includes/upload/UploadBase.php

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