }
1817                                 $blacklist[] = trim( $line );
1818                         }
1819                 }
1820
1821                 return $blacklist;
1822         }
1823
1824         /**
1825          * Gets image info about the file just uploaded.
1826          *
1827          * Also has the effect of setting metadata to be an 'indexed tag name' in
1828          * returned API result if 'metadata' was requested. Oddly, we have to pass
1829          * the "result" object down just so it can do that with the appropriate
1830          * format, presumably.
1831          *
1832          * @param ApiResult $result
1833          * @return array Image info
1834          */
1835         public function getImageInfo( $result ) {
1836                 $file = $this->getLocalFile();
1837                 /** @todo This cries out for refactoring.
1838                  *  We really want to say $file->getAllInfo(); here.
1839                  * Perhaps "info" methods should be moved into files, and the API should
1840                  * just wrap them in queries.
1841                  */
1842                 if ( $file instanceof UploadStashFile ) {
1843                         $imParam = ApiQueryStashImageInfo::getPropertyNames();
1844                         $info = ApiQueryStashImageInfo::getInfo( $file, array_flip( $imParam ), $result );
1845                 } else {
1846                         $imParam = ApiQueryImageInfo::getPropertyNames();
1847                         $info = ApiQueryImageInfo::getInfo( $file, array_flip( $imParam ), $result );
1848                 }
1849
1850                 return $info;
1851         }
1852
1853         /**
1854          * @param array $error
1855          * @return Status
1856          */
1857         public function convertVerifyErrorToStatus( $error ) {
1858                 $code = $error['status'];
1859                 unset( $code['status'] );
1860
1861                 return Status::newFatal( $this->getVerificationErrorCode( $code ), $error );
1862         }
1863
1864         /**
1865          * @param null|string $forType
1866          * @return int
1867          */
1868         public static function getMaxUploadSize( $forType = null ) {
1869                 global $wgMaxUploadSize;
1870
1871                 if ( is_array( $wgMaxUploadSize ) ) {
1872                         if ( !is_null( $forType ) && isset( $wgMaxUploadSize[$forType] ) ) {
1873                                 return $wgMaxUploadSize[$forType];
1874                         } else {
1875                                 return $wgMaxUploadSize['*'];
1876                         }
1877                 } else {
1878                         return intval( $wgMaxUploadSize );
1879                 }
1880         }
1881
1882         /**
1883          * Get the current status of a chunked upload (used for polling).
1884          * The status will be read from the *current* user session.
1885          * @param string $statusKey
1886          * @return array|bool
1887          */
1888         public static function getSessionStatus( $statusKey ) {
1889                 return isset( $_SESSION[self::SESSION_STATUS_KEY][$statusKey] )
1890                         ? $_SESSION[self::SESSION_STATUS_KEY][$statusKey]
1891                         : false;
1892         }
1893
1894         /**
1895          * Set the current status of a chunked upload (used for polling).
1896          * The status will be stored in the *current* user session.
1897          * @param string $statusKey
1898          * @param array|bool $value
1899          * @return void
1900          */
1901         public static function setSessionStatus( $statusKey, $value ) {
1902                 if ( $value === false ) {
1903                         unset( $_SESSION[self::SESSION_STATUS_KEY][$statusKey] );
1904                 } else {
1905                         $_SESSION[self::SESSION_STATUS_KEY][$statusKey] = $value;
1906                 }
1907         }
1908 }