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