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